Parlar Scientific Publications, Angerstr

FEB – Fresenius Environmental Bulletin founded jointly by F. Korte and F. Coulston

Production by PSP – Parlar Scientific Publications, Angerstr. 12, 85354 Freising, Germany in cooperation with Lehrstuhl für Chemisch-Technische Analyse und Lebensmitteltechnologie, Technische Universität München, 85350 Freising - Weihenstephan, Germany Copyright © by PSP – Parlar Scientific Publications, Angerstr. 12, 85354 Freising, Germany. All rights are reserved, especially the right to translate into foreign language. No part of the journal may be reproduced in any form- through photocopying, microfilming or other processes- or converted to a machine language, especially for data processing equipment- without the written permission of the publisher. The rights of reproduction by lecture, radio and television transmission, magnetic sound recording or similar means are also reserved.

Printed in GERMANY – ISSN 1018-4619

FEB - EDITORIAL BOARD Environmental Toxicology: Prof. Dr. H. Greim Chief Editor: Senatskomm. d. DFG z. Prüfung gesundheitsschädl. Arbeitsstoffe TU München, 85350 Freising-Weihenstephan, Germany Prof. Dr. H. Parlar Institut für Lebensmitteltechnologie und Analytische Chemie Prof. Dr. A. Kettrup TU München - 85350 Freising-Weihenstephan, Germany Institut für Lebensmitteltechnologie und Analytische Chemie e-mail: [email protected] TU München - 85350 Freising-Weihenstephan, Germany

FEB - ADVISORY BOARD Co-Editors: Environmental Analytical Chemistry: Environmental Analytical Chemistry: K. Ballschmitter, D - K. Bester, D - K. Fischer, D - R. Kallenborn, N Dr. D. Kotzias D.C.G. Muir, CAN - R. Niessner, D - W. Vetter, D – R. Spaccini, I Via Germania 29 21027 Barza (Va) ITALY Environmental Proteomic and Biology:

D. Adelung, D - G.I. Kvesitadze, GEOR A. Reichlmayr-Lais, D - C. Steinberg, D Environmental Proteomic and Biology:

Environmental Chemistry: Prof. Dr. A. Görg Fachgebiet Proteomik J.P. Lay, D - J. Burhenne, D - S. Nitz, D - R. Kreuzig, D TU München - 85350 Freising-Weihenstephan, Germany D. L. Swackhammer, U.S.A. - R. Zepp, U.S.A. – T. Alpay, TR V. Librando; I

Prof. Dr. A. Piccolo Università di Napoli “Frederico II”, Environmental Management: Dipto. Di Scienze Chimico-Agrarie Via Università 100, 80055 Portici (Napoli), Italy L.O. Ruzo, U.S.A - U. Schlottmann, D

Prof. Dr. G. Schüürmann UFZ-Umweltforschungszentrum, Environmental Toxicology: Sektion Chemische Ökotoxikologie Leipzig-Halle GmbH, K.-W. Schramm, D - H. Frank, D - D. Schulz-Jander, U.S.A. - Permoserstr.15, 04318 Leipzig, Germany H.U. Wolf, D – M. McLachlan, S

Environmental Chemistry: Prof. Dr. M. Bahadir Managing Editor: Institut für Ökologische Chemie und Abfallanalytik TU Braunschweig Dr. G. Leupold Hagenring 30, 38106 Braunschweig, Germany Editorial Chief-Officer:

Prof. Dr. M. Spiteller Selma Parlar Institut für Umweltforschung Universität Dortmund PSP- Parlar Scientific Publications Otto-Hahn-Str. 6, 44221 Dortmund, Germany Angerstr.12, 85354 Freising, Germany

e-mail: [email protected] - www.psp-parlar.de Prof. Dr. Ivan Holoubek RECETOX_TOCOEN Marketing Chief Manager: Kamenice 126/3, 62500 Brno, Czech Republic Max-Josef Kirchmaier MASELL-Agency for Marketing & Communication, Public-Rela- tions Environmental Management: Angerstr.12, 85354 Freising, Germany Dr. H. Schlesing e-mail: [email protected] - www.masell.com

Secretary General, EARTO, Rue de Luxembourg,3, 1000 Brussels, Belgium

Prof. Dr. F. Vosniakos T.E.I. of Thessaloniki, Applied Physics Lab. Abstracted/ indexed in: Biology & Environmental Sciences, P.O. Box 14561, 54101 Thessaloniki, Greece BIOSIS, C.A.B. International, Cambridge Scientific Abstracts, Chemical Abstracts, Current Awareness, Current Contents/ Agricul- Dr. K.I. Nikolaou ture, CSA Civil Engineering Abstracts, CSA Mechanical & Trans- Organization of the Master Plan & portation Engineering, IBIDS database, Information Ventures, NISC, Environmental Protection of Thessaloniki (OMPEPT) Research Alert, Science Citation Index Expanded (SCI Expanded), 54636 Thessaloniki, Greece SciSearch, Selected Water Resources Abstracts

0 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

CONTENTS

REVIEW ARTICLE

THE USE OF DIFFERENT REACTOR SYSTEMS FOR 3695 BIOLOGICAL TREATMENT OF PETROLEUM REFINERY WASTEWATERS Lucija Foglar, Sanja Papić, Dunja Margeta and Katica Sertić-Bionda

A REVIEW OF AGRO-FOOD WASTE TRANSFORMATION 3703 INTO FEEDSTOCK FOR USE IN FERMENTATION Paulina Gutiérrez-Macías, Brenda Montañez-Barragán and Blanca E Barragán-Huerta

ORIGINAL PAPERS

A STATISTICAL EXPERIMENTAL DESIGN TO DETERMINE THE AZO DYE 3717 DECOLORIZATION AND DEGRADATION BY THE HETEROGENEOUS FENTON PROCESS Yeliz Aşçı, Ezgi A. Demirtas, Cansu Filik Iscen and A. Sermet Anagun

DETERMINE SOME MORPHOLOGICAL CHARACTERISTICS OF 3727 CRAYFISH (Astacus leptodactylus ESCHSCHOLTZ, 1823) WITH TRADIONAL METHODS AND ARTIFICIAL NEURAL NETWORKS IN DIKILITAS POND, ANKARA, TURKEY Semra Benzer and Recep Benzer

FIRST IDENTIFICATION OF THE CYLINDROSPERMOPSIN (CYN)-PRODUCING CYANOBACTERIUM 3736 Cylindrospermopsis raciborskii (WOLOSZYŃSKA) SEENAYYA & SUBBA RAJU IN SERBIA Nevena B. Đorđević, Snežana B. Simić and Andrija R. Ćirić

GREENHOUSE GAS (GHG) EMISSIONS AND THE OPTIMUM OPERATION 3743 MODEL OF TIMBER PRODUCTION SYSTEMS IN SOUTHERN CHINA Yuan Zhou, Li-feng Zheng, Xin-nian Zhou, Xi-sheng Hu, Zhi-long Wu, Cheng-jun Zhou and Dan Li

MECHANISM OF GALLIC ACID EXTRACTION IN THE 3754 TRIBUTYL PHOSPHATE/KEROSENE EXTRACTING SYSTEM Yundong Wu, Kanggen Zhou, Shuyu Dong and Wei Yu

CONTENT OF ZINC IN DIFFERENT GRASS SPECIES GROWING ALONG A FAST HIGHWAY 3759 Elżbieta Malinowska, Kazimierz Jankowski, Beata Wiśniewska-Kadżajan, Jolanta Jankowska, Jacek Sosnowski, Grażyna Anna Ciepiela, Wiesław Szulc and Roman Kolczarek

IMPACT OF POSTPONED WATER FEEDING TIME ON THE 3766 INFILTRATION-REDUCING EFFECT OF CEMENT INTO ALLUVIAL SOILS Jinlan Ji and Guisheng Fan

STUDY ON THE SORPTION OF CHROMIUM(III) IONS IN AQUEOUS 3774 SOLUTION WITH DIATOMITE TREATED WITH MANGANESE OXIDE Xiangying Zeng and Er Li

LOSS CHARACTERISTICS OF NITROGEN AND PHOSPHORUS 3780 IN SURFACE RUNOFF WITH DIFFERENT LAND USE TYPES OF A SMALL WATERSHED IN FREEZE-THAW AGRICULTURAL AREA Xuehui Lai, Fanghua Hao and Xiaoli Ren

3693 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

FeCl3/NaNO2-CATALYZED OXIDATIVE DEGRADATION OF 3794 AQUATIC STEROID ESTROGENS WITH MOLECULAR OXYGEN Lianzhi Wang, Zhengchao Duan, Taoli Qu, Dongmei Fu and Xinmiao Liang

EVOLUTION OF THE MICROBIAL COMMUNITY 3802 RESPONSE TO DIFFERENT INDUSTRIAL WASTEWATERS Qing-Xiang Yang, Rui-Fei Wang, Ying Xu, Kai-Yue Zhuo, Hong-Li Zhao, Fei Zhang and Liu Yang

OPTIMIZATION OF DERIVATIZATION OF ACIDIC DRUGS FOR ANALYSIS BY GC-MS 3813 AND ITS APPLICATION FOR DETERMINATION OF DRUG RESIDUES IN WASTEWATER Richard Sýkora, Milada Vávrová, Petr Lacina and Ludmila Mravcova

ADSORPTION BEHAVIOR OF VICTORIA BLUE DYE ON 3822 HALLOYSITE NANOTUBES FROM AQUEOUS SOLUTIONS Pınar Turan Beyli

EFFECTS OF POTASSIUM SUPPLY ON PHYSIOLOGICAL 3831 PROPERTIES IN PEANUT UNDER ENHANCED ULTRAVIOLET-B RADIATION Yan Han, Yunsheng Lou, Chenguang Han, Meng Li and Chengda Hu

CHEMICAL, MICROBIAL AND PHYSICAL EVALUATION OF 3836 COMMERCIAL BOTTLED DRINKING WATER AVAILABLE IN BUSHEHR CITY, IRAN Elham Shabankareh Fard, Sina Dobaradaran and Roughayeh Hayati

EFFECTS OF CHLORPYRIFOS INSECTICIDE ON CYP1 FAMILY GENES EXPRESSION 3842 AND ACETYLCHOLINESTERASE ENZYME ACTIVITY IN ORYZIAS JAVANICUS Trieu Tuan, Yoshio Kaminishi, Aki Funahashi and Takao Itakura

CHEMICAL CHARACTERISTICS, SOURCES AND HEALTH IMPACTS OF URBAN 3853 AMBIENT PM2.5 IN WUHAN DURING THE CHINESE SPRING FESTIVAL POLLUTION EPISODE Li Liu, Hui Hu, Yiping Tian, Nan Chen, Jihong Quan and Hong Liu

APPLICATION OF THE SELF-ORGANIZING MAP FOR AQUEOUS PHOSPHORUS 3865 MODELING AND MONITORING IN A NATURAL FRESHWATER WETLAND: A CASE STUDY Liang Zhang, Yanhua Zhuang and Yun Du

SHORT NOTE

EFFECTS OF ZINC AND CADMIUM ON CONDITION FACTOR, 3871 HEPATOSOMATIC AND GONADOSOMATIC INDEX OF Oreochromis niloticus Nuray Çiftçi, Özcan Ay, Fahri Karayakar, Bedii Cicik and Cahit Erdem

INDEX 3875

3694 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

THE USE OF DIFFERENT REACTOR SYSTEMS FOR BIOLOGICAL TREATMENT OF PETROLEUM REFINERY WASTEWATERS

Lucija Foglar*, Sanja Papić, Dunja Margeta and Katica Sertić-Bionda

Faculty of Chemical Engineering and Technology, University of Zagreb, Marulićev trg 19, 10000 Zagreb, Croatia

ABSTRACT Furthermore, new technologies such as membrane technology [12, 13] and microwave-assisted catalytic wet In order to protect the environment and due to the strict air oxidation [14] are investigated. In general, these meth- regulations, harmful organic and inorganic compounds ods involve the transfer of pollutants from one medium to must be removed from refinery wastewaters. In this paper, another, thus requiring further treatment for complete re- the application of different reactor systems for biological moval of organic components. Some of these methods have treatment of petroleum refinery wastewaters was reviewed reduced efficiency, some generate sediments or sludge, and the performance of the different biological treatment whereas certain methods are performed in a narrow pH processes with suspended and immobilized biomass was range [15, 16], which limits their application. Therefore, described with emphasis on working parameters, such as the different bioreactor systems are still being investigated chemical oxygen demand (COD), organic load, hydraulic to achieve effective reduction of harmful organics present retention time, the presence of different carriers and micro- in refinery wastewaters. Biological treatment is the most organisms, the pH and the working temperature. Further- widely used treatment technologies for the removal of dis- more, some combined treatment processes were presented solved organic compounds from petroleum refining indus- and their use was analysed. According to the review of the try wastewater [17, 18]. The biological treatment can be literature data, it can be concluded that biological treat- classified into two main categories that include processes ment, with the selection of a suitable reactor system, fa- with suspended biomass and processes with immobilized vourable microorganisms and the optimal working param- biomass (Table 1.) eters, enables very efficient removal of organic substances present in the refinery wastewaters. TABLE 1 - Survey of different reactor systems and their abbrevia- tions used for the biological treatment of petroleum refinery

wastewaters. KEYWORDS: biological treatment, biomass, hydraulic retention time, chemical Reactor system Abbreviation oxygen demand, refinery wastewater, reactor systems Continuous flow suspended CSBR

biomass reactor 1. INTRODUCTION Up-flow anaerobic sludge blan- UASB ket bioreactor Reactors Sequencing batch reactors SBR The ever-increasing global energy demand makes the with suspended Membrane bioreactors MBR biomass growth processing and consumption of crude oil as raw material in Membrane sequencing batch re- MSBR petroleum industry important issue that significantly con- actors reactor tribute to pollution of water. The petroleum refinery and Wetlands ecosystems similar industries generate large volume of effluents which Active sludge reactor are discharged into water bodies [1, 2]. Such effluents con- Immobilized biological aerated IBAF tain many different substances including hydrocarbons, filters phenol, sulphides and other sulphur and nitrogen contain- Fluidised bed reactors FBR ing compounds. Numerous of these pollutants have ad- Fixed-film bioreactor system FFR verse effect on environment, especially on aquatic organ- Reactor with a biological aer- RBAF Reactors isms and therefore should be removed from wastewaters. ated filter with attached Trickling filter reactor The effective removal of these compounds with the use of biomass growth Rotating biological contactors RBC different physico-chemical and biological methods was in- Bio electrochemical reactor tensively studied during the last decade [3-11]. system Spouted bed bioreactor SBBR * Corresponding author Microbial electrolysis reactors

3695 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Generally, the microorganisms use and degrade the or- activated sludge treatment study in which effectiveness in ganic components of the wastewater and form new micro- cleaning up a petrochemical wastewater in Iran was evalu- bial biomass and degradation products. In the reactor with ated. Bacterial strains mainly belonged to Pseudomonas, suspended growth, microorganisms are suspended in the Flavobacterium, Comamonas, Cytophaga, Acidovorax, solution with organic material and during the treatment Sphingomonas, Bacillus and Acinetobacter genera [20]. The form aggregated active biomass called activated sludge. use of different engineering bacteria was studied in a pilot- The process of biological treatment with activated sludge scale two-stage anoxic-oxic process of biological treatment is one of the most effective biological systems, used in of petrochemical wastewater under low temperatures [21]. many refineries around the world for biological wastewater The different types of suspended biomass reactors ap- treatment. Microorganisms use organic matter from refin- plied for the RWW treatment, the values of applied process ery wastewater (RWW) as a source of carbon and energy parameters, the process efficiency and literature data were for growth and development and simultaneously convert shown in Table 2. The RWW treatment was also conducted these substances in cell tissue, water and other oxidized in up-flow anaerobic sludge blanket bioreactor - UASB [22]. products. In processes with immobilized biomass, micro- The operation of UASB includes anaerobic process with for- organisms are attached, bonded or immobilized to the inert mation a blanket of granular sludge which suspends in the carrier. The carriers, usually used in those processes are tank. Wastewater flows upwards through the blanket and is small rocks, gravel, plastic and other synthetic materials processed (degraded) by the anaerobic microorganisms. that have different size and properties that enable survival The monitoring of anaerobic RWW treatment from refin- and proliferation of microbes. [18]. ery in Teheran, Iran revealed 30-81% COD removal, with the highest efficiency observed at organic loading rate (OLR) of 0.4 kg/m3d and at hydraulic retention times (HRT) 2. APPLICATIONS OF DIFFERENT REACTOR of 48 h. After preliminary investigations, the process optimi- SYSTEMS WITH SUSPENDED BIOMASS FOR sation model (response surface methodology) was applied to BIOLOGICAL TREATMENT OF REFINERY predict the behaviours of influent COD, upflow velocity and WASTEWATERS HRT in the bioreactor and the determined optimal range of process performance was 630 mg/dm3, 0.27 m/h, and 21.4 h The most commonly used process that includes sus- respectively, at which was achieved 76.3% COD removal pended biomass is treatment in batch and continuous-flow efficiency [22]. reactors with activated sludge, and very often for treatment of RWW sequencing batch reactors and membrane biore- The batch reactors that have different sequences such as actors were applied. Furthermore, suspended biomass pro- fill/settling/decant during operation are in basic a modifica- cesses include the treatment in aerated lagoons and sys- tion of the traditional activated sludge processes are known tems, which are constructed as wetlands ecosystem [18]. as sequencing batch reactors - SBR [23]. The SBR technology is becoming more popular as an alternative for treating During the treatment, refinery wastewater entering the industrial wastewater due to use of a single unit, production aeration tank with microorganisms (reactor). To maintain of reduced amount of sludge, low aeration costs, minimal the structure of the aerobic active sludge and a homoge- footprint, and flexible operation and control [24-26]. nous suspension of microbes with solids present in the solution, air is continuously introduced into reactor. In gen- In a recently published paper [27], sequencing batch eral, in the RWW treatment systems, the biomass in the reactor was used for treatment of petroleum refinery waste- mixture of wastewater and sludge (mixed liquor suspended water (Kuala Lumpur) and the effects of toxic, hydraulic, solids, MLSS) contain 70-90% of active organic portion of and organic shocks on the performance of a lab-scale SBR the biomass (mixed liquor volatile suspended solids, were investigated. The results of this study indicate that with MLVSS) and 10-30% inert solids [18]. The treatment of an OLR in the range of 0.53-0.93 kg COD/kg MLSS d, COD RWW from six different refineries with use of the activated removal efficiency was 86-68.9% (Table 2). During the sludge was recently studied in the United States [17]. The toxic shock tests, it was observed that addition of 10 and +6 3 observed results suggest that the treatment was effective in 20 mg Cr /dm decrease efficiency to 79.4% and 77.1% reducing the chemical oxygen demand (COD) and the respectively, but the process regained stability and effi- naphthenic acid concentration. Furthermore, it was note- ciency increased to 86%. On the contrary, the simultaneous worthy that a significant toxicity reduction was achieved application of increased OLR and decreased HRT caused a by the biological treatment units in all six refineries. Simi- loss of biomass activity, biomass wash out and system fail- larly, Mizzouri and Shaaban [19] monitored the kinetics of ure [27]. degradation of RWW in continuous flow suspended bio- In addition to the above mentioned reactor systems, in a mass reactor (CSBR) and 89.9-96.5% reduction in COD recently published study [28] the possibility of RWW treat- was observed. The identification of bacteria present in bi- ment was investigated in the reactor system which combines omass indicated that prevailed bacteria Pseudomonas upward flow anaerobic sludge bioreactor (UASB) with and putida, Acidovorax delafieldii and Aeromonas hydrophila. immobilized biological aerated filters (IBAF). The pilot plant Furthermore, in recent investigation 67 different aerobic was set up in the vicinity of the oil fields and COD, ammonia bacterial cultures were isolated and identified during the concentration, suspended solids, and presence and degra-

3696 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 2 - The literature review of different types of suspended biomass reactors applied for the RWW treatment with processing parameters and the process efficiency data.

OLR*/ pH COD removal Microorganisms/ HRT Reactor system efficiency Reference Carrier (h) COD** T (°C) (%)  7-7.6 Active sludge reactor Refinery waste sludge  24.17-75.6 17 467-1117(700) 22-24 0.177-0.744  19,2- CSBR Mixed culture 89.9-96.5 19   25,6 0.2-1.2 6.5-7.7/ UASB Waste sludge 10-48 30-81 22 500-1200 38±1 0.3-0.93 6.7-7.3 SBR Active sludge 12,8-4,8 75-95 27 920-1620 27±2 Combination of  Active sludge (UASB) 7.8-8.3 UASB and IBAF and commercial culture 40-12 20-90 28 reactor 129.8-1238 (IBAF)/Policin urepan 25-35 /COD:616±231 Active sludge / 15-79 8.0 MBR /COD:1010±45 Membrane- polyeth- 10 48-77 29 /COD:765±187 erimide fibers 25 41-69 30-65 mg/d 16 93.8-94.3 Active sludge (3000 mg 6-8 20 94-95 (160-2300)103 MLSS/L)  33 94.5-95.5 MBR 12 24-67 mg/d / 17 93.5-94 Active sludge (5000 mg  22 93.5-94.5 (370-2300)103 MLSS/L) 34 94.7-95.8 ***Toluene: 25 Active sludge / 8  Removal of CH: MSBR Etilbenzene: 25 Poletilen mikrofiltration 16 30 27±1 97 Nonane: 25-3000 module 24 COD:165-347 Sand-gravel 8-10.5 45-78 Wetlands ecosystems Phenol: 3-9  34 Oil and grease: 24-66 Sand-compost 5-20 33-62 OLR*-organic loading rate (kg COD/kg MLSS/d) COD**-influent COD (mg COD/dm3) ***Concentration of different hydrocarbons (CH) (mg/dm3)

dation of alkanes were monitored during 252 days. The re- 3000 mg MLSS/dm3 and 5000 mg MLSS/dm3 was efficient duction of COD, NH3-N concentration and the reduction in and 82-97% reduction in COD was observed. The highest ef- suspended solids were 74%, 94% and 98% respectively ficiency of more than 95% was achieved at HRT of 34 h [12]. (Table 2). The analysis of the samples determined that most of the alkali compounds decomposed in the UASB-reactor, In addition to the above mentioned membrane system, in a recent survey [30], the effect of different HRT on the while the IBAF has an important role in the degradation of course of RWW treatment was determined in membrane other organic compounds and in reduction of NH3-N and suspended solids. Analysis of biomass indicated that in the sequencing batch reactors reactor (MSBR). Stages of operation of these reactors, depending on HRT (8, 16 and 24 h) UASB reactor dominated bacteria belonging to the genera Bacillales and Rhodobacterales, while in the IBAF prevail were filling phase - 11 min, the reaction phase (149, 389 and presence of an unidentified bacterial species [28]. 629 min), and drawing phase - 80 min. The process was monitored at 27 ± 1 °C and the prepared RWW contained Membrane bioreactors (MBR) are biological treatment mineral salts necessary for the optimal growth and develop- methods that represent the variation of application acti- ment of the microbial biomass and toluene, ethylbenzene vated sludge system. MBR combines membrane process and nonane as hydrocarbon components. Although the in- (eg. microfiltration) with suspended growth biomass [18]. crease of HRT from 8 h to 16 h and then to 24 h had resulted Application of MBR for RWW treatment [29] was studied in a statistically significant reduction in the value of MLSS, at different OLR and HRT of 10 h at 25 °C for 93 days and it was observed that the hydrocarbon removal efficiency was the results suggest the effective removal of organic matter more than 97% (Table 2) [30]. The SBR technology was (Table 2). The comparison of the results obtained with the used for treatment of RWW in other studies and effective use of MBR, with the results observed only with the use of reduction of COD was also observed [31, 32]. biomass, indicated that the application of the membrane increased COD removal and total organic carbon (TOC) re- Wetlands ecosystems can be designed as natural and moval efficiency for 17% and 20% respectively [29]. With engineered ecosystems, the natural wetland ecosystems are the application of cross flow membrane bioreactor [12] mainly used to remove suspended solids, nitrogen, phos- treatment of RWW at HRT of 16-34 h, in the presence of phorus, trace elements and microorganisms present in

3697 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

wastewater, while the constructed wetlands are used for mass, under unsteady and steady state conditions. The waste- treatment of municipal, industrial and agricultural wastewater and air stream were introduced at the bottom of re- water [33]. In order to explore the practical application of actor to achieve efficient fluidization and obtained results constructed wetland for RWW treatment, the performance indicated complete removal of diesel fuel (100%) and 90- of two parallel wetlands ecosystems and their effectiveness 97% reduction of COD [40]. The use of different types of were monitored in Pakistan [34]. Two identically engi- attached growth processes in the RWW treatment along neered wetland ecosystems contained different filter media with values of applied process parameters and process ef- (sand/compost and sand/gravel) and planted marsh cane ficiency are reviewed and listed in Table 3. shoots. In the first period, the growth of plants was moni- The aerobic treatment of RWW in a three-phase fluid- tored with the addition of fresh water, then the flow of ised bed bioreactor with polypropylene KMT® particles RWW was initiated and after a period of adjustment, the used as a biomass support was investigated at various ratios efficiency of this system was investigated. The COD re- of bed (settled) volume to bioreactor volume (Vb/VR) and at moval efficiency amounted to 45-78% for the system with different air velocities and the optimal COD removals of sand/compost and 33-62% for the system of sand/gravel. 90% was achieved at Vb/VR ratio of 0.55 and at air veloci- The reduction of total suspended solids in the wetlands was ties of 0.029 m/s [43]. A similar reactor system was used 48-73% in a system with sand/compost and 39-58% in the for RWW treatment in pilot study [45], and aerobic degra- system of sand/gravel (Table 2). In addition, an effective dation was conducted in the bioreactor containing 7 fixed reduction of biological oxygen demand (BOD) and the re- tubes placed in main reactor tube, and spherical particles duction of heavy metal concentrations were observed. Re- made of synthetic resin was used as biomass carrier. The sults of this study indicate that the use of wetlands ecosys- wastewater and air was pumped at the bottom of bioreactor tems enables effective reduction of the COD and BOD val- to enable aeration and fluidisation of carrier particles. The ues of RWW, and also enables significant reduction in the influences of pH value, air flow rate and HRT on COD and concentration of heavy metals such as Fe, Cu and Zn. [34]. ammonia nitrogen reductions were monitored and revealed Treatment of RWW in ecosystem designed as wetlands has that the optimum operation conditions were obtained at air been investigated in other studies and observed results in- flow rate of 3.2 m3 air /m3 liquid h, at HRT of 6.5 h and at dicated that the removal of harmful substances from the pH value of 7.0-8.0 (Table 3). Under the optimal operation RWW could be achieved with application of wetlands [35]. condition, during more than 40 days, the effluent COD and 3 NH4-N were lower than 100 and 15 mg/dm , respectively and the overall efficiency clearly exceeded the values ob- 3. BIOLOCGICAL TREATMENT OF REFINERY tained by traditional methods of wastewater degradation. WASTEWATER IN REACTOR SYSTEMS WITH Furthermore, this pilot-bioreactor generated only one- ATTACHED BIOMASS third of the sludge waste compared to the traditional activated sludge process. The processes with reactor systems that contain mi- Biological treatment of industrial oil refinery waste-wa- crobes attached, entrapped or immobilised on carrier, are ter is a well-established method for removal of organics but generally called attached growth processes. The carriers of in order to increase efficiency a new bioreactor system was microbes could be as mentioned sand, gravel, clay or zeo- designed and studied [46]. A fixed-film bioreactor system lite particles, hydro gels such as carrageenans and Ca-algi- (FFR) was constructed in order to allow higher organic nates, polymer granules and numerous different plastic or loads, to minimize production of sludge waste by-products synthetic materials. The use of small, porous, fluidized me- and to increase process stability and resistance to shock load- dia in the reactor enables the retaining of biomass and op- ing. The fixed-film bioreactor pilot unit was equipped with eration at reduced HRTs. In dependence to performance of a 4-chamber horizontal tank, the support frame was built the process there are different types of batch and continu- from cylindrical plastic pall rings to form a packed bed of ous flow reactors, which in correlation to flow could be re- ‘mixed-media’ and highly porous polyurethane foam was actors with fixed or fluidised bed with or without mixing. used to incubate microorganisms. Monitoring of input/out- The fluidised bed reactors (FBR) were intensively used put values of COD and phenols indicate 85-90% removal and investigated for efficient removal of diverse organic pol- of COD and complete reduction of phenols [46]. In this pa- lutants [36-39]. These reactor systems seems to be more ef- per a conventional treatment with activated sludge was also ficient than suspended biomass reactors since they enable monitored and reduction of COD and degradation of phe- process performance without mechanical moving parts, at nol yielded 50-60% and 99% respectively. The comparison higher concentration of biomass, at lower HRT and the of the results observed with FFR and conventional treat- footprint requirements are small [40]. Literature data sug- ment indicate that FFR during treatment generated only 1/3 gest that aerobic biodegradation by natural populations of of sludge waste produced during a conventional treatment microorganisms was a favourable for effective degradation [46]. Among many various systems even a reactor with a of diesel fuel [41-44]. In recently conducted study [40], biological aerated filter (RBAF) has been studied for treat- aerobic biological degradation of wastewater contaminated ment of RWW [47]. The performance of two aerobic, up- with diesel fuel was carried out in a three-phase fluidized flow, submerged polymethyl methacrylate reactors with bed reactor with lava rock particles used as support for bio- poly-ammoniacal carriers and two different commercial cul-

3698 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

tures of microorganisms (B350M and B350) were monitored degrade the organics. A trickling bed biofilm reactor at OLR of 0.4-1.07 kg/m3d and at different HRT (20-4 h) and packed with siliceous granular material (PORAVER parti- 83-99% reduction in COD was achieved. By monitoring of cles) was used to evaluate phenol and total organic carbon the RBAF treatment during 142 days and comparing the removal efficiencies by Pseudomonas putida DSM 548 results, a higher efficiency was observed in bioreactor with [48]. In another study, two identical biotrickling filters immobilized B350M culture (Table 3). Furthermore, the were loaded with aluminosilicate beads (molecular sieve) monitoring of polycyclic aromatic hydrocarbons concen- and polyurethane foam as filter media to compare the tolu- trations revealed that in both reactors their concentrations ene removal efficiency after seeding with Bacillus cereus were reduced and the efficiency of degradation in reactors S1 and based on observed higher removal capacity of tolu- with B350M and B350 cultures amounted 90% and 84% ene and shorter recovery time, aluminosilicate beads are respectively [47]. Additionally, analysis of microbial com- reported as a better choice than the polyurethane foam for munity proved presence of 20 different bacteria in the re- packing material used for biotrickling filter [49]. actor with B350M culture and only 13 different bacteria in In practice, submerged biofilm filter systems including the reactor with B350 culture, indicating that the reactor rotating biological contactors (RBC) are also used for inoculated with B350M achieves richer diversity than that RWW treatment. The RBC consists of closely spaced pol- with B350. ystyrene or polyvinyl chloride plastic discs mounted on a The trickling filter reactor system usually consisting horizontal shaft. The plastic discs are submerged in the distributors for spraying the influent wastewater to the wastewater and are continuously rotated by the horizontal surface of the filter bed, a bed of packing material such as shaft through an air driven motor. Microorganisms adhere rock or plastic on which the wastewater is distributed con- to the plastic surface and form a layer of microbial biomass tinuously and an under drain system that carried out treated (slime) on the discs. As the discs are rotating, the attached water. A slime layer of microorganisms develops on the microorganisms react and degrade the contaminants in the packing material of the trickling filter. As wastewater wastewater and convert them to new biomass and oxidation passes through the trickling filter bed, the microorganisms products, and the excess of sludge is sloughed off the discs.

TABLE 3  The review of process parameters used in different types of attached biomass growth processes during the RWW treatment and observed process efficiency.

COD removal OLR*/ Microorganisms/ pH HRT Reactor system efficiency Reference Carrier (h) COD** T (°C) (%)  6.7-7.8 90-97 547-4025 FBR Lava rock particles 4 h [40]   96 200 and 1237 diesel: 100% 6.5-7.0  Polypropylene KMT® FBR  90-95 [43] particles 45000 28-30

 Polypropylene parti- 6.5-7.0 FBR 3.33-30 40-95 [44] 36650 cles 28-30  7-8 FBR Syntic resin 6.5 > 53-84 [45] 215-613 25-35  6.0-8.5 16 85-90 FFR 510±401.9 Polyurethane foam [46] 15-39 8 phenol: 100 Phenol***: 30±6.2 0.4-1.07 Commercial culture 8.0 B350 and B350M / RBAF 20-4 83-99 [47] 124 Polyamoniacal mate- 18-24 rial 27,33 g/m2d Burkholderia cepacia 7.5 24 97.19-78.56 RBC and phototrophic mi- [51] 21 84.6-97.8 2677,30-5406,38 croorganisms 28±2  7,2-8,2 COD: 97 Combined reactor Pseudonymous putida 30 and ambi-  phenol: 100 [53] system 3600-5300 /Polyvinyl alcohol gel ent T Cresol: 100 Microbial electrol-  Mixed culture present 7.2-8.9  40-79 [54] ysis cells 136-1400 in RWW 30 OLR*-organic loading rate (kg COD/kg MLSS/d) COD**-influent COD (mg COD/dm3) ***Concentration of different hydrocarbons (CH) (mg/dm3)

3699 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

The basic advantages of RBC over other types of filter re- RWW sample or a mixture of RWW with domestic actors are that the addition of oxygen isn’t required and wastewater in microbial electrolysis cells resulted in 79% process is not dependent on the concentration of oxygen in reduction of COD and 82% reduction of BOD values. the water, furthermore, the microorganisms are attached to the surface of the disc and thus clogging of the filter are very rare, and process performance have a small energy re- 4. CONCLUSIONS quirements. Biodegradability of RWW was investigated in laboratory scale by applying the modified RBC with poly- According to the literature data on the performance of urethane foam [50]. Degradation of RWW was monitored the biological treatment of refinery wastewaters, the selec- in two parallel-connected RBC at various input hydraulic tion of an appropriate reactor system and selection of fa- loads (from 0.01 to 0.04 m3/m2d) with a rotation speed of vourable conditions of the process implementation, as well 10 rpm, and the efficiency of COD and the oil reduction as the selection of suitable microorganisms can signifi- was greater than 87% and 80%, respectively. Additionally, cantly affect the efficiency of the biological treatment of observed removal of NH3-N and phenol was 99% and 85% RWW. The presented and reviewed data indicated that respectively [50]. Application of RBC was also investi- among suspended growth bioreactor, the application of gated for treatment of wastewater containing diesel oil and membrane bioreactors is very efficient and provides more the presence of bacteria Burkholderia cepacia and cyano- than 93% reduction of COD, but the use of immobilized bacteria (Phormidium, Chroococcus and Oscillatoria) was cells also resulted in a very high efficiency of the process, determined [51]. After the formation of biofilm and the pe- particularly in the fluidized bed reactor system, in which riod of adjustment, in the outlet stream, the presence of up to 96% COD removal efficiency is observed. In addi- only 0.003% of residual diesel was determined. The effec- tion, application of the combined reactor system consisted tiveness of wastewater treatment was monitored at differ- of electro coagulation cell, fluidized bed reactor and ad- ent N:P ratios (19:1 to 47.4:1), and at HRT of 21 h, at OLR sorption column, also provides 97% reduction of COD and of 27.33 g/m2d. The observed reduction of the COD was the 100% removal of phenol and cresol. 84-97.8% and the reduction efficiency of the total petro- The treatment of RWW is significantly affected by op- leum hydrocarbons was more than 99% (Table 3). Further- erational parameters such as temperature, pH and HRT. more, the association of B. cepacia and phototrophic mi- The analysis of the data presented, indicate that the treat- croorganisms at a reactor scale was reported as favourable ment of RWW are mainly conducted at the pH neutral or for hydrocarbon degradation. The major advantages of this close to neutral (pH 6.0-8.9) and at temperatures of 25-35 RBC system apart from high total petroleum hydrocarbons °C, which are the values that enable optimal growth and removal efficiency include good settleability of sludge, development of microorganisms that can degrade organic low phosphorus requirement, no soluble carbon source re- substances present in the RWW. The comparison of differ- quirement and pH stability. Thus, although RBCs were ent reactor system with suspended and immobilized bio- commonly used for treatment of low-strength wastewater, mass, the comparison of process efficiency and the HRT RBCs utilizing the association of phototrophic microor- values provide reliable observation that the effectiveness ganisms and bacteria can effectively treat wastewater with of the process is higher, and HRT is mostly lower in the high organic loading. The RBC was investigated for possi- reactor systems with immobilized biomass, which contrib- 2 ble biodegradation of toluene at OLR of 1.5-0.6 g/m d and utes to the cost-effective removal of harmful pollutants observed reduction of COD ranged from 77 to 88.3% [52]. from refinery wastewaters during the biological treatment. Application of different bio electrochemical system, and in particular, the application of combined reactor sys- The authors have declared no conflict of interest. tems in the treatment and purification of wastewater, could result in effective removal of suspended solids, oil and grease [53]. A novel three-step process was developed and REFERENCES evaluated for the treatment of highly contaminated refinery wastewater. The process consisted of an electro-coagula- [1] Wake, H., (2005) Oil refineries: a review of their ecological tion cell (EC), a spouted bed bioreactor (SBBR) with Pseu- impacts on the aquatic environment. Estuarine, Coastal and Shelf Science, 62, 131–140. donymous putida immobilized in polyvinyl alcohol gel, and an adsorption column packed with granular activated [2] Diyauddeen, B.H., Wan Daud, W.M.A. and Aziz, A.R.A. (2010) Treatment technologies for petroleum refinery efflu- carbon produced from agricultural waste. During opera- ents: A review. Process Safety and Environmental Protection, tion, the samples from each of the output stream were ana- 89, 95–105. lyzed and the concentrations of phenol, cresol and COD [3] El-Naas, M.H., Al-Zuhair, S., Al-Lobaney, A. and Makhlouf, were determined and 100% removal of phenol and cresol, S. (2009) Assessment of electrocoagulation for the treatment and 97% reduction of COD was achieved (Table 3). Mi- of petroleum refinery wastewater. Journal of Environmental crobial electrolysis reactors in which the anode is graphite Management, 91, 180–185. plate and cathode is a mesh of stainless steel have been [4] Pardeshi, S.K. and Patil, A.B. (2008) A simple route for pho- used in a recent paper for the treatment of various RWW tocatalytic degradation of phenol in aqueous zinc oxide sus- [54]. In anaerobic batch conditions, the introduction of the pension using solar energy. Solar Energy, 82, 700–705.

3700 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[5] Xu, X. and Zhu, X., (2004) Treatment of refectory oily [22] Rastegar, S.O., Mousavi, S.M., Shojaosadati, S.A. and wastewater by electro-coagulation process. Chemosphere, 56, Sheibani, S. (2011) Optimization of petroleum refinery efflu- 889–894. ent treatment in a UASB reactor using response surface methodology. Journal of Hazardous Materials, 197, 26–32. [6] Abdelwahab, O., Amin, N.K. and El-Ashtoukhy, E.-S.Z. (2009) Electrochemical removal of phenol from oil refinery [23] Suresh, S., Tripathi, R.K. and Gernal Rana M. (2011) Review wastewater. Journal of Hazardous Materials, 163, 711–716. on treatment of industrial wastewater using sequential batch reactor. International Journal of Technology Management, 2, [7] Yang, S., Zhu, W., Wang, J. and Chen, Z. (2008) Catalytic wet 64–84. air oxidation of phenol over CeO2–TiO2 catalyst in the batch reactor and the packed-bed reactor. Journal of Hazardous Ma- [24] US EPA (1999) Wastewater Technology Fact Sheet: Sequenc- terials, 153, 1248–1253. ing Batch Reactor, 832-F-99-073, US EPA, Washington DC. [8] Altas, L. and Büyükgüngör, H. (2008) Sulfide removal in pe- [25] Gerardi, M.H. (2009) Troubleshooting the Sequencing Batch troleum refinery wastewater by chemical precipitation. Journal Reactor, Wiley, Hoboken, New Jersey. of Hazardous Materials, 153, 462–469. [26] Shammas, N.K., Wang, L.K., Pereira, N.C. and Hung, Y.T. [9] Hami, M.L., Al-Hashimib, M.A. and Al-Dooric, M.M. (2007) (2009) Biological treatment processes. Handbook of Environ- Effect of activated carbon on BOD and COD removal in a dis- mental Engineering, 8, The Humana Press, Totowa, New Jersey. solved air flotation unit treating refinery wastewater. Desali- nation, 216, 116–122. [27] Mizzouri, N.S. and Shaaban, M.G. (2013) Individual and combined effects of organic, toxic, and hydraulic shocks on se- [10] Santos, F.V., Azevedo, E.B., Sant’Anna Jr., G.L. and Dezotti, quencing batch reactor in treating petroleum refinery M. (2006) Photocatalysis as a tertiary treatment for petroleum wastewater. Journal of Hazardous Materials, 250-251, 333– refinery wastewaters. Brazilian Journal of Chemical Engineer- 344. ing, 23, 450–460. [28] Liu, G., Ye, Z., Tong, K. and Zhang, Y. (2013) Biotreatment [11] Ma, F., Guo, J.-B., Zhao, L.-J., Chang, C.-C. and Cui, D. (2009) of heavy oil wastewater by combined upflow anaerobic sludge Application of bioaugmentation to improve the activated sludge blanket and immobilized biological aerated filter in a pilot- system into the contact oxidation system treating petrochemical scale test. Biochemical Engineering Journal, 72, 48–53. wastewater, Bioresource Technology 100, 597–602. [29] Viero, A.F., de Melo, T.M., Torres, A.P.R., Ferreira, N.R., [12] Rahman, M.M.and Al-Malack, M.H. (2006) Performance of a Sant’Anna Jr., G.L., Borges, C.P. and Santiago, V.M.J. (2008) crossflow membrane bioreactor (CF–MBR) when treating re- The effects of long-term feeding of high organic loading in a finery wastewater. Desalination, 191, 16–263. submerged membrane bioreactor treating oil refinery wastewater. Journal of Membrane Science, 319, 223 –230. [13] Li, Y., Yan, L., Xiang, C. and Hong, L.J. (2006) Treatment of oily wastewater by organic–inorganic composite tubular ultra- [30] Shariati, S.R.P., Bonakdarpour, B., Zare, N. and Ashtiani, F.Z. filtration (UF) membranes. Desalination, 196, 76–83. (2011) The effect of hydraulic retention time on the performance and fouling characteristics of membrane sequencing [14] Sun, Y., Zhang, Y. and Quan, X. (2008) Treatment of petro- batch reactors used for the treatment of synthetic petroleum leum refinery wastewater by microwave-assisted catalytic wet refinery wastewater. Bioresource Technology, 102, 7692– air oxidation under low temperature and low pressure. Separa- 7699. tion and Purification Technology, 62, 565–570. [31] Lee, L.Y., Hu, J.Y., Ong, S.L., Ng, W.J., Ren, J.H. and Wong, [15] Laoufi, N.A., Tassalit, D. and Bentahar, F. (2008) The degra- S.H. (2004) Two-stage SBR for treatment of oil refinery dation of phenol in water solution by TiO2 photocatalyst in a wastewater. Water Science and Technology, 50, 243–249. chemical reactor. Global NEST Journal, 10, 404–418. [32] Hudson, N., Doyle, J., Lant, P., Roach, N., de Bruyn, B. and [16] Kuyukina, M.S., Ivshina, I.R., Serebrennikova, M.K., Krivo- Staib, C. (2001) Sequencing batch reactor technology: the key rutchko, A.B., Podorozhko, E.A., Ivanov, R.V. and Lozinsky, to a BP refinery (Bulwer Island) upgraded environmental pro- V.I. (2009) Petroleum-contaminated water treatment in a fluid- tection system – A low cost lagoon based retro-fit. Water Sci- ized-bed bioreactor with immobilized Rhodococcus cells. Inter- ence and Technology, 43, 339–346. national Biodeterioration & Biodegradation, 63, 427–432. [33] Pardue M.J., Castle J.W., Rodgers J.H.Jr. and Huddleston III, [17] Misiti, T., Tezel, U.and Pavlostathis, S.G. (2013) Fate and ef- G.M. (2014) Treatment of oil and grease in produced water by fect of naphthenic acids on oil refinery activated sludge a pilot-scale constructed wetland system using biogeochemi- wastewater treatment systems. Water Research, 47, 449–460. cal processes. Chemosphere, 103, 67–73. [18] IPIECA (2010) Petroleum refinery waste management and [34] Aslam, M.M., Malik, M., Baig, M.A., Qazi, I.A. and Iqbal, J. minimization An IPIECA Good Practice Guide, IPIECA, Lon- (2007) Treatment performances of compost-based and gravel- don, UK [available at: http://www.ipieca.org; 28.04.2014.] based vertical flow wetlands operated identically for refinery wastewater treatment in Pakistan. Ecological Engineering, 30, [19] Mizzouri, N.S. and Shaaban, G. (2013) Kinetic and hydrody- 34–42. namic assessment of an aerobic purification system for petroleum refinery wastewater treatment in a continuous regime. [35] Simi, A.L., Mitchell, C.A. (1999) Design and hydraulic per- International. Biodeterioration & Biodegradation, 83, 1–9. formance of a constructed wetland treating oil refinery wastewater. Water Science and Technology, 40, 301–307. [20] Shokrollahzadeh, S., Azizmohseni, F., Golmohammad, F., Shokouhi, H. and Khademhaghighat, F. (2008) Biodegrada- [36] Sutton, P.M., Mishra, P.N. (1994) Activated carbon based bi- tion potential and bacterial diversity of a petrochemical ological fluidized beds for contaminated water and wastewater wastewater treatment plant in Iran. Bioresource Technology, treatment: A state of the art review. Water Science and Tech- 99, 6127–6133. nology, 29, 309–317. [21] Zhao, L., Ma, F., Guo, J. and Zhao, Q. (2007) Petrochemical [37] Buchtmann, C., Kies, U., Deckwer, W.D. and Hecht, V. (1997) wastewater treatment with a pilot-scale bioaugmented biolog- Performance of three phase fluidized bed reactor for quinoline ical treatment system. Journal of Zhejiang University Science degradation on various supports at steady state and dynamic A, 8, 1831–1838. conditions. Biotechnology and Bioengineering, 56, 295–303.

3701 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[38] Haribabu, K. and Sivasubramanian, V. (2014) Treatment of Wastewater in Fluidized Bed Bioreactor Using Low Density Biosupport. Energy Procedia, 50, 214–221.

[39] Gonzales, G., Herrera, G., Garcia, Ma.T. and Pena M. (2001) Biodegradation of phenolic wastewater in a fluidized bed bioreactor with immobilized cells of Pseudomonas putida. Biore- source Technology, 80, 137–142. [40] Lohi, A., Alvarez Cuenca, M., Anania, G., Upreti, S.R., and Wan, L. (2008) Biodegradation of diesel fuel-contaminated wastewater using a three-phase fluidized bed reactor. Journal of Hazardous Materials, 154, 105–111. [41] Margesin, R. and Schinner, F. (1997) Bioremediation of diesel fuel-oil-contaminated alpine soils at low temperature. Applied Microbiology and Biotechnology, 47, 462–468. [42] Erikson, M., Swartling, A. and Dalhammar, G. (1998) Biolog- ical degradation of diesel fuel in water and soil monitored with solid-phase micro-extraction and GC–MS. Applied Microbi- ology and Biotechnology, 50, 129–134. [43] Sokol, W. (2003) Treatment of refinery wastewater in a three- phase fluidised bed bioreactor with a low density biomass support. Biochemical Engineering Journal, 15, 1–10.

[44] Sokół, W., Ambawb, A. and Woldeyes, B. (2009) Biological wastewater treatment in the inverse fluidised bed reactor. Chemical Engineering Journal, 150, 63–68.

[45] Xianling, L., Jianping, W., Qing, Y. and Xueming, Z. (2005) The pilot study for oil refinery wastewater treatment using a gas–liquid–solid three-phase flow airlift loop bioreactor. Bio- chemical Engineering Journal, 27, 40–44.

[46] Jou, C.J.G. and Huang, G.C. (2003) A pilot study for oil refinery wastewater treatment using a fixed-film bioreactor. Ad- vances in Environmental Research, 7, 463–469.

[47] Zhao, X., Wang, Y., Ye, Z., Borthwick, A.G.L. and Ni, J. (2006) Oil field wastewater treatment in Biological Aerated Filter by immobilized microorganisms. Process Biochemistry, 41, 1475–1483. [48] Sá, C.S.A. and Boaventura, R.A.R. (2001) Biodegradation of phenol by Pseudomonas putida DSM 548 in a trickling bed reactor. Biochemical Engineering Journal, 9, 211–219. [49] He, Z., Zhou, L., Li, G., Zeng, X., An, T., Sheng, G., Fu, J. and Bai, Z. (2009) Comparative study of the eliminating of waste gas containing toluene in twin biotrickling filters packed with molecular sieve and polyurethane foam. Journal of Hazardous Materials, 167, 275–281.

[50] Tyagi, R.D., Tran, F.T. and Chowdhury, A.K.M.M. (1993) Bi- odegradation of petroleum refinery wastewater in a modified Received: November 26, 2014 rotating biological contactor with polyurethane foam attached Revised: March 30, 2015 to the disks. Water Research, 27, 91–99. Accepted: April 24, 2015 [51] Chavan, A. and Mukherji, S. (2008) Treatment of hydrocarbon-rich wastewater using oil degrading bacteria and phototrophic microorganisms in rotating biological contactor: Effect of N:P ratio. Journal of Hazardous Materials, 154, 63–72. CORRESPONDING AUTHOR [52] Alemzadeh, I. and Vossoughi, M. (2001) Biodegradation of Lucija Foglar toluene by an attached biofilm in a rotating biological contactor, Process Biochemistry, 36, 707–711. Faculty of Chemical Engineering and Technology University of Zagreb [53] El-Naas, M.H., Alhaija, M.A. and Al-Zuhair, S. (2014) Eval- Marulićev trg 19 uation of a three-step process for the treatment of petroleum refinery wastewater. Journal of Environmental Chemical En- 10000 Zagreb gineering, 2, 56–62. CROATIA

[54] Ren, L., Siegert, M., Ivanov, I., Pisciotta, J.M. and Logan, B.E. (2013) Treatability studies on different refinery wastewater E-mail: [email protected] samples using high-throughput microbial electrolysis cells (MECs). Bioresource Technology, 136, 322–328. FEB/ Vol 24/ No 11a/ 2015 – pages 3695 - 3702

3702 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

A REVIEW OF AGRO-FOOD WASTE TRANSFORMATION INTO FEEDSTOCK FOR USE IN FERMENTATION

Paulina Gutiérrez-Macías, Brenda Montañez-Barragán and Blanca E. Barragán-Huerta*

Instituto Politécnico Nacional. Escuela Nacional de Ciencias Biológicas. Departamento de Ingeniería en Sistemas Ambientales. Av. Wilfrido Massieu S/N, Unidad Profesional Adolfo López Mateos, CP 07739, México D.F. México

ABSTRACT tainable development based on the concept of biorefinery has produced alternatives that facilitate waste recovery [4,5]. Food processing by-products and wastes are generated The International Energy Agency [6] defines the term from direct consumption or industrialization of primary biorefinery as “the sustainable processing of biomass into products that are no longer useful for food processing. Cur- a spectrum of marketable products (e.g., food, feed, mate- rently, in several developing countries there is insufficient rials, and chemicals) and energy (e.g., fuel, power, and infrastructure, technology, financial resources, and specific heat)”. There are several crops that are mainly used as feed- legislation to facilitate proper disposal and provide a final stock for biorefinery (perennial grasses, starch, sugarcane, destination for the waste, which can create pollution. Sus- oil, and forestry wastes), but the utilization of by-products tainable development based on the concept of biorefinery from the food industry has recently increased, because they has produced alternatives that facilitate waste recovery. are rich in nutrients and their reuse could also solve prob- The food processing by-products can be transformed into lems related to their disposal [7,8]. biomaterials or biofuels by fermentation; however, the The use of biomass as part of the biorefinery system in- main limitation is availability of fermentable compounds volves a series of sequential processes, including its prepa- because of the complex chemical structure of biomass. For ration and transformation into products of interest [9]. In that reason, some pretreatments are initially applied to the most cases, pretreating the biomass is required to increase biomass to increase the degree of saccharification and ob- the availability of its components and obtain fermentable sub- tain fermentable sugars. Recent advances in technology de- strates that are used for direct or indirect production of biofuel veloped for utilizing food biomass waste for production of and/or biomaterials [10]. Thus, hydrolysis efficiency is meas- fermentable sugars focusing in treatment conditions and ured in terms of free sugars (e.g., glucose, xylose, and galac- the limitations of each are discussed in this review. tose) and fermentation inhibitors [e.g., furfural and hydroxy- methylfurfural (HMF) yields].

KEYWORDS: biorefinery, pretreatment, biomass, lignocellulosic, food waste 2. BIOMASS COMPOSITION

Lignocellulosic biomass is mainly composed of cellu- 1. INTRODUCTION lose (38–50%), hemicellulose (23–32%), and lignin (10– 25%) [11, 12]. Cellulose is a homopolymer composed of Processed agro-food waste materials, both in solid and glucose monomers that are linked by glycoside bonds β- liquid form, are generated from the direct consumption or 1and β-4 [13]. Most of the glucose present in the lignocellu- industrialization of primary products that are no longer use- losic biomass is a component of the cellulose, which is a pol- ful for food processing [1]. This waste is mainly utilized as ymer that is hard to hydrolyze because of its crystalline and animal feed and compost or is disposed of in landfills; how- linear conformation. Moreover, it is surrounded by hemicel- ever, it is rich in chemical components with commercial lulose and lignin that further hinder its hydrolysis [14]. interest that can be obtained by extraction or transformation to generate valuable products [2]. Hemicellulose is a heteropolymer that is constituted by pentoses (D-xylose and L-arabinose), hexoses (D-glucose, The problems associated with agro-food processing D-mannose and D-galactose), and uronic acids, such as D- wastes include insufficient technology, financial resources, glucuronic acid, D-galacturonic acid, and galacturonic acid and specific legislation to provide an appropriate destination methyl esters [15]. Because of its amorphous structure, for the wastes [3]. To ensure proper management from gen- multiple branching, and relatively low molecular weight, eration to disposal, these problems need to be addressed. Sus- hemicellulose is easier to hydrolyze than cellulose [16].

But, because the diversity of sugars, it requires a wide range * Corresponding author of enzymes to complete hydrolysis of the biomass [17].

3703 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

The third major component of lignocellulosic biomass can incorporate combinations of different pretreatment clas- is lignin, which is composed of multiple units of three alkyl sifications [24, 36]. phenols: p-coumaryl, coniferyl, and sinapil, which can be solubilized and inhibit fermentation process at different 3.1 Physical pretreatments grade [18]. This polymer is tightly bound to cellulose and Physical pretreatments are mainly applied to increase hemicellulose, making them less accessible to hydrolysis the surface area, reduce particle size, degree of polymeriza- [19, 20]. Because of the large quantity of sugars in the lig- tion and crystallinity of the lignocellulosic material [18, 37]. nocellulosic biomass, it is an excellent source of fermentable The energy required in these pretreatments is relatively high materials. Moreover, lignocellulosic components of some and depends on the desired particle size and biomass char- food processing by-products, such as citrus peels, contain a acteristics [38]. However, very small particle sizes may sub- high content of essential oils or galacturonic acid, which sequently cause negative effects, such as agglomeration of could be useful in recovery of additional value-added prod- the biomass during enzymatic hydrolysis treatment [34]. ucts [21]. Potato peels contain until 52% d.w. starch, which Milling (ball mills, vibratory mills, roller mills, ham- is more easily hydrolysable than cellulose and produces a mer mills, knife mills, colloid mills, and attrition mills) is relatively clean glucose stream for fermentation process the most widely used mechanical method for physical pre- [22]. Apple pomace contains large amounts of specific pol- treatment of lignocellulosic biomass [39], although ultra- yphenols, such as chlorogenic acid, epicatechine, and phlo- sonication [40, 41], microwave irradiation [42, 43, 44], ridzin [23] that can act as fermentation and enzyme inhibi- gamma rays [45, 46], and electron bombardment [36] have tors and therefore as an additional step for the removal. recently been studied.

The choice of mill type also depends upon the desired 3. PRETREATMENT AND particle size and biomass moisture content. Colloidal mills SACCHARIFICATION OF BIOMASS and extruders are only suitable for grinding wet materials with a moisture content of more than 15–20%, while two- roll mills, attrition, hammers, or knives are only suitable The main limitation that prevents optimal use of bio- for grinding dry biomass with a moisture content of up to mass is the availability of fermentable compounds [24]. For 10–15%. Ball and vibratory ball mills are universal types that reason, it is necessary to apply the appropriate pretreat- of blasters, and they can be used for either dry or wet ma- ment to remove the complex network formed by lignin and terials [39]. hemicellulose that inhibits effective enzymatic action on cellulose [25, 26]. Likewise, it is required to disrupt the hydro- The particle size obtained with these pretreatments gen bonds between cellulose units as well as increase the po- varies from 0.2–2.0 mm after grinding and 10–30 mm after rosity and surface area for subsequent hydrolysis [27, 28]. chopping. Thus, size reduction leads to an increased overall hydrolysis yield of 5–25% as well as reduced digestion However, biomass characteristics differ depending on time by 23–59% [24, 47, 48]. Application of these pretreat- the source from which it is obtained and choice of pretreatments requires a large amount of energy, and implementa- ment employed, which are different for each situation. The tion at the industrial level is not cost-effective [18, 24], also main aspects that strongly influence enzymatic hydrolysis energy needs raises grow up with moisture content [24]. are the index of cellulose crystallinity [29], degree of Moreover, combination with another physical or chemical polymerization of cellulose[30], area of contact surface be- treatment is required [49, 50]. tween the substrate and enzyme [31], content and distribution of lignin [32], hemicellulose content, porosity of the 3.2 Chemical pretreatment substrate, and thickness of the cell wall [33]. Chemical pretreatments have been extensively investi- Depending on the physicochemical characteristics of gated because of their high efficacy in delignification of lignocellulosic biomass, pretreatment selection should take biomass as well as de-polymerization and decrystallization into consideration the following aspects: i) maximum sugar of cellulose [51]. Moreover, these pretreatments are easily yields, ii) loss and/or low sugar degradation [34], iii) mini- scaled to industrial levels and application is generally cost mum formation of toxic products such as furfural and effective. HMF, which are inhibitory compounds that affect the fer- There are a variety of chemicals that can be used for mentation process [24, 35], iv) it must not require particle this kind of pretreatment, such as oxidizing agents, acids size reduction, v) effectiveness of the process when there and bases, which may also be implemented in combination is a low moisture content [31], vi) compatibility of fermen- with other pretreatments (Table 1). tation processes, vii) minimum heat and/or energy requirements [31, 34], and viii) maximum recovery of lignin [35]. 3.2.1 Acid hydrolysis Currently, there is no universal pretreatment because The use of acid is the most common chemical pretreat- of the diverse biomass characteristics. Therefore, a variety ment in which the constituent monomers of hemicellulose of methods based on different principles have been devel- (mainly xylose) are hydrolyzed. Moreover, the effect of acid oped. These pretreatments are classified as: physical, chem- for pretreatment of the biomass structure facilitates enzy- ical, physicochemical, and biological; however, approaches matic action of cellulase [74, 75].

3704 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 1 - Chemical pretreatments applied to the exploitation of lignocellulosic biomass

Type Pretreat- Waste Pretreatment conditions Enzymatic treatment Yield Reference ment

Acid hydrolysis Sugarcane H2SO4 (1%) and acetic acid 2.32 g Celluclast 1.5 L (65 FPU/ml to GLU 93 g/100 g hemicellu- [52] bagasse (1%), 90 ºC, 10 min 17 IU/ml of β-glucosidase) and 0.52 g lose of Novozyme 188 (376 IU/g of β- glucosidase), 10 g solid, 24 h, pH 4.8 Rice hulls H2SO4 (0.3 %), S/L 5:95, 152 CEL (40 FPU/g), 50ºC, 48 h, pH 4.8 GLU 47.2 g/100 g glucan; [53] ºC, 33 min XYL 86.3 g/100 xylan Wheat straw HCl (1.5% w/w), S/L 1:10 10 FPU/(NS50013) and 10 CBU/g GLU 1.25 g/100 g RM [54] w/w, autoclaved at 121 °C, (NS50010) of cellulose,10% (w/w) XYL 16.49/100 g RM 60 min solid, 50 °C, 72 h, pH 5 Cashew Apple H2SO4 (0.6 mol/L), solid 30% Not applied GLU 256.3 mg/g cellulose [55] bagasse (w/v), 121 ºC, 30 min XYL 987.1 mg/g hemicellulose Corn stover H3PO4 (0.5% v/v), 180 °C, 15 Celluclast 1.5 L (15 FPU/g glucan), GLU 35.35 g/100 stover [56] min Novozyme 188, (1578 U/g glucan), XYL 16.83 g/100 g stover Fiberzyme (1578 U/g hemicellulose), solid 10% (w/w), 45 °C, 72 h, pH 5 Rice straw Step 1: H2SO4 (1%), 60 °C, Commercial enzyme blend (6 ppm), GLU 380 g/L [57] 24 h. Step 2: autoclave at 1% solid, 60 °C, 24 h. 121 °C, 15 min, 15 lb pressure Sugarcane H2SO4 (2%), S/L 1:20, 200 Not applied GLU 69.06 g/100g glucan [58] bagasse ºC, 45 min, assisted with ul- XYL 81.35 g/100 xylan trasonic energy

Alkaline hy- Sorghum Ammonium hydroxide (28% CEL (60 FPU/g glucan) GLU 42 g/100 g RM [59] drolysis bagasse v/v), 130 °C, 1 h Novozyme 188 (64 CBU/g glucan), 5 g solid, 55 °C, 24 h, pH 4.8 Rice straw NaOH (12% w/v), 5% solid, Celluclast 1.5 L (80 FPU/ml), GLU 163 g/kg RM [60] 0 °C, 3 h Novozyme 188 (240 IU/ml), 2% solid, 45 °C, 72 h, pH 4.8 Sorghum straw NaOH (2% w/w), 10% (w/v) CEL (2.5 FPU/g), β-glucosidase GLU 5.33 g/100 g RM; [61] solid, 121 °C, 60 min (3.75 CBU/g), xylanase (1.5 FXU/g), XYL 2.34 g/100 g RM 50 °C, 48 h Rapeseed straw NaOH 2M (5% in solution), HCEL (0.3% v/v), β-glucosidase GLU: 12 g/100 g RM; [62] H2O2 (35% m/m), 50⁰C, 1 h, (0.3% v/v), 50 ºC, 24 h, pH 4.8 XYL: 3 g/100 g RM pH 11.5 Kapok fiber NaOH (2%), S/L 1: 12, 120 ºC, Celluclast 1.5 L (68.3 FPU/ml), TS 7.5 g/100 g RM [63] 60 min 50 ºC, 60 min Soybean straw NaOH (12 % w/w), S/L 1:10, Spezyme CP loading 2.5% (w/w) GLU 64.55/100 g solid [64] RT, 24 h, pH 8 and 20 FPU/g solid, 50 °C, 48 h mass

Ionic liquids Rice straw 1-ethyl-3-methylimidazolium Celluclast 1.5L (60 FPU/ ml), GLU 450 g/kg rice straw [65] acetate, 5% solid, 120 ºC, 5 h Novozyme 188 (190 IU/ml), 10 g/l solid, 45 ºC, 72 h, pH 4.8 Corn stover 5% 1-ethyl 3-methyl imidazo- Novozyme 188 (5 FPU/g), 50 ºC, GLU 60 g/100 glucan, [66] lium acetate, 100 ºC, 4 h 24 h, pH 4.8 XYL 55 g/100 g xylan Triticale straw 1-ethyl-3-methylimidazolium ac- CEL (35 U/ml), 10 mg solid, 50 °C, GLU 81/100 g glucan [67] etate (50% w/v), 150 °C, 90 min pH 4.8 Rice straw 1-ethyl-3-methylimidazolium- 1.5 mg CEL (20 FPU/g biomass), GLU 10.4 mg/ml [68] DMSO, 50 mg/ml solid, 130 ºC 30 mg solid, 50 ºC, pH 4.8 XYL 5.1 mg/ml microwave irradiation, 50 W, 1 min

Organosolv Sitka spruce EtOH:water 60:40 (H2SO4, 1%), Celluclast 1.5 L (44 FPU/ml), 7.2 µl GLU 38.6 g/100 g biomass [69] S/L 1:10, 180 ºC, 60 min Novozyme 188, 200 mg solid, 45 ºC, Ethylglycosides 12.5 g/100 68 h g RM Empty palm EtOH:water 65:35 (2% H2SO4), Celluclast 1.5 L (15 FPU/g), GLU 88.3 g/100 g RM [70] fruit bunch S/L 1:8, 180 ºC, 60 min Novozym 188 (15 IU/g), 2% solid, 48 ºC, 48 h, pH 4.8 Sugarcane EtOH (30%):NaOH (3%) Celluclast 1.5 L (15 FPU/g), GLU 67.3 g/100 g RM [72] bagasse aqueous solution (S/L 7:1 w:w), Novozyme 188 (15 IU/g), 5% solid, XYL 6.1 g/100 g RM 195 ºC, 60 min 50 ºC, 24 h, pH 4.8 Wheat straw EtOH:water (50:50), 30 mM CEL (20 FPU/g), 3% solids, Digestibility of 86% of [73] H2SO4, 210 ºC, 60 min 50 ºC, 72 h, pH 4.8 wheat straw with a lignin yield of 84% RT: room temperatura, GLU: glucose, XYL: xylose, TS: total sugar, EtOH: ethanol, RM: raw material, S/L Solid:liquid ratio, CEL: Cellulase, HCEL: hemicellulase.

3705 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Acid hydrolysis may be performed in two ways: i) with The addition of an oxidizing agent, such as H2O2, can dilute acid (0.2–2.5%) and high temperatures (140–210 °C), increase lignin removal when using NaOH or Ca(OH)2 and or ii) with concentrated acid (30–70%) and low temperatures significantly inhibit the formation of furfural and HMF [39, (80–100 °C) for periods ranging from seconds to several 62]. Pretreatment of lignocellulosic waste from the produc- minutes [76]. In both cases, sulfuric acid is the most com- tion of furfural with 2% NaOH and 20.4% H2O2 produced monly used [77]. However, because of equipment corro- higher saccharification performance. Percentage of xylose sion and the high concentration of the fermentation inhibi- did not significantly increase; but, glucose yield was 9% tors, which are generated by the use of concentrated acid, higher [90]. pretreatment with dilute acid is preferred [18]. 3.2.3 Ionic liquids Furthermore, other acids that have been studied include HCl, HNO3 [78], H3PO4 [79, 80], and organic acids The use of ionic liquids is a trend that has been gaining [81, 82] (Table 1). Evaluation of acid pretreatment of attention as a pretreatment of lignocellulosic biomass [91]. wheat straw showed that high concentrations of maleic acid Ionic liquids are salts that are typically liquid at relatively had a negative effect on glucose yield, because the acid low temperatures and consist of large organic cations and promoted solubility of glucose in the aqueous phase, which small inorganic anions. In addition, these solvents have low is discarded. In this case, about 46 mM acid concentration viscosity, hydrophobicity, and vapor pressure; chemical and is sufficient to achieve yields of 85% glucose. In addition, thermal stability; and are non-flammable. These characteris- it was found that only 1% of these sugars degraded to tics vary depending on the nature of the anion and cation. HMF. Pretreatment with maleic acid showed similar yields Moreover, they are called "green" solvents, since they are to those obtained with sulfuric acid, with the advantage that neither toxic nor explosive [92]. Additionally, because of less sugar degradation occurs [82]. their low vapor pressure, recovery of more than 99% is possible [75]. In general, the main drawback of this pretreatment is the production of fermentation inhibitors, such as acetic The action mechanism of ionic liquids is based on the acid, furfural and HMF, which can be further transformed fact that they can dissolve the carbohydrates and lignin, to formic acid and levulinic acid if conditions are too ex- since the anionic part forms hydrogen bonds with the hy- treme; therefore, the obtained acid hydrolysates must be droxyl groups of cellulose in a 1:1 ratio, thus breaking the detoxified [83-85] which can increases the process cost. crystalline structure of the cellulose and making it more accessible to enzymatic hydrolysis [93]. Nevertheless, some 3.2.2 Alkaline hydrolysis studies revealed that the structures of the lignin and hemi- In alkaline hydrolysis, bases such as NaOH [85, 86], cellulose remain unchanged after treatment with ionic liq- Ca(OH) , KOH, and NH OH [87] are used in order to re- uids [94]. Da Costa et al. [95] developed a methodology for 2 4 the pretreatment of wheat stubble using 1-ethyl-3-me- move lignin, acetyl groups, and uronic acid, which are known to block the enzymatic reaction during cellulose hy- thylimidazolium acetate. The pretreatment performed at drolysis [36]. The saponification that takes place facilitates 120 °C for 6 h with 5% of biomass yielded a cellulose residue that was 12% higher than the pretreatment conducted the rupture of cross linkages between hemicellulose with lignin or other compounds [47].This bonding elimination at 120 °C for 2 h with 2% of biomass; however, pretreat- also leads to an increase of porosity and inner surface area; ment at 6 h with 6 % biomass was preferable for the hydrolysis of hemicellulose and lignin. moreover, it diminishes the cellulose crystallinity index and degree of polymerization. However, the cellulose sol- Typically, pretreatments with ionic liquids are carried ubility is less than is obtained with acid or hydrothermal out under temperatures between 90–130 °C and periods pretreatments. Additionally, in the case of hemicellulose, from 1–24 h [96] (Table 1). Afterward, the biomass is recov- the main solubilized fraction corresponds to xylan [87]. ered by precipitation through the addition of water or other, solvents such as ethanol, methanol, and acetone [97]. Cellu- NaOH is the most commonly used base for alkaline lose dissolves very well in ionic liquids that contain chloride pretreatment [77], although Ca(OH) has recently been 2 ions, formate, acetate, and alkylphosphonates [98]. considered important because it is safe as well as cost effective, as it is cheaper than NaOH [88]. On the other hand, Pretreatment of sugarcane bagasse that had previously the principal factors that affect the efficiency of the process been conditioned by grinding with bis(2-hydroxyethyl-am- are temperature, reaction time, and base concentration (Ta- monium) acetate for 4 h at 100 °C yielded 84% saccharifi- ble 1). Khuong et al. [89] found that higher yields of sac- cation [99]. charification were obtained with a greater concentration of Groff et al. [100] analyzed pretreatment of switchgrass NaOH during the pretreatment of sugarcane bagasse. With for 3 h with 1-butyl-3-methylimidazolium chloride in com- 0.2% NaOH, only 3% of the solids were saccharified; in bination with Amberlyst-15, which is a protic acid resin contrast, with a concentration of 5% NaOH, the soluble that catalyzes the hydrolysis of the ether bonds between solids increased up to 51%. Likewise, this pretreatment had glucose molecules, which are adjacent in cellulose. Their a greater effect on lignin and xylan fractions than on those results notably improved saccharification performance, of glucan; therefore, with 5% NaOH, the residual percentage which was 10 times higher than the use of only 1-butyl-3- of each fraction was 11%, 3.8%, and 81.6%, respectively. methylimidazolium. However, this technology is expensive

3706 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

and further development should focus on ionic liquids that then biomass is quickly depressurized [51, 74]. This pro- are available for their application at an industrial level cess combines mechanical forces and chemical effects. The [101]. Additionally, it has also been reported in some stud- mechanical effects are caused by quick depressurization; ies that certain solvents might irreversibly inhibit the activ- alternatively, the chemical effects occur by auto-hydrolysis ity of hydrolytic enzymes [102]. of hemicellulose acetyl groups, which facilitate the release of hemicellulose sugars and the transformation of lignin 3.2.4 Organosolv and cellulose exposure during enzymatic hydrolysis [129].

The organosolv process consists of breaking the inter- Some studies revealed that the use of H2SO4, CO2, or nal bonds of the lignocellulosic material using an organic SO2 improves biomass hydrolysis, which in turn increases solvent or its aqueous mixtures [103]. At a structural level, the yield of the hemicellulose sugars and decreases produc- the organosolv pretreatment allows hydrolysis of the internal tion of inhibitory compounds [51]. Corrales et al. [130] bonds of lignin, between lignin and hemicellulose, and the compared the effect of CO2 and SO2 with regard to the hemicellulose glycosidic bonds with cellulose [104, 105]. structure of sugarcane bagasse that was exposed to a vapor Usually, these pretreatments are conducted at temper- explosion pretreatment. Moreover, it was found by X-ray atures between 180 and 200 °C; however, the addition of diffraction that the index of untreated samples [index crys- an inorganic acid catalyzer, such as HCl, H2SO4, or H3PO4, tallinity (IC)] was 40.0%; on the other hand, samples that allows the pretreatment to work at temperatures less than were pretreated with SO2 and CO2 showed ICs of 65.5% 180 °C [70] (Table 1). Some studies have evaluated other and 56.4% because of the elimination of 68.3% and 40.5% neutral and basic catalysts to increase the yield in the orga- of hemicellulose, respectively. Likewise, SO2 showed nosolv process. In particular, a study conducted on pitch greater efficiency in the pretreatment of biomass because pine (Pinus rigida) compared the effects of H2SO4 (an acid the required energy and process time is less (190 °C for 5 catalyst), NaOH (a basic catalyst), and MgCl2 (a neutral min) compared with CO2 (205 °C for 15 min). catalyst). In optimal time and temperature conditions, it Like other treatments that require high temperatures was found that the maximum yield of saccharification was and acidic conditions, furfural and HMF form in steam ex- obtained with 1% MgCl2 (69.5%) followed by 1% H2SO4 plosion; however, their concentrations can be minimized if (61.5%). Nevertheless, the use of 1% NaOH did not signif- the pretreatment conditions are carefully controlled [131]. icantly improve the yield. However, when the concentra- Moreover, with this pretreatment, there is also risk of con- tion was increased to 2% NaOH, saccharification increased densation and precipitation of the soluble lignin compo- to 58.8% [106]. nents that result in a less digestible biomass and therefore On the other hand, organic acids, such as acetic and reduces their exploitation [18]. formic acids, can also be used. Pretreatment of wheat straw Working in birch wood, Vivekanand et al. [132] found with an solution of water, acetic and formic acid (15:55:30) an increase in Klason lignin when pretreatment severity in- at 105 °C over 3 h produced a composition of 48% cellu- creased with respect to time and temperature. This was lose, 27% hemicellulose, and 21% lignin [71]. probably due to the production of pseudo-lignin from the One of the main disadvantages of organosolv is that degradation products of polysaccharides, primarily furfural the solvent system needs to be completely removed, since or other aromatic compounds that can polymerize again this can inhibit enzymatic action and growth of microor- following a resulting rearrangement reaction, hence, be- ganisms in the fermentation process. For this reason as well coming insoluble and forming pseudo-lignin. as their low cost, methanol and ethanol are the solvents of choice. However, several others have also been used, in- 3.3.2 Ammonia fiber explosion (AFEX) cluding acetone, glycol, glycerol, aqueous phenol, and The AFEX process is a pretreatment where biomass is aqueous n-butanol [98]. subjected to high pressures and temperatures between 60– 100 °C in the presence of anhydrous liquid ammonia dur- 3.3 Physicochemical pretreatments ing various periods of time. Subsequently, a sudden pres- The most common physicochemical treatments are sure drop takes place, resulting in the formation of ammo- steam explosion and ammonia fiber explosion AFEX [107]; nia gas that breaks the fibers of the biomass and produces however, there have also been studies on the use of liquid a partial loss of the crystalline conformation of cellulose hot water [108]. Table 2 lists some studies that were con- [133]. During AFEX, the deacetylation of the hemicellu- ducted on lignocellulosic biomass using physicochemical lose and some specific regions of the lignin polymer also methods pretreatments. take place. Only a small amount of solid material is solubilized [120]. 3.3.1 Steam explosion Factors that exert greater effects in pretreatment are tem- In the pretreatment with steam explosion (previously perature, water load, and ammonia concentration, but pressure conditioned by mechanical methods), biomass is exposed and time should also be checked and controlled (Table 2). The to relatively high pressures (0.7 and 4.8 MPa) of saturated use of high temperatures promotes saccharification yield; vapor at temperatures of about 160–240 °C (Table 2). The however, excessive heat can result in the degradation of com- process can last from a few seconds to several minutes and pounds as well as furfural and HMF formation [134].

3707 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 2 - Physicochemical pretreatments applied to lignocellulosic biomass

Type Waste Pretreatment conditions Enzymatic treatment Product (Yield) Reference Pretreatment

Steam explosion Mandarin peel Hot popping: 150°C, Pectinase (5 mg/kg), xylanase GLU 45.1 g/100 g RM; Fructose [109] pressure: 15 kgf/cm (5mg/kg), β-glucosidase (2 mg 18.4 g/100 g RM /kg), 45 ºC, 24 h, pH 4.8 Oat straw Chopped oat straw Accellerase 1000 (2707 CMC/g), GLU 157 g/kg solid [110] (3-5 mm), 190 ºC, 10 min 80 g solid, 55 ºC, 72 h, pH 4.5 XYL 76 g/kg solid Wheat straw Milled wheat straw Celluclast 1.5 L, Novozym 188 GLU 21.6 g/L [111] (10 mm), 210 ºC, 10 min (20 FPU/g), 50 g/L solid, 50 ºC, pH 5.0 Corn stover Steam chamber, 2.5 MPa, Not applied GLU 3.3 g/L [112] 200 XYL 6.46 g/L Corn stover 100 g solid (2 cm), steam Accellerase 1500 (60 FPU/g GLU 33.4 g/100 g glucan [113] 200 ºC, 5 min glucan), Novozyme 188 (64 XYL 3.17 g/100g xylan pNPGU/g glucan), 6% solid, 50 ºC, 72 h, pH 4.8 Roasted coffee Hot popping: 150°C, pres- CEL (18.3 mg/g), 37 ⁰C, 96 h, FS 40.2 g/L [114] beans sure: 1.47 MPa. pH 5 Rice straw 1.4 MPa, 4 min, Ratio CEL (20 FUP/g), 50 ºC, 36 h, GLU 600 mg/g RM [115] solid:aqueous NH4Cl pH 4.8 (90 g/l) 1:1, RT, overnight Rice straw S/L: 40:60, 2 MPa, 2 min Celluclast 1.5 L (15 FPU/g), GLU 30.08 mg/g RM [116] Novozym 188 (15 IU/g), 0.5 g XYL 55.35 mg/g RM solid, 50 ºC, 48 h, pH 4.5 Corn stalk 200 g rm, S/L 1:1, CEL produced by Trichoderma GLU 150 g/kg RM [117] 1.7 MPa,1 min reesei YG3 (50 FPU/cm), 50 ºC, 48 h, pH 4.8

Ammonia Fiber Sweet sorghum Ratio 2:1 ammonia:bio- CEL: 31.3 mg/g glucan, β-gluco- GLU 25g/100g glucan [118] Explosion bagasse mass loading (w/w), sidase: 17.1 mg/g glucan and xy- XYL 12.7g/100g xylan 140°C, 30 min lanase: 15.6 mg/g, 72 h Corn stover Ratio 1:1 ammonia:bio- Accellerase 1000 (6 mg/g glucan), GLU 0.54 g/L [119] mass loading (w/w), Multifect xylanase (3 mg/g glu- XYL 2.04 g/L 130 ºC,15 min can), 24 h, 50 ºC, pH 4.8 Rice straw Ratio 2:1 ammonia:bio- Spezyme (15 FPU/g glucan), GLU 7.9 g/L [120] mass loading (w/w), Novozyme (64 pNPGU/g glucan), XYL 4.3 g/L 100 ºC, 30 min 6% glucan, 50 ºC, 168 h, pH 4.8 Switchgrass Ratio: 1.5:1 ammonia:bio- Novozyme (735 CBU/g), Spezyme GLU 69.3 g/100 g glucan [121] mass loading (w/w), (15 FPU/g), Multifect Xylanase XYL 67.7 g/100 g xylan 140 ºC, 20 min (27 mg protein/ml), 1% glucan, 50 ºC, 72 h, pH 4.8

Liquid hot water Sugarcane with high pressure CO2: Cellulose (Celluclast 1.5 L), β- Glcose 30.43g/L [122] bagasse S/L: 1:12, 115 ºC, 60 min glucosidase (Novozym 188), solid XYL 9.82g/L 10% (w/v), 50 ºC, 72 h, pH 4.8 Corn stover 100 g solid, 230 ºC, CEL (170 FPU/g), hemicellulose GLU 376 mg/100 g RM; [123] 24 min, 4.3% NaOH (11.3 IU/g), 1 g solid, 50 ºC, 120 h, XYL 160 mg/100 g RM pH 4.8 Sugarcane S/L 1:3, 70 ºC, 60 min Not applied XYL 41.28 mg/100 g RM [124] bagasse

Supercritical Sugarcane 45 g RM, 300 ml water, CEL (20 FPU/ g solid), 2% solid, GLU 29.17 g/100 g RM; [125] CO2 explosion bagasse 160 ºC, 60 min, 7 MPa 50 ºC, 60 h, pH 4.8, XYL 3.76 g/100 g RM

Wheat straw 200 ºC , 12 MPa, 60 min CEL (25 FPU/g wheat bran), GLU 234.6 g/kg RM [126] 50 °C, 72 h, pH 4.8. Corn stover 5 g RM, 165 ºC , 1 g CEL (0.95 g TS 37 g/ 100 g RM [127] 34.5 MPa, 60 min NS50013 & 0.05 g NS50010), 1 g solid, 50 ºC, 48 h, pH 4.8 Sugarcane ba- 20 g RM, 80 ºC, 25 MPa, Cellulolitic complex from Tricho- FS380 g/kg RM [128] gasse 120 min (ultrasonic bath: derma reesei (NS50013), 1 g solid, 154 W) 50 ºC, 8 h, 4.5 Note: GLU: glucose, XYL: xylose, TS: total sugar, RM: raw material, S/L: Solid:liquid ratio, CEL: Cellulase, HCEL: hemicellulose, RT: room temperature.

3708 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

According to Yoo et al. [135] bound water molecules could not to be quantified in the first 10 min [142]. More- in the biomass react with ammonium ions and create hy- over, acetic acid, which is generated by ionization of the drogen bonds with cellulose, which effectively results in hemicellulose acetyl groups, can also inhibit the fermenta- the swelling of the cellulose structure and a consequent loss tion process. However, it was found that certain concentra- of its crystallinity. This also results in increased enzyme tions of acetic acid can act as catalysts during saccharifica- accessibility. On the other hand, ammonia concentration tion, since its hydronium ions sufficiently decrease the pH strongly influences the obtained sugar yield. However, the of the medium and allow the hydrolysis of polymers [122]. ammonia load must be approximately 2.0. A study exam- Hongdan et al. [140] showed that pretreatment at 220 °C ined the impact of the different conditions of AFEX pre- for 20 min yielded a concentration of 2.25 g/L of acetic treatment on the ultrastructure of the corn stubble’s cellular acid and showed no negative effect on the production of wall and found that the average structure of the blades, ethanol. Generally, the pH must be between 4 and 7 to fa- which are rich in lignin, was significantly affected with vor the hemicellulose to be in its oligomeric form and avoid loads of 3.0 ammonia. Additionally, the formation of large the formation of inhibitors from monomers [78]. pores and the deposition of hemicellulose were also observed [136]. 3.3.4 Supercritical CO2 explosion This pretreatment uses CO in a supercritical state, Previous or simultaneous treatment of biomass with 2 which removes the lignin, increasing hemicellulose and H O as an oxidizing agent significantly improves the sac- 2 2 cellulose digestibility. In aqueous solution, the CO forms charification yield after treatment with AFEX. Samples of 2 carbonic acid, which promotes the hydrolysis of polymers corn stover were previously conditioned with a load of 0.5 [31]. The CO molecules are comparable in size to water H O at 30%, which then underwent pretreatment for sac- 2 2 2 molecules and under the high pressure can penetrate the charification at 130 °C for 10 min with a load of 0.7 water pores of lignin. Subsequently, a rapid drop of pressure oc- and 1.0 ammonia. It was observed that the H O had greater 2 2 curs, which generates the breakdown of the cellulose and effect on the polysaccharide structure in lignin with hydrol- hemicellulose structures and in turn increases their suscep- ysis yields of 88.08% and 90.64% for glucan and xylan, tibility to enzymatic hydrolysis [127]. Supercritical CO respectively [137]. This occurs because H O generates 2 2 2 explosion has been performed at temperatures of 80-200 ºC radical free reagents, including the hydroxyl radical (OH-), - - and pressures of 7-34.5 MPa for 60 minutes (Table 2). perhydroxyl anion (OOH ), and superoxide anion (O2 ), which oxidizes and degrades lignocellulose. At the same Glucose yields obtained in corn stover pretreatment time, the crystalline structure of cellulose breaks, which in- with CO2 were higher as the temperature increased, alt- creases its surface area and therefore improves the enzy- hough such increments were not directly proportional. A matic hydrolysis [138]. low temperature (80 °C) has less impact on the saccharification yield because there is less diffusivity of supercritical 3.3.3 Liquid hot water CO2 in the biomass structure. The same study showed, by X- ray diffraction, that there was no change in the cellulose frac- In pretreatment of lignocellulosic biomass with liquid hot water high pressure is used to keep water in a liquid tion crystallinity of the biomass treated with supercritical state at temperatures of 160-240 °C and pressures above 5 CO2 compared with the untreated biomass, while scanning electron microscopy images revealed an increase in surface MPa for 20 min [34], with the aim of hydrolyzing hemicellulose and making the cellulose more accessible to enzy- area using a pressure of 3500 psi at 150 °C for 1 h [143]. matic action [139, 140]. Hemicellulose is the main polymer 3.4 Biological pretreatments that is hydrolyzed during this process, for which the major yield is xylose [141]. Lignocellulosic biomass degradation can also be conducted using microorganisms. Currently, fungi have been Time, heat, and pressure supplied to the process are studied the most, and the main fungi include brown rot, critical factors, because insufficient saccharification yields white rot, and soft rot fungi [144, 145]. Brown rot is the occur at very low temperatures during short periods of time most effective for cellulose hydrolysis, while white and [142] (Table 2). In contrast, very high temperatures for ex- soft rot are involved in both cellulose and lignin hydrolysis tended periods of time favors the formation of inhibitors, [146-148]. It has been observed that the white fungi with such as furfural and HMF, which are generated from the the highest efficiency of delignification are Phanerochaete degradation of sugars, including xylose and glucose but chrysosporium, Ceriporia lacerata, Cyathus stercoerus, production of these inhibitory compounds can be mini- Ceriporiopsis subvermispora, Pycnoporus cinnabarinus, mized by maintaining the pH between 4 and 7 [139]. and Pleurotus ostreatus [146]. Additionally, in a study of One study showed that the yield of saccharification de- sugarcane bagasse that was exploited as substrate to obtain creased from 86.16 to 62.60% by increasing the tempera- fermentable sugars, it was found that Pleurotus florida in- ture in the pretreatment of sugarcane bagasse from 160 to creased the delignification of the substrate by 7.91% com- 200 °C [140]. On the other hand, during the pretreatment pared with samples that were only treated with H2SO4 of eucalyptus kindling at 180 °C, it was observed that the (0.5%, w/v) and NaOH (0.5%, w/v); similarly, Ganoderma furfural concentration increased from 0.42 to 2.05 g/L by sp. rckk-02 and Coriolopsis caperata RCK 2011 showed extending the process from 10 to 40 min; however, HMF 5.58 and 5.48% increments, respectively [149].

3709 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Despite the fact that biological pretreatments have the choice of a suitable method or selection of the most con- advantage of being cost effective, require little energy, and venient operation conditions must take into account mainly do not use chemical reagents, they require a large amount the nature and composition of the lignocellulosic material of space, long periods of time, and constant monitoring of [153]. However, when more than one type of pretreatment the growth of the microorganism [150, 151]. can be used for a specific type of biomass, it would be useful to consider the advantages and disadvantages offered 3.5 Advantages and disadvantages of the pretreatment tech- by the reported pretreatment methods (Table 3). Correct nologies for saccharification of lignocellulosic material. implementation of the pretreatment technology used for Nowadays, there is an intensive research on the devel- saccharification of food biomass waste is crucial in the pro- opment of technologies to take advantage of agro-food cess of biorefinery, because the efficiency of selected pre- waste for biofuels or biomaterial production, due to envi- treatment will significantly define the profit of the process ronmental problems associated to their disposal and the ne- and therefore the yield of fermentable sugars for biofuels cessity to have alternatives for use of fossil fuels. Cur- and biomaterials production. Intensive research has been rently, it has been developed a variety of technologies to dedicated mainly on the use of waste from commodities, optimize the yield of saccharification products, so that the such as sugar cane, maize, rice, and wheat as feedstock, but

TABLE 3 - Advantages and disadvantage of pretreatments used for saccharification of lignocellulosic material

Pretreatment Advantage Disadvantage method

Physical No waste production or fermentation inhibitor compounds Requires a chemical, physico-chemical or biological pretreat- [26] ment complementary

Acid hydrolysis High rate glucan conversion to glucose (at least 90%) [26] Corrosive, production of fermentation inhibitor compounds Higher performance saccharification that alkaline hydrolysis [26] Easily scalable to industrial level [36] High energy requirement of the process that increases cost [36] Alkaline hydrolysis Requires lower temperatures than other pretreatments [26, 78, Production of toxic effluents [26] 152] Requires longer reaction times (from hours to days) and neutral- Reagents used are inexpensive ization of slurry is required [154] Easily scalable to industrial level [153] Ionic liquids Can dissolve efficiently lignin and carbohydrate simultane- High cost of ionic liquids ously [155] Ionic liquids are incompatible with cellulase leading to the in- Ionic liquids can be easily reused activation requires effective recuperation [156] Non-flamability [11] Organosolv Easy recovery of lignin [157] High cost process Requires a thorough solvent recovery (fermentation inhibitor) [30] Steam explosion Easily scalable to industrial level [11] Incomplete disruption of the lignin–carbohydrate matrix [34] Nor chemical is required Acidic conditions favor formation of fermentation inhibitors Environmentally attractive [34] compounds Requires up to 70% less energy than mechanical methods to Requires proper equipment to work with high pressures obtain a similar particle size [36] Less effective for softwoods [36] Low cost [38]

AFEX Small particle size is not required for efficacy [78] Ineffective in biomass with high lignin content Short times treatment [38] Requires efficient ammonia recovery to be economical due to Unlike steam explosion produced no slurry, but solid material the high cost of ammonia [11, 38] [11] Very low solubilizing hemicellulose [11] Liquid hot water Low or no inhibitors production [11] No chemicals or catalysts are required No washing or neutralization is required [158] Easily scalable to industrial level Not require particle size reduction of the biomass [159] Supercritical CO2 Easy removal of CO2 by depressurization [11] Yields of fermentable sugars are lower compared to steam ex- explosion Requires less energy than AFEX plosion and AFEX [31] More cost-effective than AFEX [31] Requires infrastructure to operate at high pressures [11, 36] No waste products and does not requires further recovery Saccharification yields are higher compared to steam explosion method Not cause the formation of inhibitors as in steam explosion [34, 36] Biological Not requires chemicals Long times treatment [26] Decreases both lignin and cellulose [34] Loss of sugars (used as substrate by microorganisms) Low energy requirements Requires infrastructure with controlled conditions [36] Low waste production [36]

3710 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

local crops which produce high quantities of waste have [4] Yepes, S.M., Montoya, N.L.J. and Orozco, S.F. (2008) been poorly studied [160-162]. It is necessary to develop Agroindustrial waste valorization fruits in Medellin and the south of Valle de Aburra, Colombia. Revista Facultad technologies to use non-food crops and biomass waste in Nacional de Agronomía, Medellín. 61(1): 4422-4431. large scale to obtain fermentable sugars for commercial [5] Cherubini, F. (2010) The biorefinery concept: using biomass production of biomaterials or biofuels. Particularly, un- instead of oil for producing energy and chemicals. Energy complicated chemical and physicochemical treatments Conversion and Management. 51(7): 1412-1421. with reduced cost and high conversion efficiencies can be [6] International Energy Agency. Technology Roadmap-Bioen- profitable within the biorefinery concept. ergy for Heat and Power (2012) Accessed: Jan. 14, (2014), available in: http://www.iea.org/publications/freepublica- tions/publication/2012_Bioenergy_Roadmap_2nd_Edi- tion_WEB.pdf

[7] Galanakis, C.M. (2011) Recovery of high added-value com- 4. CONCLUSIONS ponents from food wastes: Conventional, emerging technologies and commercialized applications. Trends in Food Science Food processing by-products and wastes are a source and Technology. 26: 68-87. of fermentable sugars that can be used by specific micro- [8] Dyk, J.S., Gama, R., Morrison, D., Swart, S. and Pletschke, organisms to produce value-added products. The amount B.I. (2013) Food processing waste: Problems, current management and prospects for utilization of the lignocellulose com- and kind of sugars produced from this waste depends on ponent through enzyme synergistic degradation. Renewable the residue source and pretreatment process. In contrast Sustainable Energy Reviews. 26: 521–531. with forestry waste, food by-products contain additional [9] Mirabella, N., Castellani, V. and Sala, S. (2014) Current op- components, such as essential oils, alkaloids, or polyphe- tions for the valorization of food manufacturing waste: a re- nolic compounds that can interfere with the fermentation view. Journal of Cleaner Production. 65: 28-41. process. In this case, it is necessary to carry out previous [10] FitzPatrick, M., Champagne, P., Cunningham, M.F. and Whit- extraction of the bioactive compounds before the pretreat- ney, R.A. (2010) A biorefinery processing perspective: Treat- ment process that is used to obtain hydrolysates. Thus, food ment of lignocellulosic materials for the production of value- added products. Bioresource Technology. 101: 8915-8922. waste can be turned into value-added compounds by a combination of extraction and hydrolysis technologies, ei- [11] Haghighi-Mood, S., Golfeshan, A.H., Tabatabaei, M., Salehi- Jouzani , G., Najafi, G.H., Gholami, M. et al. (2013) Lignocel- ther by using physicochemical or enzymatic treatments. lulosic biomass to bioethanol, a comprehensive review with a The presence of enzyme and fermentation inhibitors of the focus on pretreatment. Renewable Sustainable Energy Re- obtained fermentable liquor depends on the pretreatment views. 27: 77-93. used, but selection of resistant microbial strains could pre- [12] Sawatdeenarunat, C., Surendra, K.C., Takara, D., Oechsner, vent the need for additional clean-up steps prior to the fer- H. and Khanal, S.K. (2015). Anaerobic digestion of lignocellulosic biomass: Challenges and opportunities. Bioresource mentation processes. Technology. 178: 178-186.

[13] Hu, F. and Ragauskas, A. (2012) Pretreatment and lignocellulosic chemistry. Bioenergy Research. 5: 1043-1063.

[14] Li, C., Knierim, B., Manisseri, C., Arora, R., Scheller, H.V., ACKNOWLEDGMENTS Auer, M., et al. (2010) Comparison of dilute acid and ionic liquid pretreatment of switchgrass: Biomass recalcitrance, del- The authors express their gratitude to Consejo Nacional ignification and enzymatic saccharification. Bioresource de Ciencia y Tecnología (CONACyT Project CB-2011-01- Technology. 101: 4900-4906. 169779) and Instituto Politécnico Nacional (Project SIP [15] Limayem, A. and Ricke, S. (2012) Lignocellulosic biomass for bioethanol production: Current perspectives, potential issues 20151298) for the financial support. P.Gutiérrez-Macías re- and future prospects. Progress in Energy and Combustion Sci- ceived graduate scholarship from the CONACyT. ence. 38: 449-467. [16] Zhou, C.H., Xia, X., Lin, C.H., Tong, D.S. and Beltramini. The authors have declared no conflict of interest. (2011) Catalytic conversion of lignocellulosic biomass to fine chemicals and fuels. Chemical Society Reviews. 40: 5588- 5617.

[17] Himmel, M.E., Ding, S.Y., Johnson, D.K., Adney, W.S., Nim- los, M.R., Brady, J.W. and Foust, T.D. (2007) Biomass recal- REFERENCES citrance: engineering plants and enzymes for biofuels production. Science. 315(5813): 804-807. [1] Laufenberg, G., Kunz, B. and Nystroem, M. (2003) Transfor- [18] Hendriks, A.T.W.M. and Zeeman, G. (2009) Pretreatments to mation of vegetable waste into value added products: (A) the enhance the digestibility of lignocellulosic biomass. Biore- upgrading concept; (B) practical implementations. Biore- source Technology. 100: 10-18. source Technology. 87: 167-198. [19] Pandey, M.O. and Kim, C.S. (2011) Lignin depolymerization [2] Tuck, C.O., Pérez, E., Horváth, I.T., Sheldon, R. A. and Polia- and conversion: a review of thermochemical methods. Chem- koff, M. (2012) Valorization of biomass: deriving more value ical Engineering and Technology. 34: 29-41. from waste. Science. 337(6095): 695-699. [20] Sun, Y.S., Lu, X.B., Zhang, S.T., Zhang, R. and Wang, X.Y. [3] Saval, S. (2012) Aprovechamiento de residuos agroindus- (2011) Kinetic study for Fe(NO3)3 catalyzed hemicellulose hy- triales: Pasado, presente y futuro. Revista de la Sociedad Mex- drolysis of different corn stover silages. Bioresource Technol- icana de Biotecnología y Bioingeniería, AC. 14-46. ogy. 102: 2936-2942.

3711 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[21] Sahraoui, N., Vian, M.A., El Maataoui, M., Boutekedjiret, C. [40] Iskalieva, A., Yimmou, B.M., Gogate, P.R., Horvath, M., and Chemat, F. (2011). Valorization of citrus by-products us- Horvath, P.G. and Csoka, L. (2012) Cavitation assisted delig- ing microwave steam distillation (MSD). Innovative Food Sci- nification of wheat straw: A review. Ultrasonics Sonochemis- ence and Emerging Technologies. 12(2): 163-170. try. 19: 984-993. [22] Arapoglou, D., Varzakas, T., Vlyssides, A. and Israilides, C. [41] Rehman, M.S., Kim, I., Chisti, Y. and Han, J.I. (2013) Use of (2010) Ethanol production from potato peel waste ultrasound in the production of bioethanol from lignocellulosic (PPW). Waste Management. 30(10): 1898-1902. biomass. Energy Education Science and Technology Part A: [23] Parmar, I. and Rupasinghe, H.P. (2012) Optimization of dilute Energy Science and Research. 30: 1391-1410. acid-based pretreatment and application of laccase on apple [42] Lu, X., Xi, B., Zhang, Y. and Angelidaki, I. (2011) Microwave pomace. Bioresource Technology. 124: 433-439. pretreatment of rape straw for bioethanol production: Focus on [24] Kratky, L. and Jirout, T. (2011) Biomass size reduction ma- energy efficiency. Bioresource Technology. 102: 7937-7940. chines for enhancing biogas production. Chemical Engineer- ing and Technology. 3: 391-399. [43] Pang, F., Xue, S., Yu, S. Zhang, C., Li, B. and Kang, Y. (2012) Effects of microwave power and microwave irradiation time [25] Zhang, Y.H., Berson, E., Sarkanen, S. and Dale, B.E. (2009) on pretreatment efficiency and characteristics of corn stover Pretreatment and biomass recalcitrance: fundamentals and pro- using combination of steam explosion and microwave irradia- gress. Applied Biochemistry and Biotechnology. 153: 80-83. tion (SE–MI) pretreatment. Bioresource Technology. 118: [26] Barakat, A., de Vries, H. and Rouau, X. (2013) Dry fractiona- 111-119. tion process as an important step in current and future lignocellulose biorefineries: a review. Bioresource Technology. [44] Binod, P., Satyanagalakshmi, K., Sindhu, R., Janu, K.U., 134: 362-373. Sukumaran, R.K. and Pandey, A. (2012) Short duration microwave assisted pretreatment enhances the enzymatic sacchari- [27] Balat, M., Balat, H. and Oz, C. (2008) Progressin bioethanol fication and fermentable sugar yield from sugarcane bagasse. processing. Progression Crystal Growth and Characterization Renewable Energy. 37: 109-116. of Materials. 34: 551-73. [45] Yoon, M., Choi, J.I., Lee, J.W. and Park, D.H. (2012) Im- [28] Li, M.F., Fan, Y.M., Xu, F., Sun, R.C. and Zhang, X.L. (2010) provement of saccharification process for bioethanol produc- Cold sodium hydroxide/urea based pretreatment of bamboo tion from Undaria sp. by gamma irradiation. Radiation Phys- for bioethanol production: characterization of the cellulose ics and Chemistry. 81: 999-1002. rich fraction. Industrial Crops and Products. 32: 551-559. [29] Chang, V.S. and Holtzapple, M. (2000) Fundamentals factors [46] Hong, S.H., Lee, J.T., Lee, S., Wi, S.G., Cho, E.J., Singh, S. et affecting biomass reactivity. Applied Biochemistry and Bio- al. (2014) Improved enzymatic hydrolysis of wheat straw by technology. 5(37): 284-286. combined use of gamma ray and dilute acid for bioethanol production. Radiation Physics and Chemistry. 94: 231-235. [30] Kumar, P., Barrett, D.M., Delwiche, M.J. and Stroeve, P. (2009) Methods for pretreatment of lignocellulosic biomass [47] Sun, Y. and Cheng, J. (2002) Hydrolysis of lignocellulosic ma- for efficient hydrolysis and biofuel production. Industrial and terials for ethanol production: a review. Bioresource Technol- Engineering Chemistry Research. 48: 3713-3729. ogy. 83: 1-11. [31] Alvira, P., Tomás, P.E., Ballesteros, M. and Negro, M.J. [48] Niemi, P., Faulds, C., Sibakov, J., Holopainen, U., Poutanen, (2010) Pretreatment technologies for an efficient bioethanol K. and Buchert, J. (2012) Effect of a milling pre-treatment on production process based on enzymatic hydrolysis: A review. the enzymatic hydrolysis of carbohydrates in brewer’s spent Bioresource Technology. 101: 4851-4861. grain. Bioresource Technology. 116: 155-160. [32] Demirbas, A. (2012) Green Energy and Technology-Biorefin- [49] Bruni, E., Jensen, A.P. and Angelidaki, I. (2010) Comparative eries for Biomass Upgrading Facilities. British Library, study of mechanical, hydrothermal, chemical and enzymatic Springer: London. treatments of digested biofibers to improve biogas production. [33] Chandra, R.P., Bura, R., Mabee, W.E., Berlin, A., Pan, X. and Bioresource Technology. 101: 8713-8717. Saddler, J.N. (2007) Substrate pretreatment: the key to effec- [50] Yu, J., Zhang, J., He, J., Liu, Z. and Yu, Z. (2009) Combina- tive enzymatic hydrolysis of lignocellulosics?. Advances in tions of mild physical or chemical pretreatment with biological Biochemical Engineering/Biotechnology. 108: 67-93. pretreatment for enzymatic hydrolysis of rice hull. Biore- [34] Sarkar, N., Kumar, G.S., Bannerjee, S. and Aikat, K. (2012) source Technology. 100: 903-908. Bioethanol production from agricultural wastes: An overview. Renewable Energy. 37: 19-27. [51] Mtui, G.Y.S. (2009) Recent advances in pretreatment of lignocellulosic wastes and production of value added products. [35] Yang, B. and Wyman, C.E. (2008) Pretreatment: the key to African Journal of Biotechnology. 8: 1398-1415. unlocking low-cost cellulosic ethanol. Biofuels, Bioproducts and Biorefining. 2: 26-40. [52] de-Moraes-Rocha, G.J., Martin, C., Barbosa, S.I., Souto, M.A.M., Macedo, B.H., Moraes, A.C.A. (2011) Dilute mixed- [36] Behera, S., Arora, R., Nandhagopal, N. and Kumar, S. (2014) acid pretreatment of sugarcane bagasse for ethanol production. Importance of chemical pretreatment for bioconversion of lig- Biomass and Bioenergy. 35: 663-670. nocellulosic biomass. Renewable Sustainable Energy Re- views. 36: 91-106. [53] Dagnino, E.P., Chamorro, E.R., Romano, S.D., Felissia, F.E. [37] Tedesco, M., Barroso, T.M. and Olabi, A.G. (2014) Optimiza- and Area, M.C. (2013) Optimization of the acid pretreatment tion of mechanical pre-treatment of Laminariaceae spp. bio- of rice hulls to obtain fermentable sugars for bioethanol pro- mass-derived biogas. Renewable Energy. 62: 527-534. duction. Industrial Crops and Products. 42: 363-368. [38] Talebnia, F., Karakashev, D. and Angelidaki, I. (2010) Pro- [54] Toquero, C. and Bolado, S. (2014) Effect of four pretreatments duction of bioethanol from wheat straw: an overview on pre- on enzymatic hydrolysis and ethanol fermentation of wheat treatment, hydrolysis and fermentation. Bioresource Technol- straw. Influence of inhibitors and washing. Bioresource Tech- ogy. 101: 4744-4753. nology. 157: 68-76. [39] Taherzadeh, M.J. and Karimi, K. (2008) Pretreatment of ligno- [55] Ponte, R.M.V., Soares, R., Lima, A.T., Gonçalves, R.L. and cellulosic wastes to improve ethanol and biogas production: a de-Macedo, G.R. (2014) Evaluation of dilute acid pretreat- review. International Journal of Molecular Science. 9: 1621- ment on cashew apple bagasse for ethanol and xylitol produc- 1651. tion. Chemical Engineering Journal. 243: 234-243.

3712 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[56] Avci, A., Saha, B.C., Dien, B., Kennedy, G.J. and Cotta, M.A. [73] Wildschut, J., Smit, A.T., Reith, J.H. and Huijgen, W.J.J. (2013) Response surface optimization of corn stover pretreat- (2013) Ethanol-based organosolv fractionation of wheat straw ment using dilute phosphoric acid for enzymatic hydrolysis and for the production of lignin and enzymatically digestible cel- ethanol production. Bioresource Technology. 130: 603-612. lulose. Bioresource Technology. 135: 58-66. [57] Ranjan, A. and Moholkar, V.S. (2013) Comparative study of [74] Agbor, V.B., Cicek, N., Sparling, R., Berlin, A. and Levin, various pretreatment techniques for rice straw saccharification D.B. (2011) Biomass pretreatment: Fundamentals toward ap- for the production of alcoholic biofuels. Fuel. 112: 567-571. plication. Biotechnology Advances. 29: 675-685. [58] Velmurugan, R. and Muthukumar, K. (2011) Utilization of [75] Brodeur, G., Yau, E., Badal, K., Collier, J., Ramachandran, sugarcane bagasse for bioethanol production: Sono-assisted K.B. and Ramakrishnan, S. (2011) Chemical and physico- acid hydrolysis approach. Bioresource Technology. 102: chemical pretreatment of lignocellulosic biomass: A review. 7119-7123. DOI: http://dx.doi.org/10.4061/2011/787532. [59] Chen,C., Boldor, D., Aita, G. and Walker, M. (2012) Ethanol [76] Saha, B.C., Iten, L.B., Cotta, M.A. and Wu, Y.V. (2005) Di- production from sorghum by a microwave-assisted dilute am- lute acid pretreatment, enzymatic saccharification and fermen- monia pretreatment. Bioresource Technology. 110: 190-197. tation of wheat straw to ethanol. Process Biochemistry. 40: 3693-3700. [60] Moradi, F., Amiri, H., Soleimanian-Zad, S., Ehsani, M.R. and [77] Zheng, Y., Zhao, J., Xu, F. and Li, Y. (2014) Pretreatment of Karimi, K. (2013) Improvement of acetone, butanol and etha- lignocellulosic biomass for enhanced biogas production. Pro- nol production from rice straw by acid and alkaline pretreatments. Fuel. 112: 8-13. gress in Energy and Combustion Science. 42: 35-53. [78] Mosier, N., Wyman, C., Dale, B., Elander, R., Lee, Y.Y., Holt- [61] McIntosh, S. and Vancov, T. (2010) Enhanced enzyme sac- azapple, M. and Ladisch, M. (2005) Features of promising charification of Sorghum bicolor straw using dilute alkali pre- technologies for pretreatment of lignocellulosic biomass. Bio- treatment. Bioresource Technology. 101: 6718-6727. resource Technology. 96: 673-686. [62] Karagöz, P., Rocha, I.V., Özkan, M. and Angelidaki, I. (2012) [79] Gámez, S., González, C.J.J., Ramírez, J.A., Garrote, G. and Alkaline peroxide pretreatment of rapeseed straw for enhanc- Vázquez, M. (2006) Study of the hydrolysis of sugar cane ba- ing bioethanol production by same vessel saccharification and gasse using phosphoric acid. Journal of Food Engineering. 74: co-fermentation. Bioresource Technology. 104: 349-357. 78-88. [63] Tye, Y.Y., Lee, K.T., Abdullah, W.N.W. and Leh, C.P. (2012) [80] Vázquez, M., Oliva, M., Téllez, L.J. and Ramírez, J.A. (2007) Potential of Ceiba pentandra (L.) Gaertn. (kapok fiber) as a Hydrolysis of sorghum straw using phosphoric acid: Evalua- resource for second generation bioethanol: Effect of various tion of furfural production. Bioresource Technology. 98: simple pretreatment methods on sugar production. Biore- 3053-3060. source Technology. 116: 536-539. [81] Kootstra, A.M.J., Beeftink, H.H., Scott, E.L. and Sanders, [64] Wan, C., Zhou, Y. and Li, Y. (2011) Liquid hot water and al- J.P.M. (2009a) Comparison of dilute mineral and organic acid kaline pretreatment of soybean straw for improving cellulose pretreatment for enzymatic hydrolysis of wheat straw. Bio- digestibility. Bioresource Technology. 102: 6254-6259. chemical Engineering Journal. 46: 126-131. [65] Poornejad, N., Karimi, K. and Behzad, T. (2014) Ionic liquid [82] Kootstra, A.M.J., Beeftink, H.H., Scott, E.L. and Sanders, pretreatment of rice straw to enhance saccharification and bi- J.P.M. (2009b) Optimization of the dilute maleic acid pretreat- oethanol production. Journal of Biomass to Biofuel. 1: 8-15. ment of wheat straw. Biotechnology for Biofuels. 2(31): 1-14 [66] Barr, C.J., Hanson, B.L., Click, K., Perrotta, G. and Schall, [83] Liu, Z.L. (2011) Molecular mechanisms of yeast tolerance and C.A. (2014) Influence of ionic-liquid incubation temperature in situ detoxification of lignocellulose hydrolysates. Applied on changes in cellulose structure, biomass composition, and Microbiology and Biotechnology. 90: 809-825. enzymatic digestibility. Cellulose. 21: 973-982. [84] Alriksson, B., Cavka, A. and Jönsson, L.J. (2010) Improving [67] Fu, D. and Mazza, G. (2011) Aqueous ionic liquid pretreat- the fermentability of enzymatic hydrolysates of lignocellulose ment of straw. Bioresource Technology. 102: 7008-7011. through chemical in-situ detoxification with reducing agents. Bioresource Technology. 102: 1254-1263. [68] Mai, N.L., Ha, S.H. and Koo, Y.M. (2014) Efficient pretreatment of lignocellulose in ionic liquids/co-solvent for enzy- [85] Wang, Z., Keshwani, D.R., Redding, A.P. and Cheng, J.J. matic hydrolysis enhancement into fermentable sugars. Pro- (2010) Sodium hydroxide pretreatment and enzymatic hydrol- cess Biochemistry. 49: 1144-1151. ysis of coastal Bermuda grass. Bioresource Technology. 101: 3583-3585. [69] Bouxin, F.P., Jackson, S.D. and Jarvis, M.C. (2014) Organo- solv pretreatment of Sitka spruce wood: Conversion of hemi- [86] Silverstein, R.A., Chen, Y. and Sharma, S.R.R. (2007) A com- celluloses to ethyl glycosides. Bioresource Technology. 151: parison of chemical pretreatment methods for improving sac- 441-444. charification of cotton stalks. Bioresource Technology. 98: 3000-3011. [70] Goh, CS, Tan, HT, Lee, KT and Brosse, N. (2011) Evaluation [87] Chaudhary, G., Singh, L.K. and Ghosh, S. (2012) Alkaline and optimization of organosolv pretreatment using combined pretreatment methods followed by acid hydrolysis of Sac- severity factors and response surface methodology. Biomass charum spontaneum for bioethanol production. Bioresource Bioenergy.; 35: 4025-4033. Technology. 124: 111-118. [71] Snelders, J., Dornez, E., Benjelloun, M.B., Huijgen, W.J.J., [88] Reilly, M., Dinsdale, R. and Guwy, A. (2015). Enhanced biode-Wild, P.J., Gosselink, R.J.A., Gerritsma, J., Courtin, C.M. methane potential from wheat straw by low temperature alka- (2014) Biorefining of wheat straw using an acetic and formic line calcium hydroxide pre-treatment. Bioresource Technol- acid based organosolv fractionation process. Bioresource ogy. 189: 258-265. Technology. 156: 275–282. [89] Khuong, L.D., Kondo, R., de-Leon, R., Anh, T.K., Shimizu, [72] Mesa, L., González, E., Cara, C., González, M., Castro, E. and K. and Kamei, I. (2014) Bioethanol production from alkaline- Mussatto, S.I. (2011) The effect of organosolv pretreatment pretreated sugarcane bagasse by consolidated bioprocessing variables on enzymatic hydrolysis of sugarcane bagasse. using Phlebia sp. MG-60. International Biodeterioration and Chemical Engineering Journal. 168: 1157-1162. Biodegradation. 88: 62-68.

3713 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[90] Wang, K., Yang, H., Chen, Q. and Sun, R. (2013) Influence of [107] Bals, B., Wedding, C., Balan, V., Sendich, E. and Dale, B. delignification efficiency with alkaline peroxide on the digest- (2011) Evaluating the impact of ammonia fiber expansion ibility of furfural residues for bioethanol production. Biore- (AFEX) pretreatment conditions on the cost of ethanol produc- source Technology. 146: 208-214. tion. Bioresource Technology. 102: 1277-1283. [91] Mora P.M., Meli, L., Doherty, T.V., Linhardt, R.J. and [108] Pu Y., Jiang N., Ragauskas A.J. (2007) ionic liquid as a green Dordick, J.S. (2011) Room temperature ionic liquids as emerg- solvent for lignin. Journal of Wood Chemistry and Technol- ing solvents for the pretreatment of lignocellulosic biomass. ogy. 27(1):23–33 . Biotechnology and Bioengineering. 108: 1229-1245. [109] Choi, I.S., Kim, J.H., Wi, S.G., Kim, K.H. and Bae, H.J. [92] Zheng, Y., Pan, Z. and Zhang, R. (2009) Overview of biomass (2013) Bioethanol production from mandarin (Citrus unshiu) pretreatment for cellulosic ethanol production. International peel waste using popping pretreatment. Applied Energy. 102: Journal of Agricultural and Biological Engineering. 2: 51-68. 204-210. [93] Brandt, A., Ray, M.J., To, T.Q., Leak, D.J., Murphy, R.J. and [110] Dererie, D.Y., Trobro, S., Momeni, M.H., Hansson, H., Welton, T. (2011) Ionic liquid pretreatment of lignocellulosic Blomqvist, J., Passoth, V. et al. (2014) Improved bio-energy biomass with ionic liquid-water mixtures. Green Chemistry. yields via sequential ethanol fermentation and biogas digestion 13: 2489-2499. of steam exploded oat straw. Bioresource Technology. 102: 4449-4455. [94] Samayamand, I.P. and Schall, C.A. (2010) Saccharification of ionic liquid pretreated biomass with commercial enzyme mix- [111] Horn, S.J., Nguyen, Q.D., Westereng, B., Nilsen, P. and tures. Bioresource Technology. 101: 3561-3566. Eijsink, V.G.H. (2011) Screening of steam explosion conditions for glucose production from non-impregnated wheat [95] Da-Costa, L.A., João, G.K., Rubik, D.F., Bogel-Łukasik, E., straw. Biomass and Energy. 35: 4879-4886. Duarte, L.C., Andreaus, J. and Bogel-Łukasik, R. (2013) Pre- treatment of lignocellulosic biomass using ionic liquids: [112] Chang, J., Cheng, W., Yin, Q., Zuo, R., Song, A., Zheng, Q. et Wheat straw fractionation. Bioresource Technology. 142: 198- al. (2012) Effect of steam explosion and microbial fermenta- 208. tion on cellulose and lignin degradation of corn stover. Biore- source Technology. 104: 587-592. [96] Heinze, T., Schwikal, K. and Barthel, S. (2005) Ionic liquids [113] Liu, Z.H., Qin, L., Pang, F., Jin, M.J., Li, B.Z., Kang, Y. et al. as reaction medium in cellulose functionalization. Macromo- (2013) Effects of biomass particle size on steam explosion pre- lecular Bioscience. 5: 520-525. treatment performance for improving the enzyme digestibility [97] Prado R., Erdocia X., Labidi J. (2013b) Lignin extraction and of corn stover. Industrial Crops and Products. 44: 176-184. purification with ionic liquids. Journal of Chemical Technol- [114] Choi, I.S., Kim, J.H., Wi, S.G., Kim, K.H. and Bae, H.J. ogy and Biotechnology. 88(7):1248–1257. (2012) Conversion of coffee residue waste into bioethanol [98] Zhao, H., Jones, C.L., Baker, G.A., Xia, S., Olubajo, O. and with using popping pretreatment. Bioresource Technology. Person, V.N. (2009) Regenerating cellulose from ionic liquids 125: 132-137. for an accelerated enzymatic hydrolysis. Journal of Biotech- [115] Chen, H., Li, G. and Li, H. (2014) Novel pretreatment of steam nology. 139: 47-54. explosion associated with ammonium chloride preimpregna- [99] De-Azevedo, R.E.G., da-Costa, C.A. and Aznar, M. (2014) tion. Bioresource Technology. 153: 154-159. Use of protic ionic liquids as biomass pretreatment for ligno- [116] Wang, K., Xiong, X., Chen, P., Chen, L., Su, X. and Liu, Y. cellulosic ethanol production. Chemical Engineering Transac- (2012) Comparison of gamma irradiation and steam explosion tions. 37: 397-402. pretreatment for ethanol production from agricultural residues. [100] Groff, D., George, A., Sun, N., Sathitsuksanoh, N., Bokinsky, Biomass Bioenergy. 46: 301-308. G., Simmons, B.A., et al. (2013) Acid enhanced ionic liquid [117] Li, G. and Chen, H. (2014) Synergistic mechanism of steam pretreatment of biomass. Green Chemistry. 15: 1264-1267. explosion combined with fungal treatment by Phellinus baumii for the pretreatment of corn stalk. Biomass Bioenergy. [101 Hayes, D.J. (2009) An examination of biorefining processes, 67: 1-7. catalysts and challenges. Catalysis Today. 145: 138-151. [118] Li, B.Z., Balan, V., Yuan, Y.J. and Dale, B.E. (2010a) Process [102] Turner, M.B., Spear, S.K., Huddleston, J.G., Holbrey, J.D. and optimization to convert forage and sweet sorghum bagasse to Rogers, R.D. (2003) Ionic liquid salt-induced inactivation and ethanol based on ammonia fiber expansion (AFEX) pretreat- unfolding of cellulase from Trichoderma reesei, Green Chem- ment. Bioresource Technology. 101: 1285-1292. istry. 5(4): 443–447. [119] Humpula, J.F., Uppugundla, N., Vismeh, R., Sousa, L., [103] Geng, A., Xin, F. and Ip, J.Y. (2012) Ethanol production from Chundawat, S.P.S., Jones, D., et al. (2014) Probing the nature horticultural waste treated by a modified organosolv method. of AFEX-pretreated corn stover derived decomposition prod- Bioresource Technology. 104: 715-721. ucts that inhibit cellulase activity. Bioresource Technology. [104] Obama, P., Ricochon, G., Muniglia, L. and Brosse. (2012) 152: 38-45. Combination of enzymatic hydrolysis and ethanol organosolv [120] Harun, S., Balan, V., Takriff, M.S., Hassan, O., Jahim, J. and pretreatments: Effect on lignin structures, delignification Dale, B.E. (2013) Performance of AFEX™ pretreated rice yields and cellulose-to-glucose conversion. Bioresource Tech- straw as source of fermentable sugars: the influence of particle nology. 112: 156-163. size. Biotechnology for Biofuels. 6: 1-17. [105] Lai, C., Tu, M., Li, M. and Yu, S. (2014) Remarkable solvent [121] Garlock, R.J., Balan, V. and Dale, B.E. (2012) Optimization and extractable lignin effects on enzymatic digestibility of or- of AFEX™ pretreatment conditions and enzyme mixtures to ganosolv pretreated hardwood. Bioresource Technology. 156: maximize sugar release from upland and lowland switchgrass. 92-99. Bioresource Technology. 104: 757-768. [106] Park, N., Kim, H.Y., Koo, B.W., Yeo, H. and Choi, I.G. (2010) [122] Alves, G.L.V., Borges, P.M.T. and da-Silva, A.A. (2014) En- Organosolv pretreatment with various catalysts for enhancing hancing liquid hot water (LHW) pretreatment of sugarcane ba- enzymatic hydrolysis of pitch pine (Pinus rigida). Bioresource gasse by high pressure carbon dioxide (HP-CO2). Industrial Technology. 101: 7046-7053. Crops and Products. 57: 141–149.

3714 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[123] Li, H.Q., Jiang, W., Jia, J.X. and Xu, J. (2014) pH pre-cor- [139] Yu, Q., Zhuang, X., Lv, S., He, M., Zhang, Y., Yuan, Z., Qi, rected liquid hot water pretreatment on corn stover with high W., Wang, Q., Wang, W. and Tan, X. (2013a) Liquid hot water hemicellulose recovery and low inhibitors formation. Biore- pretreatment of sugarcane bagasse and its comparison with source Technology. 153: 292-299. chemical pretreatment methods for the sugar recovery and structural changes. Bioresource Technology. 129: 592-598. [124] Vallejos, M.E., Zambon, M.D., Area, M.C. and da-Silva, C.A.A. (2012) Low liquid–solid ratio (LSR) hot water pre- [140] Hongdan, Z., Shaohua, X. and Shubin, W. (2013) Enhance- treatment of sugarcane bagasse. Green Chemistry. 14: 1982- ment of enzymatic saccharification of sugarcane bagasse by 1989. liquid hot water pretreatment. Bioresource Technology. 143: 391-396. [125] Zhang, H. and Wu, S. (2013) Subcritical CO2 pretreatment of sugarcane bagasse and its enzymatic hydrolysis for sugar pro- [141] Yu, Q., Zhuang, X., Yuan, Z., Qi, W., Wang, W., Wanga, Q. duction. Bioresource Technology. 149: 546–550. and Tan, X. (2013b) Pretreatment of sugarcane bagasse with [126] Alinia, R., Zabihi, R., Esmaeilzadeh, F. and Kalajahi, J.F. liquid hot water and aqueous ammonia. Bioresource Technol- ogy. 144, 210-215. (2010) Pretreatment of wheat straw by supercritical CO2 and its enzymatic hydrolysis for sugar production. Biosystems En- [142] Weiqi, W., Shubin, W. and Liguo, L. (2013) Combination of gineering. 107: 61-66. liquid hot water pretreatment and wet disk milling to improve [127] Huisheng, L., Miaomiao, R., Minhua, Z. and Ying, C. (2013) the efficiency of the enzymatic hydrolysis of eucalyptus. Bio- Pretreatment of corn stover using supercritical CO2 with wa- resource Technology. 128, 725–730. ter-ethanol as co-solvent. Chinese Journal of Chemical Engi- [143] Narayanaswamy, N., Faik, A., Goetz, D.J. and Gu, T. (2011) neering. 21: 551-557. Supercritical carbon dioxide pretreatment of corn stover and [128] Benazzi, T., Calgaroto, S., Astolfi, V., Rosa, D.C., Oliveira, switchgrass for lignocellulosic ethanol production. Biore- J.V. and Mazutti, M.A. (2013) Pretreatment of sugarcane ba- source Technology. 102: 6995-7000. gasse using supercritical carbon dioxide combined with ultra- [144] Eisenhuber, K., Krennhuber, K., Steinmüller, V. and Jäger, A. sound to improve the enzymatic hydrolysis. Enzyme and Mi- (2013) Comparison of different pre-treatment methods for sep- crobial Technology. 52: 247-250. arating hemicellulose from straw during lignocellulose bioeth- [129] Pan, X., Gilkes, N. and Saddler, J.N. (2006) Effect of acetyl anol production. Energy Procedia. 40: 172-181. groups on enzymatic hydrolysis of cellulosic substrates. [145] Singh, S., Harms, H., and Schlosser, D. (2014) Screening of Holzforschung. 60: 398-401. ecologically diverse fungi for their potential to pretreat ligno- [130] Corrales, R.C.N.R., Mendes, F.M.T., Perrone, C.C., cellulosic bioenergy feedstock. Applied Microbiology and Bi- Sant’Anna, C., de-Souza, W., Abud, Y., Bon, da S.E.P. and otechnology. 98: 3355-3370. Ferreira-Leitão, V. (2012). Structural evaluation of sugar cane [146] Kumar, R. and Wyman, C.E. (2009) Effect of cellulase and bagasse steam pretreated in the presence of CO2 and SO2. Bi- otechnology for Biofuels. 5: 1-8. xylanase enzymes on the deconstruction of solids from pretreatment of poplar by leading technologies. Biotechnology [131] García A.M.P., Ballesteros, I. and González, A. (2006) Effect Progress. 25: 302-314. of inhibitors released during steam-explosion pretreatment of barley straw on enzymatic hydrolysis. Applied Microbiology [147] Mathews, S.L., Pawlak, J. and Grunden, A.M. (2015) Bacterial and Biotechnology. 129: 278-288. biodegradation and bioconversion of industrial lignocellulosic streams. Applied microbiology and biotechnology. 99(7), [132] Vivekanand, V., Olsen, E.F., Eijsink, V.G.H. and Horn, S.J. 2939-2954. (2013) Effect of different steam explosion conditions on methane potential and enzymatic saccharification of birch. Biore- [148] Wang, W., Yuan, T. and Cui, B. (2014) Biological pretreat- source Technology. 127: 343-349. ment with rot fungi and their co-culture to overcome lignocellulosic recalcitrance for improved enzymatic digestion. Bio- [133] Hoover, A., Tumuluru, J.S., Teymouri , F., Moore, J. and Resources. 9: 3968-3976. Gresham, G. (2014) Effect of pelleting process variables on physical properties and sugar yields of ammonia fiber expan- [149] Deswal, D., Gupta, R., Nandal, P. and Kuhad, R.C. (2014) sion pretreated corn stover. Bioresource Technology. 164: Fungal pretreatment improves amenability of lignocellulosic 128-135. material for its saccharification to sugars. Carbohydrate Poly- mers. 99: 264-269. [134] Langan, P., Petridis, L., O’Neill, H., Pingali, V., Foston, M., Nishiyama, Y., et al. (2014) Common processes drive the ther- [150] Wyman, C.E., Dale, B.E., Elander, R.T., Holtzapple, M., La- mochemical pretreatment of lignocellulosic biomass. Green disch, R. and Lee, Y.Y. (2005) Coordinated development of Chemistry. 16: 63-68. leading biomass pretreatment technologies. Bioresource Tech- nology. 96: 1959-1966. [135] Yoo, C.G., Nghiem, N.P., Hicks, K.B. and Kim, T.H. (2011) Pretreatment of corn stover using low-moisture anhydrous am- [151] Shi, J., Sharma S.R.R., Chinn, M. and Howell, N. (2009) Ef- monia (LMAA) process. Bioresource Technology. 102: fect of microbial pretreatment on enzymatic hydrolysis and 10028–10034. fermentation of cotton stalks for ethanol production. Biomass Bioenergy. 33: 88-96. [136] Chundawat, S.P.S., Bellesia, G., Uppugundla, N., da-Costa, S.L., Gao, D., Cheh, A.M., et al. (2011) Restructuring the crys- [152] Digman, M.F., Shinners, K.J., Dien, B.S., Hatfield, R.D., Li, talline cellulose hydrogen bond network enhances its depoly- X.L., Muck, R.E., Weimer, P.J. (2007). On-farm pretreatment merization rate. Journal of the American Chemical Soci- technologies for improving enzymatic degradability of cellu- ety.133: 11163-11174. lose and hemicellulose present in perennial grass. In ASABE [137] Zhao, C., Ding, W., Chen, F., Cheng, C., Shao, Q. (2014) Ef- Annual International Meeting, Minneapolis, MN. fects of compositional changes of AFEX-treated and H- [153] García, A., Alriols, M. G., and Labidi, J. (2014). Evaluation of AFEX-treated corn stover on enzymatic digestibility. Biore- different lignocellulosic raw materials as potential alternative source Technology. 155: 34-40. feedstocks in biorefinery processes. Industrial Crops and [138] Shen, G., Tao, G., Zhao, M., Yang, B., Wen, D.,Yuan, Q. and Products. 53: 102-110. Rao, G. (2011) Effect of hydrogen peroxide pretreatment on [154] Wan, C., Zhou, Y., Li, Y. (2011). Liquid hot water and alkaline the enzymatic hydrolysis of cellulose. Journal of Food Process pretreatment of soybean straw for improving cellulose digest- Engineering. 3: 905-921. ibility. BioresourceTechnology.102: 6254-6259.

3715 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[155] Tan, H.T., Lee, K.T., Mohamed, A.R. (2010). Pretreatment of lignocellulosic palm biomass using a solvent-ionic liquid [BMIM] Cl for glucose recovery: An optimization studyusing response surface methodology. Carbohydrate Polymer. 83:1862-1868. [156] Yang, F., Li, L., Li, Q., Tan, W., Liu, W., and Xian, M. (2010). Enhancement of enzymatic in situ saccharification of cellulose in aqueous-ionic liquid media by ultrasonic intensifica- tion. Carbohydrate Polymers. 81(2): 311-316. [157] Mesa, L., Gonzalez, E., Cara, C., González M., Castro, E., Mussatto, S.I. (2011). The effect of organosolv pretreatment variables on enzymatic hydrolysis of sugarcane bagasse. Chemical Engineering Journal. 168:1157–1162.

[158] Girio, M., Fonseca, C., Carvalheiro, F., Duarte, L.C., Marques, S., Bogel-Lukasik, R. (2010). Hemicelluloses for fuel ethanol: a review. Bioresource Technology. 101: 4775-4800.

[159] Kratky, L., Jirout, T., Nalezenec, J. (2012). Lab-scale technology for biogas production from lignocellulosic wastes. Acta Polytech. 52: 54–59.

[160] Barragán-Huerta, B.E., Cabrera-Soto,M. L., Cruz, M.T., Ped- roza-Rodríguez, A.M. (2015). The valorization of tequila industry agricultural waste throught its use as substrate for laccase production in a fluidized bed reactor. Fresenius Environ- mental Bulletin. 24: 2026-2034. [161] Nuñez, H.M., Rodríguez, L., Khana, M., N. (2011). Agave for tequila and biofuels: an economic assessment and potential opportunities. GCB-Bioenergy. 3: 43–57. [162] Gustavsson, J., Cederberg, C., Sonesson, U., van Otterdijk, R., Meybeck A. (2011). Global food losses and food waste – Ex- tent, causes and prevention. FAO. Rome.

Received: May 04, 2015 Revised: August 31, 2015 Accepted: September 10, 2015

CORRESPONDING AUTHOR

Dr. Blanca E. Barragán-Huerta

Instituto Politécnico Nacional

Escuela Nacional de Ciencias Biológicas

Departamento de Ingeniería en Sistemas Ambientales

Av. Wilfrido Massieu S/N, Unidad Profesional

Adolfo López Mateos

CP 07739 México

D.F. MEXICO

Phone +52 (55)-57296000 Ext 52310

E-mail: [email protected]

[email protected]

FEB/ Vol 24/ No 11a/ 2015 – pages 3703 - 3716

3716 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

A STATISTICAL EXPERIMENTAL DESIGN TO DETERMINE THE AZO DYE DECOLORIZATION AND DEGRADATION BY THE HETEROGENEOUS FENTON PROCESS

Yeliz Aşçı1,*, Ezgi A. Demirtas2, Cansu Filik Iscen3 and A. Sermet Anagun2

1Eskişehir Osmangazi University, Department of Chemical Engineering, 26480, Meselik, Eskisehir, Turkey 2Eskişehir Osmangazi University, Department of Industrial Engineering, 26480, Meselik, Eskisehir, Turkey 3Eskişehir Osmangazi University, Department of Elementary Education, 26480, Meselik, Eskisehir, Turkey

ABSTRACT The traditional methods are being used to decolorize dye in textile wastewaters, such as chemical or electrochemical Full factorial design of experiment was used to screen precipitation, coagulation/flocculation, membrane separa- the factors affecting the color and COD removal efficiency tion (ultrafiltration, reverse osmosis), biological treatment for an azo dye, Direct Orange 26 (DO26). The obtained lin- processes and adsorption [4-5]. However, most of the dyes ear model was statistically tested using analysis of variance are found to be resistant to conventional treatment process (ANOVA), Student’s t-test, lack of fit test, and test of curva- [5] and due to the variability of the organic dyes it is diffi- ture. The percentage removal of color and COD was exam- cult to treat this kind of wastewater using traditional bio- ined by varying experimental conditions with center points. chemical treatment process and coagulation treatment pro- The factors and levels used during the experiments were; cess is inadequate [6]. Most azo dyes are not biodegradable initial pH (2, 4 and 6), catalyst amount (0.2 g/250 ml, 0.6 g/ by aerobic treatment processes, but they can be decolorized 250 ml and 1g/250 ml), stirring speed (150 rpm, 200 rpm and by anaerobic treatment. However, this treatment results in 250 rpm) and reaction time (10 min., 35 min. and 60 min.). A the cleavage of the nitrogen double bond, and the resulting desirability function based simultaneous optimization proce- fragments, which are aromatic amines, are proven carcino- dure was implemented to seek better conditions in terms of gens [7].The disposal problem of precipitated wastes is the maximizing the color and COD removal from wastewaters. main disadvantage of precipitation methods. For dye ad- The results showed that 98.5% color and 77.5% COD re- sorption, activated carbon, biomass derivatives, and other moval was obtained when initial pH, catalyst amount, stir- adsorbents were found to remove dye efficiently. Activated ring speed, and reaction time are roughly set to 2, 1 g/250 ml, carbon adsorption process for the removal of dyes is an ac- 250 rpm, and 60 min., respectively. cepted practice, but the cost of treatment is high [8]. Also, the regeneration of the adsorbents is not easy. Some low- cost adsorbents may be directly disposed of, but in many countries disposal of the wastewater treatment sludge and KEYWORDS: Experimental Design, Heterogeneous Fenton, spent adsorbents are prohibited by laws [9]. Therefore, re- Fe(III)-sepiolite catalyst, Direct Orange 26, Oxidation cently, advanced oxidation processes (AOP) using ozone,

titanium dioxide (TiO2), ultra violet (UV), and Fenton’s re-

agent (H2O2 and ferrous ion) have received considerable attention as effective pretreatment processes of less biode- 1. INTRODUCTION gradable wastewater [10]. Among them, Fenton’s reagent has been widely used because had proved to be highly effec- Wastewaters from textile and dye industries constitute tive in the degradation and mineralization of most of the pol- a substantial amount of pollutants such as high Chemical lutants in wastewaters. Fenton’s process has not only trans- Oxygen Demand (COD), broad fluctuating pH, high amount formed chemically polluting agents, but present very attrac- of suspended particles and are strongly colored and direct tive advantages, such as the complete mineralization of some discharge of textile industry wastewater into the receiving compounds, their oxidation at very low concentrations, the media causes serious environmental pollution by imparting generation of environmentally friendly by products and the intensive color and toxicity to aquatic environment [1, 2]. low consumption of energy, in comparison with other Azo dyes are the most common pollutants of textile waste- wastewater treatment methods [11]. The Fenton process also waters and can cause several problems to the aquatic life [3]. could be adopted rapidly in a textile wastewater treatment system, without the need for reconstructing the existing co- * Corresponding author agulation unit. The only change in operating the process

3717 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

2 will be the addition of H2O2 and Fe+ as well as pH adjust- In this work, Fenton treatment in heterogeneous 2+ ment. The use of Fe /H2O2 as an oxidant for wastewater phase is used to remove organic dye present in synthetic treatment is attractive since iron is highly abundant and solutions. In order to investigate the influence of different non-toxic, and a 30% hydrogen peroxide aqueous solution variables 24 full factorial experimental design with center is easy to handle and environmentally not harmful [6]. Nu- points was used and then the different models were de- merous studies on the treatment of textile dye effluents in- scribed for the organic dye decolorization and degradation clude Fenton processes [12-19]. The Fenton system also process. To the best of our knowledge, simultaneous opti- has been applied in real industrial wastewaters such as mization of color and COD removal of DO26 using Fe(III)- treatment of olive oil mill wastewater in southwestern sepiolite clay as a catalyst has not been reported previously. Spain [20], removal of organic contaminants in paper pulp treatment effluents in Spain [21] chemical industry for pesticides removal in southern Poland, El-Nasr pharmaceuti- 2. MATERIALS AND METHODS cal industry for antibiotic removal in South-East of Cairo, tannery industrial treatment plant in Brazil [22]. 2.1. Materials Although Fenton process has received much attention Direct Orange 26 commonly used in textile industries in the past few years, the use of Fe(II)/Fe(III) as a homo- was obtained from Burboya Co. Textile Industry, Bursa, geneous catalyst has a significant disadvantage. The homo- Turkey, at 45 % purity and was used without further puri- geneous Fenton requires complementary steps, such as pre- fication. The structure and characteristics of DO26 are precipitation and flocculation to recover the iron catalyst, presented in Table 1. Hydrogen peroxide (30%, w/w) and vent contamination and enable catalyst reuse [11-23]. As a Fe(NO3)3·9H2O were obtained from Sigma-Aldrich, Ger- consequence, heterogeneous Fenton processes are of par- many. All chemicals were analytical-grade reagents and ticular interest since most of the iron remains in the solid were used as received without further purification. Deion- phase. In addition to showing limited leaching of iron ions, ized water was used and the heterogeneous catalyst was the the catalysts can easily be recovered after the reaction and natural clay without pre-treatment, which was supplied from remain active during successive operations [24]. sepiolite mines in the Eskisehir region of Turkey.

The development of supported Fenton catalysts has re- TABLE 1 - The characteristics of DO26 cently become important in the emerging field of advanced oxidation technologies [25]. A number of heterogeneous catalysts developed by immobilizing Fe(II)/Fe(III) ions on various supports had been reported in the literature [26-27]. The use of clays as catalysts for Fenton-like reactions is a promising alternative because they are abundant in nature and, they have high specific surface area, high chemical and mechanical stabilities, and a variety of surface and structural properties [22]. Among these clays, sepiolite is gaining considerable attention because of its sorptive, rheological and catalytic properties, as well as its needle-like morphology. Moreover, the relatively low cost of sepiolite guarantees its Molecular Formula: C33H22N6Na2O9S2 continued utilization in the future [24]. The heterogeneous Formula Weight: 756.67 Fenton techniques depend on various parameters such as the Colour:Orange catalyst amount, H2O2 concentration, pH, reaction time, stir- λmax (nm): 494.5 ring speed and temperature. The number of the influential variables in the selected system is high, and for this reason it 2.2. Preparation of the catalyst is useful to apply experimental design techniques. The ex- After washing and drying, the selected 0.038–0.053 perimental design is a helpful tool to identify significant var- mm-sized sepiolite was prepared through a cation ex- iables that affect the process and to determine optimal con- change process as follows: 10 g sepiolite and 100mL of 1 ditions in several processes with minimal experimental runs g/L Fe(NO3)3·9H2O solution were mixed and shaken at 175 [28]. Additionally, the relative significance of the factors and rpm for 6 h at 55°C. The samples were washed with deion- possible synergistic or antagonistic interactions that may ex- ized water and dried at 105°C. The characterization of this ist between them can be evaluated [29]. A two-level factorial catalyst was performed previously [24]. design (2k) was selected in the present work due to the fact that this design technique is very useful in experiments stud- 2.3. Experimental Procedure ying the effect of several variables and how these variables Chemical oxidation of DO26 was carried out in batch interact with each other over a response factor. Full factorial mode, using a beaker filled with 250mL of the DO26 solu- design is considered to be the most suitable design when the tion at a given concentration. In a typical run, the pH of the study is directed towards gaining knowledge of the influence DO26 solution has been adjusted to the desired pH value of the variables on the process [30]. by NaOH or H2SO4. After stabilization of the temperature

3718 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

and pH, the powder catalyst (Fe(III)-sepiolite) was added to where C0 is the initial dye concentration in ppm and Ct the DO26 solution. The beginning of the reaction was con- is dye concentration at any time t. COD removal was meas- sidered when the required amount of H2O2 (0.5 mM) was ured at different time intervals by COD kits. The calcula- added. All experiments were carried out under constant stir- tion of the extent of COD removal is shown in Eq.2. ring to ensure good dispersion of the catalyst. Thereafter, COD COD samples were withdrawn periodically and centrifuged to re- %()100COD removal 0 t (2) move suspended particles for 5 min and analyzed using a COD0 UV–vis spectrophotometer (Shimadzu, model UV-120-01). where COD0 is the initial COD in ppm and CODt is COD in ppm at any time t. 2.4. Analysis

The maximum absorbance wavelength (λmax) of DO26 was found to be 494.5nm using a UV–vis spectro- 3. RESULTS AND DISCUSSION photometer (Thermo Electron Corporation, model Helios Aquamate). 3.1 Experimental Design and Modeling The color removal was determined using Eq.1. Initially, preliminary experiments were conducted to CC decide the most influential operating parameters and their %Color removal  (0 t ) 100 (1) levels. The effect of pH was investigated in the range of 2- C0 6. The results indicated that color and COD removal was significantly influenced by the pH of the solution.

%Color Removal %COD Removal %Color Removal %COD Removal 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 0 102030405060 0 102030405060 Reaction Time (min.) Reaction Time (min.)

(a) A=2, B= 0.2g/250mL, C= 150 rpm (b) A=2, B= 1g/250mL, C= 150 rpm

%Color Removal %COD Removal %Color Removal %COD Removal 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 0 102030405060 0 102030405060 Reaction Time (min.) Reaction Time (min.)

(c) A=2, B= 0.2 g/250mL, C= 250 rpm (d) A=2, B= 1g/250mL, C= 250 rpm FIGURE 1 - Preliminary Experiment Results for Color and COD Removal

3719 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Some results of preliminary experiments are shown in culated through regression analysis and tested for signifi- Figure 1 at the optimum level of pH (pH=2). According to cance by Student’s t-test at 95% confidence level. The es- the preliminary experiments maximum color or COD re- timated effects and coefficients are given in Table 4 and moval was reached at pH=2, 1 g/250 ml catalyst concentra- Table 5 for color and COD removal, respectively. Table 4 tion, 250 rpm stirring speed, and 60 min. reaction time under shows that the coefficients of all the main effects and AB, the following conditions: a temperature of 250C, 100 ppm ABD, BCD interactions effects are statistically significant initial dye concentration and 0.5 mM H2O2 concentration. (p<0.05) and the remaining coefficients are statistically in- High temperature more than 250C (room temperature) was significant at 95% confidence level. not preferred because of its high cost in macro scale. Accord- ing to the preliminary experiments, initial dye concentra- TABLE 2 - Variable levels of the factors tion was determined as 100 ppm to measure COD accu- Factors -1 0 +1 rately. On the other hand, high H2O2 concentration more than 0.5 mM causes interference with COD. pH of the reaction (A) 2 4 6 Catalyst concentration (B) g/250 mL 0.2 0.6 1 To evaluate the influence of operating parameters on Stirring speed (C) rpm 150 200 250 color and COD removal efficiency of DO26 contaminated Reaction time (D) min. 10 35 60 wastewater, four different factors were chosen for analysis: pH of the reaction (A), catalyst amount (B), stirring speed In order to test whether the reduced model is adequate re- (C) and reaction time (D) under the following conditions: a garding with lack of fit and curvature effects, analysis of var- temperature of 250C, 100 ppm initial dye concentration and iance (ANOVA) was carried out. The results of ANOVA for 0.5 mM H2O2 concentration. After the preliminary experi- percentage of color and COD removal are given in Table 6 ments, 24 full factorial design with center points (20 experi- and 7. According to the analysis of variance results for color ments), was employed for the optimization of color and removal, the impact of the curvature effect was not found COD removal as response variables. Experimental data to be significant, meaning that the first order model is suit- were analyzed using Design Expert software from Stat able for the process concerned. All the main effects and Ease Inc., USA. The levels and ranges of factors are given AB, ABD, BCD interactions effects are statistically signif- in Table 2. icant (p<0.05). On omitting the coefficients not significant Base level experiments were carried out to determine at 95% confidence level, Y1 was given in terms of coded critical factors and respected levels, and to test adequacy of variables and represents % color removal in Eq. (3): the model (Table 3). The coefficients of the empirical . . . . model for percentage of color and COD removal were cal- . . . . (3)

TABLE 3 - 24 full factorial experiments for color and COD removal

Color Removal COD Removal Run A B C D Actual Predicted Actual Predicted 1 -1 -1 -1 -1 34.318 36.852 31.25 32.656 2 1 -1 -1 -1 12.682 10.534 10.00 5.7812 3 -1 1 -1 -1 67.591 69.455 56.25 55.156 4 1 1 -1 -1 42.864 42.932 41.25 45.781 5 -1 -1 1 -1 49.227 43.738 52.50 51.406 6 1 -1 1 -1 17.410 17.420 12.50 14.531 7 -1 1 1 -1 87.818 88.091 68.75 65.781 8 1 1 1 -1 58.682 61.568 45.00 46.406 9 -1 -1 -1 1 61.773 66.818 45.00 46.094 10 1 -1 -1 1 23.409 21.522 17.50 19.218 11 -1 1 -1 1 95.364 92.193 70.00 68.593 12 1 1 -1 1 86.955 84.647 61.25 59.218 13 -1 -1 1 1 87.545 85.454 66.25 64.843 14 1 -1 1 1 36.136 40.159 27.50 27.969 15 -1 1 1 1 98.045 99.080 73.75 79.219 16 1 1 1 1 92.182 91.534 63.75 59.844 17 0 0 0 0 53.591 59.500 40.00 46.406 18 0 0 0 0 58.682 59.500 45.00 46.406 19 0 0 0 0 59.227 59.500 47.50 46.406 20 0 0 0 0 55.955 59.500 46.25 46.406

3720 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 4 - Coefficients for color removal TABLE 5 - Coefficients for COD removal

Factor Effect Coef SE Coef T p Factor Effect Coef SE Coef T p Intercept -- 59.500 0.652 91.2 0.000 Intercept -- 46.41 0.822 56.46 0.000 A -26.42 -13.210 0.652 -20.25 0.000 A -23.12 -11.56 0.822 -14.07 0.001 B 38.37 19.188 0.652 29.41 0.000 B 27.19 13.59 0.822 16.54 0.000 C 12.76 6.381 0.652 9.78 0.002 C 9.69 4.84 0.822 5.89 0.01 D 26.35 13.176 0.652 20.20 0.000 D 13.44 6.72 0.822 8.18 0.004 AB 9.39 4.693 0.652 7.19 0.006 AB 8.75 4.38 0.822 5.32 0.013 AC -3.14 -1.570 0.652 -2.4 0.096* AC -5.0 -2.5 0.822 -3.04 0.056* AD 0.41 0.200 0.652 0.31 0.774* AD 1.88 0.94 0.822 1.14 0.337* BC -1.77 -0.890 0.652 -1.36 0.267* BC -4.06 -2.03 0.822 -2.47 0.090* BD 2.55 1.270 0.652 1.95 0.146* BD 0.94 0.47 0.822 0.57 0.608* CD -1.16 -0.580 0.652 -0.89 0.440* CD -0.31 -0.16 0.822 -0.19 0.861* ABC 2.67 1.340 0.652 2.05 0.133* ABC 2.5 1.25 0.822 1.52 0.226* ABD 9.49 4.744 0.652 7.27 0.005 ABD 3.13 1.56 0.822 1.90 0.153* ACD 0.51 0.260 0.652 0.39 0.721* ACD 1.88 0.94 0.822 1.14 0.337* BCD -5.87 -2.938 0.652 -4.5 0.020 BCD -2.19 -1.09 0.822 -1.33 0.275* ABCD 1.23 0,610 0.652 0.94 0.416* ABCD 0.00 0.00 0.822 0.00 1.000* Center Point -- -2.636 1.459 -1.81 0.168* Center Point -- -1.72 1.838 -0.94 0.419* *insignificant coefficients at 95% confidence level (p>0.05) *insignificant coefficients at 95% confidence level (p>0.05)

TABLE 6 - ANOVA results for color removal

Sum of Mean F p-value Source Squares df Square Value Prob > F Model 12962.52 7 1851.788 271.9252 < 0.0001 significant A-pH 2792.162 1 2792.162 410.014 < 0.0001 B-Catalyst amount 5890.563 1 5890.563 864.9976 < 0.0001 C-Stirring speed 651.4096 1 651.4096 95.65602 < 0.0001 D-Time 2777.769 1 2777.769 407.9006 < 0.0001 AB 352.4153 1 352.4153 51.7503 0.0004 ABD 360.1369 1 360.1369 52.88418 0.0003 BCD 138.0625 1 138.0625 20.27374 0.0041 Curvature 22.24132 1 22.24132 3.266019 0.1207 not significant Residual 40.8595 6 6.809917 Lack of Fit 20.42975 3 6.809917 1 0.5000 not significant Pure Error 20.42975 3 6.809917 Cor Total 13025.62 14 Model 12962.52 7 1851.788 271.9252 < 0.0001 significant R2: 0.997 R2(adj): 0.993 R2(pred):0.961

TABLE 7 - ANOVA results for COD removal

Sum of Mean F p-value Source Squares df Square Value Prob > F Model 6665.625 7 952.2321 70.93878 < 0.0001 significant A-pH 2139.063 1 2139.063 159.3545 < 0.0001 B-Catalyst amount 2956.641 1 2956.641 220.2619 < 0.0001 C-Stirring speed 375.3906 1 375.3906 27.96561 0.0003 D-Time 722.2656 1 722.2656 53.80688 < 0.0001 AB 306.25 1 306.25 22.81481 0.0006 AC 100 1 100 7.449735 0.0196 BC 66.01562 1 66.01562 4.917989 0.0486 Curvature 9.453125 1 9.453125 0.704233 0.4192 not significant Residual 147.6563 11 13.4233 Lack of Fit 115.2344 8 14.4043 1.332831 0.4478 not significant Pure Error 32.42188 3 10.80729 Cor Total 6822.734 19 R2: 0.990 R2(adj): 0.979 R2(pred):0.923

3721 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

According to the Table 5, the coefficients of all the lyst amount provides higher percentage. On the other hand, main effects and AB interaction effect are statistically sig- high level of time provides an important effect in terms of nificant (p<0.05) and the remaining coefficients are statis- percentage of color and COD removal than stirring speed. tically insignificant at 95% confidence level. However, by In addition, catalyst amount is considered the most im- considering the ANOVA results of COD removal for all portant factor among others due to a steep slope appear in the second and third order interactions and to ensure the Fig. 2 and 5. hierarchical model structure AC and BC interaction effects It is stated from AB, ABD and BCD interaction plots were included to the model given in Eq. (4): from Fig. 3 and Fig. 4 higher percentage of color removal

. . . . would appear to be obtained when pH is at the low levels . . . . (4) and the other factors are at the high levels. (A = 2 and B = In the case of color removal, by focusing on the F val- 1.0 g/250 mL, C=250 rpm, D=60 min., respectively). On the ues in Table 6, B is considered as the most important factor, other hand, according to the Fig.6 higher percentage of COD while A is the second and D is the third ones. C, ABD, AB removal would appear to be obtained when pH is at the low and BCD follow them. On the other hand, In the case of levels and the other factors are at the high levels again.

COD removal, the most important factors and interactions 3.2. Simultaneous Optimization and Confirmation are in order of B, A, D, C, AB, AC and BC. Individual goals can be combined into an overall de- 2 In addition, both of the models provide higher R , ad- sirability function to find the local solutions Desirability 2 2 justed R and predicted R . These values are 0.997, 0.993 functions are used to obtain qualitative and quantitative re- and 0.961 for the color removal and 0.990, 0.973 and 0.923 sponses by the simple and quick transformation of different 2 - for the COD removal, respectively. Predicted R values responses to one measurement [31]. Generally speaking, 2 are in a good agreement with adjusted R . Thus, it can be the first step when using desirability functions is to convert concluded that the reduced models may be used for pre- each response into an individual desirability function (di) cisely estimating the percentage of both color and COD re- the values of which vary from 0 to 1 (lowest desirability to moval. highest desirability). In order to estimate the percentage of color and COD The individual desirability scores for the predicted val- removal using the reduced model given in Eq. (3) and (4), ues of each dependent variable are then combined in an the appropriate levels for statistically significant main and overall desirability function, D, by computing the geomet- interaction effects should be determined by means of main ric mean of different di values as shown in Eq. (5): and interaction effect plots. 1 The main and interaction effect plots are depicted for n (5) Ddd 12 dn  color and COD removal in Figs. 2-6, respectively. It is ob-   served from Fig. 2 and 5 that the percentage of color and where di indicates the desirability of response i and n COD removal is affected by each level of the factors in a is the number of responses in the measure. If any of the different way. While low level of initial pH provides higher responses do not satisfy their desirability requirement(s), percentage of color and COD removal, high level of cata- the overall function becomes zero.

Main Effects Plot for Colour Data Means

pH Catalyst amount Point Type 80 Corner Center 70

40 2 4 6 0,2 0,6 1,0

Mean Stirring speed Time 80

40 150 200 250 10 35 60

FIGURE 2 - The Main Effect Plots for Color Removal

3722 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Interaction Plot for Colour Data Means

0,2 0,6 1,0 150 200 250 10 35 60 90 pH Point Type 2Corner 60 pH 4Center 6Corner 30

90 Catalyst amount Point Type 60 Catalyst amount 0,2 C orner 0,6 C enter 30 1,0 C orner 90 Stirring speed Point Type 60 Stirring speed 150 C orner 200 C enter 30 250 C orner

Time

FIGURE 3 - The interaction effects for color removal

FIGURE 4 - (ABD) and (BCD) interaction effects for color removal

3723 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Main Effects Plot for COD Data Means

pH Catalyst amount Point Type 60 Corner Center 50

30 2 4 6 0,2 0,6 1,0

Mean Stirring speed Time 60

30 150 200 250 10 35 60

FIGURE 5 - The Main Effect Plots for COD Removal

Interaction Plot for COD Data Means

0,2 0,6 1,0 150 200 250 10 35 60

60 pH Point Type 2Corner

pH 40 4Center 6Corner 20

60 Catalyst amount Point Type Catalyst amount 40 0,2 C orner 0,6 C enter 20 1,0 C orner

60 Stirring speed Point Type

Stirring speed 40 150 C orner 200 C enter 20 250 C orner

Time

FIGURE 6 - The interaction effects for COD Removal

For the simultaneous optimization of color and COD removals of color and COD are 98.398 and 78.351, respec- removal, the optimization function in the Design Expert tively. Similarly, the other non-dominated solutions which software was used. Optimization was carried out by gener- have the same desirability value from Table 8 could be ating a global desirability function (D) that assigned a used one over the other. value ranging from 0 to 1 to each solution (see Table 8). A useful measure of confirmation is to ensure that the The first highlighted solution is selected with the predicted responses fall within the prediction interval (PI) highest desirability value of 1. According to the first so- for each response at the optimal point [32]. The calcula- lution, optimized reaction conditions were established as tions for prediction interval can be found in the study. a pH of 2, 1 g/250 mL catalyst dose, 250 rpm stirring speed For the optimized reaction conditions, an additional and 60 minute reaction time, approximately. Estimated % experiment was carried out under the same conditions as

3724 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 8 - Local solution sets with desirability values

Number pH Catalyst amount Stirring speed Time Color COD Desirability 1 2.04 0.990 249.560 58.570 98.398 78.351 1 2 2.07 0.970 247.940 59.280 98.073 77.912 1 3 2.01 0.990 238.400 59.960 98.046 77.712 1 4 2.07 0.990 248.320 58.620 98.271 78.099 1 5 2.02 0.980 242.470 59.990 98.241 78.037 1 6 2.05 1.000 238.160 59.730 98.060 77.618 1 7 2.1 0.960 248.650 59.760 98.076 77.852 1 8 2.5 1.000 249.930 59.770 98.061 76.720 1 9 2.02 0.990 249.780 57.410 98.224 78.200 1 10 2.02 1.000 241.190 58.840 98.066 77.801 1 11 2.13 0.990 248.360 58.240 98.147 77.790 1 12 2.17 0.980 249.760 59.790 98.413 78.051 1 13 2 0.990 249.430 58.080 98.411 78.434 1 14 2 1.000 233.690 60.000 97.956 77.486 0.999 15 2 1.000 250.000 54.390 97.841 77.705 0.999 16 2 1.000 230.440 60.000 97.731 77.139 0.998 17 2 1.000 230.060 60.000 97.706 77.100 0.998 18 2.11 1.000 250.000 54.740 97.655 77.260 0.998 19 2 0.910 249.990 60.000 97.515 77.561 0.997 20 2 1.000 250.000 52.640 97.458 77.237 0.997

TABLE 9 - Results of the confirmation

Response Prediction 95% PI low 95% PI high Confirmatory trial % Color Removal 98.398 90.73 106.07 98.5 % COD Removal 78.351 68.64 88.06 77.5

the original experiment. Percent removal of color and COD COD removal, individual effects of all the factors are signif- were measured for the confirmation and the results are icant. Interaction effects of pH & catalyst amount, pH and shown in the last column of Table 9. According to the con- stirring speed, pH & reaction time are also significant. For firmatory trial, 98.5% color and 77.5% COD removals fell the simultaneous optimization of color and COD removal, within the PI with a 95% confidence level. The results ver- desirability functions were used and optimized reaction con- ify that the model predicts the optimal point well. ditions were established as a pH of 2, 1 g/250 mL catalyst dose, 250 rpm stirring speed and 60 minute reaction time. Under the optimum conditions, color and COD removal 4. CONCLUSION were predicted to be 98.398% and 78.351%, respectively, and the corresponding confirmatory trial values (98.5 % and The aim of the work was to investigate the color and 77.5 %) fit well to the dataset of predicted values. COD removal by the heterogeneous Fenton-type oxidation of DO26 azo dye. In order to determine the effects of various operating conditions (pH of the reaction, catalyst ACKNOWLEDGEMENTS amount, stirring speed and reaction time) a full 24 factorial design with center points was performed. The study was This paper is a part of the results of project number conducted by Fe(III)-sepiolite catalyst in the presence of 201215D13 funded by Eskisehir Osmangazi University. hydrogen peroxide (H2O2) in a batch process. According to The authors wish to thank their university for all the sup- the regression models and ANOVA results, the influential port provided. factors such as pH of the reaction, catalyst amount, stirring speed and reaction time are highly significant. Moreover, The authors have declared no conflict of interest. the interaction between pH of the reaction & catalyst amount, the interaction among pH & catalyst amount & reaction time, the interaction among catalyst amount & stirring speed & reaction time are significant. The models fit- REFERENCES ted very well to the experimental data as confirmed by the 2 2 2 [1] Lim, C.L., Morad, N., Teng, T.T. and Norli, I. (2011) Chemical high R , adjusted R and predicted R values. Similarly, ac- Oxygen Demand (COD) reduction of a reactive dye wastewater cording to the regression model and ANOVA results of using H2O2/pyridine/Cu (II) system. Desalination, 278, 26–30.

3725 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[2] Ay, F., Catalkaya, E. C. and Kargi, F. (2009) A statistical experi- [21] Perez, M., Torrades, F., Hortal J.A., Domenech, X. and Peral, J. ment design approach for advanced oxidation of Direct Red azo- (2002) Removal of organic contaminants in paper pulp treatment dye by photo-Fenton treatment. J Hazard Mater., 162, 230–236. effluents under fenton and foto-fenton conditions. Applied Cata- lyst B: Environmental. (1), 63-74. [3] Rodrigues C.S.D., Madeira, L.M. and Boaventure, R.A.R. (2009) Optimization of the azo dye Procion Red H-EXL degradation by [22] Soon, A.N. and Hameed, B.H. (2011) Heterogeneous catalytic Fenton’s reagent using experimental design. J Hazard Mater., 164, treatment of synthetic dyes in aqueous media using Fenton and 987-994. photo-assisted Fenton process. Desalination, 269, 1-16. [4] Lucas, M.S. and Peres, J.A. (2006) Decolarization of the azo dye [23] Muthukumari, B., Selvam, K., Muthuvel, I. and Swaminathan, M. Reactive Black 5 by Fenton and photo-Fenton oxidation. Dyes and (2009) Photoassisted hetero-Fenton mineralisation of azo dyesby Pigments, 71, 236-244. Fe(II)-Al2O3 catalyst. Chemical Engineering Journal, 153, 9–15. [5] Karataş, M., Argun, Y.A. and Argun M.E. (2012) Decolorization [24] Aşçı, Y. (2013) Decolorization of Direct Orange 26 by Heteroge- of antraquinonic dye, Reactive Blue 114 from synthetic neous Fenton Oxidation. Desalination and Water Treatment, 51, wastewater by Fenton process: Kinetics and thermodynemics. 7612–7620. Journal of Industrial and Engineering Chemistry, 18, 1058-1062. [25] Mosteo, R., Ormad, P., Mozas, E., Sarasa, J. and Ovelleiro, J. L. [6] Modirshahla, N., Behnajady, M.A. and Ghanbary, F. (2007) De- (2006) Factorial experimental design of winery wastewaters treat- colorization and mineralization of C.I. Acid Yellow 23 by Fenton ment by heterogeneous photo-Fenton process, Water Research, 40, and photo-Fenton processes. Dyes and Pigments, 73, 305-310. 1561– 1568. [7] Ntampegliotis, K., Riga, A., Karayannis, V., Bontozoglou, V. and [26] Neamtu, M., Zaharia, C., Catrinescu, C., Yediler, A. and Ma- Papapolymerou G. (2006) Decolorization kinetics of Procion H- coveanu, M. (2004) Kettrup, A. Fe exchanged Y zeolite as catalyst exl dyes from textile dyeing using Fenton-like reactions. J Hazard for wet peroxide oxidation of reactive azo dye Procion Marine H- Mater.,136, 75-84. EXL, Appl Catal B, 48, 287–294. [8] Hameed, B.H. and Lee, T.W. (2009) Degradation of malachite [27] Song, Y.L. and Li, J.T. (2009) Degradation of C.I. Direct Black green in aqueous solution by Fenton process. J Hazard Mater. 164, 168 from aqueous solution by fly ash/H2O2 combining ultrasound. 468-472. Ultrason. Sonochem., 16, 440–444. [9] Chang, M.W. and Chern, J.M. (2010) Decolorization of peach red [28] Cerino-Córdova, F.J., García-León, A.M., Soto-Regalado, E., azo dye, HF6 by Fenton reaction: Initial rate analysis. Journal of M.N. Sánchez-González, Lozano-Ramírez, T., García-Avalos, Taiwan Institute of Chemical Engineers, 41, 221-228. B.C. and Loredo-Medrano, J.A. (2012) Experimental design for [10] De León, M.A., Castiglioni, J., Bussi, J. and Sergio, M. (2008) the optimization of copper biosorption from aqueous solution by Catalytic activity of an iron- pillared montmorillonitic clay mineral Aspergillus terreus. Journal of Environmental Management, 95, in heterogeneous photo-Fenton process, Catalysis Today, 133– 577-582. 135, 600–605. [29] Cristovaoa, R. O., Goncalves, C., Botelho, C. M., Martins, R. J. E. [11] Ayodele, O.B., Lim, J.K. and Hameed, B.H. (2012) Pillared mont- and Boaventura, R. A.R. (2014). Chemical oxidation of fish can- morillonite supported ferric oxalate as photo-Fenton catalyst for ning wastewater by Fenton’s reagent. Journal of Environmental degradation of amoxicillin, Applied Catalysis A: General, 413- Chemical Engineering, 2, 2372-2376. 414, 301-309. [30] Rodrıguez-Chueca, J., Mosteo, R., Ormad, M.P. and Ovelleiro, [12] Gao, A., Gao, H. and Wan, J. (2013) Treatment pharmaceutical J.L. (2012) Factorial experimental design applied to Escherichia wastewater using fenton oxidation, ultraviolet and ultrasound pro- coli disinfection by Fenton and photo-Fenton processes. Solar En- cesses. Fresenius Environmental Bulletin, 22 (2), 449-454. ergy, 86, 3260–3267. [13] Argun, H. (2013) Electrochemical oxidation of reactive blue 221 [31] Derringer G. and Suich, R. (1980) Simultaneous Optimization of based on fenton’s process. Fresenius Environmental Bulletin, 22 Several Response Variables. J. Qual. Tech., 12, 214-219. (2), 310-317. [32] Montgomery, D. C. (2013) Design and Analysis of Experiments, [14] Wen, C.C., Wang, C.T. and Chou, W.L. (2012) Degradation of 8th ed, New York, John Wiley & Sons. DMSO in aqueous solutions by the Fenton reaction based advanced oxidaiton processes. Fresenius Environmental Bulletin, 21(3), 644-650. [15] Turhan, G.D. and Kartal, Ö.E. (2010) Photo-Fenton treatment of CI Reactive black 5 by response surface methodology. Fresenius Environmental Bulletin, 19(11A),2736-2743. [16] Solmaz, S. K.A, Ustun, G. E. and Birgul, A. (2009) Advanced ox- Received: January 21, 2015 +2 idation of textile dyeing effluents: comparison of Fe /H2O2, Revised: May 26, 2015 +3 -3 Fe /H2O2, O and chemical coagulation processes. Fresenius En- Accepted: June 23, 2015 vironmental Bulletin, 18(8), 1424-1433. [17] Arslan-Alaton, I. and Tureli, G. (2008) Degradation of aqueous Acid Red 183 textile dye and acid dye-bath effluent by Fenton-like CORRESPONDING AUTHOR and Photo-Fenton-like advanced oxidation processes. Fresenius Environmental Bulletin, 17(7B), 915-926. Yeliz Aşçı [18] Ertugay, N. and Acar F.N. (2013) Removal of COD and color from direct blue 71 azo dye wastewater by Fenton’s oxidation: Kinetic Eskişehir Osmangazi University study. Arabian Journal of Chemistry. Department of Chemical Engineering DOI:101016/j.arabjc.2013.02.009. Article in Press. Meselik Kampusu [19] Su, C.C., Pukdee-Asa, M., Ratanatamskul C. and Lu, M.C. (2011) Eskisehir, 26480 Effect of operating parameters on decolorization and COD re- TURKEY moval of three reactive dyes by Fenton’s reagent using fluidized- bed reactor. Desalination, 278, 211-218. [20] Rivas, F.J., Beltran, F.J., Gimeno, O. and Frades, J. (2001) Treat- E-mail: [email protected] ment of olive oil wastewater by Fenton’s reagent. Journal of Agri- cultural of Food Chemistry, 4, 1873-1880. FEB/ Vol 24/ No 11a/ 2015 – pages 3717 - 3726

3726 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

DETERMINE SOME MORPHOLOGICAL CHARACTERISTICS OF CRAYFISH (Astacus leptodactylus ESCHSCHOLTZ, 1823) WITH TRADIONAL METHODS AND ARTIFICIAL NEURAL NETWORKS IN DIKILITAS POND, ANKARA, TURKEY

Semra Benzer1,*and Recep Benzer2

1Gazi University, Gazi Faculty of Education, 06500, Teknikokullar, Ankara, Turkey 2Gazi University, Institute of Information, 06500, Teknikokullar, Ankara, Turkey

ABSTRACT 1. INTRODUCTION

This study aimed to determine some morphological The importance of crayfish in the food webs of fresh- characteristics of freshwater crayfish (Astacus leptodacty- water habitats has been recognised for a long time; thus, lus Eschscholtz 1823) populations 2014, in Dikilitaş Pond. they are regarded as a flagship species for comprehensive We present the relationships between total length (TL), to- water protection [1]. tal weight (TW); carapace length (CL), total weight (TW); carapace length (CL), total length (TL); abdomen length Astacus leptodactylus is a widespread species distrib- (AL), total weight (TW); chelea length (ChL), total weight uted throughout Europe, eastern Russia, and the Middle East (TW) for Astacus leptodactylus from Dikilitaş Pond. The [2]. Freshwater crayfish has had an economic importance in research was used 260 (105 female, 155 male). The re- Turkish aquatic food market. Recently, there has been an in- search was found as 40.38 % male, 59.62 % female of cray- crease, gradually. In 1990s, the annual crayfish production fish thought investigation female and male ratios was of was between 300 and 500 tonnes. In 2011, crayfish produc- determined as to 0.68 / 1.00. Results of the research can be tion in Turkish freshwater was 1681 tonnes. 609.6 tonnes seemed as follows; average total length 110.60 mm for fe- crayfish were caught from Turkish freshwater [3]. male 113.50 mm for male, average total weight 38.52 g for Water bodies in British Isles, Norway and Turkey re- female 50.08 g for male. Length-weight relation equation mained long the last safe havens for the European crayfish was found for females W = 0.031 x L 2.97, for males W = [2] until mid 1980’s, after which many productive native 0.059 x L 2.75 and for all gender W = 0.024 x L 3.01. The narrow clawed crayfish (Astacus leptodactylus) stocks col- results obtained by artificial neural networks and length- lapsed in Turkey [4]. As a result, the crayfish trapping and weight relation equation are compared to those obtained by export ceased from Turkey to Europe. The stock collapses the growth rate of the crayfish caught from the natural en- in Turkey were partial in some water bodies with up to 25% vironment. Length-weight relation and artificial neural net- survival [5]. The collapsed narrow clawed crayfish stocks work MAPE results were examined. It was found MAPE have later recovered substantially in Turkey [6]. value of the forecast of ANNs as a 0.46 and 1.30, while MAPE value of relationship results as a 6.07 and 2.98 for The most frequently used dimensions for crustaceans length – weight of all gender. Artificial neural networks are carapace length, body length, total length, body width, gives better results than length-weight relation. Artificial and wet weight [7]. To demonstrate the morphological neural networks can be alternative as a evaluated for changes between the male and female crayfish species are growth estimation. utilized to differences in length between individual parts of the body [8]. These differences also in the determination of the crayfish populations, relative growth, comparing the populations of the same species, the comparison of the KEYWORDS: Artificial neural networks, Dikilitas Pond, estimated, morphology of crayfish species, and in determining the length-weight relation size of crayfish to be placed on the market and is used in the systematic classification of crayfish [8-10]. It may be convenient to able to convert into the desired length measurement when only one of the other length measurements is known and the length-weight regression may be used to * Corresponding author estimate length from weight [11].

3727 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

The use of ANNs in ecosystem modeling began in the There is a very wide literature upon the properties of early 1990s, particularly for situations where data do not fit crayfish in various water system [23-31]. The main pur- statistical suppositions and non-linear relationships prevail. pose of the present investigation was to study length- ANNs behave better than linear models and have a good ca- weight relation and artificial neural networks for growth in pacity for generalizing when entering new data [12]. crayfish. Thus, the present study provides on the properties of the crayfish in Dikilitaş Pond in the last years. In fisheries, a wide range of multivariate techniques have been used to this end, including several methods of ordination and canonical analysis, univariate and multivar- 2. MATERIALS AND METHODS iate linear, curvilinear and logistic regressions [15]. Most of the models used to predict species distribution assume Dikilitaş Pond is located about 54 km south of Ankara, that relationships are smooth, continuous and either linear capital of Turkey, and lies within the coordinates of 39º31' or simple polynomials [16]. However, in real nature, any 22.9332'' N and 32º43' 7.8312'' E longitudes. It is near the changes in distributional boundaries or location cen tres of Dikilitaş Village of Haymana town. Crayfish samples were a fish stock are unlikely to be either monotonic or linear. collected from Dikilitaş Pond. During the study, 260 cray- Traditional methods of statistic may therefore be defective fish specimens (105 females and 155 males) were caught for detecting and successfully quantifying such changes 2014. The total length (TL), total weight (TW), carapace [13]. ANNs offer a promising alternative to traditional sta- length (CL), carapace width (Cw), abdomen length (AL) tistical approaches for predictive modeling when non-lin- and abdomen witdh (Aw), chela length (ChL) and chela ear patterns exist [14]. width (Chw) of each specimen were measured with a digi- Artificial neural networks (ANNs) are widely used in sci- tal caliper to the nearest 0.10 mm, while weighted to the ence and technology with applications in various branches of nearest 0.01 g, and each specimen was sexed [32]. Some chemistry, physics, and biology. ANN have long been widely parameters of specimen (TL, TW, CL, AL and ChL) were used in in many disciplines, such as predicting the distribu- used in ANNs. The crayfish obtained from the lake were tions of demersal fish species [13], predicting the presences immediately transported to the laboratory. The length and of small-bodied fish in a river [15], modeling population dy- weight (min-max) of the crayfish were 85– 140 mm and namics of aquatic insects [17] and modeling and spatially 23.80–86.65 g. mapping freshwater fish and assemblies of decapods [14]. Sex and length composition, the average length and ANNs have been trained to overcome known limitations of weight, and the length-weight relationship for each sex and conventional approaches in solving complex problems. combined sexes were determined from samples. Much work has been performed in predicting future The length-weight relationships were estimated from data with artificial neural networks (ANNs) as they exhib- the formula, W = a Lb, where W is total body weight (g), L ited better results than the traditional methods from the lit- the total length (mm), a and b are the coefficients of the erature [18-20]. functional regression between W and L [33]. The equation The fact that the ANN provides a better model was high- was log transformed to estimate the parameters ‘a’ and ‘b’. lighted by better predictions for lower values, the normality When b is equal to three (3), isometric pattern of growth of the residuals and their independence from the predicted occurs but when b is not equal to 3, allometric pattern of variable. Several authors have reported greater perfor- growth occurs, which may be positive if >3 or negative if mances of ANNs compared to linear regressions [21]. ANNs <3. have another advantage in that the ANN modeling approach A multilayer feed-forward neural network was used is fast and flexible [22]. In this study, the ANN has demon- for ANN. A schematic representation of a typical ANN is strated a new and alternative approach for its application in shown in Figure 1 and consists of 4 interconnected layers of predicting the growth and weight of crayfish. ‘nodes’ or ‘neurons’, including an input layer containing 1

Length /

Weight / Length

Sex

Output Layer

Input Layer Hidden Layer

FIGURE 1 - The ANN consisted of an input layer with 2 nodes, hidden layer, and an output layer with 1 node to be predicted for length and weight

3728 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Input 1 (x1) Output Input 2 (x2) ∑ ......

Input n (xn)

FIGURE 2 - Biological and mathematical explanation for ANNs design

node per independent variable (i.e., length, weight and sex During the training of the network, the input data and of crayfish), hidden layer, and finally, an ‘output layer’ with the input-output relationship between the learning of the 2 node (i.e., the weight or length of the crayfish samples). network is provided. This method, which is generally ANNs are computational systems that simulate biolog- called supervised learning, is one of the preferred methods [37]. The supervised learning method trained with the net- ical neural networks and can be defined as a specific type of parallel processing system based on distributional or work structure (Back-propagation Networks) will be used connectionist methods [34]. The use of ANNs to predict to solve problems in this study. the future is one a key area of study. ANNs can reveal the In this study, the sum squared error (SSE) and mean power relations between unknown and unnoticed data. absolute percentage error (MAPE) are used as the two performance criteria SSE was used as a criterion for determine ANNs are simulations of biological nervous systems training during the training of the network. In addition, we using mathematical models. They are networks with sim- can make comparisons involving more than one method ple processor units, interconnections, adaptive weights and because the MAPE of each provides information about the scalar measurement functions (e.g., summation and activa- average relative size of their errors. tion functions) [35]. ANN mathematical expression is seen in Figure 2. y is the neuron’s output, x is the vector of in- SSE and MAPE are described by equations 2 and 3, puts, and w is the vector of synaptic weights. respectively. In case of biological neuron information comes into n SSE  e2 the neuron via dendrite, soma processes the information  i and passes it on via axon. In case of artificial neuron the i  1 (2) information comes into the body of an artificial neuron via inputs that are weighted (each input can be individually 1 n e MAPE   i *100 multiplied with a weight). The body of an artificial neuron n Y then sums the weighted inputs, bias and “processes” the i  1 i (3) sum with a transfer function. At the end an artificial neuron Where Y is the actual observation value, e is the dif- passes the processed information via output(s). Benefit of i i ference between the actual vale and prediction value, and n artificial neuron model [36] simplicity can be seen in its is the number of total observations. mathematical description below: Neural Network Toolbox of MATLAB was used for the ANN calculations. This study was performed on 260 cray- . (1) fish (105 females and 155 males) caught 2014 in Dikilitaş Pond. Data were divided into three equal parts: training, Where: validation and test sets. The Matlab functions were used for “training”, “testing”, and “validation”. They were used ran- w (k) is weight value in discrete time k where i goes i domly: 70% in training, 15% in testing, and 15% in the val- from 0 to m, idation of the fishes. xi(k) is input value in discrete time k where i goes from 0 to m, F is a transfer function, 3. RESULTS AND DISCUSSION y (k) is output value in discrete time k. i There were about 40.38 % females, 59.62 % males. The As seen from a model of an artificial neuron and its female: male ratio was found to be 0.68:1 for the general equation (1) the major unknown variable of our model is its population. The length and weight (min-max) of the crayfish transfer function. Transfer function defines the properties of were 85 – 140 mm and 23.80 –86.65 g. The average length artificial neuron and can be any mathematical function. (mm) and weight (g) of samples were 110.60 ± 1.05 and

3729 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

60 Female Male Female+Male 50 40 30 20 10 0 Number of distribution of Number 83-88 89-94 95-100 101-106 107-112 113-118 119-124 125-130 131-136 137-142 Range of length (mm)

FIGURE 3 - Number of distribution of all individuals

38.52 ± 10.00 mm for female, 113.50 ± 1.24 and 50.08 ± female, W = 0.059 x L 2.75 for male and W = 0.024 x L 3.01 13.16 for males and 112.30 ± 1.17 and 46.87 ± 13.72 for the for female+male. combined sex, respectively (Table 1). The number of males The carapace length - weight relations (CL - TW) for the were greater than females in Dikilitaş Pond. It is seemed that crayfish in Dikilitaş Pond were found as W= 1.499 x L 1.93 for number of distribution all individual in Figure 3. female, W = 2.168 x L 1.84 for male and W = 0.981 x L 2.25 Total length (TL), total weight (TW), carapace length for female+male. The carapace length - length relations (CL (CL), abdomen width (Aw), chelea length (ChL) and chelea - TL) for the crayfish in Dikilitaş Pond were found as W = width (Chw) of females are smaller than males, while cara- 2.980 x L 0.80 for female, W = 2.675 x L 0.83 for male and W pace width (Cw), and abdomen length (AL) of females are = 3.179 x L 0.74 for female+male (Table 2 and Figure 4). bigger than males (Table 1). TL, TW, CL, Cw, AL, Aw, ChL and Chw of females and males are were insignificant (p < TABLE 2 - Length-weight relation parameters, equations and corre- 0.05). lation coefficients.

Species Sex Relationship r2 TABLE 1 - Metric properties for crayfish in Dikilitaş Pond ♀ W = 0.031x L 2.97 0.98 TL – TW ♂ W = 0.059 x L 2.75 0.99 Metric 3.01 Sex Mean ± Sx Min-Max t test ♀♂ W = 0.024 x L 0.96 species ♀ W = 1.499 x L 1.93 0.98 ♀ 110.6+10.5 95–130 CL – TW ♂ W = 2.168 x L 1.84 0.98 TL ♂ 113.5+12.4 85– 140 P<0.05 ♀♂ W = 0.981 x L 2.25 0.97 ♀♂ 112.3+11.7 95- 140 ♀ W = 2.980 x L 0.80 0.99 ♀ 38.52+10 23.80 - 59 CL – TL ♂ W = 2.675 x L 0.83 0.99 TW ♂ 50.08+13.16 29.65– 86.65 P<0.05 ♀♂ W = 3.179 x L 0.74 0.99 ♀♂ 46.87+13.72 23.80 - 86.65 ♀ W = 10.308 x L 0.70 0.97 ♀ 52.0±5.1 45-60 AL-TW ♂ W = 16.749 x L 0.66 0.98 CL ♂ 56.5±5.7 45-67 P<0.05 ♀♂ W = 17.583 x L 0.53 0.96 ♀♂ 54.7±5.9 45-67 ♀ W = 5.083 x L 1.04 0.98 ♀ 55.1±4.9 45-60 ChL-TW ♂ W = 59.137 x L 0.06 0.94 Cw ♂ 53.7±6.6 30-60 P<0.05 ♀♂ W = 28.037 x L 0.23 0.95 ♀♂ 54.3±6.0 30-65 ♀ 57.5±5.0 45-65 AL ♂ 56.4±8.7 42-75 P<0.05 In the present study, Table 3 shows that there are ♀♂ 56.8±7.5 42-75 strong positive correlations between TW and TL and there ♀ 19.10±6.6 18-40 are high correlations between CL and TL, CL and TW (p < Aw ♂ 20.10±5.6 17-40 P<0.05 ♀♂ 19.8±6.0 17-40 0.01). ♀ 65.2±10.3 55-100 ChL ♂ 83.2±20.7 35-115 P<0.05 Figure 5 illustrate the graphical presentation of the fit ♀♂ 75.9±19.4 35-115 between the actual and predicted values. Performance of ♀ 11.7±3.5 10-25 ANNs between forecast values for TL and TW is seen in Chw ♂ 15.7±3.8 10-25 P<0.05 Figure 6. ♀♂ 14.1±4.2 10-25 The values observed, ANNs and length-weight rela- The length - weight relations (TL - TW) for the cray- tion data are presented in Table 4. The observed data, fish in Dikilitaş Pond were found as W = 0.031 x L 2.97 for which is collected from the Dikilitaş Pond, were presented

3730 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

according to the gender of group with length and weight. for comparison of data of the crayfish in Dikilitaş Pond The calculated data observed from the artificial neural net- with length-weight relation and ANNs method. works and length-weight relations. Table 4 were prepared

120 160 2.90 W = 0.041 L 2.88 100 W = 0.032 L 140 r2 =0.96 120 r2 = 0.97 80 n=130 100 n=195 60 80 40 60 Weight (g) Weight

Weight (g) Weight 40 20 20 0 0 0 50 100 150 200 0 50 100 150 200 Length (mm) Length (mm) (a) (b)

160 3.02 140 W = 0.022 L 120 r2 = 0.96 100 n=325 80 60

Weight (g) Weight 40 20 0 0 50 100 150 200 Length (mm) (c) FIGURE 4 - Length–weight relationships in female (a), male (b) and all gender (c)

TABLE 3 - Pearson correlation coefficients between metric properties.

TL TW CL AL Aw Cw ChL Chw TL 1.000 TW 0.901** 1.000 CL 0.758** 0.796** 1.000 AL 0.658** 0.496** 0.377** 1.000 Aw 0.353** 0.256** 0.366** 0.390** 1.000 Cw 0.300 0.289 -0.007 0.332 0.043 1.000 ChL 0.350** 0.375** 0.435** 0.123* 0.247** 0.284** 1.000 Chw 0.592** 0.652** 0.487** 0.265** 0.194** 0.466** 0.484** 1.000 ** correlation is significant at the 0.01 level; * correlation is significant at the 0.05 level;

FIGURE 5 - Relationship between observed and forecast values for TL - TW

3731 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

the females while mean AL and Cw of the female individuals was greater than those of the males (Table 1). TW, Cw, ChL and Chw of males was greater than those of females in Keban Dam Lake [27]. CL, TW, Cw and ChL of males were greater than females, while AL, TL, AW and Chw of females are greater males in Eğirdir Lake [28]. TW, ChL and Chw of males are greater than females in İznik Lake and Hirfanlı Dam Lake [28]. TW, TL, CL, Cw, ChL of males are greater than females, while AL, Aw of females are greater than males [39]. TL, TW, CL, AL, Cw, Aw of females are greater than males, while Chw of males are greater than females [29]. Variation in morphometric traits may also be largely

affected by environmental factors such as eding, behavior, FIGURE 6 - Performans of ANNs between forecast values for TL - foraging efficiency, and the availability and quality of food TW resources [8]. Environmental conditions influence crusta- cean growth by effecting molt intervals and incremental in- The number of males were greater than females in Dik- creases in length and weight. The relationship between ilitaş Pond. Some researcher reported that the males were body length and weight is an important and widely used greater than females [24, 29, 31, 38], while some re- equation in fishery studies, fish length being the easiest pa- searcher reported that the females were greater than males rameter to measure, particularly in the field. During their (Table 5) [23, 26-28]. development, crustaceans are known to pass through stages The average lengths of males and weight was higher in their life history which are defined by different length- than females in Dikilitaş Pond. Some researchers found weight relationships. The parameter b is characteristic of that the average length and weight of males were higher species and generally does not vary significantly through- than females [23, 24] while some researchers found that the out the year, unlike the parameter, which may vary daily, average length and weight of females were higher than seasonally, between different habitats, water temperature males [28, 29]. It was found that the average length of fe- and salinity, sex, food availability, differences in the num- males was higher than the average length of males [38]. It ber of specimens examined, as well as in the observed was found that the average length and weight were higher length ranges of the species 195 caught [40]. than in some literature [23,24, 28, 29, 38]. The slope (b) values of the length–weight relationship It was determined that mean TL, TW, CL, Aw, ChL in both gender is found as a 3.09. It was found intercept of and Chw of the male individuals was greater than those of the relationship (a) as a 0.02481870. In the present study,

TABLE 4 - Observed and calculated values for ANNs and length-weight relation.

Type* Observed data Forecast MAPE (%) L W Relation MAPE (%) Sex (1) - (2) 1 2 1 2 1 2 1 2 1 2 ♀ 110.6 36.64 109.9 37.18 0.63 1.47 109.6 37.52 0.90 2.40 TL (1) TW (2) ♂ 113.5 53.77 115.9 52.41 2.11 2.53 116.4 50.98 2.56 5.19 ♀♂ 112.3 46.86 113.0 46.94 0.62 0.17 114.1 45.41 1.60 3.09 ♀ 52.0 36.64 51.5 35.68 0.96 2.62 52.6 35.83 1.15 2.21 CL (1) TW (2) ♂ 56.5 53.77 56.7 52.81 0.35 1.79 57.2 52.6 1.24 2.18 ♀♂ 54.7 46.86 54.6 45.96 0.18 1.92 55.6 45.22 1.65 3.50 ♀ 52.0 110.6 51.2 109.2 1.54 1.27 52.0 110.4 0.00 0.18 CL (1) TL (2) ♂ 56.5 113.5 56.6 113.2 0.18 0.26 56.7 113.2 0.35 0.26 ♀♂ 54.7 112.3 54.4 111.9 0.55 0.36 54.8 112.1 0.18 0.18 ♀ 57.5 36.64 56.8 37.18 1.22 1.47 59.9 35.63 4.17 2.76 AL (1) TW (2) ♂ 56.4 53.77 56.7 55.22 0.53 2.70 58.4 52.54 3.55 2.29 ♀♂ 56.8 46.86 56.8 48.21 0.00 2.88 61.6 44.86 8.45 4.27 ♀ 65.2 36.64 64.8 37.01 0.61 1.01 66.5 35.91 1.99 1.99 ChL (1) TW (2) ♂ 83.1 53.77 83.3 52.1 0.24 3.11 47.1 51.93 43.32 3.42 ♀♂ 75.9 46.86 75.2 46.3 0.92 1.20 89.9 45.04 18.45 3.88 Mean MAPE 0.46 1.30 6.07 2.98 (%) *Type (1) – (2): is a description of the parameters used in the calculation for ANNs and and length-weight relation.

3732 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 5 - Comparison of growth parameters in different locations.

Regression parameters Sex ratio References Sex TL TW (F/M) a b r2 Karabatak and Tüzün [23] ♀ 104.45 31.92 0.00002 3.05 - 1.22/1 Mogan Lake ♂ 105.44 36.98 0.00001 3.18 Kuşat and Bolat [24] ♀ 92.67 29.18 0.000087 2.78 - 0.60/1 Eğirdir Lake ♂ 107.21 52.91 0.000019 3.13 Köksal et al. [26] ♀ 102.04 32.24 0.00002 3.08 0.99 1.08/1 Dikilitaş Pond ♂ 102.50 33.11 0.00005 3.01 0.99 ♀ 103.29 32.17 2.11 0.82 Harlioğlu and Harlioğlu [28] - 1.14/1 ♂ 101.81 33.07 2.51 0.92

♀ 104.54 29.19 2.66 0.94 Eğirdir Lake - 1.15/1 ♂ 100.47 29.34 2.72 0.94 İznik Lake ♀ 105.93 19.36 2.22 0.78 Hirfanlı Dam Lake - 0.82/1 ♂ 104.76 20.17 3.66 0.89 Balık et al. [38] ♀ 92.88 24.19 0.00002 3.06 0.97 0.49/1 Demirköprü Dam Lake ♂ 90.18 25.43 0.00001 3.27 0.98 ♀ 89.07 21.85 0.0003 2.94 0.99 Berber and Balık [29] ♂ 0.53/1 82.12 19.57 0.0003 2.98 0.97 Manyas Lake Berber and Balık [30] ♀ 20.62 0.0003 2.96 0.94 - 0.68/1 Apolyont Lake ♂ 21.92 0.0002 3.03 0.95 ♀ 0.00003 2.96 0.87 Deniz et al. [31] - - 0.84/1 ♂ 0.00003 2.97 0.88 Inland water in Turkey ♀ 0.00007 2.78 0.88 - - 0.91/1 ♂ 0.00001 3.21 0.93 Eğirdir Lake ♀ 0.0001 2.71 0.90 Hirfanlı Dam Lake - - 0.57/1 ♂ 0.000008 3.28 0.92 Keban Dam Lake ♀ 0.0001 2.42 0.75 Porsuk Dam Lake - - 0.87/1 ♂ 0.000005 3.38 0.83 ♀ 126.00 52.80 0.032933 2.91 0.96 This Study ♂ 128.50 75.52 0.041409 2.89 0.97 0.66:1 ♀♂ 127.50 66.43 0.022299 3.02 0.96

growth was a negative allometry for females and males, while The estimated length-weight relation are model de- growth was a positive allometry for all gender. It is found that pendent; as a result, model selection uncertainty may be b values are smaller than Karabatak and Tüzün [23]; Köksal quite high in certain data sets. Ignoring model selection un- et al. [26]; Balık et al. [38] (Table 5). Similar results were re- certainty may cause substantial overestimation of the pre- ported for Harlioğlu and Harlioğlu [28] in Eğirdir and İznik cision and estimation of the confidence intervals of the pa- Lake; Berber and Balık [29] in Manyas Lake. rameters below the nominal level. This uncertainty has se- The reason for these differences may be factors such rious implications, e.g., when comparing the growth pa- as the environmental, food, population density and the se- rameters of different crayfish populations. lectivity of the traps or fykenets used in the studies. In ad- Length-weight relation and artificial neural network dition, male crayfish were heavier than females. This dis- MAPE results were examined. It was found MAPE value parity is due primarily to the accelerated development of of the forecast of ANNs as a 0.46 and 1.30, while MAPE the male chelae with the onset of sexual maturity, while the value of relationship results as a 6.07 and 2.98 for length – chelae of the females remain isometric throughout their life weight of all gender. According to Table 4; artificial neural [41]. Furthermore, males are rougher and more thick-set networks gives better results than length-weight relation. than females [9]. Lengths of both male and female crayfish In conclusion in the study the use of ANNs as a fore- ranged from 40 to 150 mm. The catching of crayfish casting tool was found to provide fairly good results. Arti- smaller than 90 mm in total length is prohibited in all Turk- ficial neural networks can be an alternative as a evaluated ish lakes [42]. for growth estimation. The relationships between carapace length and weight The morphometric characters could be helpful in com- have many uses. They are, for example, indicators of con- paring the same species in different locations. It was shown dition and can be used to calculate biomass and to estimate that the study of morphometric characters could be used to the recovery of edible meat from crayfish of various sizes. describe populations. This study also provides some infor- On the other hand, body weight and total length, carapace mation on the length-weight relationships that would be length and carapace width are the most frequently used di- useful for sustainable fisheries management in Dikilitaş mensions in the study of crustaceans [43]. There are several Pond. In addition, for sustainable economic yield, the cray- possible explanations for such a pattern but the most likely fish population should be carefully monitored in the future in this case being different growth rates between sexes. as well.

3733 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

ACKNOWLEDGMENTS [16] Mann, K.H. (1993) Physical oceanography, food chains, and fish stocks: a review. ICES J Mar Sci 50:105–119 We would like to thank all the referees, who have [17] Obach, M., Wagner, R., Werner, H., Schmidt, H.H. (2001) added value to our paper with their thorough reviews and Modelling population dynamics of aquatic insects ith artificial neural networks. Ecological Modeling, 146:207-217. recommendations. [18] Tureli Bilen C., Kokcu, P., Ibrikci, T. (2011) Application of Artificial Neural Networks (ANNs) for Weight Predictions of The authors have declared no conflict of interest. Blue Crabs (Callinectes sapidus Rathbun, 1896) Using Predic- tor Variables. Mediterranean Marine Science 12(2):439-446. [19] Suryanarayana I, Braibanti, A., Rao, R.S., Ramamc, V.A., Su- REFERENCES darsan, D. and Rao, G.N. (2008) Neural networks in fisheries research. Fisheries Research 92:115–139. [1] Füreder, L., Oberkofler, B., Hanel, R., Leiter, J. and Thaler, B. [20] Benzer, R. (2015) Population Dynamics Forecasting Using (2003) The freshwater crayfish Austropotamobius pallipes in Artificial Neural Networks. Fresenius Environmental Bulletin. South Tyrol: Heritage species and bioindicator. Bull. Fr. Pêche 24(2):460-466. Piscic., 370-371:79−95. [21] Sun L, Xiao, H., Li, S. and Yang, D. (2009) Forecating fish stock [2] Souty-Grosset, C., Holdrich, D.M., Noel, P.Y., Reynolds, J.D., recruitment and planning optimal harvesting strategies by using Haffner, P. (2006) Atlas of crayfish in Europe. Publications neural network. Journal of Computers 4(11):1075-1082. Scientifiques du MNHN-Paris. [22] Brosse, S., Guegan, J., Tourenq, J. and Lek, S. (1999) The Use [3] TÜİK (2011) Turkey Statistical. Fisheries Statistics. of Artificial Neural Networks to Assess Fish Abundance and spatial occupancy in the littoral zone of a mesotrophic lake. [4] Harlioğlu M.M. (2004) The present situation of freshwater Ecological Modelling 120(2-3):299-311. crayfish, Astacus leptodactylus (Eschscholtz, 1823) in Turkey. Aquaculture, 230:181–187. [23] Karabatak, M. and Tüzün, İ. (1989) Some aspects of the crayfish (Astacus leptodactylus Eschscholtz, 1823) population in [5] Harlioğlu, M.M. and Harlioglu, A.G. (2006) Threat of non-na- the Mogan Lake, Ankara (in Turkish). Akdeniz Üniversitesi, tive crayfish introductions into Turkey: global lessons. Rev. Su Ürün., Müh., Derg., 2:1-34. Fish Biol. Fisher., 16:171–181. [24] Kuşat, M. and Bolat, Y. (1995) Length-weight distribution of [6] Güner, U. and Harlioğlu, M.M. (2010) Status of freshwater the crayfish (Astacus leptodactylus salinus, Esch., 1823) in crayfish distribution in Thrace region of Turkey. Reviews Eğirdir Lake an a study on the crayfish plaque (in Turkish). Fish. Sci., 18:1–6. E.U. Journal of Fisheries & Aquatic Sciences 12(1-2):69-74. [7] Primavera, J.H., Parado-Estepa, F.D. and Lebata, J.L. (1998) [25] Erdem, Ü., Cebeci, M., Selçuk, S., Tunç, N., Özbay, A. and Morphometric relationship of length and weight of giant tiger Çildem, B. (2001) Some bio-echologic features of crayfish prawn Penaeus monodon according to life stage, sex and (Astacus leptodactylus Eschscholtz, 1823) in Iznik Lake (in source. Aquaculture, 164: 67–75 Turkish). XI. Ulusal Su Ürünleri Sempozyumu 04-06 Eylül, Hatay, 449-456. [8] Lindqvist, O.V. and Lahti, E. (1983) On the sexual dimorphism and condition index in the crayfish Astacus astacus L. [26] Köksal, G., Korkmaz, A.Ş. and Kırkağaç, M. (2003) Investi- in Finland. Freshwater Crayfish 5: 3-11. gation of freshwater crayfish population (Astacus leptodactylus Eschscholtz, 1823) in Ankara Dikilitaş Pond. A.Ü., Ziraat [9] Skurdal, J. and Qvenild, T. (1986) Growth, maturity, and fe- Fakültesi, Tarım Bilimleri Dergisi, 9(1):51-58. cundity of Astacus astacus in lake Steinfjorden, S.E. Norway. Freshwater Crayfish 6:182-186. [27] Harlioğlu, M.M. (1999) The Relationships between Length- Weight, and Meat Yield of Freshwater Crayfish, Astacus lep- [10] Gillet, C. and Laurent, P.J. (1995) Tail length variations todactylus Eschscholtz, in the Ağın Region of Keban Dam among noble crayfish (Astacus astacus (L)) populations. Lake. Tr. J. of Zool. 23: 949-958. Freshwater Crayfish 10: 31-36, 1995. [28] Harlioğlu, M.M. and Harlioğlu, A.G. (2005) The comparison [11] Tosunoğlu, Z., Aydın, C., Özaydın, O. and Leblebici, S. of morphometric analysis and meat yield contents of freshwa- (2007) Trawl codend mesh selectivity of braided PE material ter crayfish, Astacus leptodactylus (Esch 1823) caught from for Parapenaeus longirostris (Lucas, 1846) (Decapoda, Pe- İznik, Eğirdir Lakes and Hirfanlı Dam Lake, Science and En- naeidae). Crustaceana, 80: 1087-1094. gineering Journal of Fırat University, 17(2):412-423. [12] Yanez, E., Plaza, F., Gutierrezestrada, J.C., Rodriquez, N., [29] Berber, S. and Balık, S. (2006) Determination of traits some Barbieri, M.A., Pulido-Calvo, I. and Borquez, C. (2010) An- growth and morphometric of crayfish (Astacus leptodactylus chovy (Engraulis ringens) and sardine (Sardinops sagax) Eschscholtz, 1823) at Manyas Lake (Balıkesir), E.U. Journal abundance forecast off northern Chile: A multivariate ecosys- of Fisheries & Aquatic Sciences. 23(1-2):83-91. tem neural network approach. Oceanography, 87: 242-250. [30] Berber, S. and Balık, S. (2009) The length-weight relation- [13] Maravelias, C.D., Haralabous, J. and Papaconstantinou, C. ships, and meat yield of crayfish (Astacus leptodactylus (2003) Predicting demersal fish species distributions in the Eschcholtz, 1823) population in Apolyont Lake (Bursa, Tur- Mediterranean Sea using artificial neural networks. Marine key). Journal of Fisheries Sciences, 3(2):86-99. Ecology, 255:249-258. [31] Deniz, T.B, Aydın, C. and Ateş, C. (2013) A study on some [14] Joy, K.M. and Death, R.G. (2004) Predictive modelling and morphological characteristics of Astacus leptodactylus (Esch- spatial mapping of freshwater fish and decapod assemblages scholtz 1823) in seven different inland waters in Turkey. J. using GIS and neural Networks. Freswater Biology, Black Sea/Mediterranean Environment. 19(2):190˗205. 49(8):1306-1052. [32] Rhodes, C.P. and Holdich, D.M. (1979) On size and sexual [15] Mastrorillo, S., Lek, S., Dauba, F. and Belaud, A. (1997) The dimorphism in Austropotamobius pallipes (Lereboullet) - A use of artificial neural networks to predict the presence of step in assessing the commercial exploitation potential of the small-bodied fish in river. Freshwater Biology 38: 237-246. native British freshwater crayfish. Aquaculture 17: 345-358.

3734 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[33] Ricker, W.E. (1973) Linear regressions in fishery research. J. of Fisheries Research Board of Canada 30:409-434. [34] Hopgood, A.A. (2000) Intelligent systems for engineers and scientists, Florida: CRC Press. [35] Rumelhart D.E., Hinton, G.E. and Williams, R.J. (1986) Learning internal representations by error propagation in Par- allel Distributed Processing. Explorations in the Microstruc- ture of Cognition MIT Press, 1:318-362.

[36] Krenker A, BešTer, J. and Kos, A. (2011) Introduction to the Artificial Neural Networks, Artificial Neural Networks - Methodological Advances and Biomedical Applications, Prof. Kenji Suzuki (Ed.), ISBN: 978-953-307-243-2.

[37] Haykın, S. (1999) Neural Networks: A Comprehensive Foun- dation, Perenctice Hall, New Jersey.

[38] Balık, S., Ustaoğlu, M.R., Sarı, H.M. and 4Berber, S. (2005) Determination of traits some growth and morphometric of crayfish (Astacus leptodactylus Eschscholtz, 1823) at Demirköprü Dam Lake (Manisa). E.U. Journal of Fisheries & Aquatic Sciences, 22(1-2):83-89. [39] Güner, U. (2008) Some Morphometric Properties and Growth Parameter of crayfish (Astacus leptodactylus Eschscholtz, 1823) at Kavaklı Lake (Edirne, Meriç), Research Journal of Biological Sciences, 181:37-42. [40] Tesch, F.W. (1971) Age and growth. In: Methods for Assess- ment of Fish Production in Fresh Waters (ed., W. E. Ricker), Blackwell Scientific Publications, Oxford, 99˗130 pp

[41] Romaire, R.P., Forester, J.S. and Avault, J.W.Jr. (1977). Length˗weight relationships of two commercially important crayfishes of the genus Procambarus. Freshwater Crayfish 3: 463˗470.

[42] TKB. (2002) In the seas and inland waters of Commercial Fisheries 2000-2002 Hunting Season Hunting Regulating Cir- cular No. 34/1. TKB General Directorate of Protection and Control, Ankara. [43] Atar, H.H. and Seçer, S. (2003) Width/length˗weight relationships of the blue crab (Callinectes sapidus Rathbun 1896) population living in Beymelek Lagoon Lake. Turk. J. Vet. Anim. Sci. 27: 443˗447.

Received: January 27, 2015 Revised: April 13, 2015 Accepted: May 26, 2015

CORRESPONDING AUTHOR

Assist. Prof. Dr. Semra Benzer Gazi University Education Faculty Department of Science Education Ankara TURKEY

Phone: (+90) 5056603181 E-mail: [email protected]; [email protected]

FEB/ Vol 24/ No 11a/ 2015 – pages 3727 - 3735

3735 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

FIRST IDENTIFICATION OF THE CYLINDROSPERMOPSIN (CYN)-PRODUCING CYANOBACTERIUM Cylindrospermopsis raciborskii (WOLOSZYŃSKA) SEENAYYA & SUBBA RAJU IN SERBIA

Nevena B. Đorđević1, Snežana B. Simić1,* and Andrija R. Ćirić2

1University of Kragujevac, Faculty of Science, Department of Biology and Ecology, 34000 Kragujevac, Serbia 2University of Kragujevac, Faculty of Science, Department of Chemistry, 34000 Kragujevac, Serbia

ABSTRACT 90 years, the presence and blooming of cyanobacteria has been described for different aquatic ecosystems of Serbia. In this study, is reported the first occurrence of cylin- The research so far has shown that the accumulation drospermopsin (CYN) in Serbia. CYN is a newly emerging blooming most commonly occurs due to the excessive mul- carcinogenic hepatotoxic alkaloid, originally identified in tiplication of the genera Anabaena, Aphanizomenon, Mi- tropical cyanobacteria Cylindrospermopsis raciborskii crocystis, Oscillatoria and Planktothrix [7]. Previous stud- (Woloszyńska) Seenayya et Subba Raju. The research was ies of cyanotoxins in Serbia are related to the presence and conducted on a monthly basis from June 2012 to February detection of microcystins [8-11]. In Serbia C. raciborskii 2013 in Aleksandrovac Lake, Serbia (42°29′22″ N, 21°53′54″ was detected in a salt marshes and a fish ponds [12, 13] E). C. raciborskii showed the value of abundance between while the occurrence of massive development of this spe- 0.1 x 102 and 2.38 x 106 trichomes ml-1. According to cies was detected in an eutrophic lowland river [14] and an HPLC-PDA analysis CYN concentration ranged between alkaline microaccumulation [15-17]. Since C. raciborskii 1.91 and 24.28 g l-1. Maximum concentration of CYN was produces CYN, it is a highly-ranked species on the Watch in November 2012, and the fish kill occurred in the follow- List of toxic cyanobacteria for water managers [4]. ing month. The massive presence of C. raciborskii in Aleksan- drovac Lake [16, 17] and the use of the lake (fishing, swimming, sports and recreation activities) guided our choice of

KEYWORDS: Cylindrospermopsin, Cylindrospermopsis raci- the research related to CYN. borskii, HPLC, Serbia. The aim of the present study is to describe the first detection and quantification of CYN in Serbia using HPLC/PDA technique and it also demonstrates a poten-

1. INTRODUCTION tial hazard to the environment (fish kill) due to the presence of CYN. The cyanotoxin cylindrospermopsin (CYN) is an alkaloid with hepatotoxic, cytotoxic, dermatotoxic, genotoxic and carcinogenic properties [1]. Because of that the presence 2. MATERIALS AND METHODS of CYN in surface waters and reservoirs is observed as a potential risk for human health [2]. CYN was first isolated from Aleksandrovac Lake is an oligosaline, shallow lake Cylindrospermopsis raciborskii (Woloszyńska) Seenayya et (maximum depth 4 m) in the southernmost part of Serbia Subba Raju by Ohtani et al. [3]. CYN-producing species of (42°29′22″ N, 21°53′54″ E). It is an artificial reservoir the genera: Cylindrospermopsis, Aphanizomenon, Ana- formed in 1963-1964; the lake was emptied in 2009. The baena, Umezakia, Raphidiopsis and Lyngbya have been de- lake was completely restored and refilled in 2010. This res- tected worldwide [4, 5]. ervoir was originally used for the irrigation of agricultural areas and later for sports and recreational purposes and, Kinner [6] showed a map provided with the global dis- above all, for fishing [15-17]. Sampling from Aleksandrovac tribution of blooms of CYN-producing species, but the ter- Lake was conducted monthly, from June 2012 to February ritory of Serbia was not marked on the map. Over the last 2013. Twenty seven samples of phytoplankton were gath- ered by standard methods - phytoplankton net (net frame * Corresponding author 25 cm, mesh net e.g. 22 µm) and Ruttner bottle (2 l) in all

3736 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

aspects. All samples were immediately preserved in 4 % of World Health Organization [21]. Sample of 5 l water fil- formaldehyde. C. raciborskii was identified on morpholog- tered through pre-weighed Glass fibre filters 70 mm, GF/C, ical grounds on a Nikon Eclipse E100 microscope at 40- 1.2 μm under the vacuum. Filter papers were dried using an 100x magnification. The identification was performed by oven, under the temperature below 50°C until constant using literature [18]. weight. For cyanotoxin detection, dry samples on filter papers The quantitative analysis of phytoplankton was made (0.699 g) extracted with ten times higher volume of meth- by using Utermöhl method [19] with a Carl Zeiss inverted anol then was weight of cells. Cells in methanol were soni- microscope (No. 724 395) and it is expressed as the number cated for 3 min in 100 Hz and then allow cells to extract of trichomes per ml. for 1 h in dark on room temperature. The suspension was then filtered through Glass fibre filters and the solvent was Simultaneously with gathering phytoplankton samples removed under vacuum in a rotary evaporator, yielding in the field and samples for CYN detection, the following 0.093 g of the dry crude extract. Extract suspended in Mili- physical and chemical parameters were measured: water Q water (Watford, UK) and dissolved by sonication for temperature (ºC), pH, oxygen concentration (mg l-1), satu- 5 min. Trifluoracetic acid (TFA), 0.1% v/v (Sigma-Aldrich ration (%), conductivity (µs cm-1), phosphorus concentra- St. Louis, USA) was added after the centrifugation. Sam- tion (mg l-1) and water transparency (Secchi disc) (m) [20]. ples were mixed for 1 h and decanted at room temperature Physical and chemical parameters, phytoplankton samples for 3 h, according to Welker et al. [22]. and samples for CYN detection were taken from three sites (dam, middle and end of the lake) and at greater depths The prepared samples were analysed according to the (surface 0–0.3 m, 2 m and 3.5-4 depending on the depth of protocol of Welker et al. [22] and Bláhová et al. [23]. The the lake at the moment of sampling), but because of both samples were analysed by HPLC (Shimadzu, Kyoto, Japan) the size of the lake and insignificant deviations mean val- consisting of a DGU-20A3 degasser, LC-20AT analytical ues are given for all parameters. Samples of water bloom pumps, 7125 injector and a diode array detector SPD-M20A biomass of C. raciborskii were collected from Aleksan- and a CBM-20A system controller on a Supelcosil ABZ Plus, drovac Lake when blooming observed. Sampling of the bi- 150 x 4.6 mm, 5 μm column (Supelco) at 30C. The binary omass was done 1th November, 2012. gradient of the mobile phase consisted of (A) H2O + 0.05 % TFA and (B) methanol + 0.05 % TFA (linear increase from For the analysis of CYN, water samples were collected 0 % B at 0 min to 50 % B at 20 min), the flow rate was 1 ml in 1-litre bottles and transported to the laboratory in refrig- min-1. The chromatograms at 262 nm were recorded with a erators. The samples were stored in the dark at -20ºC, until SPD-M20A diode array detector. CYN was identified by analysis. characteristic UV absorption spectra (200 - 300 nm) and re- Chemicals of appropriate purity were used in the investi- tention time and quantified using external calibrations 1 - gation (HPLC grade). The CYN standard was adopted from 60 mg l-1 (LOD = 0.188, LOQ = 0.572 mg l-1), a injection Enzo Life Sciences (Lausen, Switzerland). Trifluoroacetic volume 20 µl. The chromatographic data were processed acid ( 99.0 %) was purchased from Sigma-Aldrich (St. using LC Solution computer software (Shimadzu). Louis, USA). Methanol (HPLC grade was from J.T. Baker, Deventer, Holland), Millipore Milli-Q system (Watford, UK). Water samples were prepared according to the proto- 3. RESULTS AND DISCUSSION col of Meriluoto and Codd [5], suggesting that solid phase extraction (SPE) is necessary to concentrate the toxin to Physical and chemical data for Aleksandrovac Lake concentrations capable of being detected by HPLC in order to during the study period are summarized in Table 1. Maxi- determine the concentration of CYN in the extracellular frac- mum water temperature was 27.9ºC in July and minimum tion of filtered environmental waters. Water samples (1 l) was 3.1ºC in December. High pH levels were recorded with were filtered through GF/C filter. A Supelco (Bellefonte, PA, the maximum value of 9.36 in September. Total phospho- USA) vacuum tank and Supelco LC-18 (500 mg, 6 ml Super- rus levels varied, the maximum value of 0.69 mg l-1was clean) cartridges connected to a PGC cartridge (200 mg, 3 ml measured in June, and there was a gradual decrease until for Enviro Clean) (UCT, Bristol, PA) were used for the September and after that it was uniformly measured at SPE. The samples were applied to the cartridges onto the <0.02 mg l-1. Dissolved oxygen values were between 7.7 conditioned SPE system with 10 ml methanol containing and 11.6 mg l-1. Water transparency ranged from 1.3 m to 0.1 % (v/v) TFA followed by 10 ml of water. The car- 0.2 m and it decreased transitioning from early summer to tridges were not allowed to dry during condition and sam- late autumn. ple application. After that, the PGC cartridge should be air- Cyanobacteria C. raciborskii (Figure 1) was present in dried. Cylindrospermopsin eluted with 3 ml 0.1 % (v/v) the phytoplankton in Aleksandrovac Lake over the entire TFA in methanol. Methanol solution of sample was evap- study period (Table 2). The abundance of C. raciborskii var- orated under nitrogen and reconstructed in 500 l of Milli- ied through months; in June and July it was very low if com- Q water. pared to the autumn months, while the highest abundance of Cyanotoxins extraction from C. raciborskii biomass for C. raciborskii was recorded in the month of October (2.38 x cyanotoxin analysis were performed according to the protocol 106 trichomes ml-1). In that period of blooming C. raci-

3737 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 1 - Mean values of physical and chemical parameters at Aleksandrovac Lake June 2012 - February 2013

Month Jun Jul Aug Sep Oct Nov Dec Feb Water temperature (ºC) 26.8 27.9 26.3 23.26 17.16 12.26 3.1 5.1 pH 8.68 8.47 9.02 9.36 8.68 9.05 8.21 8.5 Oxygen (mg l-1) 9.68 11.60 10.25 8.07 9.04 9.05 7.7 8.1 Saturation (%) 129 157.6 135.4 98.53 100.76 89.9 60.2 65.3 Phosphorus (mg l-1) 0.69 0.56 0.42 0.46 <0.02 <0.02 <0.02 <0.02 Conductivity (µs cm-1) 434 436 449 473 475 478 428 430 Water transparency (m) 1.3 0.8 0.35 0.3 0.25 0.20 ice 0.3

Lake both individually and in series (number of 2-4) [15, 17]. It is known that under favourable conditions akinetes have the potential of germination [17, 25]. The conditions in this oligosaline, alkaline, shallow lake, where the water temperature in summer months increased up to 26.2ºC, favoured the development of this species, so it reappeared in Septem- ber 2011 in large populations (2.37 x 102 trichomes ml-1) [17]. Unchanged and still suitable conditions (Table 1) remained favourable so that the species occurred in June of the next year, and its abundance significantly increased each month in 2012 (Table 2). In this study the massive development of C. raciborskii was noticed during the period and after the period of high temperature (over 26ºC) (Table 1 and 2). High temperature is one of the most important factor for massive development of C. raciborskii [26]. This species prefers environments with temperatures ranging from 20 to 35ºC, but it also has a wider temperature tolerance range [27]. At lower temperature values there was a gradual reduction in the abundance of this species in Aleksandrovac Lake (Ta- bles 1 and 2).

FIGURE 1 - Photomicrographs of Cylindrospermopsis raciborskii High pH values also favoured the development of C. from Aleksandrovac Lake. Scale bar represent 10 m. Legends: h- heterocyst, a-akinete. raciborskii. It occurs in water ecosystems with high pH values – 8.0-8.7 [26]. borskii looked like a brown-green mash. The abundance Phosphorus also appears to play an important role in was significantly reduced over winter months (Table 2). the dominance of C. raciborskii. According to Padisák C. raciborskii is an invasive species that has been ob- [26], inorganic phosphorus content, as well as total phos- served in many tropical, subtropical and, recently, temper- phorus concentrations, can vary within a wide range in ate regions [24]. In Serbia C. raciborskii was detected in a lakes where C. raciborskii occurs in abundant populations, salt marshes and a carp ponds [12-13] while the occurrence as presented in this study. In the waters of Aleksandrovac of massive development of this species was detected in an Lake phosphorus was reduced gradually starting June. It eutrophic river [14]. Simić et al. [15] provided first data on was observed that with increasing abundance, the value of the occurrence of this species in the Aleksandrovac Lake, phosphorus started to decline from late spring through au- just three months after the restoration and filling the micro- tumn. After reaching the maximum value abundance in Oc- acumulations with fresh water (September 2010). The tober, there was a large reduction in phosphorus concentra- abundance of C. raciborskii was low (37 trichomes ml-1). tion. C. raciborskii has a high affinity and storage capacity Akinetes were observed on trichomes in the Aleksandrovac for phosphorus if compared to other cyanobacteria [28].

TABLE 2 - Mean values of abundance of C. raciborskii and distribution of CYN at Aleksandrovac Lake (Serbia) June 2012 - February 2013

Jun Jul Aug Sep Oct Nov Dec Feb C. raciborskii (trichomes ml-1) 1.2 x 103 1.7 x 103 6.8 x 104 8.7 x 104 2.38 x 106 2.13 x 105 1.1 x 103 0.1 x 102 CYN (g l-1) Not detected Not detected 1.91 9.75 10.70 24.28 5.25 4.22

3738 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

These adaptations allow C. raciborskii to grow in low raciborskii was 98% in relation to the entire phytoplankton levels of phosphate [25] and may provide a competitive ad- of the Aleksandrovac Lake. Because of the absolute domi- vantage to C. raciborskii. nance of this species, it was possible to collect biomass of C. After all, recent research has found that the one of the raciborskii. HPLC-determined CYN content was 6.65 μg -1 main drivers behind the dominance of the cyanobacterium CYN mg dw. C. raciborskii is also in low light requirements [29]. C. Maximum CYN concentration in the water was in No- raciborskii blooms in Aleksandrovac Lake under low light vember 24.28 g l-1, while the maximum value of abun- levels (Table 1 and 2). dance was in the previous month. This was expected be- In this study, the cyanotoxin CYN was first detected cause the concentration of cyanotoxins significantly in- and quantified in Serbia, using HPLC/PDA method (Table creased as a defence mechanism in stressful conditions 2). CYN was identified by retention time and characteristic (lack of nutrients/light) [30], and especially because of the UV absorption spectra. The retention time of CYN stand- cell age and after the extinction of cyanobacteria and the ard was 5.392 min (Figure 2). The chromatogram of a sam- release of cell content in the water. ple from Aleksandrovac Lake showed that retention time Previous research [31, 32] showed that environmental was 5.371 min (Figure 3). The concentrations of CYN concentrations of CYN in freshwater habitats can vary con- ranged between 1.91 - 24.28 g l-1. CYN was not detected siderably. Cellular abundance of CYN-producing cyano- only in June and July, while it was found in the water sam- bacteria can differ and does not necessarily correlate with ples taken over the remaining months. CYN concentrations [33]. The value of abundance and CYN concentrations in- A high toxin concentration of 589 μg l-1 was detected creased until November, when the abundance started to de- in an aquaculture pond with the abundance up to 3.25 x 105 crease. In November the representation of the species C. trichomes ml-1 [34]. In Lake Albano (Italy) concentrations

FIGURE 2 - Chromatogram of cylindrospermopsin standard (5 mg l-1) (262 nm, retention time 5.392 min) and UV spectra

FIGURE 3 - Chromatogram of sample - Aleksandrovac Lake (262 nm, retention time 5.371 min) and UV spectra (retention time 5.371 min)

3739 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

of CYN was in the range 1.6 - 15 μg l-1 while toxic cyano- locations clearly show that C. raciborskii is now a species bacteria had abundance between 0.1 x 106 - 30 x 106 tri- of concern in recreational, sport and other water bodies in chomes ml-1 [35]. Low concentrations of toxin of 1 μg l-1 Serbia and it extends its already known geographic distri- were reported with the blooms of 9 x 104 trichomes ml-1 bution. The first detection of CYN by HPLC/PDA provides [36]. McGregor and Fabbro [37] also discussed the corre- evidence for the presence of this toxin in Serbia. Further lation between abundance and CYN concentrations. Low monitoring of toxin-producing cyanobacteria is needed, densities with an abundance of 2 x 104 trichomes ml-1 were both worldwide and in Serbia, to minimize possible harm- associated with 1 μg l-1 CYN. The relationship between cell ful effects on human health and the environment. Monitor- number and toxin concentration varied in reservoirs and ing of water bodies that humans use in agriculture or for blooms. The majority of the reservoirs had C. raciborskii entertainment and recreation is necessary by all means. abundance of around 3 x 104 - 10 x 104 with recorded CYN concentrations below 10 μg l-1 [37]. This was exactly the case in this research, suggesting that the relationship (CYN concentrations - abundance) ACKNOWLEDGEMENTS varied through months. This investigation was supported by the Ministry of When cyanotoxins are released into the water, they can Education, Science and Technological Development of the be absorbed by different aquatic organisms. CYN may ad- Republic of Serbia – Project Grants TR 31011 and III versely affect all food components (phytoplankton, zoo- 43002. plankton and fish) [38, 39]. Human consumption of contaminated fish could therefore lead to significant health hazards. The authors have declared that no conflict of interest Fish are CYN-exposed directly by feeding on phytoplankton exists. or contaminated organisms and/or passively via their epithe- lium (gills and skin) when CYN is dissolved in the water [40].

In December 2012 fish kill occurred in Aleksandrovac Lake. The fish kill happened in one day, when 3 tones of cyprinid fish died. It happened after the period when the REFERENCES maximum CYN value was detected in the lake; in the same period the lake was completely frozen and oxygen concen- [1] Cruz, A.A., Hiskia, A., Kaloudis, T., Chernoff, N., Hill, D., Antoniou, G.M., He, X., Loftin, K., OˈShea, K., Zhao, C., Pe- tration was reduced (Table 1). Potential causes of fish death laez, M., Han, C., Lynch, J.T., Dionysios, D.D. (2013) A re- could be CYN in combination with other environmental view on cylindrospermopsin; the global occurrence, detection, factors [40]. A similar situation indicated that the toxicity toxicity and degradation of a potent cyanotoxin. Environ Sci of C. raciborskii caused the fish kill in 1990 and 1993 in Processes Impacts 15, 1979-2003. Rio Pequeno in Brazil [41]. Berry et al. [42] studied the [2] Chorus, I., Cavalieri, M. (2000) Monitoring bathing waters. developmental toxicity of CYN, as well as extracts from C. Chapter 10. A practical quide to the desing and implementation of assessments and monitoring programmes. Edited by Ja- raciborskii on zebra fish (Danio rerio) embryos. They con- mie Bartram and Gareth Rees, London. cluded that pure CYN was toxic only when it was injected directly into embryos. In contrast, direct immersion of em- [3] Ohtani, I., Moore, R.E., Runnegar, M.T.C. (1992) Cylin- drospermopsin: A potent hepatotoxin from the blue-green alga bryos in all the extracts resulted in both increased mortality Cylindrospermopsis raciborskii. J Am Chem Soc 114, 7941- and reproducible developmental dysfunctions. The occur- 7942. rence of toxic blooms or dense growths of cyanobacteria [4] WHO (1999) Toxic Cyanobacteria in Water: A guide to their has been related to fish and other animals mortalities and public health consequences, monitoring and management. human illness [1]. There are studies on different effects of Routledge, London, New York. CYN in fish (accumulation of CYN in fish, oxidative pa- [5] Meriluoto, J., Codd, G.A. (2005) Toxic; Cyanobacterial Mon- rameters and histological parameters) moreover, LD50 itoring and Cyanotoxin Analysis. Åbo Akademi University data for CYN has not been reported on fish [40]. Press, Finland. It is worth mentioning that CYN occurred at low con- [6] Kinner, S. (2010) Cylindrospermopsin: A Decade of Progress centrations in February after the period when the lake was on Bioaccumulation Research. Mar Drugs 8, 542-564. completely frozen (Table 2). This is in accordance with the [7] Svirčev, Z., Tokodi N., Drobac, D., Codd, G. (2014) Cyano- research indicating the stability of this toxin; CYN is stable bacteria in aquatic ecosystems in Serbia: effects on water qual- when exposed to a range of light intensities, temperatures ity, human health and biodiversity. Syst Biodivers 12(3), 261- 270. and pH [30]. [8] Simeunović, J., Svirčev, Z., Karaman, M., Knežević, P., Me- lar, M. (2010) Cyanobacterial blooms and first observation of microcystin occurences in freshwater ecosystems in Vojvo- 4. CONCLUSIONS dina region (Serbia). Fresen Environ Bull 19(2), 198-207. [9] Natić, D., Jovanović, D., Knežević, T., Karadžić, V., Bulat, Z., The occurrence and dominance of the invasive species Matović, V. (2012) Mikrocistin – LR u površinskim vodama C. raciborskii in the Aleksandrovac Lake as well as other reke Ponjavice. Vojnosanit Pregl 69(9), 753-758.

3740 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[10] Pantelić, D., Svirčev, Z., Simeunović, J., Vidović, M., [26] Padisák, J. (1997) Cylindrospermopsis raciborskii (Wolo- Trajković, I. (2013) Cyanotoxins: Characteristics, production szynska) Seenayya et Subba Raju, an expanding, highly adap- and degradation routes in drinking water treatment with refer- tive cyanobacterium: worldwide distribution and a review of ence to the situation in Serbia. Chemosphere 91, 421–441. its ecology. Arch Hydrobiol – Suppl 107(4), 563–593.

[11] Svirčev, Z., Simeunović, J., Subakov-Simić, G., Krstić, S., [27] Kling, J.H. (2009) Cylindrospermopsis raciborskii (Nosto- Pantelić, D., Dulić, T. (2013) Cyanobacterial Blooms and their cales, Cyanobacteria): a brief historic overview and recent dis- Toxicity in Vojvodina Lakes, Serbia. Int J Environ Res 7(3), covery in the Assiniboine River (Canada). Fottea 9, 45–47. 745-758. [28] Istvánovics, V., Shafik, H.M., M., Presing, M., Juhos, S. [12] Cvijan, M., Fužinato, S. (2011) The first finding of Cylin- (2000) Growth and phosphate uptake kinetics of the cyanobac- drospermopsis raciborskii (Woloszynska) Seenaya et Subba terium Cylindrospermopsis raciborskii (Cyanophyceae) in Raju (Cyanoprokaryota) in Serbia. Arch Biol Sci 63, 507–510. throughflow cultures. Freshwater Biol 43, 257–275.

[13] Cvijan, M., Fužinato, S. (2012) Cylindrospermopsis raci- [29] Burford, M.A. and Davis, T.W. (2011) Physical and chemical borskii (Cyanoprokaryota) – potential invasive and toxic spe- processes promoting dominance of the toxic cyanobacterium cies in Serbia. Bot Serb 36, 3-8. Cylindrospermopsis raciborskii. Chinese J Oceanol Limnol 29, 883-891. [14] Karadžić, V., Subakov-Simić, G., Natić, D., Ržaničanin, A., Ćirić, M., Gačić, Z. (2013) Changes in the phytoplankton com- [30] Chiswell, R.K., Shaw, G.R., Eaglesham, G.K., Smith, M.J., munity and dominance of Cylindrospermopsis raciborskii Norris, R.L., Seawright, A.A., Moore, M.R. (1999) Stability (Wolosz.) Subba Raju in a temperate lowland river (Ponjavica, of cylindrospermopsin, the toxin from the cyanobacterium Serbia). Hydrobiologia 711, 43–60. Cylindrospermopsis raciborskii: effects of pH, temperature and sunlight on decomposition. Environ Toxicol 14, 155–161. [15] Simić, S., Miščević, M., Đorđević, N., Popović, N. (2011). Cy- anobacteria in Aleksandrovac Lake, before and after revitali- [31] Griffiths, D.J., Saker, M.L. (2003) The Palm Island mystery sation. 16th Academy of Studenica: Cyanobacteria and Human disease 20 years on: A review of research on the cyanotoxin Health, Novi Sad, Serbia. Abstract book pp 17-18. cylindrospermopsin. Environ Toxicol 18, 78–93.

[16] Simić, S., Komárek, J., Đorđević, N. (2014) The confirmation [32] Rücker, J., Stüken, A., Nixdorf, B., Fastner, J., Chorus, I., of the genus Glaucospira (Cyanobacteria) and the occurrence Wiedner, C. (2007) Concentrations of particulate and dis- of Glaucospira laxissima (G. S. West) comb. nova in Serbia. solved cylindrospermopsin (CYN) in 21 Aphanizomenon Cryptogamie Algol 35(3), 259-267. dominated lakes of North East Germany. Toxicon 50, 800– 809. [17] Đorđević, N., Simić, S. (2014) Cyanobacterial blooms in oligosaline and alkaline microaccumulation before and after re- [33] Bittencourt-Oliveira, M.C., Piccin-Santos, V., Moura, A.N., habilitation. Pol J Environ Stud 23(6), 1975-1982. Aragao-Tavares, N.K.C., Cordeiro-Araujo, M.K. (2014) Cya- nobacteria, microcystins and cylindrospermopsin in public [18] Komárek, J. (2013) Cyanoprokaryota. 3. Teil Heterocytous drinking supply reservoirs of Brazil. An Acad Bras Cienc Genera. In: Büdel B, Gärtner G, Krienitz L Schagerl M (eds) 86(1), 297-309. Freshwater Flora of Central Europe. Springer Spektrum, [34] Saker, M.L., Eaglesham, G.K. (1999) The accumulation of Springer - Verlag Berlin, Heidelberg. pp 1 - 1130. cylindrospermopsin from the cyanobacterium Cylindrosper- [19] Utermöhl, H. (1958) Zur Vervollkommnung der quantitativen mopsis raciborskii in tissues of the redclaw crayfish Cherax Phytoplankton – Methodik. Mitt Int Ver Limnol 9, 1–38. quadricarinatus. Toxicon 37, 1065–1077.

[20] APHA (2005) Standard Methods (Standard methods for exam- [35] Messineo, V., Melchiorre, S., Di Corcia, A., Gallo, P., Bruno, ination of water and wastewater. 21st Edn.). Am. Publ. Health M. (2010) Seasonal succession of Cylindrospermopsis raci- Assoc., Washington DC, USA, 2005. borskii and Aphanizomenon ovalisporum blooms with cylindrospermopsin occurrence in the Volcanic Lake Albano, Cen- [21] Harada, K., Kondo, F., Lawton, L., (1999) Laboratory analysis tral Italy. Environ Toxicol 25(1), 18-27. of cyanotoxins, chapter 13. In: Chorus, I., Bartram, J., (Eds.), Toxic Cyanobacteria in Water: A guide to their public health [36] Saker, M.L., Griffiths, D.J. (2001) Occurrence of blooms of consequences, monitoring and management (WHO). The the cyanobacterium Cylindrospermopsis raciborskii (Wolo- World Health Organization, E. and F.N. Spon, London, UK. szynska) Seenayya and Subba Raju in a north Queensland do- pp. 369-405. mestic water supply. Mar Freshwater Res 52, 907–915.

[22] Welker, M., Bickel, H., Fastner, J. (2002) HPLC-PDA detec- [37] McGregor, G.B., Fabbro, L.D. (2000) Dominance of Cylin- tion of cylindrospermopsin–opportunities and limits. Water drospermopsis raciborskii (Nostocales, Cyanoprokaryota) in Res 36, 4659–4663. Queensland tropical and subtropical reservoirs: Implications for monitoring and management. Lakes. Reservoirs. Res Man- [23] Bláhová, L., Babica, P., Adamovský, O., Kohoutek, J., age 5, 195–205. Maršálek, B., Bláha, L. (2008) Analyses of cyanobacterial tox- [38] Leflaive, J., Ten-Hage, L. (2007) Algal and cyanobacterial ins (microcystins, cylindrospermopsin) in the reservoirs of the secondary metabolites in freshwaters: a comparison of alle- Czech Republic and evaluation of health risks. Environ Chem lopathic compounds and toxins. Freshw Biol 52, 199–214. Lett 6, 223–227. [39] Ibelings, B., Havens, K. (2008) Cyanobacterial toxins: a qual- [24] Chapman, A., Schelske, C. (1997) Recent appearance of Cyl- itative meta-analysis of concentrations, dosage and effects in indrospermopsis (Cyanobacteria) in five hypereutrophic Flor- freshwater, estuarine and marine biota. Adv Exp Med Biol ida lakes. J Phycol 33, 191-195. 619, 675–732.

[25] Briand, J.F., Robillot, C., Quiblier-Lloberas, C., Humbert, [40] Sotton, B., Domaizon, I., Anneville, O., Cattanéo, F., Guillard, J.F., Coute, A. (2002) Environmental context of Cylindrosper- J. (2014) Nodularin and cylindrospermopsin: a review of their mopsis raciborskii (cyanobacteria) blooms in a shallow pond effects on fish. Rev Fish Biol Fisheries DOI 10.1007/s11160- in France. Water Res 36, 3183–3192. 014-9366-6.

3741 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[41] De Souza, C.R., Carvalho, M.C., Truzzi, A.C. (1998) Cylin- drospermopsis raciborskii (Wolosz.) Seenaya and Subba Raju (Cyanophyceae) dominance and a contribution to the knowledge of Rio Pequeno Arm, Billings Reservoir, Brazil. Environ Toxic Water 13, 73-81. [42] Berry, J.P., Gibbs, P.D.L., Schmale, M.C., Saker, M.L. (2009) Toxicity of cylindrospermopsin, and other apparent metabolites from Cylindrospermopsis raciborskii and Aphani- zomenon ovalisporum, to the zebrafish (Danio rerio) embryo. Toxicon 53, 289–299.

Received: February 09, 2015 Revised: May 12, 2015 Accepted: June 17, 2015

CORRESPONDING AUTHOR

Associate Professor Snežana B. Simić

University of Kragujevac

Faculty of Science

Department of Biology and Ecology

Radoja Domanovića 12

Kragujevac, 34000

SERBIA

Phone: (+381) 34 336223 ex. 248

Fax: (+381) 34 335040

E-mail: [email protected]

FEB/ Vol 24/ No 11a/ 2015 – pages 3736 - 3742

3742 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

GREENHOUSE GAS (GHG) EMISSIONS AND THE OPTIMUM OPERATION MODEL OF TIMBER PRODUCTION SYSTEMS IN SOUTHERN CHINA

Yuan Zhou, Li-feng Zheng*, Xin-nian Zhou, Xi-sheng Hu, Zhi-long Wu, Cheng-jun Zhou and Dan Li

School of Transportation and Civil Engineering, Fujian Agriculture and Forestry University, Fuzhou 350002 Fujian, China

ABSTRACT 1. INTRODUCTION

Deforestation as an activity of forest management is It has been hypothesised that increasing CO2 concen- an intermediate link in timber production. Different models trations in the atmosphere may be increasing global warm- of timber production make a difference in GHG emissions ing, and a rise in mean global surface temperature of 1.4- and to a certain extent affect the sustainable development 5.8℃ by the year 2100 has been forecast [1-4]. There is a of the ecological environment. Accordingly, due to global general consensus that the increasing emissions of CO2 and climate change, energy conservation and emission reduc- other greenhouse gases (GHG) have led to changes in the tion have become critical issues concerning all humans. earth’s climate and the warming of the earth’s surface [5]. This paper taking the operating systems of timber produc- The use of carbon footprints as the quantitative parameter tion in the southern China, as a case study, and GHG emis- of energy conservation and emission reduction has been re- sions and the optimum operation model of timber produc- sponded to and extensively researched by many scholars. tion are analysed with the B2B model of life cycle assess- Among the research on carbon footprints, many investiga- ment (LCA) and the current forest industry standard. The tions have focused on individual carbon footprints [6, 7], results indicate that the total emissions of GHG range be- production or service carbon footprints [8-11] and com- tween 11.600 9 kg/m3 and 15.303 6 kg/m3, accounting for pany carbon footprints [12, 13], which involve the energy approximately 4.5 percent to 5.5 percent of their carbon se- production of electricity and coal [14, 15], material produc- questration abilities. Direct GHG emissions generated by tion of steel and cement [16-18], product production of fuel consumption during harvesting, skidding, loading, pulp and paper [19], usage of transport [20, 21] and so on. transportation and production are the primary emissions, Energy consumption plays the critical role in modern accounting for approximately 69.1 percent to 91.5 percent economic development, and currently even more in eco- of the total emissions. Moreover, the emission of every system. The IPCC Guidelines for National Greenhouse type of GHG is lowest under model IV (which included Gas Inventories has reported that emissions of GHG can be chainsaw cutting, cableway skidding, winch loading, diesel generated by energy consumption, industrial production, vehicle transportation and vehicle transmission). From the agricultural and forestry production, waste disposal and so perspective of ecology, therefore, model IV is the optimal on [22]. On the other hand, energy departments are the operation model. On the other hand, the emission from tim- most important departments in the GHG inventories. Table ber transportation is highest, accounting for approximately 1 reflects the energy consumption of the timber production 61.5 percent to 72.3 percent. Therefore, appropriate opera- in China. It shows the energy consumption both physical tion models should be choose according to the mechanical quantity and standard quantity. Results denote that the in- production and management, production tasks and scales, dustry has consumed a significant amount of energy. In de- foundations of forest mechanization and economic condi- veloped countries, their emission often accounts for more tions. In terms of carbon reduction, improvement of the than 90 percent of CO2 emissions and 75 percent of the to- mechanical equipment efficiency, operation skills and uti- tal emission of GHG [23]. According to the available liter- lization of new energy resources should be considered. ature, studies on energy carbon emission have primarily

been on the mechanism of energy-related carbon emissions KEYWORDS: as well as dynamic systems and point-source research [24- Timber production, Operation models, GHG emissions, Life cycle 27], and the levels of research involve the industrial fea- assessment, Emission reduction, Energy consumption tures of energy-related carbon emissions, low-carbon economies, carbon emission and other relevant issues [28-32]. Research and Development emphasise that dual func- * Corresponding author tions of improving energy efficiency and reducing pollution

3743 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 1 - Type of energy consumption in timber production tion, human activities act as intermediaries in forest eco- (10 000 tce) system and creating a new balance of power between local Industry division Physical quantity Standard quantity communities, the government authorities and business con- Fuel & electricity total 675.93 624.63 cessions [40]. Electricity 264.08 324.55 For the timber production is a primary activity of forest Gasoline 7.47 10.99 Kerosene 0.08 0.12 management, its operational process will not only under- Crude oil 0.17 0.24 mine the carbon sink function of forests but also increase the Diesel oil 14.32 20.87 emission of GHG [45]. Therefore, the sustainable develop- LPG 0.09 0.15 ment of forest ecology is viewed as an effective way for re- Raw coal 389.35 267.35 Coke 0.37 0.36 lieving ecological crises. The broad objective of this study is to estimate the GHG emissions and confirm the optimum Source: NBS (National Bureau of Statistics [33]) operation model of the mechanized timber production in the are mostly neglected, especially in the logging sector. Eval- southern China. Additionally, we adopt the B2B model of uations of the environmental sustainability of lifestyles and life cycle assessment (LCA) and the current forestry-indus- energy consumption have been taking centre stage in the try standard to derive the estimates of GHG emissions. The research projects in recent years, with final goal of attain- following objectives guided the research project: ing GHG emissions reduction [34, 35]. Considerable work (i) Identify both diagram of steps for the commodity has been undertaken on the environmental assessment of process and flow chart of the timber production operating energy consumption patterns, and several analytical tools system; and methodologies have been proposed to quantify the en- (ii) Identify the type of GHG emission as well as cal- vironmental burden of production and consumption [36]. culation methods; In order to minimize harmful impact on the environment, ISO standards were introduced. Certified management stand- (iii) Assess the GHG emission and optimum operation ards also came into existence for ensuring of the worker- model of timber production; health and safety [37]. With all major mitigation measures (iv) Set reasonable emission reduction goal and implemented, energy consumption and GHG emissions in choose the appropriate operation models for the logging in- 2050 can be reduced by 30 percent and 32 percent, respec- dustry. tively [38]. Forests are the most extensively distributed vegetation type in the world and cover approximately 1/3 of the land 2. MATERIALS AND METHODS surface of the earth [4, 39]. As an organic component of 2.1 Flow chart the global environment and human well-being, forest plays an important role in addressing climate change, preserving Forest harvesting transferred the stumpage into timber biodiversity and lowering the risk of disasters. To a certain production as well as transported them to the timber logis- extent, the quantity of forest resources represents the qual- tics distribution centre. While GHG emission of this pro- ity of the ecological environment. In addition, forests con- cess is equivalent to the B2B (Business to Business) model stitute an important component of terrestrial ecosystems of life cycle assessment, which is a method of “cradle to that exerts a great influence on the global carbon cycle [4]. gates”. In the life cycle of commodity, it only contained As a consequence, some policies were necessary to balance raw material reach to a new organization via production, the forest ecosystem. While, much of current policy dis- and distributes and transport to the place of customer [46]. course on reducing emissions from forest ecosystem is still However, it does not included the final product distribu- framed in aspects of forest management types, or as a tri- tion, retail, consumer use and disposal/recycling. The steps angle in the stakeholder groups [40-42]. Necessarily, poli- for the commodity process are shown below in Fig. 1. cies makers require a good understanding of the complex Timber production involves five steps for the process connections between four key stakeholder groups: state, lo- of changing from standing trees to timber as raw materials, cal communities, human activities, and state-sanctioned including cutting, skidding, loading, transporting and lum- concessions [40]. Earlier studies have examined the role of beryard like production. During the process, steps such as human activities in forest ecosystem, and the likely conse- cutting, lopping and bucking serve to change the shape of quences for global climate change [43, 44]. In this interac- timber, and skidding, loading, transporting and unloading

Raw material Production Distribution

FIGURE 1 - The diagram of steps for the commodity process

3744 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Tree species

Forest type Cutting Chainsaw Forest Conditions Stand volume

Rotting rate

Mill run Tractor Skidding To p o grap h y Cableway

Forest farm level

Climate Loading Winch Natural Conditions Elevation

Slop

Soil Automobile Transportation Road grade Diesel vehicle

Technology

Labor Market Conditions Lumberyard Vehicle transmit production Electric power

Material

Impact factor Operation process Mechanical equipment

FIGURE 2 - Flow chart of the timber production operating system

serve to displace the timber. A flow chart based on the unit trolled by steel consumption for the equipment for cutting, for the operation system for the mechanization of timber skidding, transportation and lumberyard production. products is shown in Fig. 2. The initial boundary demarca- tion at the harvesting and the terminate boundary demarca- 2.3 Calculation methods tion at the lumberyard production stage are shown. 2.3.1 Calculation of energy consumption

2.2 Category In this research, the comprehensive energy consumption for timber production under standard operation is cal- The carbon footprint standard ISO 14064 has defined the culated by regarding Energy Consumption for Wood Pro- GHG emission type as being divided into three categories: duction in Forest Region [47] as the baseline.

Direct GHG Emission. This refers to the emission of (1) Fuel consumption of chainsaw cutting GHG generated from fuel consumption by equipment for In a lumbering operation, the fuel consumption of skidding, transportation and production during logging. chain saws for producing a unit of wood is related to every Indirect Electric Power GHG Emissions. This refers to log volume and mill run per hectare as well as temperature, the emission of GHG generated from electric power con- elevation, varieties of trees, grades of forest farms and the sumption over the lifetime of production workers during rotting rate of wood. The lower the temperature and pro- logging, fuel production and steel production. portion of coniferous trees and the higher the elevation, forest farm level and rotting rate, the higher the energy con- Other Indirect GHG Emissions. This refers to the emis- sumption will be: sions of indirect GHG unrelated to energy in the process of logging. Instead, the greenhouse gases pertain to or are con- C1  qa1  Kt1  Kh1  Kl1  Ka1  Kb1 (i)

3745 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

where C1 refers to the fuel consumption of chainsaws refers to the total energy consumption of lumberyard pro- 3 (kg/m ), qa1 refers to the basic fuel consumption of chain- duction during the statistical period (measured with stand- 3 saws (kg/m ), Kt1 refers to the correction factor of temper- ard coal) (kg) and Ma5 refers to the total lumberyard pro- ature, Kh1 refers to the correction factor of elevation, Kl1 reduction during the statistical period (measured with stand- 3 fers to the correction factor of the forest farm level, Ka1 re- ard coal) (m ). fers to the correction factor of tree species, and Kb1 refers The basic fuel consumption at the formula is the fuel to the correction factor of the rotting rate. consumption under basic conditions, that is, the correction factor is 1. (2) Fuel consumption of skidding In the industry standard, only the fuel consumption for 2.3.2 Calculation of GHG emission tractors for logging or small-scale skidding in mountain In the timber production process, mainly energy use is farms with cableways is listed. The fuel consumption is re- one of the primary production inputs as a production pro- lated to temperature, elevation, skidding slope, grades of cess. During this process, the substances emitted are CO2, forest farms and skidding distance. The bigger the skidding CH4 and N2O .The GHG emission of timber production slope and distance, the higher the consumption will be: was estimated with the process analysis method.

C2  qa2  Kt 2  Kh2  K p2  Kl 2  K j2 (ii) As highlighted in IPCC Guidelines for National Greenhouse Gas Inventories, carbon emission equals ac- where C2 refers to the fuel consumption of skidding 3 tivity data multiplied by emission factor. When converting (kg/m ), qa2 refers to the basic fuel consumption of skid- 3 through the heat value of fuel, we can obtain a formula for ding (kg/m ), Kt2 refers to the correction factor of temper- calculating the emission equivalent (MC) of CO2: ature, Kh2 refers to the correction factor of elevation, Kl2 refers to the correction factor of the forest farm level, Kp2 re- M  E  A  r (vi) C  i i fers to the correction factor of the skidding slope, and Kj2 refers to the correction factor of the skidding distance. where Ei refers to the energy consumption of standard 3 coal (kg/m ), Ai refers to the carbon emission factor and r (3) Fuel consumption of winch loading is the calorific value of standard coal, 29 271 kJ/kg [48]. The fuel consumption of the loading of 2105 diesel The emission of other types of GHG can be converted winches is specified in the standard. The fuel consumption to emissions of CO2 through the GWP value. Various ref- is related to temperature and elevation: erential coefficients are shown in Table 2 [23, 49]. There- fore, simply, effective load of the energy consumption was C3  qa3  Kt3  Kh3 (iii) composed of energy usage, emission factor and GWP where C3 refers to the fuel consumption of loading value. 3 (kg/m ), qa3 refers to the basic fuel consumption of loading 3 (kg/m ), Kt3 refers to the correction factor of temperature TABLE 2 - Energy emission factor and GWP and Kh3 refers to the correction factor of elevation. Species GHG Emission factor/(kg·TJ-1) GWP CO 74 100 1 Gasoline 2 (4) Fuel consumption of log transportation CH 3.0 25 diesel 4 The fuel consumption of EQ140 vehicles and diesel N2O 0.6 298 vehicles is specified in the standard. The fuel consumption Sources: IPCC (Intergovernmental Panel on Climate Change). [23], Wang et al. [49]. of log transportation for a unit of transportation distance is related to temperature, elevation and grade of road. The Because different models of operation consume differ- lower the grade of the road, the higher the fuel consump- ent types of fossil fuels such as oil and electric power and it tion will be: would be incomplete or inaccurate to compare only certain

C4  qa4  Kt 4  Kh4  Kr 4 (iv) types of fossil fuel, there is a need to convert the consumption of all types of energy into standard coal for calculating where C4 refers to the fuel consumption of log trans- 3 CO2. In other words, the converted consumption of standard portation [kg/(m ·km)], qa4 refers to the basic fuel consumption of log transportation [kg/(m3·km)], K refers to coal is equal to physical quantity multiplied by the converted t4 standard coal coefficient, as shown in Table 3 [23, 50]. the correction factor of temperature, Kh4 refers to the correction factor of elevation and K refers to the correction r4 TABLE 3 - Energy conversion to standard coal factor of road grade. Converted standard Average low calorific Name coal coefficient (5) Energy consumption of lumberyard production value (KJ/kg) (kg/kg) W C  a5 (v) gasoline 43 124 1.4714 5 M diesel 42 705 1.4571 a5 electric power 11 840 0.404 0 where C5 refers to the energy consumption of lumber- Source: 2006 IPCC date [23]. Qingdao energy consumption factor contri- 3 bution analysis and energy conservation countermeasures. [50]. yard production (measure with standard coal) (kg/m ), Wa5

3746 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Once the GHG emission of every activity is attained in timber manufacturing, a forestry-industry standard in the through calculation, GWP (refer to Guide to PAS 2050) is People’s Republic of China, the basic fuel consumption of used to convert it into the equivalent of CO2. The carbon all steps of timber production by timber production enter- 3 emission equivalent generated by CH4 and N2O is MˊC: prises is the following [47]: 0.180 kg/m for cutting with gasoline chain saws, 0.691 kg/m3 for tractor skidding, M   E  A GWP  r (vii) C  ij ij ij j 0.615 kg/m3 for cableway skidding, 0.145 kg/m3 for winch 3 3 where Eij refers to the standard coal fuel consumption loading, 0.103 kg/m for EQ140 transportation, 0.073 kg/m 3 3 of gas j in activity i (kg/m ), Aij refers to the emission factor for diesel vehicles and 0.073 kg/m for lumberyard produc- of gas j in activity I and GWPj refers to the warming poten- tion with vehicles. Moreover, all of the fuel consumption tial of gas j. fixed for all operation models are the maximum limits. The carbon emission of timber production operating systems under basic conditions was calculated according to 3. RESULTS basic conditions and the basic fuel consumption specified in the industry standard. This paper uses southern China as 3.1 Data the subject, while where the basic conditions of timber pro- We assumed that all operation models are practiced duction operations are as follows: the temperature ranges under the same conditions based on the flow chart of the from 5 to 28, the elevation is below 1 000 m; the forest operation system of each unit of timber product; that is, the farm level ranges from 5 to 7, and the forest road reaches conditions of total skidding and transportation, area of cut- level 2. The skidding slope of the tractors ranges from 5 to ting area, total mill run and the length of existing log trans- 15°, and the cableways a in the range of 15 to 22°. The portation branch roads (combined with the actual situation, skidding distance of the tractors range from 200 to 500 m, such as the 42 km) are identical [51, 52]. Moreover, we and cableways are in the range of 400 to 700 m. assumed that the cutting area meets the conditions for skid- Because the distance of log transportation is not pro- ding and transporting logs of the four operation models vided in the standard and there are limitations from the re- shown in Table 4. The place that is approximately 3 km search conditions and research scale, the author has referred away from the centre of the cutting area can be used for log to the values of field surveys of previous research [51, 52] transportation. and set the log transportation distance as 42 km, 3 km of which is the skidding distance of tractors and 350 m of which 3.2 Data analysis under basic conditions is the skidding distance of cableways. When timber produc- (1) Direct GHG Emissions tion proceeds under basic conditions, fuel consumption dur- Direct GHG emissions are used primarily for collect- ing operations is viewed as basic fuel consumption. The re- ing the emission of GHG resulting from fuel consumption. sults of the calculation are shown in Table 5. According to national level statistics for energy consumption

TABLE 4 - Operation models of timber production

Operation model Logging equipment Ways of skidding Loading equipment Models of transportation Lumberyard production Ⅰ Chainsaw Tractor Winch EQ140 Vehicle transmission Ⅱ Chainsaw Tractor Winch Diesel vehicle Vehicle transmission Ⅲ Chainsaw Cableway Winch EQ140 Vehicle transmission Ⅳ Chainsaw Cableway Winch Diesel vehicle Vehicle transmission

TABLE 5 - Direct emissions of GHG/ (kg·m-3)

Cutting Skidding Transportation Lumberyard Model Type Loading Total Chainsaw Tractor Cableway EQ140 Diesel vehicle Vehicle transmit CO2 0.390 4 1.498 8 0.314 5 9.383 0 1.151 7 12.738 4 CH 0.000 4 0.001 5 0.000 3 0.009 5 0.001 2 0.012 9 Ⅰ 4 N2O 0.000 9 0.003 6 0.000 8 0.022 7 0.002 8 0.030 8 Subtotal 0.391 8 1.503 9 0.315 6 9.415 2 1.155 7 12.782 2 CO2 0.390 4 1.498 8 0.314 5 6.650 1 1.151 7 10.005 5 CH 0.000 4 0.001 5 0.000 3 0.006 7 0.001 2 0.010 1 Ⅱ 4 N2O 0.000 9 0.003 6 0.000 8 0.016 1 0.002 8 0.024 2 Subtotal 0.391 8 1.503 9 0.315 6 6.672 9 1.155 7 10.039 9 CO2 0.390 4 1.333 9 0.314 5 9.383 0 1.151 7 12.573 5 CH 0.000 4 0.001 4 0.000 3 0.009 5 0.001 2 0.012 8 Ⅲ 4 N2O 0.000 9 0.003 2 0.000 8 0.022 7 0.002 8 0.030 4 Subtotal 0.391 8 1.338 5 0.315 6 9.415 2 1.155 7 12.616 8 CO2 0.390 4 1.333 9 0.314 5 6.650 1 1.151 7 9.840 6 CH 0.000 4 0.001 4 0.000 3 0.006 7 0.001 2 0.010 0 Ⅳ 4 N2O 0.000 9 0.003 2 0.000 8 0.016 1 0.002 8 0.023 8 Subtotal 0.391 8 1.338 5 0.315 6 6.672 9 1.155 7 9.874 5

3747 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Table 5 shows that operation model IV generates the (2) Electric power Indirect GHG Emissions 3 least amount of GHG emission (9.874 5 kg/m ). Model I Electric power indirect GHG emissions refer to the generates the highest amount of GHG emission (1.3 times of consumption of electric power presented in electric power that by model IV). Among direct greenhouse gases gener- consumption over the life of a production worker and the ated by fuel consumption, CO2 is the primary one, account- electric power consumption for fuel and steel production. ing for 99.7 percent of the total emission. Loading winches In this article, electric power consumption is estimated ac- generate the least direct GHG (0.315 6 kg/m3). The second 3 cording to the number of workers at every stage of operation. lowest emission of direct greenhouse gases (0.391 8 kg/m ) Moreover, the required number of workers and consumption is generated by chainsaws. The direct GHG emissions gen- of steel at every stage are sorted according to the statistical erated by the skidding segment and cableway skidding are results of available data [52], as shown in Table 6. lower than that by tractor skidding, approximately 89 percent of that by tractor skidding. The direct GHG emissions gener- According to the statistics, the domestic electric power ated by the production of the timber yard are 1.155 7 kg/m3. consumption per capita across the country is 0.8 kWh per Due to longer transportation distances, the direct GHG emis- day [52]. Therefore, the electric power consumption at sion generated by the log transportation segment is highest, every stage can be estimated as 0.8 multiplied by the num- and the direct GHG emission generated by EQ140 log ber of work days. Moreover, the electric power consump- transportation (is 9.415 2 kg/m3) is higher than that by tion for producing fuels is approximately 0.06 kWh/kg, and transporting with diesel vehicles (6.672 9 kg/m3). that for producing steel is approximately 4 kWh/kg. The

TABLE 6 - Artificial demand and unit of steel consumption of harvesting operations

Skidding Winch load- Transportation Lumberyard Section Cutting Tractor Cableway ing EQ140 Diesel vehicle production Artificially demand 0.55 0.01 0.05 0.25 0.35 0.60 (work days / m3) Steel consumption 0.000 3 0.133 9 0.021 9 0.000 2 0.218 4 0.159 6 0.077 9 (kg/m3) Source: Study on the influence of plantation operation on resource-environment-economics based on Life Cycle Assessment theory. [52].

TABLE 7 - Electric power indirect GHG emissions

Lumberyard Cutting Skidding Transportation production Model Type Loading Total Vehicle Chainsaw Tractor Cableway EQ140 Diesel vehicle transmission Fuel electric power 0.003 4 0.013 1 — 0.002 7 0.081 9 — 0.010 0 0.111 1 consumption Life electric power 0.138 8 0.002 5 — 0.063 1 0.088 3 — 0.151 4 0.444 1 Ⅰ consumption Steel electric power 0.000 4 0.168 9 — 0.000 3 0.275 5 — 0.098 3 0.543 4 consumption Subtotal 0.142 5 0.184 5 — 0.066 1 0.445 7 — 0.259 7 1.098 5 Fuel electric power 0.003 4 0.013 1 — 0.002 7 — 0.058 0 0.010 0 0.087 2 consumption Life electric power 0.138 8 0.002 5 — 0.063 1 — 0.088 3 0.151 4 0.444 1 Ⅱ consumption Steel electric power 0.000 4 0.168 9 — 0.000 3 — 0.201 4 0.098 3 0.469 3 consumption Subtotal 0.142 5 0.184 5 — 0.066 1 — 0.347 7 0.259 7 1.000 5 Fuel electric power 0.003 4 — 0.011 6 0.002 7 0.081 9 — 0.010 0 0.109 6 consumption Life electric power 0.138 8 — 0.012 6 0.063 1 0.088 3 — 0.151 4 0.454 2 Ⅲ consumption Steel electric power 0.000 4 — 0.027 6 0.000 3 0.275 5 — 0.098 3 0.402 1 consumption Subtotal 0.142 5 — 0.051 9 0.066 1 0.445 7 — 0.259 7 0.965 9 Fuel electric power 0.003 4 — 0.011 6 0.002 7 — 0.058 0 0.010 0 0.085 7 consumption Life electric power 0.138 8 — 0.012 6 0.063 1 — 0.088 3 0.151 4 0.454 2 Ⅳ consumption Steel electric power 0.000 4 — 0.027 6 0.000 3 — 0.201 4 0.098 3 0.328 0 consumption Subtotal 0.142 5 — 0.051 9 0.066 1 — 0.347 7 0.259 7 0.867 9

3748 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

global average electric power carbon emission coefficient emissions of CH4 and N2O are nearly 0. Similar to direct is 315.4×10-3 [53]. Therefore, the results of the electric GHG emissions, the emission of GHG reaches a mini- power emission calculation are shown in Table 7. mum (0.858 6 kg/m3) in model IV. Moreover, winch load- 3 From Table 7, we can see that electric power indirect ing generates minimal emissions at 0.000 6 kg/m . The GHG emissions primarily consist of fuel electric power con- chain saw segment generates the second lowest emissions 3 sumption, life electric power consumption and steel electric (0.000 9 kg/m ). The direct GHG emissions generated by power consumption. Among them, life electric power con- the skidding segment and cableway skidding are much sumption and steel electric power consumption generate lower than that by tractor skidding, only approximately more emissions. Moreover, fuel electric power consumption 16.4 percent of that by tractor skidding. The log transpor- generates the highest emissions (0.111 1 kg/m3) in model I tation segment generates the highest emission, of which 3 and the lowest emissions (0.085 7 kg/m3) in model IV. EQ140 log transportation generates 0.721 6 kg/m and log 3 Life electric power consumption generates less emissions transportation by diesel vehicles generates 0.527 3 kg/m . in model I and model II than in model III and model IV Through comprehensively sorting Table 6 to Table 8, (0.454 2 kg/m3). Moreover, steel electric power consump- we have attained Table 9. 3 tion generates minimal emissions (0.328 0 kg/m ) in model From Table 9, we can see that GHG emissions gener- IV. Therefore, total electric power indirect GHG emissions ated by harvesting are primarily direct GHG emissions 3 reach a minimum (0.867 9 kg/m ) in model IV. generated by fuel consumption accounting for approximately 69.1 percent to 91.5 percent of the total emission. (3) Other Indirect GHG Emissions Moreover, indirect GHG primarily consist of electric power Although logging operation does not generate the di- consumption and steel consumption. From the whole opera- rect consumption of steel, the manufacturing of forestry tions process, winch loading generates a minimal amount of mechanical equipment consumes a great deal of steel. To emission at 0.382 3 kg/m3, and the cutting segment generates attain an accurate analysis result, steel was also considered the second lowest emission (approximately 1.4 times that by in the research as an emission of other greenhouse gases. winch loading). Moreover, the emissions by cableway skid- The calculation results of emissions from steel consump- ding, tractor skidding and lumberyard production are 3.8, tion are shown in Table 8. 5.6 and 4.4 times that generated by winch loading, respectively. The log transportation segment generates the highest Table 8 indicates that CO2 remains the primary greenhouse gas among other indirect greenhouse gases, account- emissions, of which EQ140 vehicle and diesel vehicle log 3 3 ing for 96.7 percent of the total emission. In the cutting transportation generates 10.582 5 kg/m and 7.547 9 kg/m , segment and the loading segment with winches, the respectively.

TABLE 8 - Other Indirect GHG Emissions

Cutting Skidding Transportation Lumberyard production Model Type Loading Total Chainsaw Tractor Cableway EQ140 Diesel vehicle Vehicle transmission CO2 0.000 9 0.427 8 0.000 6 0.697 7 0.248 9 1.375 9 CH 0 0.004 3 0 0.007 1 0.002 5 0.013 9 Ⅰ 4 N2O 0 0.010 3 0 0.016 8 0.006 0 0.033 1 Subtotal 0.000 9 0.442 4 0.000 6 0.721 6 0.257 4 1.422 9 CO2 0.000 9 0.427 8 0.000 6 0.509 9 0.248 9 1.188 1 CH 0 0.004 3 0 0.005 2 0.002 5 0.012 0 Ⅱ 4 N2O 0 0.010 3 0 0.012 3 0.006 0 0.028 6 Subtotal 0.000 9 0.442 4 0.000 6 0.527 3 0.257 4 1.228 6 CO2 0.000 9 0.070 0 0.000 6 0.697 7 0.248 9 1.018 1 CH 0 0.000 7 0 0.007 1 0.002 5 0.010 3 Ⅲ 4 N2O 0 0.001 7 0 0.016 8 0.006 0 0.024 5 Subtotal 0.000 9 0.072 4 0.000 6 0.721 6 0.257 4 1.052 9 CO2 0.000 9 0.070 0 0.000 6 0.509 9 0.248 9 0.830 3 CH 0 0.000 7 0 0.005 2 0.002 5 0.008 4 Ⅳ 4 N2O 0 0.001 7 0 0.012 3 0.006 0 0.020 0 Subtotal 0.000 9 0.072 4 0.000 6 0.527 3 0.257 4 0.858 6

TABLE 9 - CHG emissions of operations at various stages (kg/m3)

Direct GHG emission Indirect GHG emission Operation process Total emissions Fuel Electric power Steel Subtotal Cutting Chainsaw 0.391 8 0.142 5 0.000 9 0.143 4 0.535 2 Cableway 1.338 5 0.051 9 0.072 4 0.124 3 1.462 8 Skidding Tractor 1.503 9 0.184 5 0.442 4 0.626 9 2.130 8 Loading Winch 0.315 6 0.066 1 0.000 6 0.066 7 0.382 3 EQ140 9.415 2 0.445 7 0.721 6 1.167 3 10.582 5 Transporting Diesel vehicle 6.672 9 0.347 7 0.527 3 0.875 0 7.547 9 Lumberyard production 1.155 7 0.259 7 0.257 4 0.517 1 1.672 8

3749 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

18 ) 15.303 6

3 14.635 5 16 14 12.269 0 11.609 9 12 10

6 4 Total emissions (kg/m emissions Total 2 0 ⅠⅡⅢⅣ

cutting skidding loading Operationtransportation model lumberyard production

FIGURE 3 - Total carbon emissions of each operating model

To further analyse the GHG emissions of forestry har- 4. DISCUSSION AND CONCLUSION vesting operations, a comparative analysis of the total emissions of the above four operation models has been The forest is a library of carbon cycles on the earth, made, as shown in Fig. 3. and the carbon sequestration per m3 of wood reaches as high as 0.21-0.34 t [54]. According to the energy consump- Figure 3 indicates that the total emissions of all operation tion of timber production in timber yards specified in cur- models range between 11.600 9 kg/m3 and 15.303 6 kg/m3. rent forestry industry standards, the author has analysed When further analysing the proportions of carbon emissions GHG emissions and the optimum operating model in the of all operational steps in the total carbon emissions of the op- five stages (cutting, skidding, loading, and transporting and eration models, we find that the carbon emissions gener- lumberyard production) of mechanical timber production ated by the log transportation segment are maximal in all based on the B2B model of life cycle assessment. The re- operational models, and the proportion is lowest (61.5 per- sults indicate that the total emissions of GHG from timber cent) in model II and highest (72.3 percent) in model III. production under basic conditions are approximately The skidding segment generates the second highest emis- 11.600 9 kg/m3 to 15.303 6 kg/m3, accounting for approxi- sions, and the proportions of carbon emissions of tractor mately 4.5 percent to 5.5 percent of the carbon sequestra- skidding in model I and model II are 13.9 percent and 17.4 tion ability. Timber production inevitably produced the percent, respectively. The proportion of carbon emissions carbon emissions such as fuel consumption, power con- generated by the cableway skidding in model III and model sumption, both of which offset can achieve the zero emis- IV is 10.0 percent and 12.6 percent, respectively. The pro- sions of carbon cycle is worthy studying. The GHG effect duction of the timber yard generates the third highest emis- is one of our most severe current environmental problems. sions, and the emission reaches a maximum (14.4 percent) Forests make up large ecosystems and can play an im- in model IV and a minimum (10.9 percent) in model I. The portant role in mitigating the emissions of GHG. Dong et lowest proportion of carbon emission generated by chain al. [55] emphasized that there is no shirking the responsi- saw felling is 3.5 percent, and the highest is 4.6 percent. bility for the forestry sector, especially for state-owned Winch loading generates the lowest carbon emissions, and Forest Region to respond to global climate change and de- the proportion is 2.5 percent (the lowest) in model I and 3.3 velop low-carbon economy. Forest industry low-carbon percent (the highest) in model IV. GHG emission gener- economy and constantly enhance ecological civilization ated by model IV is minimal (11.600 9 kg/m3), that is, the construction level, we must achieve a new leap in carefully optimal model of operation. As with other carbon emis- cultivating forest resources, effectively protecting forest sions, model II generates more than model III, which gen- resources and reasonably using forest resources. Jiang [56] erates more than model I. Furthermore, the proportions of analyzed the need for low-carbon economic development carbon emissions change with transportation distance. The and the measurement of state-owned Forest Region. The longer the transportation distance, the higher the propor- results indicated that low-carbon economic development is tion of carbon emission of the log transportation, and the an important way for state-owned Forest upgrading devel- lower the proportions of other steps. Therefore, the log opment. Moreover, Alig [57] recognized that forest can not transportation segment is the key to achieving energy sav- only reduce its GHG, but also provides environmental, eco- ings and emission reduction for log production and opera- nomic and social benefits. Tavoni [58] utilized the energy- tion systems. economy-climate model to show that forestry is a determi- nant abatement option and could lead to significantly lower carbon emissions.

3750 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

The GHG emissions calculated in this paper are pri- of log transportation and using new energy resources. Fur- marily direct GHG emissions, electric power indirect GHG thermore, the government should also making policies in emissions and other indirect GHG emissions. Among reaching emission reduction. them, direct GHG emissions resulting from fuel consump- Generally, for the items specified in this standard are tion are the primary emissions, accounting for approxi- the maximum limits of fuel consumption, the results of the mately 69.1 percent to 91.5 percent of the total emission. calculation are a little higher than the actual emissions. Moreover, model IV generates the least GHG emission Moreover, fuel consumption during mechanical operation 3 (9.874 5 kg/m ). Model I generates the highest emission, is affected by temperature, elevation, slope and other oper- approximately 1.3 times of that by model IV. Among indi- ation conditions. In addition, certain measurement errors rect electric power GHG emissions, lifetime electric power and calculation errors exist in the data attained through lit- consumption and steel electric power consumption are pri- erature survey. Due to different tree species, terrains, cli- mary, and their total emissions are the lowest (0.867 9 mates and other conditions in forest areas, different opera- 3 kg/m ) in model IV. In addition, indirect steel GHG emis- tion models as well as different equipment for cutting, skid- 3 sions are the lowest (0.858 6 kg/m ) in model IV. Hence, ding and transporting and different degrees of mechaniza- the model IV is the optimum operation model of timber tion should be applied. The calculation result of this reproduction system. Among the four operation models, log search was attained after considering the GHG emissions transportation generates the highest emission, accounting of all mechanical operations. In current timber production, for approximately 61.5 percent to 72.3 percent. The pro- however, push cart skidding, animal skidding and manual portion of carbon emission generated by log transportation loading which do not directly generate emission are uni- of the total carbon emissions increases with the increasing versal. Nevertheless, GHG emissions in practice are less transportation distance. The segments that generated less than the emissions in this research. Therefore, in practice, carbon emissions were the skidding segment (the tractor it should work in the same environment, taking into ac- skidding accounts for 13.9 percent to 17.4 percent and the count the GHG emissions generated by both of the human cableway skidding accounts for 10.0 percent to 12.6 per- activities and the mechanical operation, and the correction cent), lumberyard production accounts for 10.9 percent and coefficient should also be considered into the calculation 14.4 percent, cutting with chain saws accounts for 3.5 per- in order to minimize the error calculation. cent and 4.6 percent and winch loading accounts for 2.5 percent and 3.3 percent. Zhang et al. [51] analyzed and evaluated eight kinds of harvesting operation models in ACKNOWLEDGEMENTS southern China, and the results demonstrated that under the same operation conditions, the economic, ecologic, social This study was supported by the National Natural Sci- and comprehensive benefits of different operation models ence Foundation of China (31070567 and 30972359) , Fu- are different, the model II (chainsaw cutting-walking trac- jian province natural science foundation project of Fujian tor skidding-boat transportation) should have priority to the Province, China (2013J01072) and the University Devel- optimum operation model with the condition of waterway opment Foundation of Fujian Agriculture and Forestry transportation, while the model III (chainsaw cutting-ca- University (113-612014018). bleway skidding-farm vehicle transportation) should be the optimum model without the condition of waterway trans- The authors have declared no conflict of interest. portation. Certainly, we should choose appropriate operation models featured with large-scale and mechanized production according to the mechanical production and management model of forestry, scales of production, produc- REFERENCES tion tasks, foundations of forest mechanization and economic conditions. [1] Kirschbaum, M.U.F. (1995) The temperature dependence of soil organic matter decomposition, and the effect of global Due to the carbon emission generated by log transpor- warming on soil organic C storage. Soil Biology and Biochem- tation accounts for the largest proportion, reasonable selec- istry 27 (6), 753-760. tion of log transportation machines is an effective way for [2] IPCC TAR WG1 (2001) Climate Change 2001: The scientific reducing logging work and achieving ecological balance in basis. Contribution of working group I to the third assessment the timber production system. Ortiz et al. suggested that report of the intergovernmental panel on climate change. Cam- environmental studies should not only aim to promote and bridge, Cambridge University Press. ensure the quality of physical components of nature, but [3] Bagherzadeh, A., Brumme, R. and Beese, R. (2008) Tempera- should also consider the ecosystem properties and dynam- ture dependence of carbon mineralization and nitrous oxide emission in a temperate forest ecosystem. Journal of Forestry ics that emerge from complex networks [59]. Chen et al. Research 19(2), 107-112. introduced a new concept of “controlled energy”, that adds to the analysis of energy consumption, indicating the ad- [4] Lu, X.Y., Cheng, G.W., Xiao, F.P. and Huo, C.F. (2008) Simu- lating carbon sequestration and GHG emissions in Abies fabric justable energy consumed by each steps [60]. In practice, forest on the Gongga Mountains using a biogeochemical pro- we can start by improving the efficiency of vehicles, skills cess model Forest-DNDC. Journal of Mountain Science 5 (3), of drivers and exhaust emission by optimizing the process 249-256.

3751 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[5] Miehle, P., Livesley, S.J., Feikema, P.M., Li, C. and Arndt, [21] Postorino, M.N. and Mantecchini, L. (2014) A transport car- S.K. (2006) Assessing productivity and carbon sequestration bon footprint methodology to assess airport carbon emissions. capacity of Eucalyptus globulus plantations using the process Journal of Air Transport Management 37, 76-86. model Forest-DNDC: Calibration and validation. Ecological Modelling 192(1-2), 83-94. [22] Liu, H.Y., Wang, Z. and Ma, X.Z. (2011) Predictions of net carbon emissions based on the emissions and forest carbon [6] Bhoyar, S.P., Dusad, S.,, Shrivastava, R., Mishra, S., Gupta, N. sinks in Yunnan Province. Acta Ecologica Sinica 31(15), 4405- and Rao, A.B. (2013) Understanding the impact of lifestyle on 4414. individual carbon footprint. Procedia·Social and Behavioral Sciences 133, 47-60. [23] IPCC (Intergovernmental Panel on Climate Change). (2006) IPCC country listing guide the second volume of greenhouse [7] Yang, D.J. and Timmernans, H. (2014) Effects of urban spatial gases. Japan, Japan Global Environmental Strategies Institute. form on individuals’ footprints, Empirical study based on per- sonal GPS panel data from Rotterdam and Eindhoven area. [24] Ma, C., Ju M.T., Zhang, X.C. and Li, H.Y. (2011) Energy con- Procedia Environmental Sciences 22, 169-177. sumption and carbon emissions in a coastal city in China. Pro- cedia Environmental Sciences 4, 1-9. [8] Ruviaro, C.F., Léis, C.M,D., Lampert, V.D,N., Barcellos, J.O.J.B. and Dewes, H. (2015) Carbon footprint in different [25] Mao, G.Z., Dai, X., Wang, Y., Guo, J.H., Cheng, X., Fang, beef production systems on a southern Brazilian farm: a case D.M., Song, X.F., He, Y.Y. and Zhao, P. (2013) Reducing car- study. Journal of Cleaner Production 96, 435-443. bon emissions in China, Industrial structural upgrade based on system dynamics. Energy Strategy Reviews 2, 199-204. [9] Kiefer, L.R., Menzel, F. and Bahrs, E. (2015) Integration of ecosystem services into the carbon footprint of milk of South [26] Mensah, J.T. (2014) Carbon emissions, energy consumption German dairy farms. Journal of Environmental Management and output, A threshold analysis on the causal dynamics in 152, 11-18. emerging African economies. Energy Policy 70, 172-182.

[10] Hogne, N.L. and Edgar, G.H. (2011) Analyzing the carbon [27] Huang, G.X., Ouyang, X.L. and Yao, X. (2015) Dynamics of footprint from public services provided by counties. Journal of China’s regional carbon emissions under gradient economic Cleaner Production 19, 1975-1981. development model. Ecological Indicators 51, 197-204.

[11] Cheng, J.C. (2011) A web service framework for measuring [28] Timothy, J.F. (2011) A coevolutionary framework for analys- and monitoring environmental and carbon footprint in con- ing a transition to a sustainable low carbon economy. Ecolog- struction supply Chains. Procedia Engineering 14, 141-147. ical Economics 70, 2258-2267. [12] Lee, K.H. (2011) A Web Service Framework for Measuring [29] Paul, K.I., Reeson, A., Polglase, P., Crossman, N., Freudender- and Monitoring Environmental and Carbon Footprint in Con- ger, D. and Hawkins, C. (2013) Economic and employment struction Supply Chains. Journal of Cleaner Production 19, implications of a carbon market for industrial plantation for- 1216-1223. estry. Land Use Policy 30, 528-540.

[13] Vujanović, A., Čuček, L., Pahor, B. and Kravanja, Z. (2014) [30] Tian, Y., Zhang, J.B. and He, Y.Y. (2014) Research on spatial- Multi-objective synthesis of a company’s supply network by temporal characteristics and driving factor of agricultural car- accounting for several environmental footprints. Process bon emissions in China. Journal of Integrative Agriculture Safety and Environmental Protection 92, 456-466. 13(6), 1393-1403. [14] Fairuz, S.M.C., Sulaiman, M.Y., Lim, C.H., Mat, S., Ali, B., [31] Lin, J.Y., Jacoby J., Cui, S.H. Liu, Y. and Lin, T. (2014) A Saadatian, O., Ruslan, M.H., Sslleh, E. and Sopian, K. (2013) model for developing a target integrated low carbon city indi- Long term strategy for electricity generation in Peninsular Ma- cator system, The case of Xiamen, China. Ecological Indica- laysia–Analysis of cost and carbon footprint using MES- tors 40, 51-57. SAGE. Energy Policy 62, 493-502. [32] Wang, Z.H. and Yang, L. (2015) Delinking indicators on re- [15] Messagie, M., Mertens, J., Oliveira, L., Rangaraju, S., Sanfe- gional industry development and carbon emissions, Beijing– lix, J., Coosemans, T., Mierlo, J.V. and Macharis, C. (2014) Tianjin–Hebei economic band case. Ecological Indicators 48, The hourly life cycle carbon footprint of electricity generation 41-48. in Belgium, bringing a temporal resolution in life cycle assessment. Applied Energy 134, 467-476. [33] NBS (National Bureau of Statistics). (2013) China energy sta- [16] Salmi, O. and Wierink, M. (2011) Effects of waste recovery on tistical yearbook. Beijing: China Statistic Press. carbon footprint, a case study of the Gulf of Bothnia steel and [34] Novak, P. (2015) Sustainable energy system with zero emis- zinc industries. Journal of Cleaner Production 19, 1857-1864. sions of GHG for cities and countries. Energy and Building 98, [17] Cagiao, J., Gómez, B., Doménech, J.L., Mainar, S.G. and 27-33. Lanza, H.G. (2011) Calculation of the corporate carbon foot- [35] Pairotti, M.B., Cerutti, A.K., Martini, F., Vesce, E., Padovan, print of the cement industry by the application of MC3 meth- D. and Beltramo, R. (2015) Energy counsumption and GHG odology. Ecological Indicators 11, 1526-1540. emissions of the Mediterranean diet: a systemic assessment us- [18] Xu, T.F., Galama, T. and Sathaye, J. (2013) Reducing carbon ing a hybrid LCA-IO method. Journal of Cleaner Production footprint in cement material making, characterizing costs of 103, 507-516. conserved energy and reduced carbon emissions. Sustainable [36] Delivand, M.K., Cammerino, A.R.B., Garofalo, P. and Mon- Cities and Society 9, 54-61. teleone, M. (2015) Optimal locations of bioenergy facilities, [19] Dias, A.C. and Arroja, L. (2012) Comparison of methodolo- biomass spatial availability, logistics costs and GHG (green- gies for estimating the carbon footprint case study of office house gas) emissions: a case study on electricity productions paper. Journal of Cleaner Production 24, 30-35. in South Italy. Journal of Cleaner Production 99, 129-139. [20] Wu, X.M., Hu, S. and Mo, S.J. (2013) Carbon footprint model [37] Dayaratne, S.P. and Gunawardana, K.D. (2015) Carbon foot- for evaluating the global warming impact of food transport re- print reduction: a critical study of rubber production in small frigeration systems. Journal of Cleaner Production 54, 115- and medium scale enterprises in Sri Lanka. Journal of Cleaner 124. Production 103, 87-103.

3752 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[38] Hao, H., Geng, Y., Li, W.Q. and Guo, B. (2015) Energy con- [56] Jiang, S. (2013) Analysis of low-carbon economy development sumption and GHG emissions from China’s freight transport strategy for state-owned forest. Journal of Agricultural Mech- sector: Scenarios through 2050. Energy Policy 85, 94-101. anization Research 8, 241-244.

[39] Li, C.S. and Aber, J. (2000) A processs-oriented model of N2O [57] Alig, R., Latta, G., Adams, D. and McCarl, B. (2010) Mitigat- and NO emissions from forest soils, 1. model development. ing greenhouse gases: The importance of land base interac- Journal of Geophysical Research 105(D4), 4369-4384. tions between forests, agriculture, and residential development in the face of changes in bioenergy and carbon prices. Forest [40] Galudra, G., Noordwijk, M.V., Agung, P., Suyanto, S. and Pra- Policy and Economics 12(1), 67-75. dhan, U. (2014) Migrants, land markets and carbon emissions in Jambi, Indonesia: Land tenure change and the prospect of [58] Tavoni, M., Sohngen, B. and Bosetti, V. (2007) Forestry and emission reduction. Mitigation and Adaptation Strategies for the carbon market response to stabilize climate. Energy Policy Global Change 19, 715-731. 35(11), 5346-5353.

[41] Ribot, J. and Larson, A.M. (2012) Reducing REDD risks: af- [59] Ortiz, M., Berrios, F., Campos, L., Uribe, R., Ramirez, A., Her- firmative policy on an uneven playing field. International Jour- mosillo-Núñez, B., González, J. and Rodriguez-Zaragoza, R. nal of the Commons 6(2), 233-254. (2015) Mass balanced trophic models and short-term dynam- ical simulations for benthic ecological systems of Mejillones [42] Cassels, S., Curran, S.R. and Kramer, R. (2005) Do migrants and Antofagasta bays (SE Pacific) : comparative network degrade coastal environments? Migration, natural resource ex- structure and assessment of human impacts. Ecological Mod- traction and poverty in North Sulawesi, Indonesia. Human and elling 309-310, 153-162. Ecological Risk Assessment 33(3), 329–363. [60] Chen, S.Q. and Chen, B. (2015) Urban energy consumption: [43] Gaveau, D.L.A., Linkie, M., Suyadi, L.P. and Williams, N.L. different insights from energy flow analysis, input-output (2009) Three decades of deforestation in Southwest Sumatra: analysis and ecological network analysis. Applied Energy 138, effects of coffee prices, law enforcement and rural poverty. Bi- 99-107. ological Conservation 142(3), 597–605.

[44] Elmhirst, R. (2011) Migrant pathways to resource access in Lampung’s political forest: Gender, citizenship and creative conjugality. Geoforum 42, 173–183.

[45] Zhou, Y., Zheng, L.F., Zhou, X.N., Wu, Z.L., Zhou, C.J. and Li, D. (2014) Carbon emission of timber production operating system based on industry standards. Journal of Beihua Univer- sity (Natural Science) 15(6), 815-820.

[46] Zhou, Y., Li, D., Wu, Z.L., Zhou, C.J. and Zheng, L.F. (2014) Carbon emission in forest logging operation process. Forest Engineering 30(3), 1-5.

[47] Logging research institute in Heilongjiang Province in China. (2005) Energy consumption for wood production in forest region (LY/T 1444.1-1444.6-2005). Heilongjiang, China Stand- ards Press.

[48] China National Standards Management Committee. (2004) Thermal power plant technical and economic index calculation method (DL/T 904-2004). Beijing, China Standards Press.

[49] Wang, W., Lin, J.Y., Cui, S.H. and Lin, T. (2010) An overview of carbon footprint analysis. Environmental Science & Tech- nology 33(7), 71-77.

[50] Yang, R.S. (2011) Qingdao energy consumption factor contribution analysis and energy conservation countermeasures. Qingdao, Qingdao University. Received: February 28, 2015 Revised: April 28, 2015, June 22, 2015, July 08, 2015 [51] Zhang, Z.X., Zhou, X.N., Zhao, C. and Chen, Y.F. (2008) Se- Accepted: August 06, 2015 lecting of the optimum operation model of ecological harvesting and transportation in southern artificial forest area in China. Scintia Silvae Sinicae 44(5), 128-134. CORRESPONDING AUTHOR [52] Chen, J.S. (2011) Study on the influence of plantation operation on resource-environment-economics based on Life Cycle Li-feng Zheng Assessment theory. Jiangsu, Nanjing Forestry University. School of Transportation and Civil Engineering [53] Fang, K., Zhu, X.J. and Gao, K. (2012) Carbon footprint of Fujian Agriculture and Forestry University global electricity and its equivalent calculation. Chinese Jour- Fuzhou 350002, Fujian Province nal of Ecology 31(12), 3160-3166. P.R. CHINA [54] He, Y. (2005) Summary of estimation methods of the carbon stored in forests. World Forestry Research 18(1), 22-27. Phone, +86 15280085840 [55] Dong, J., Jiang, Y.J. and Zhang, S.L. (2010) Developing forest E-mail: [email protected], [email protected] industry low-carbon economy to strengthen ecological civilization construction. Forestry Economics 6, 67-69. FEB/ Vol 24/ No 11a/ 2015 – pages 3743 - 3753

3753 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

MECHANISM OF GALLIC ACID EXTRACTION IN THE TRIBUTYL PHOSPHATE/KEROSENE EXTRACTING SYSTEM

Yundong Wu1, Kanggen Zhou1,*, Shuyu Dong1 and Wei Yu1

1 School of Metallurgy and Environment, Central South University, Changsha 410083, Hunan, China

ABSTRACT A significant amount of studies has been done on the separation and recovery of GA from wastewater. Of the nu- The extraction of gallic acid (GA) with tributyl phosphate merous removal methods that have been suggested [8-11], (TBP) is a neutral complex extraction process with good ex- solvent extraction was easy to perform and the evaluation traction performance. In this study, the mechanism of GA ex- was high. As reported in our previous study, the removal ef- traction with TBP was studied, the composition of the GA- ficiency of GA from GA processing wastewater with tributyl TBP-complex was determined by the double logarithmic phosphate (TBP)/kerosene was higher than 93% [12], which slope ratio and continuous variables methods, the structure of indicated that separation of GA by extraction had high appli- the GA-TBP-complex was determined by Fourier transform cation value. Currently, no study had reported the mech- infrared spectroscopy (FT-IR), and the enthalpy of the anism of extraction of GA with any extractant. extraction was measured. Our results showed that the Extraction mechanisms published with TBP as extract- proportion of gallic acid and TBP in the extracted complex ant had generally focused on the extraction of metal ions [13], was 2:1. The GA and TBP molecules were connected by which were basically connected by covalent bonds, so the intermolecular hydrogen bonds in the GA-TBP-complex. mechanism reported could not be used to explain the mecha- The enthalpy of the reaction was -1.064 kJ mol-1, and the nism of extraction of GA with TBP. Because the GA mole- nature of the extraction was an exothermic process. The re- cule had one carboxyl group (-COOH) and three hydroxyl sults of this work were useful for the design of separating groups (-OH), and the TBP molecule had four available sites GA from wastewater. to form hydrogen bonds, we speculated that the extracted complex generated by GA and TBP might be connected by

KEYWORDS: tributyl phosphate; extraction; gallic acid; extracted one hydrogen bond or more. complex; enthalpy Research on the structure of the extracted complex

formed by GA and TBP was of great theoretical and practical 1. INTRODUCTION significance. In this study, the double logarithmic slope method and continuous variable method were used to study The separation of phenolic compounds from industrial the composition of the extracted complex; the FT-IR spectra wastewater has received considerable attention in recent of GA, TBP, and the GA-TBP-complex were obtained to years [1-5]. Gallic acid (GA) (Fig. 1) is a phenolic compound study the possible connections between TBP and GA in the that is also well-known as an important environmental extracted molecular complex. In addition, the enthalpy of contaminant because of its toxicity [6] and widespread ex- the TBP/kerosene gallic acid extraction system was analysed. istence in industrial wastewater and natural water-bodies, especially in the Chinese nutgall processing wastewater. Widespread concerns have been aroused about the emission 2. MATERIALS AND METHODS of GA. The biodegradability of GA is poor [7]; after being 2.1 Reagents discharged as waste, it could cause serious damage to the local environment. The GA standard was provided by China Pharmaceutical and Biological Products. GA, TBP, kerosene, NaOH, HCl, methanol and phosphoric acid were of analytical grade. All chemicals were used as received without further purification.

2.2 Analytical methods The GA in the aqueous phase was analysed by an Ag- ilent Series 1100 LC system (Agilent Technologies Inc., FIGURE 1 - The molecular structure of GA. Waldbronn, Germany) [14, 15], and the GA in the organic

phase was calculated through quality balance. TBP was * Corresponding author monitored by an Agilent 5890 (Agilent Technologies Inc.,

3754 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Wilmington, DE, USA) [16]. The Fourier transform infrared lgDM ( -1)lg[ GA ] constant Eq. (5) spectra (FT-IR) of TBP, GA and the GA-TBP-complex were ()aq recorded on a Nicolet IS10 FT-IR (Thermo Fisher Scientific The D value could be obtained by changing the GA con- Inc., Waltham, MA, USA) spectrophotometer in the 400- tent in the aqueous phase. The correspondent curve was ob- -1 4000 cm wavenumber range, and 128 scans were taken at tained by plotting the values of lgD and lg[GA](aq) (Fig. 2). −1 a resolution of 4 cm . From Eq. (3) we could also deduce the relationship between D and TBP content as Eq. (6): 2.3 Experimental setup

Experiments were carried out in separating funnels. lgDNTBP lg[ ]()org constant Eq. (6) Both the extraction time and hierarchical time were 10 min, The D value was obtained when the GA level was fixed and the temperature was 25 °C. and TBP content was different. The relationship between

2.4 Preparation of the GA-TBP-complex lgD and lg[TBP](org) is shown in Fig. 3. In addition, the D value of the extraction of GA by kerosene is shown in As the composition of kerosene was complex which Fig. 2. Because the D value was much smaller in the kerosene would interfere with the FT-IR results, to study the struc- extraction system than the TBP/kerosene extraction system, ture of the extracted complex accurately, pure TBP was the effect of kerosene on the extraction of either fixed GA used as the organic phase to extract GA. GA solution (10 or TBP levels was ignored. g·L-1) was prepared as the aqueous phase. The extraction was not stopped until the crystals precipitated. After that, the 0.8 0.0 separating funnels were allowed to stand for stratification for lg D 3 h, and then, the aqueous phase and the crystals were dis- lgD(Kerosene) -0.5 carded, and the remaining organic phase was the GA-TBP- 0.6 complex. The FT-IR spectra of TBP, GA, and the GA-TBP- complex were scanned and recorded. -1.0 0.4 D

lg -1.5 (Kerosene)

3. RESULTS AND DISCUSSION D lg 0.2 -2.0 3.1. Double logarithmic slope method We assumed that M molecules of GA were connected 0.0 -2.5 with N molecules of TBP in a GA-TBP-complex; and the 0.0 0.2 0.4 0.6 0.8 1.0 1.2 reaction could be expressed as follows: lg [GA] (aq) FIGURE 2 - Change of D value under different GA levels. MNGA(aq ) TBP (org) (GA M TBP N ) (org)

The equilibrium constant (K) and the distribution ratio 1.8 (D) of the extraction could be calculated as Eq. (1) and Eq. lg D (2), respectively: 1.5

([GA ]MN [ TBP ] )()org K  MN Eq. (1) 1.2 []GA()aq [ TBP ] ( org )

0.9

MGA([ ] [ TBP ] ) D

MN()org lg D  Eq. (2) 0.6 []GA ()aq

By merging Eq. (1) and Eq. (2), we got Eq. (3): 0.3 MN-1 DKMGATBP ··[] [ ] Eq. (3) ()aq ( org ) 0.0 -0.2 0.0 0.2 0.4 0.6 By taking the logarithm of both sides of Eq. (3), the lg [TBP] (org) relationship among the concentrations of D, K and GA was FIGURE 3 - Change of D value under different TBP levels. described by Eq. (4): The slope of the lg [GA](aq)-lg D line was 0.959, the lgDKMMGANTBP lg lg ( - 1) lg[ ]()aq  lg[ ] ( org ) value was taken as 1; so, we could get that M = 2 from Eq. Eq. (4) (5). The slope of the lg [TBP](org)- lg D line was 1.248, and Kerosene was used as the diluent, the TBP level was we could also take the value as 1. Thus, M: N = 2: 1, which 30%, and the relationship between D and GA content could indicated that the extracted complex was composed of one be derived as Eq. (5): TBP molecule and two GA molecules.

3755 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

3.2 Continuous variables method complex, and thus, the extraction formula could be expressed The GA levels in the aqueous phase and TBP contents as follows: in the organic phase were different in each treatment, but 2CHO7566()aq +TBP () org (CHO)72 5 (TBP) () org the total amount was 60 mmol in each group.

3.3 FT-IR spectra of GA, TBP and the GA-TBP-complex 6 The absorption of the carbonyl band group of the GA

) -1

-1 and GA-TBP-complex was 1706 cm (Fig. 5). The absorption 5 of the resonance of the aromatic ring was 1611 cm-1 and 1536 cm-1. The absorption of the 3,4,5-substituted benzene 4 ring was 692 cm-1. The absorption peak of the GA-TBP- complex at 1284 cm-1 was weaker and moved to 1249 cm-1; 3 meanwhile, a new absorption peak was formed at 1194 cm-1. An absorption peak at 1249 cm-1 was the absorption of P-O- 2 H, and one at 1194 cm-1 was P-O-Ar. The absorption peak of P=O in the GA-TBP-complex derived from 1284 cm-1 1 -1

GA in organic phase (mol·mol to 1253 cm , presumably because intermolecular hydrogen

0 bonds had been generated between the P=O bond of TBP 012345 and the -OH of GA, and the electron clouds had moved. n : n (mmol·mmol-1) Compared with the curve of pure TBP, the composite ab- GA TBP sorption peak of P-O and P=O in the GA-TBP-complex re- FIGURE 4 - Extraction equilibrium under different GA and TBP levels. mained the same, indicating that no covalent bond was generated during the extraction. Neither pure GA nor pure TBP -1 The GA levels dissolved in the organic phase did not had an absorption at 3210 cm , which was the absorption of increase after the GA: TBP ratio reached 2:1 (Fig. 4); thus, intermolecular hydrogen bonds, whereas the GA-TBP- we concluded that the ratio of GA to TBP molecules in the complex did have one, indicating that there were inter-mo- extracted complex was 2:1. Also from the result of the con- lecular hydrogen bonds generated between the GA and TBP tinuous variables method, it could be calculated that when molecules during the extraction process. Combined with the 30% TBP/kerosene was used as extractant, the capacity for previous results, we concluded that the GA and TBP mole- GA extraction was 374.30 g L-1 in the ideal case. cules were connected by the intermolecular hydrogen bonds, and the mechanism of GA extraction by TBP was The results of both the double logarithmic method and completion extraction. the continuous variables method showed that two molecules From the above results, we conjectured the possible of GA and one molecule of TBP composed the extracted structure of the extracted complex as drawn in Fig. 6.

FIGURE 5 - FT-IR spectra of GA, TBP and the GA-TBP-complex

3756 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Regression analysis of Eq. (9) was carried out with the least squares method. The slope of lnD - T-1 was 1.064, as H was the negative value of the slope; thus, the H value was -1.064 kJ mol-1, which indicated that the extraction was an exothermic process. The result showed that the extraction of gallic acid with TBP/kerosene was a spontaneous exothermic reaction.

4. CONCLUSION FIGURE 6 - Structure of the GA-TBP-complex.

3.4 Determination of the reaction heat (H) In this paper, the mechanism of the TBP/kerosene gallic extraction acid system was studied. The GA-TBP-com- We assumed that the extraction process of GA with plex was found to be composed of two GA molecules and TBP was a macro process, and the distribution coefficient one TBP molecule, and the equation of the extraction was (k) could be denoted as follows: as follows: H () 2C H O +TBP 2(C H O ) (TBP) kke RT Eq. (7) 7566()aq () org7 5 () org 0 By taking the logarithm of both sides of Eq. (7), we got The results of FT-IR showed that inside the extracted Eq. (8): complex, the GA and TBP molecules were connected by two intermolecular hydrogen bonds. The enthalpy of the H 1 reaction was found to be -1.064 kJ mol-1, and the extraction lnkk ln ( )( ) Eq. (8) 0 R T reaction was an exothermic process. Under the same extraction conditions, D and k would show a linear relationship as can be seen from Eq. (3), so Eq. (8) could be transformed into Eq. (9): ACKNOWLEDGMENTS H 1 lnDDn ln ( )( ) Eq. (9) 0 R T This study was supported by the Strategic Project of Sci- GA solution (10 g·L-1) was prepared and extracted ence and Technology of Hunan Province, China [Grant No with 30% TBP/kerosene and kerosene under different tem- 2014GK1059]. The authors would like to thank the members peratures. The relationship between temperature and D of the Research Group for Extraction and Separation value was presented with T-1 as the abscissa and lnD values Chemistry who co-operated with the present study. as the vertical axis. The authors have declared no conflict of interest.

2.0

TBP/kerosene 1.5 kerosene REFERENCES 1.0 [1] Yang, C.F., Qian, Y., Zhang, L.J, Feng, J.Z. (2006) Solvent 0.5 extraction process development and on-site trial-plant for phenol removal from industrial coal-gasification wastewater. D

ln -3.5 Chemical Engineering Journal. 117, 179-185. [2] Keshav, A., Chand, S., Wasewar, K.L. (2009) Reactive extrac- -4.0 tion of acrylic acid using tri-n-butyl phosphate in different dil- uents. Journal of Chemical & Engineering Data. 54, 1782- -4.5 1786.

-5.0 [3] Brinda L.A., Balasubramanian, A., Venkatesan, S. (2013) Ex- 28 29 30 31 32 33 34 35 36 traction of Phenol and Chlorophenols Using Ionic Liquid T-1(10-4K-1) [Bmim]+[BF4]− Dissolved in Tributyl Phosphate. CLEAN– Soil, Air, Water. 41, 349-355. FIGURE 7 - The extraction of GA with TBP/ kerosene and kerosene under different temperatures. [4] Kloetzer, L., Postaru, M., Galaction, A.I., Blaga, A.C., Cas- caval, D. (2013) Comparative study on rosmarinic acid sepa- Because the D value of the extraction with kerosene ration by reactive extraction with Amberlite LA-2 and D2EHPA. 1. Interfacial reaction mechanism and influencing was almost a constant (Fig. 7), thus the impact of kerosene factors. Industrial & Engineering Chemistry Research. 52, in the TBP/kerosene extracting GA system was ignored. 13785-13794.

3757 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[5] Mohammadi, S., Kargari, A., Sanaeepur, H., Abbassian, K., Najafi, A., Mofarrah, E. (2014) Phenol removal from industrial wastewaters: a short review. Desalination and Water Treatment. 53, 2215-2234.

[6] Field, J.A., Lettinga, G. (1992) Toxicity of tannic compounds to microorganisms. 59, 673-692.

[7] Yu, W., Zhou, K.G., Wu, Y.D., Du, H. (2013) Extration technology for recovery of gallic acid from galic mother liquor. CIESC Jnournal. 64, 1660-1664.

[8] Boye, B., Brillas, E., Buso, A., Farnia, G., Flox, C., Giomo, M., Sandonà, G. (2006) Electrochemical removal of gallic acid from aqueous solutions. Electrochimica acta. 52, 256-262.

[9] Lin, Y.B., Fugetsu, B., Terui, N., Tanaka, S. (2005) Removal of organic compounds by alginate gel beads with entrapped activated carbon. Journal of Hazardous Materials. 120, 237- 241. [10] Benitez, F.J., Real, F.J., Acero, J.L., Leal, A.I., Garcia, C. (2005) Gallic acid degradation in aqueous solutions by UV/H< sub> 2 O< sub> 2 treatment, Fenton's reagent and the photo-Fenton system. Journal of Hazardous Materials. 126, 31-39.

[11] Cláudio, A.F.M, Ferreira, A.M., Freire, C.S.R., Silvestre, A.J.D., Freire, M.G., Coutinho, J.A.P. (2012) Optimization of the gallic acid extraction using ionic-liquid-based aqueous two-phase systems. Separation and Purification Technology. 97, 142-149. [12] Wu, Y.D., Zhou, K.G., Dong, S.Y., Yu, W., Zhang, H.Q. (2015) Recovery of gallic acid from gallic acid processing wastewater. Environmental technology. 36, 661-666. [13] Rout, A., Binnemans, K. (2014) Separation of rare earths from transition metals by liquid–liquid extraction from a molten salt hydrate to an ionic liquid phase. Dalton Transactions. 43, 3186-3195. [14] Weidenhamer, J.D., Li, M., Allman, J., Bergosh, R.G., Posner, M. (2013) Evidence Does not Support a Role for Gallic Acid in Phragmites australis Invasion Success. Journal of Chemical Ecology. 39, 323-332.

[15] Queimada, A.J., Mota, F.L., Pinho, S.P., Macedo, E.A. (2009) Solubilities of biologically active phenolic compounds: measurements and modeling. The Journal of Physical Chemistry B. 113, 3469-3476. [16] Bajoria, S.L., Rathod, V.K., Pandey, N.K., Mudali, U.K., Na- tarajan, R. (2011) Equilibrium Study for the System Trinbutyl Phosphate, Normal Paraffin Hydrocarbon, and Nitric Acid. Journal of Chemical & Engineering Data. 56, 2856-2860.

Received: March 03, 2015 Revised: March 23, 2015 Accepted: May 22, 2015

CORRESPONDING AUTHOR

Kanggen Zhou School of Metallurgy and Environment Central South University Changsha 410083, Hunan CHINA

E-mail: [email protected]

FEB/ Vol 24/ No 11a/ 2015 – pages 3754 - 3758

3758 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

CONTENT OF ZINC IN DIFFERENT GRASS SPECIES GROWING ALONG A FAST HIGHWAY

Elżbieta Malinowska1,*, Kazimierz Jankowski1, Beata Wiśniewska-Kadżajan1, Jolanta Jankowska2, Jacek Sosnowski1, Grażyna Anna Ciepiela3, Wiesław Szulc4 and Roman Kolczarek1

1Department of Grassland and Landscape Architecture, Institute of Agronomy, Siedlce University of Natural Sciences and Humanities, Prusa 14 Street, Siedlce, Poland 2Laboratory of Agrometeorology and Fundamentals of Land Reclamation, Siedlce University of Natural Sciences and Humanities, Prusa 14 Street, Siedlce, Poland 3Department of Tourism and Recreation, Siedlce University of Natural Sciences and Humanities, Prusa 14 Street, Siedlce, Poland 4Department of Soil Environment Sciences, Warsaw University of Life Sciences, Nowoursynowska 159, Warsaw, Poland

ABSTRACT fast highways in good condition, most often different sorts of grass are used. Among heavy metals of traffic origin the This paper deals with the influence of traffic on bioac- most dangerous are: lead, cadmium, zinc, copper and cumulation of zinc in three grass species: Dactylis glomer- nickel. The content of these elements in soil along traffic ata, Arrenatherum elatius, Alopecurus pratensis. Plant ma- routes is usually monitored. terial samples were collected in 2011, along the interna- Zinc is classified as a plant nutrient of high mobility in tional road E30, a section of 9 km of the bypass of Siedlce. soil [9, 10]. Thereby, one of its features is the fact that it is Plants in the flowering stage were collected at the follow- easily absorbed by plants because it is present in the form ing distances from the road: 1, 5, 10 and 15 m. Then they of easily soluble compounds. Organic substances, as well were divided into three parts: stems, leaves and inflores- as Fe and Mn oxides, form stable compounds with zinc [9, cences. The zinc concentration was determined with the 11-13]. Generally, plants are tolerant to an increased con- AAS method, after earlier dry mineralization. Zinc bioaccu- centration of zinc as by binding it in cell membranes they mulation in tested grasses differed significantly according to exclude it from physiological processes. They can absorb the distance from the road. On average, in Arrenatherum ela- this nutrient through roots and leaves in the form of Zn2+ tius biomass, most zinc was accumulated in the immediate ions or in the form of zinc chelates [14]. A high concentra- vicinity of the road (1m) and for other species within the area tion of zinc in soil is harmful to plants because of its easi- of 15 m away. The results showed that most zinc is accumu- ness of accumulation in vegetative and generative parts, lated by leaves, then by inflorescences and stems. which causes worsening of plant quality [15, 16]. A proper selection of plant species growing by roadsides may limit KEYWORDS: bioaccumulation of metals and, consequently, make those zinc, grass species, morphological parts, highway, Poland plants less harmful to animals feeding on them. The aim of the research was to determine the relation 1. INTRODUCTION between zinc bioaccumulation in different parts of three grass species: Dactylis glomerata, Arrenatherum elatius, The development of municipal infrastructure, includ- Alopecurus pratensis and the distance of their growing ing traffic infrastructure, is an important source of heavy place from the highway. metals pollution in the environment. According to many publications [1-6], the main sources of heavy metals are dust and gasses emitted by motor vehicles. Heavy metals 2. MATERIALS AND METHODS come from wearing of different elements of the engine but also as a result of traffic infrastructure exploitation, a con- The plant material for the research was morphological sequence of tarmac surface wearing out or corrosion of parts of the following grass species in the flowering stage: acoustic screens, road railings and traffic signs. Those sub- Dactylis glomerata, Arrenatherum elatius and Alopecurus stances settle on road surfaces or are spread in the immedi- pratensis. Samples of plants were taken along the interna- ate neighborhood of the road [7, 8]. Heavy metals accumu- tional E30 road, the ring-road of Siedlce (Fig. 1) in May 2011. late in soil or in the form of falling dust are absorbed by A 9 km road section was examined with samples collected on plants growing near roads. In order to keep roadsides along both sides covering a stretch of 700 m, at the following distances from the edge of the road: 1, 5, 10 and 15 m. The total * Corresponding author number of samples was 96.

3759 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

FIGURE 1 - Schematic representation of the studied sites of the E30 road [17]

The E30 road is a part of the international traffic route selected. Next, plant material was dried in an oven (SUSLAB from Cork (Ireland) to Omsk (Russia). –PLE-406, SLW 1000 STD) at 105oC for 48 hours. It was It passes through five provinces but the range of re- ground to 0.25mm particle size with 1-g samples weighed out search covers only a section of the road in the Masovian and poured into stoneware crucibles, and then organic matter o province and is located in the central-eastern part of Po- was ashed at 450 C in a muffle furnace (L3/11) for 18 hours, land, approx. 80 km east of Warsaw. In 2010 the General diluted in 10% HNO3 and transferred to 100 ml volumetric Directorate for National Roads and Motorways (GDNR & flask. M) counted the number of vehicles (SDR) on Polish roads, Zinc concentration was marked with the AAS method, including the tested section (Table 1). The average of three using the Varian Spectra AA20 spectrophotometer (VAR- counts of the daily number of motor vehicles (SDR) on in- IAN, Australia), using the standards of the Merck com- ternational roads in 2010 was 16667 vehicles/day, whereas pany. The parameters of the spectrophotometer are: VAR- on remaining national roads 7097 vehicles/day [17]. On the IAN, Australia; range of 190 nm to 900 nm, equipped with analysed section of road (the ring-road of Siedlce), the aver- a graphite cuvette type GTA-96. age daily movement of motor vehicles was higher than the An internal quality control system was used for valida- average on national roads of Poland and it was 8136 vehi- tion of analytical methods. It was assumed that the average cles/day [17]. recovery of an internal standard should be in the range of The average concentration of organic carbon in the soil 85 to 115%) of the true value. For the needs of the research, sampled 1, 5, 10 and 15m from the road was 9.05 g.kg-1, two standards of quality control were prepared. One stand- while total nitrogen concentration was 0.57 g.kg-1, with ard was added to a known amount of analyte. A recovery pHKCl of 4.10. The average zinc concentration in the tested was calculated after the analysis of the sample with and soil was 20.41 mg.kg-1. Mostly it was bulk soil which de- without the analyte. The average recovery was dependent veloped from sand. Samples of the abovementioned grass on the analyte content in the sample. The analysis was per- were divided into three morphological parts: leaves, stems formed in each series of samples and the obtained recovery and inflorescences. Then, out of all samples, five samples was included in the assumed range. for each plant species and each distance from the road were

TABLE 1- The average daily traffic of motor vehicles (SDR) in 2010, on the ring road of Siedlce [17]

Trucks Measur- Length The name of the Total Motorcycles Automobiles Light trucks without with a Buses ing point (km) measuring point a trailer trailer Siedlce/ 11517 5.65 8938 32 5827 970 395 1665 45 Ring Road 1 Siedlce/ 11518 5.00 8202 25 5314 874 352 1597 39 Ring Road 2 Siedlce/ 11504 2.60 7267 29 4490 696 413 1594 41 Ring Road 3 Average 8136 29 5210 847 387 1619 42

3760 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

The obtained data were elaborated statistically with the A high concentration of zinc and other elementary sub- use of the Statistica software, Version 10.0 StatSoft. Arith- stances, especially lead, in plants growing along highways metical means, coefficients of variation (V) and the geoac- is the result of their morphological features and growing cumulation index (Igeo) were calculated. According to Mül- traffic [24, 25]. According to the General Directorate for ler [18], the geoaccumulation index, Igeo, is calculated ac- National Roads and Motorways, in the years 2005-2010 the cording to the formula: Igeo= logCn/1.5Bn, where Cn indi- motor vehicles traffic on the national road network in- cates concentration of elementary substance in a given el- creased by 22%. Ciećko et al. [26] say that zinc concentra- ement of the environment and Bn – geochemical back- tion in grass growing along the national road No. 7 Ol- ground of a given elementary substance. The geoaccumu- sztynek-Płońsk is almost twice lower than the concentra- lation index calculated by this method is presented by Kró- tion along the ring-road of Siedlce. lak [19]. Analyzing the general concentration of zinc in the Influence of the tested factors on zinc bioaccumulation whole biomass of three grass species in relation to the dis- was examined with the bi-factor analysis of variance. For tance from the road, it may be said that in the case of Dac- the purpose of detailed comparison of means Tuke’y test tylis glomerata there was a proportional increase of zinc was carried out at p≤0.05. concentration together with the growing distance (Table 2). Within 1m from the road the concentration of this metal was 68.3 mg . kg-1, and 15 m from the road it was 3. RESULTS AND DISCUSSION 104.2 mg . kg-1. Arrenatherum elatius reacted conversely; on average, most zinc was accumulated in plants growing The average zinc concentration in the tested species of close to the road, 81.9 mg . kg-1. In turn, in Alopecurus grass was 77.8 mg . kg-1 DM (Fig. 2). In Dactylis glomerata pratensis the higher concentration of zinc was noted at the a higher bioaccumulation of this elementary substance was farthest distance from the road, 136.0 mg . kg-1 on average. noted than in the remaining species. Wowkonowicz et al. [27] finds twice lower concentration Kabata-Pendias and Pendias [12] found that the av- of zinc in grass collected at a considerable distance (250 m erage zinc concentration in grass ranges between 3.7 and to 10 km) from highway roads to Warsaw. However, 292 mg . kg-1, and maximal natural concentration of this Słowik et al. [28] notices spatial diversification of the con- elementary substance for grass is 72 mg . kg-1 DM. In Po- centration of heavy metals in soil depending on the dis- land in areas not exposed to pollution, natural concentra- tance from the road. The distance significantly influences tion of zinc ranges from 10 to 50 mg . kg-1 DM [20], de- bioaccumulation in plants and the authors note that the pending on the plant species. According to Cottenie et al. highest concentration of heavy metals is within 40 m from [21] and Ilyin [22] the concentration of zinc in plant bio- the road. mass is 25-150 mg . kg-1 DM. This study has shown that Nevertheless, within 100-120 m from the road, there zinc concentration in Dactylis glomerata (86.1 mg . kg-1) are no increased concentration exceeding the geochemical and Alopecurus pratensis (81.7 mg . kg-1) was above the background. The literature data show that accumulation of natural concentration. According to the Guidelines for Ag- heavy metals in plants does not always increase proportion- riculture [23] admissible concentration of this metal is ally to their content in the soil. There are soil features and 41.2-67.2 mg·kg-1DM in wheat grain, potatoes and grasses. plant species features which determine it [20, 29].

100

80 DM) DM)

-1 60 kg . 40

20 Zn (mg Zn (mg

0 Dactylis Arrenatherum Alopecurus mean glomerata elatius pratensis

FIGURE 2 - The average Zn concentration (mg . kg-1 DM) in selected species of grass growing in the roadside area along the highway E30

3761 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 2 - Relation between the average concentration of Zn (mg . kg -1 DM) in selected species of grass and the distance from the road

Coefficient of Distance from the road (m) Plant species Zinc variation [%] Dactylis glomerata 68.28 30.02 1 m Arrenatherum elatius 81.88 59.95 Alopecurus pratensis 53.42 11.12 Average for 1 m 67.86 33.70 Dactylis glomerata 77.56 48.36 5 m Arrenatherum elatius 68.65 37.35 Alopecurus pratensis 71.05 34.75 Average for 5 m 72.42 40.15 Dactylis glomerata 94.19 69.53 10 m Arrenatherum elatius 43.62 22.72 Alopecurus pratensis 66.35 38.98 Average for 10 m 68.06 43.74 Dactylis glomerata 104.27 12.59 15 m Arrenatherum elatius 68.74 40.91 Alopecurus pratensis 135.98 36.93 Average for 15 m 102.99 30.14

LSD0.05 for: A – distance A = 3.28 B – grass species B = 2.57 AxB - interaction AxB = 5.68 BxA = 5.14

0,045 0,04

x 0,035 0,03 0,025 0,02 0,015

Geocummulation inde 0,01 0,005 0 1 m 5 m 10 m 15 m

Dactylis glomerata Arrenatherum elatius Alopecurus pratensis

FIGURE 3 - Value of the geoaccumulation index (I geo) for average zinc concentration in the chosen species of grass

. -1 The value of the geoaccumulation index (I geo) for aver- observed within 10 m from the road (175.1 mg kg DM), age zinc concentration in the chosen species of grass grow- contrary to the leaves in which the concentration of zinc ing at particular distances from the road was between 0.034 was the smallest (29.0 mg . kg-1 DM). In Dactylis glomer- and 0.044 (Figure 3). As a reference value, the average nat- ata samples collected from 15 m distance, the concentra- ural concentration of zinc in Polish soil was taken into con- tion of zinc significantly increased in comparison with the sideration, which is 32.4 g . kg-1 [30]. According to Müller plants growing in the nearest neighborhood of the road. [18], the values indicate that tested grasses belong to the 1st Among the analysed parts of Arrenatherum elatius, the class of the geoaccumulation index (0< I geo<1), i.e. they are highest concentration of zinc was found in stems within 1m slightly polluted. Królak [19] notes similar values of I geo for from the road (Figure 4b). Within 10m from the highway, zinc in dandelion leaves on South Podlasie Lowland. the bioaccumulation of the tested elementary substance in The distance from the highway considerably differen- stems decreased over four times and in inflorescence over tiated the content of zinc in all analysed parts of Dactylis 2 times. Within 15m zone, the content of zinc increased in glomerata (Fig. 4a). In stems the content of the tested metal inflorescence of Arrenatherum elatius whereas in leaves, grew proportionally along with the distance from the road. along with the distance from the road, the increase of zinc In inflorescences, the greatest bioaccumulation of zinc was from 20.1 to 72.5 mg . kg-1 DM was noted.

3762 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

a. b.

Dactylis glomerata Arrenatherum elatius 210 140 180 -1 120 -1 150 100 kg

. 120 80 90 60 60 mg Zn . kg mg 40

mg Zn 30 20 0 0 1 5 10 15 1 5 10 15 distance from the road (m) distance from the road (m)

stems leaves inflorescences stems leaves inflorescences

A - distance from the road A = 10.12 B – morphological part A - distance from the road A = 4.15 B – morphological part B = 7.93 AxB = 17.53 BxA = 15.87 B = 3.52 AxB = 7.19 BxA = 6.51

c. d.

Alopecurus pratensis average for grass species 210 150

-1 180 -1 120

kg 150 kg . 120 . 90 90 60

mg Zn 60 mg Zn 30 30 0 0 1 5 10 15 1 5 10 15 distance from the road (m) distance from the road (m) stems leaves inflorescences stems leaves inflorescences

A - distance from the road A = 4.65 B –morphological part A - distance from the road A = 33.43 B – morphological part B = 3.64 AxB = 8.05 BxA = 7.29 B = n.s. AxB = n.s. BxA = n.s.

FIGURE 4 - The concentration of zinc (mg . kg-1 DM) in morphological parts of selected grass species

Essential differences in the concentration of the tested Germany, Wybieralski and Maciejewska [31] observed elementary substance in morphological parts of Alopecu- that environmentally friendly values of lead were ex- rus pratensis as well as at particular distances from the road ceeded and zinc and copper concentration was a little (Figure 4c) were noticed. Compared to the concentration of too high too. zinc in the grass growing 1m from the road the same concentration was higher over 3 times in leaves, over 2 times in Many publications [32, 33] underline the danger of inflorescence and almost 2 times in stems within 15 m from heavy metals to the environment around petrol stations. the road. Therefore, it is very important to determine the scale of accumulation of those elementary substances in the plants as The average concentration of zinc in stems of the tested a result of traffic pollution. In particular, leaves accumulate grass species collected at four distances did not exceed dusts because they have bristles, grooves and cilia so it is 80 mg . kg-1 DM (Figure 4d). In comparison to the plants important to introduce proper grass species absorbing the growing closest to the road, there was a considerably least amounts of heavy metals so that they can be used as greater concentration of zinc in leaves and inflorescences fodder for animals. of grass growing 15m from the road. Concentration of zinc within 15m was, respectively, 130.3 and 101.4 mg . kg-1 DM, and within 1m 56.0 and 70.5 mg . kg-1 DM, respec- 4. CONCLUSIONS tively. Concentration of zinc was probably dependent on morphological features of plants and the intensity of traf- The tested grass species Dactylis glomerata, Arrenath- fic. Those results were confirmed by many authors [25, erum elatius, Alopecurus pratensis growing in the roadside 27]. Kabata-Pendias and Pendias [12] noted that exces- area along the fast highway E30 (the ring-road of Siedlce) sive quantities of zinc were most often in leaves or in had a high concentration of zinc. The average concentra- roots. In Rosówek, on the border between Poland and tion of this elementary substance in Dactylis glomerata and

3763 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

Alopecurus pratensis was above the natural concentration [12] Kabata – Pendias, S. and Pendias, H. (1999) Biogeochemistry of zinc for grass. of trace elements. PWN, Warsaw (in Polish). Distance from the road considerably differentiated bi- [13] Kalembasa, D. and Malinowska, E. (2013) Bioaccumulation of zinc under the influence of sewage sludge and liming and oaccumulation of the tested elementary substance in grass. its speciation in soil. Fresenius Environ. Bull., 22(11a), 3359- In Arrenatherum elatius biomass, on average the highest 3369. zinc concentration was noted in places close to the road [14] Lityński, T. and Jurkowska, H. (1982) Fertility of the soil and (1 m), for the remaining species the highest concentration plant nutrition. PWN, Warsaw (in Polish). was within 15 m from the road. [15] Doğanlar, Z. and Atmaca, M. (2011) Influence air borne pollution on Cd, Zn, Pb, Cu and Al accumulation and physiolog- For all the tested grass species, on average the highest ical parameters of plant leaves in Antakya (Turkey). Water zinc concentration was in leaves, so in parts of plants most Soil Poll. 214, 1/4, 509-523. eagerly eaten by animals, whereas in inflorescences and [16] Rutkowska, B., Szulc W., Łabętowicz, J. and Pikuła D. (2010) stems the concentration of zinc was lower. Bioaccumulation index as criteria for assessment of accumulation of copper and zinc in biomass of pasture plants. Frese- Given the increasing rate of road traffic, the chemical nius Environ. Bull., 19(4), 620-623. composition of plants growing along the roadway and their [17] General Directorate for National Roads and Motorways, use as animal fodder should be systematically monitored. (2010) The average daily traffic in Poland 2010. www.gdd- kia.gov.pl (in Polish). The authors have declared no conflict of interest. [18] Müller, G. (1979) Schwermetalle in den Sedimenten des Rheins-Veränderungen seit 1971. Umschau, 79, 778-783. [19] Królak, E. (2004) The relationship between the content of heavy metals in total precipitation, soil and plant indicator Ta- REFERENCES raxacum sp. on South-Podlasie Lowland. Habilitation thesis no. 75, pp. 123 (in Polish). [1] Ho, Y.B. and Tai, K.M. (1998) Elevated levels of lead and [20] Kabata – Pendias, A. (2002) Biogeochemistry of zinc. Zinc in other metals in roadside soil and grasses and their use to mon- the environment - environmental problems and methodical. itor aerial metal depositions in Hong Kong. Environ. Pollut., Adv. Sci. Committee of the Polish Academy of Sciences “Hu- 49, 37-51. man and the environment”, PAN, Warsaw, 33, 11-18 (in Polish). [2] Diatta, J.B., Grzebisz, W. and Apolinarska K. (2003) A study of soil pollution by heavy metals in the city of Poznań (Poland) [21] Cotteinie, A., Dhaese, A. and Camerlynck R. (1976) Plant using dandelion (Taraxacum officinale Web.) as a bio- quality response to the uptake of polluting elements. Qual. indycator. EJPAU, 6(2), #01. Plantarum, 80, 293-319. [3] Swaileh, K.M., Hussein, R.M. and Abu-Elha, S. (2004) Asses- [22] Ilyin, U.B. (1991) Heavy metals in the system soil-plants. ment of heavy metal contamination in roadside surface soil Nauka, Novosibirsk, 150 pp. and vegetation from the West Bank. Arch. Environ. Contam. Toxicol. 47, 23-30. [23] Framework guidelines for agriculture (1993): assessment of the state of pollution of soils and plants heavy metals and sul- [4] Gracia, R. and Milan, E. (2005) Assessment of Cd, Pb and Zn fur. JUNG, Puławy (in Polish). in roadside soils and grasses from Gipuzkoa (Spain). Chemo- sphere, 37, 1615-1625. [24] Arslan, H., Güleryüz, G., Leblebici, Z., Kırmıcı, S. and Aksoy, A. (2010) Verbascum bombyciferum Boiss. (Scrophulariacea) [5] Jozic, M., Peer, T. and Türk, R. (2009) The impact of the tun- as possible bio-indicator for the assessment of heavy metals in nel exhausts in terms of heavy metals to the surrounding eco- the environment of Bursa, Turkey. Environ. Monit. Assess. system. Environ. Monit. Assess. 150, 261-271. 163, 105-113. [6] Deska, J., Bombik, A., Marciniuk-Kluska, A. and Rymuza, K. [25] Jankowski, K., Ciepiela, G.A., Jankowska, J., Kolczarek, R., (2011) Trends in lead and cadmium contents in soils adjaced to Wiśniewska –Kadżajan, B., Sosnowski, J. and Deska J. (2013) Europen Track E30. Polish J. Environ. Stud., 19(2), 317-325. Lead and cadmium content in selected dicotyledonous plants from areas adjacent the fast traffic route. Wulfenia Journal [7] Ciesielczuk, T., Kusza, G., Kowalska-Góralska, M. and Senze, 20(6), 288-302. M. (2011) Aluminium of selenium content of soils of industrial area in Opole (southern Poland). Arch. Environ. Prot. [26] Ciećko, S., Żołnowski, A. and Wyszkowski, M. (2004) The 39(2), 97-104. content of heavy metals in soils and plants at the national road [8] Luginina, E.A. and Egoshina, T.L. (2013) The peculiarities of No. 7 Olsztynek-Płońsk. Integrated Monitoring of Environ- heavy metals accumulation by wild medicinal and fruit plants. ment. Operation and monitoring geoecosystems under increasing anthropopression. Environmental Monitoring Library, Ann. Warsaw Univ. of Live Sci. – SGGW, Agricult. 61, 97- Toruń, 5-17 (in Polish). 103. [9] Domańska, J. (2009) Soluble forms of zinc in profiles of se- [27] Wowkonowicz, P., Malowaniec, B. and Niesiobędzka, K. lected types of arable soils. J. Elem. 14(1), 63-70. (2011) Heavy metals in plants and soils permanent grassland in the area of Warsaw. Environ. Prot. Natural Resources, 49, [10] Rutkowska, B., Szulc, W. and Bomze, K. (2013) Plant availa- 309-319 (in Polish). bility of zinc in differentiated soil conditions. Fresenius Envi- ron. Bull., 22(9), 2542-2546. [28] Słowik, T., Jackowska, I. and Piekarski, W. (2008) The problems of environmental pollution by transport infrastructure for [11] Spiak, Z. (1998) Influence of soil pH to uptake of zinc by example Roztoczański National Park. Rozp. i Monografie, plants. Adv. Agric. Sci. Prob., 456, 439-443 (in Polish). 165(5), pp. 138 (in Polish).

3764 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[29] Jankowski, K., Jankowska, J., Ciepiela, G.A., Sosnowski, J., Wiśniewska-Kadżajan, B., Kolczarek, R. and Deska, J. (2014) Lead and cadmium content in some grasses along expressway areas. J. Elem. 1, 119-128.

[30] Terelak, H., Motowicka-Terelak, T., Stuczyński, T. and Pie- truch Cz. (2000) Trace elements (Cd, Cu, Ni, Pb, Zn) in agricultural soils Polish. Library of Environmental Monitoring. In- spection of Environmental Protection, Warsaw, pp. 69 (in Polish). [31] Wybieralski, J. and Maciejewska, M. (2001) Research the level of heavy metal contamination of soil and plants in the border areas in Rosówek near Szczecin. Ecol. Chem. Eng., 8(7), 741-748 (in Polish).

[32] Wybieralski, J. and Maciejewska, M. (1999) Assessment environmental contamination by heavy metals near the petrol station and around Szczecin. Ecolog. Techn. 7(1), 20-23 (in Polish).

[33] Antonkiewicz, J. and Macuda, J. (2005) The content of heavy metals and hydrocarbons in soils adjacent to the selected fuel station in Krakow. Acta Sci. Pol., Formatio Circumiectus 4(2), 31-36 (in Polish).

Received: March 10, 2015 Revised: May 08, 2015 Accepted: June 10, 2015

CORRESPONDING AUTHOR

Elżbieta Malinowska

Institute of Agronomy

Siedlce University of Natural Sciences and Humanities

Prusa 14 Street

Siedlce, 08-110

POLAND

Phone: (+48) 256431319

E-mail: [email protected]

FEB/ Vol 24/ No 11a/ 2015 – pages 3759 - 3765

3765 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

IMPACT OF POSTPONED WATER FEEDING TIME ON THE INFILTRATION-REDUCING EFFECT OF CEMENT INTO ALLUVIAL SOILS

Jinlan Ji1,2,*, and Guisheng Fan1,*

1 College of Environmental Science and Engineering, Taiyuan University of Technology, Taiyuan 030024, Shanxi Province, China 2 Jincheng Institute of Technology, Jincheng 048026, Shanxi Province, China

ABSTRACT water contamination has mainly been focused on the treatment of groundwater pollution and soil pollution. Helalia et Based on a series of indoor simulating infiltration tests, al. [5] and other researchers [6-9] have proven that changing which experienced different postponed water feeding times the physical properties of topsoil on the river bed could en- after fine particulate matters infiltrating into the river allu- hance the infiltration-reducing effect. Keijzer et al. [10, 11] vial sandy soils, the impact of postponed water feeding believe that small pores of clay soil show certain membra- time to infiltration-reducing in the case of cement infiltrat- nous properties, making it considerably potential in the prac- ing into alluvial soils is revealed, and its mechanism is ex- tical application of clay barriers. Some researchers [12] also plained as well. The results show preliminarily that (1) suggest the crust of soil surface decreases its infiltration ca- compared to the method of instant water feeding after ce- pacity dramatically while some others [13-14] claim that ment infiltrating into the riverbed alluvial soil, the method the infiltration properties of soil change when they are ad- of postponed water feeding may enhance the infiltration- mixed or covered with foreign substances. Zhang et al. [9] reducing effect 2 ~ 4 times on both tap water and sewage proposed the new concept of ‘isolating pollutants, and thus, infiltration process, and extending the postponed time can sewage water can be allowed to enter groundwater, and de- enhance infiltration-reducing effect; (2) the infiltration-re- velop the techniques of adding a variety of chemical curing ducing rate is exponentially positive correlated to the post- agents into the soil. However, how to enhance the infiltra- poned time and the reducing effect reaches to an ideal level tion ability of alluvial silt to block surface pollutants mi- when the postponed time rises to 12 h; however, the reduc- grating into deeper layers of soil or groundwater through ing effect is not obvious any more when the time is exceed- soil-water convection has gradually become one of the ing 24 h, especially for exceeding 36 h; (3) the trend of the main concerns in the studies of reduction and control of infiltration-reducing rate changing is consistent after ce- groundwater pollution. From our previous research results, ment infiltrating into different textures of alluvial soil, yet, it can be shown preliminarily that using cement to gradu- the postponed time that different soils reach to the ideal re- ally infiltrate or ooze into soil is effective and applicable ducing effect is not the same. The results may provide a for the infiltration-reducing effect of the soil, especially the useful reference for the study of cutting pollutants migrat- loosely-structured sand silt [15]. However, the postponed ing into the groundwater through surface water infiltration. water supply time obviously impact the infiltration-reducing effect of the sand silt. In this paper, based on laboratory

experiments of fine cement particulate substance infiltrat- KEYWORDS: Cement, alluvial silt, infiltration-reducing effect, post- ing into sand silt with different delaying times for water poned water feeding time supply, we attempt to reveal and explain how postponed time influences the infiltration-reducing effects.

1. INTRODUCTION 2. MATERIALS AND METHODS In the process of saturated/unsaturated water infiltration, pollutants in surface water could accumulate down 2.1 Test material into the soil and infiltrate into groundwater. Consequently, The Fen River is the largest tributary of the Yellow river channels or ditches that convey polluted surface water River in Shanxi Province, and is also the largest river itself inevitably become the main sources of groundwater contam- in Shanxi. Soil samples were taken from the alluvial silt ination [1-4]. Research on reducing and controlling ground- from the sections of both the main stream of Fen River and the Xiao River (the largest tributary of the Fen River). The * Corresponding author locations included the Eddy Dam of Xiao River at Yuci

3766 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 1 - Grain size distribution of the test materials

Particle Size ≤2 μm 2-20 μm 20-65 μm 65-2000 μm ≥2000 μm D90[a] /μm Dav[b]/μm Material property

Material Cement 0.26% 58.13% 40.70% 0.91% 0 39.93 20.69 Ordinary Portland cement Soil 1 14.38% 13.04% 72.3% 72.3% 0.22% 204.59 131.18 Sandy loam Soil 2 10.11% 9.44% 80.17% 80.17% 0.28% 228.67 165.05 Sandy loam Soil 3 15.90% 9.82% 74.05% 74.05% 0.23% 207.96 135.74 Sandy clay loam The data in the above table are average data calculated from multiple device measurements. [a] D90 = 90% of the grain diameter is less than this value (μm); [b] Dav = volume average grain size (μm).

City, the Fen River at Qingxu County, and the Fen River at The infiltration soil column is composed of the water supply Qingyuan Town, Qingxu County (herein referred to as Soil chamber, the infiltration column, and the drainage chamber, 1, 2, and 3, respectively). Samples were collected from 0- which are connected by flanges. There are three controllable 30 cm depth from the surface layer. In order not to destroy in-and-out holes for supplying water, adding cement slurry, the soil structure, the tested alluvial silt was not dried or and exhausting, respectively. The infiltration column (Φ 10 sifted. Only the big chunks of soil particles were pulverized cm × H 100 cm) is designed with two rows of evenly dis- and the large debris (e.g. plant roots, stones etc.) were elim- tributed holes from top to bottom for the purpose of sam- inated. The tested fine particle substance was an ordinary pling water and soil at the proper time and positions. The Portland cement purchased from the market (model: P·O drainage chamber is designed to measure and discharge the 42.5). The cement was sealed for preservation, and screened yielding water from the soil column, ensuring the aeration by a sieve with 0.075 mm round holes before use. The test of soil. The main unit of the water supply system is a Mar- water for infiltration was tap water (pH value: 7.2, TDS kov tube (Φ 3 cm × H 100 cm), which provides a constant value: 320 mg/L) and the natural sewage effluent (pH: 7.8, water head for the soil column while supplying and meas- TDS: 917 mg/L, SS: 52 mg/L, CODCr: 598 mg/L) from the uring water flow automatically. The cement adding system Fen River at Huyu Valley. Test soil and cement were ana- includes an oxygenator, an adding bottle and an equalizer lyzed by a Rise-202X laser granulo-meter and the composi- to make sure that the cement is evenly dispersed in water, tion is shown in Table 1. so as the water could ooze smoothly into the soil column.

2.2 Infiltration test rig 2.3 Test design and methods A group of self-made laboratory infiltration devices was Research design: The experiments contained a set of used to conduct this experiment. Each set of the infiltration tap water infiltration tests and a set of effluent infiltration tests, device consists of the infiltration soil column, the water sup- and both sets used the above-mentioned three alluvial silt ply system, and the cement adding system shown in Fig. 1. samples collected from different places. While keeping all

Water adding hole Cement infiltration system 90 Water supply system Adding bottle 80 70 60 50 Markov tube 40 Water outlet 30 20 10 Air inlet Exhaust hole Equalizer Water supply chamber 1 Infiltration Water supply hole device 2 Adjustable support Infiltration column Water&soil-sampling 3 hole 4

5 Drainage hole Drainage chamber

FIGURE 1 - The schematic diagram of the test rig.

3767 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

of the other experimental conditions identical, a set of five feeding water followed the principle of ‘dense to sparse’, water supply waiting times (0, 6, 12, 24 and 48 h) was ap- and based on this infiltration water amount, the instantane- plied to each type of soil. Also each sample was tested and ous infiltration rate could then be calculated. It indicates compared simultaneously both with and without cement. the end of the test when the entire soil column is com- Therefore, each kind of alluvial silt had 10 sets of soil col- pletely infiltrated, i.e. the stable infiltration phase. Other umns, and each test included 30 samples, leading to total infiltration conditions included an infiltration water head test specimens of 60. of 100 cm and an initial moisture content of about 7%.

Research method: A one-dimensional cylindrical pressurized soil column was used in the infiltration tests. 3. RESULTS AND DISCUSSION According to the existing research conducted by our research group [15], adding too a small quantity of cement, In this test, the infiltration-reducing effect means the could not meet the required amount of cement particles in- extent to which the infiltration rate is reduced, compared filtrating into the sand matrix, whereas a too large dosage with the reference sample. In order to quantify the infiltra- lead to a thick crust on the surface of the infiltrated alluvial tion-reducing effect, this paper introduces the infiltration- silt. As a result, a cement dosage of 1.0 kg/m3 was used for reducing rate ( R ), which is the percentage of infiltration this study, which could form a fairly diluted suspension liq- i rate decrease of the cement-infiltrated-soil column when uid of cement when dissolved in water. At the same time, compared with the reference sample within the same infil- an equal amount of water was added in the reference sam- tration period: ples (referred as R in Fig. 2) to ensure its initial water content was consistent with that of the tested samples. In order R  (i i )/i *100% (1) to avoid both cement accumulation on the silt surface due i r s r to excessively fast feeding and initial cement setting (45 where R is the infiltration-reducing rate (%), i is the min) and due to excessively slow feeding, the time for ce- i r ment to infiltrate into the soil column was limited to 20 infiltration rate of the reference sample (cm/min), and i s min. In the intermittent infiltration mode, after the comple- represents the infiltration rate of the tested sample (cm/min). tion of the cement slurry infiltration process, the top water chamber valve was opened to keep the tested soil fully con- 3.1 Postponed water feeding can obviously reinforce the infil- tacting the open air. tration-reducing effect of cement infiltrating into the silt The selected dry density of filling soil was 1.4 cm³ and The experiments reveal that, regardless of tap water or the thickness of the stratified filling soil column. Both were sewage, the postponed water supply had a better perfor- weighed and filled according to the pre-set soil dry density mance than instant water supply (i.e. postponed time: 0 h) and the actual initial moisture content. Surfaces between in reducing the infiltration. layers were scabbed. 1 cm was left at the top layer for ce- Figure 2 describes how infiltration rates evolve along ment infiltration. The reading of the Markov tube when time under instant water supply and 12–h postponed water

FIGURE 2 - Comparison of the infiltration-reducing effects between postponed water supply and instant water supply.

3768 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

supply conditions, after the cement infiltrates into Soil 2. the soil, especially the superficial colloidal particles of the Figure 2-a (left) represents tap water infiltration and Figure soil, which can enhance the soil particle aggregation. The 2-b (right) is sewage infiltration. fine cement particles, as a mineral mixture, and their mul- tivalent cat-ions can enhance the cohesion and adhesion ca- In Fig. 2, the infiltration rate curves of samples added pacities of soil particles, and the strong molecule-adsorp- with cement are all below that of the reference sample tion capacity of the cement itself also can adsorb the soil without cement addition, and the infiltration rate curve of moisture so as to further reduce the hydraulic convection. the 12-h postponed test case is always below that of the instant water supply one. In Fig. 2-a, when the infiltration In the mode of the instant water supply, the first mech- lasted 10 min, the infiltration-reducing effects of instant anism plays the primary role. In the postponed mode, there water supply and postponed water supply were 41.07 and is a water supply time delay after the cement oozes in the 63.21%, respectively. When the infiltration lasted 400- sand silt. According to the national standards, the final set- 600 min, the infiltration-reducing effects of steady infiltrating time of ordinary Portland cement P·O 42.5 should be tion rate were 9.93 and 37.01%, respectively. It can be ob- no more than 600 min. This period of time is sufficient to served in Fig. 2-b that, when the infiltration lasted 10 min, enable the already infiltrated cement particles in the silt and the infiltration-reducing effects of instant water supply and the deposited cement particles on the soil surface to set and postponed water supply were 12.46 and 38.34%, respec- harden with the soil particles, form a reasonably solid tively; when the infiltration lasted 400-600 min, the infil- structure for better infiltration-reducing capacity [19]. For tration-reducing effects of steady infiltration rates for the sewage, due to the higher contents of SS and TDS, the two cases were 8.03 and 18.77%, respectively. The exper- higher content of suspended and dissolved solids in water imental results reveal the following facts: 1) No matter how would increase its physical viscosity [20]. In the process of long the delay time of water supply is, adding cement in- infiltration, some suspended solids are gradually trapped filtrating into alluvial silt could decrease the infiltration ca- by soil, and then block the pores in soil and finally impede pacity of the silt; 2) compared with the instant water supply water infiltration [21]. The remaining suspended solids de- mode, the postpone water supply mode generates relatively posit on the soil surface, so the obstructing effect for water better infiltration-reducing effects when the cement infil- infiltration would become stronger with prolonged time of tration process is completed; 3) The judgment that post- infiltration [22]. This will lead to an effect on the reference poned water supply can reinforce the infiltration-reducing samples, which is similar as that produced by the fine ce- effect of infiltrating cement into river alluvial silts is also ment particles. Due to the decreased reference value, the be applicable to sewage infiltration, and the infiltration-re- infiltration-reducing effect of cement infiltrating into tap ducing effect could be raised to more than twice that of the water performs is better than that into sewage; however, instant water supply. However, the infiltration-reducing ef- the postponed water supply mode still has a stronger influ- fect of effluent is inferior to tap water under the same infil- ence on the infiltration-reducing effects than the instant tration conditions. water supply.

Analyses have shown that after infiltrating into the river alluvial silt, cement could influence the infiltration capacity of sandy soil by three mechanisms. Firstly, the fine particles of cement infiltrating into alluvial silt could block the capillary pores and the macro air-pores of sandy soil, which makes the big pores smaller, and the small pores further smaller, or even blocked. In this way, the pore structure of the topsoil to be infiltrated will be changed. Thus, the congestion of pores decreases the flow cross-section of water in the silt and increases the resistance of water flowing through the pores which, consequently, leads to a decrease in water infiltration capacity of the silt [16]. Sec- ondly, the oozing of fine particles of cement could cause the formation of an extremely dense infiltration control layer (i.e. a dense film that covers the macro-pores of the sandy soil matrix) on the surface of the silt, which plays a similar role as a soil crust (referred to as a compacted surface layer with a thickness of 0~2 mm). The crust could FIGURE 3 - Relationship between infiltration-reducing effect and block the channels for water infiltrating into sandy soil and, postponed time. thereby, dramatically decrease the infiltration capacity of sandy soil [17]. In addition, the impact will become in- 3.2 Impact of postponed time on infiltration-reducing effects creasingly remarkable with the extension of time [18]. The To reveal the impact of postponed time on the infiltra- third mechanism can be attributed to the mutual adsorption tion ability of river alluvial silt after cement infiltrating, a and adhesion between the colloidal particles of cement and series of experiments were conducted to explore the infil-

3769 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

tration-reducing effect evolution, with a postponed water Figure 3 describes that the slopes or the increasing paces of supply time varying from 0-48 h under the same infiltration the infiltration-reducing rates decline. For example, when conditions and processes. These experiments use the silts the infiltration process lasts 400-600 min, the average in- from three different cross sections. The results are pre- filtration-reducing ratios of the relatively stable infiltration sented in Fig. 3. rates for postponed time of 0, 3, 6, and 12 h were 9.59, 22.86, 27.15 and 37.01%, respectively. Figure 3 shows the relationship between the postponed time and the infiltration-reducing rate of three different 2) When the postponed water supply time is between tested soils with various postponed time at the steady infil- 12 and 24 h, the infiltration-reducing effects slowly in- tration stage (400 ~ 600 min). From Fig. 3, we can observe: crease with the extension of postponed water supply time. In Fig. 4-b, with the extension of postponed time (12, 3.2.1 Infiltration-reducing effect increases with the extension 24, and 48 h), the infiltration rate curves show a close and of the postponed time unnoticeable down-trend. It indicates that the differences After infiltrating the cement into three kinds of alluvial among all the infiltration rate curves under different post- silts, an increasing trend of the infiltration-reducing effect poned time conditions get smaller, and accordingly, the in- can be seen when the postponed water supply time was been filtration-reducing rates increase more slowly. For exam- increased. Figures 4-a (left) and 4-b (right) present two sets ple, when infiltration lasts for 10 min, the infiltration-re- of infiltration curves for Soil 2 under different water supply ducing rates for a postponed time of 12, 24, and 48 h are time delays (relatively long and short postponed time). 62.09, 62.65 and 60.30%, respectively. When infiltration Figure 4 provides us with the following statements: lasts for 400-600 min, however, the three average infiltration-reducing rates of the relatively stable infiltration rates 1) When the postponed water supply time < 12 h, in- are 37.01, 38.14 and 39.61%, respectively, showing a very filtration-reducing rate increases faster slow increase. In Fig. 4-a, with the extension of the postponed time, 3) When the postponed time is more than 24 h, the infil- the infiltration rate curves decline significantly. To put it tration-reducing effect becomes relatively stable. Figure 3 in- into another way, lower infiltration rates are reached. It indicates that when the postponed water supply time is longer dicates that, within the range of postponed time in Fig. 4-a, the than 24 h, all the infiltration-reducing rate curves of allu- infiltration-reducing rate increases with the postponed vial silt are almost approaching parallel to the horizontal time. For example, when the infiltration lasts for 10 min, axis (time). This suggests that under different sand condi- the infiltration-reducing rates for postponed time of 0, 3, 6, tions, when the time delay > 24 h, the infiltration-reducing and 12 h are 43.51, 49.24, 61.61 and 62.09%, respectively. rates will reach a relatively stable stage. They will not When the postponed time is less than 12 h, the two adjacent cause a noticeable increase with further extending the post- infiltration rate curves become closer with time extending. poned water supply time. It indicates that the increasing amplitude of the infiltration- reducing rate declines with the postponed time extending.

FIGURE 4 - Infiltration rate curves under different postponed times after cement infiltrates into the soil.

3770 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

3.2.2 With the extension of the postponed time, both the time cles produce carbonization reactions to form a solid and spent for the samples to reach relatively stable infiltration hard carbide structure [19]. Consequently, the infiltration- stages and the relatively stable infiltration rates decrease. reducing effect will be reinforced remarkably. Macroscop- From Fig. 4, it can be seen that the infiltration rate ically, the infiltration process quickly reaches a balance curve of the instant water feeding (postponed time: 0 h) is with a significant decrease in stable infiltration rate. When almost parallel to the reference sample curve. With the ex- the postponed time continues to extend, condensation and tension of the postponed time, the time needed for infiltra- aggregation of cement particles themselves and soil parti- tion rate to reach a relatively stable stage was reduced. Es- cles will slightly reinforce the infiltration-reducing. How- pecially when the postponed time is above 12 h, the infil- ever, the delay time, in this case, only contributes trivially tration rate becomes relatively stable rapidly. At the same to the infiltration-reducing effect, the decrease of stable in- time, the relatively stable infiltration rate gradually de- filtration rate, and the reduction of stable infiltration time. creases with the increase of time. Compared with the time for the reference sample (420 min), the time for the tested 3.3 Mathematical relationship between infiltration-reducing ef- samples under various postponed time (0, 3, and 6 h) to reach fects and postponed time relatively stable stages is shortened to approximately 400, 3.3.1 Curve fitting between infiltration-reducing rates of tested 360, and 300 min. When the postponed time is over 12 h, the soils during relatively stable infiltration and postponed time infiltration rate becomes relatively stable within 90 min. In Fig. 3, we could see that when the delay time in- When the infiltration process lasts for 400-600 min, the rel- creases to 6, 12, 24, and 48 h from 0, according to Eq. (1), atively stable infiltration rates of the reference sample (R) the infiltration-reducing rate of relatively stable infiltration and tested samples with different postponed times (0, 3, 6, of Soil 1 increases to 16.66, 30.89, 35.02 and 36.56%, re- 12, 24, and 48 h) are 0.098, 0.089, 0.076, 0.071, 0.062, spectively, from 8.91%; the infiltration-reducing rate of 0.060, and 0.059 cm/min, respectively. Soil 2 increases to 27.15, 37.01, 38.14 and 39.61%, respec- Possible reasons to explain the above differences are tively, from 9.93%; the infiltration-reducing rate of Soil 3 depicted as follows: When cement was in soils, the filling increases to 20.62, 37.59 and 38.45%, respectively, from functions to modify the infiltration properties of soils in- 9.84%. A significant positive exponential correlation can t clude ‘physical filling’ and ‘chemical filling (formed by be observed. The correlation formula is R i  a - b *c , and changing the bonding effects between soil particles) [23]. specifically, the correlation equations of Soils 1, 2 and 3 When postponed time is 0 h, the ‘physical filling’, which are as follows: has a weak infiltration-resisting effect, is the main re- t 2 sistance mechanism. The infiltration rate variation is also R i1  37.8621-30.0762*0.9128 (R  0.9112) (2) consistent with that of the reference sample. With a shorter t 2 R  39.3909- 29.2986*0.8431 (R  0.9834) (3) postponed time, the fine cement particles themselves, the i2 cement grains and sandy soil grains will begin the initial R  39.5085- 30.4789*0.8964t (R 2  0.9379) (4) i3 setting stage. However, water supply makes the condensed structure before the final setting in the water environment 3.3.2 Test of models under certain pressure. Since this unstable crystallized structure could not sufficiently contact with air and also In the analysis of regression, Goodness of Fit Test (R- withstands a certain degree of water head pressure, in the test) and Overall Significance Test (F test) are normally used for the overall test of linear regression models [24]. In meanwhile, the agglomerated crystalline structure accord- 1 ingly could not form a solid and hardened structure, which Eq. (1), Ri = a-b*c . By taking the logarithms, the model gives rise to unstable decrease in infiltration-reducing ef- can be converted to a linear regression model, and then, fect, unnoticeable reduction of time to reach a relatively tested by F-test. Test data is presented in Table 2. stable phase, and the decline of stable infiltration rate. With In the Goodness of Fit Test, adjusted R-Square is used. prolonging the delay time, the setting and the crystalliza- A greater value means the fitting of the model is better. tion reactions of the cement particles will be enhanced. The From Fig. 2, the R2 values of three soils > 0.9 can be seen, cement particles infiltrating into the silt pores, the cement suggesting the fitness of the model is good. By construct- particles covering the infiltrated topsoil, and the silt parti- ing F, F-test is an overall significance test for all explana-

TABLE 2 - Parametric analysis of the forecasting model for the relationship between infiltration-reducing ratio and postponed time.

0 h 6 h 12 h 24 h 48 h Adj. Reduced F Value F0.05 R-Sqr Chi-Sqr Soil1 Actual value 8.9124 16.6556 30.8851 35.0220 36.5639 0.9112 13.2153 97.0534 10.128 Fitted value 7.7859 20.4678 27.8023 34.4973 37.4856

Soil 2 Actual value 9.9265 27.1506 37.0125 38.4192 39.6082 0.9834 2.2072 867.6164 7.709 Fitted value 10.0923 28.8704 35.6175 39.1822 39.3827

Soil 3 Actual value 9.8356 20.6160 34.1932 37.5871 38.4524 0.9379 9.6817 157.0984 10.128 Fitted value 9.0297 23.6978 31.3067 37.3015 39.3487

3771 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

tory variables. It is a strict statistical test method. Its es- [3] Wang C., Li Y. and BAO Z. Q. (2002). Study on the influence sence is to test whether the variables to be explained could of polluted river on its offshore groundwater environment. Ad- vances in Water Sci..13(5), 535-541. be interpreted by these explanatory variables [24]. Accord- ing to the tested sample data, the reduced Chi-Square and [4] Bekhita H. M., Mohamed A. E. and Ahmed E. H. (2009). Con- taminant Transport in Groundwater in the Presence of Colloids the F values (follow F (1, n-2) distribution) are shown in and Bacteria: Model Development and Verification. J . of Con- Table 2. Given the significance   0.05 , the corre- taminant Hydrology. 108 (9), 152 -167. sponding F0.05 value can be found. From Table 2, all the [5] Zhang W. J., Dong W. H., and Su X.S., et al. (2006). Compre- calculated F values are greater than F0.05 value, with large hensive evaluation of groundwater remediation technologies. differences. So, the regression of this model is significant, Water Resources Protection. 22 ( 5), 1-4. and a significant positive exponential relationship between [6] Zhang Y., Liu C. L., and Lu D. Y., et al. (2013). Investigation the stable infiltration-reducing rate and postponed time of groundwater pollution prevention strategies. In Chinese So- can, thereby, be identified. ciety for Environmental Sciences Annual Conference. In: Chi- nese Society for Environmental Sciences Academic Annual Meeting of Chinese Society for Environmental Sciences. China Environmental Science Press, Kunming, 3081-3091. . 4. CONCLUSIONS [7] Ferro S. (2014). Electrokinetic Barriers for Preventing Ground- water Pollution. Encyclopedia of Applied Electrochemistry. 9, 1) A certain delay time for supplying water after add- 719-723. ing cement could allow cement particles themselves, or ce- [8] Helalia A. M. (1993). The relation between soil infiltration and ment particles and soil particles to set and harden each effective Porosity in different soil. Agric. Water Mgmt.. 24(8), other, which can greatly improve the infiltration-reducing 39-47. effect. It was applied to both tap water infiltration and sew- [9] Zou Y., Chen H. S. and Su Y. R., et al. (2005). Study on age infiltration, but the infiltration-reducing effect of the Ponded Water Infiltration and Soil Water Redistribution in latter is slightly inferior. Red Soil．J. of Soil Water Cons. 19(3), 174-177. 2) The postponed feeding water time could increase [10] Keijzer T. J. S., Loch J. P. G. (2000). Chemical osmosis in the infiltration-reducing effect by a factor of 2 to 4. How- compacted dredging sludge. Soil Sci. Soc. Am. 65, 1045. ever, the contribution of delay time to infiltration-reducing [11] Xi Y. H., Feng S. J. (2009). Development of Study on Semi- effect is limited. When the delay time is shorter than 12 h, permeable Membrane of Clayey Soil. J. of Tongji Univ. ( Nat- the infiltration-reducing effect signally increases with the ural Sci.). 37(2), 187-191 delay time extending. Once the time is above 24 h, partic- [12] Freebaim D. M., Gupta S. C.. (1990). Microrelief rainfall and ularly exceeding 36 h, the continuing enhancement of in- cover effects on infiltration.Soil. Tllage Res.. 16, 307-327. filtration-reducing effect is getting slighter. [13] Kalinski M. E., Yerra P. K. (2006). Hydraulic conductivity of 3) Herein, an exponentially positive correlation be- compacted cement-stabilized fly ash. Fuel. 85, 2330-2336. tween the infiltration-reducing rate and the postponed time [14] Song R. Q., Chu G. X., and Ye J., et al. (2010). Effects of is presented. A postponed time of 12 to 24 h can achieve surface soil mixed with sand on water infiltration and evapo- more satisfactory results to infiltration-reducing effects. ration in laboratory. Trans. of CSAE. 26 (Supp.1), 109－114. [15] Ji J. L, Fan G. S. (2014a). Study on infiltration-reducing effect of cement infiltrating into the riverbed alluvial silt. J. of Irrig. and Drainage. 33 (3), 110-114 ACKNOWLEDGEMENTS [16] Wu J. Q., Zhang J. F. and Gao R.. (2009). Physical simulation experiments of effects of macropores on soil water infiltration This work is granted by the financial supports from characteristics. Trans. of the CSAE. 25(10), 13－18. Shanxi Province Science and Technology Development [17] Sonia C., Yolanda C. (2012). Biological soil crust develop- project under No. 20150313002-2, Shanxi River Manage- ment affects physico-chemical characteristics of soil surface in ment and Protection Services Program under Sci 2012-1, semiarid ecosystems. Soil Biol Biochem. 49(6), 96–105 and Jincheng City Science and Technology Development [18] Liu M. (2011). Analysis of the Relationship among the Statis- project under No.201202238. tical Tests on Linear Regression Model. Stat. & Info. Forum. 26(4), 21-24 The authors have declared no conflict of interest. [19] Ji J. L., Fan G. S. (2014b). Study on infiltration-reducing effect of using different cement-adding methods to the riverbed alluvial silt. J. of Taiyuan Univ. of Tech. 45(7), 480-484

REFERENCES [20] Pan J., Yu L. (2015). Characteristics of subsurface wastewater infiltration systems fed with dissolved or particulate organic matter. Int. J. Environ. Sci. Technol.. 12, 479–488. [1] Alley W. M., Healy R. W. and LaBaugh J. W., et al. (2002). Flow and storage in groundwater systems. Science. 296, 985- [21] Wang Y., Wang J. F. and Wang H. (2006). Investigation of 990. seeping pattern of processed sewage in irrigated loess layer. J. of Lanzhou Univ. of Tech.. 32(6), 70-71 [2] Jin Z. F., Wang F. E. and Chen Y. X. (2004). Nitrate pollution of groundwater in urban area. Acta Pedologica Sinica. 4 (2), [22] Hiscock K. M, Grischek T. (2002). Attenuation of groundwa- 252 -257 ter pollution by bank filtration. J. of Hydrology. 266, 139-144.

3772 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[23] Yang B. (2007). Cement soil strength properties PhD diss. Shanghai Jiao Tong Univ., Shanghai, China. [24] Niu Y. Y, Shen Y. Y, and Gao C. Y., et al. (2006). Impact of Different Tillage Method on Infiltration Character to Winter Wheat in the LoessPlateau. J. of Mountain Sci.. 24(1), 13-18

Received: March 17, 2015 Accepted: May 05, 2015

CORRESPONDING AUTHOR

Guisheng Fan

College of Environmental Science and Engineering

Taiyuan University of Technology

Taiyuan 030024, Shanxi Province

P.R. CHINA

Phone: +86-0351-6018892

E-mail: [email protected]

Jinlan Ji

College of Environmental Science and Engineering

Taiyuan University of Technology

Taiyuan 030024, Shanxi Province

P.R. CHINA

E-mail: [email protected]

FEB/ Vol 24/ No 11a/ 2015 – pages 3766 - 3773

3773 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

STUDY ON THE SORPTION OF CHROMIUM(III) IONS IN AQUEOUS SOLUTION WITH DIATOMITE TREATED WITH MANGANESE OXIDE

Xiangying Zeng* and Er Li

Wuhan Municipal Engineering Design and Research Institute, Changqing Road 1037, Jianghan District, Wuhan 430015, PR China

ABSTRACT are always in dispute [1, 2]. Therefore, an alternative method appropriate for this type of heavy metals has been in study In this paper, the properties and characteristics of diat- [1-3]. Diatomite, which consists of a wide variety of shaped omite treated with manganese oxide (MTD) for the re- and sized diatoms, in a structure containing up to 80–90% moval of chromium(III) ions in aqueous solution was stud- voids, is a pale-colored, soft, lightweight sedimentary rock, ied. The characters and performance of sorption MTD were composed principally of silica micro-fossils of aquatic uni- compared with that of raw diatomite (RD). The results in- cellular algae [4]. Due to its highly porous structure, low dicate that the trivalent chromium(III)) ion sorption capac- density and high surface area, diatomite has been widely ities of diatomite were considerably improved after treated used in a number of industrial applications, e.g. as a filtra- by manganese oxide. Due to the formation of manganese tion media for various beverages and inorganic and organic oxide (MnO) on the diatomite surface, the surface area of chemicals, and as an adsorbent for the treatment of oil MTD was improved. And the sorption of chromium(III) by spills [1]. In order to improve the adsorption capacity, the MTD was more than that in RD owing to the different elec- surface of diatomite should be treated [5, 6]. Many treat- trostatic forces between the heavy metal ions and the sur- ment methods have been researched. Manganese oxide has face of the diatomite. The removal of chromium(III) ions been demonstrated to have high surface charge which is by RD and MTD were both improved with the increment useful for treatment [5, 7]. of synthetic solution concentration. The aim of this study is to study the characteristics of

diatomite treated with manganese oxide (MTD) and its per-

KEYWORDS: formance for the removal of chromium(III) ions from aque- Diatomite, manganese oxide, treat removal of chromium(III) ions ous solutions.

1. INTRODUCTION 2. MATERIALS AND METHODS

Water is considered to be an important and scarce re- 2.1. Adsorbent source in many countries. In particular, the contamination In this study, the raw diatomite (RD) was obtained of surface and ground water with heavy metals is becoming through the National Territorial Resource Authority (NTRA) a public concern of China. Heavy metal ions like trivalent Suizhou, China. The samples were washed several times chromic(III) ions are not biodegradable and tend to accu- using deionized water and dried in an oven at 100 ºC over- mulate in living organisms causing various diseases and night, and then, desiccated and stored in tightly stoppered disorders. It is becoming a severe menace to public health glass bottles. Obtained by X-ray florescence (XRF), the de- that the chromium(III) ion concentrations increase owing tailed chemical composition of raw diatomite in mass per- to fast development of some local industries, such as elec- centage was: Al2O3 13.26, SiO2 71.35, Fe2O3 5.5, CaO 1.94, troplating, leather tanning, metal finishing and chromate Na2O 6.7, TiO2 0.08, MgO 0.15, K2O 0.11, MnO<0.05, and preparation [1]. CuO <0.07. The raw diatomite was sieved into different Many conventional methods, such as chemical precip- Tyler sizes and the fraction passed through a sieve of 0.84 itation, membrane filtration, ion exchange, alum coagula- mm (+20 Tyler) and retained at 0.297 mm (-20+48 Tyler) tion and adsorption tend to be mentioned for aqueous solu- in the experiments. tions containing heavy metals, but their safe and economic efficiency, especially their effect for chromium(III) ions, 2.2 Chemicals All chemicals were of analytical grade, and used with- * Corresponding author out pre-purification.

3774 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

2.3 Adsorbate CC1 Standard chromium(III) ion (1000 mg L-1) solutions ee  (2) qqbqes s were prepared by dissolving Cr(NO3)3·9 H2O powder, respectively, in distilled water. Diluted solutions were pre- where qe and qs are the quantity of heavy metal ions pared from the stock solutions (1000 mg L-1). The concen- adsorbed per unit of mass of MTD, (mg.g-1) and Langmuir tration of heavy metals in aqueous solutions were analyzed adsorption capacity of MTD, (mg.g-1), respectively. C and by flame atomic absorption spectroscopy (FAAS) using a i C are the initial and equilibrium concentration of heavy VARIAN SpectrAA–10 PLUS, with air–acetylene and ni- e metal ions, (mg.ml-1), respectively, and b is the Langmuir trous oxide–acetylene flames. isotherm constant, V is the volume of solution (ml), and W 2.4 Surface treatment is the mass of the adsorbent (g). Samples of raw diatomite (15 g) were immersed in 6M The essential characteristics of the Langmuir equation sodium hydroxide at 90 ºC for 2 h. The mixture was then can also be expressed in terms of a dimensionless factor, R : placed in 100 ml of 2.5M manganese chloride (adjusted to 1 pH 1–2 with HCl), at room temperature, for 10 h. The man- 1 ganese-soaked diatomite was filtered and separated from R1  (3) the supernatant. The solid was then immersed in 6M so- 1 qsbC0 dium hydroxide, at room temperature, for 10 h to precipi- where C is the highest initial heavy metal ion con- tate the manganese hydroxide. The excess solution was de- 0 −1 canted off and the diatomite left exposed to the air (to fa- centration, (mg ml ). The magnitude of the factor R1 in- cilitate oxidation of manganese hydroxide to form a mix- dicates the nature of the interaction, and the isotherm type; ture of hydrated manganese oxide). The samples were then unfavorable ( R >1), linear ( R =1), favorable (0< R <1), washed with deionized water, dried at 100 ºC, desiccated 1 1 1 and stored in tightly stoppered glass bottles. The manga- and irreversible values ( R1 = 0) [10]. nese content of the samples was determined by atomic absorption spectroscopy (Perkin–Elmer UV–Vis). Prior to 2.7 pH of solution analysis, the samples were immersed in 4M HCl at 40 ºC, The pH of diatomite was determined in 60 cm3 to dissolve the deposited oxides. Manganese oxide diato- solution glass bottles where 2.5g of treated diatomite and 25 cm3 of mite (MTD) samples were prepared, using the aforemen- deionized water were added. The mixture was then agitated tioned procedure, with manganese chloride concentrations for 24 h. The pH of the mixture was recorded using a pH- from 0.20 to 3.50M. meter. 2.5 Surface area The hydroxyl groups present on the surface of the di- Based on the Zn2+ adsorption method, the surface areas atomite can gain or loose a proton, resulting in a surface of MTD were estimated. A 2-g sample was mixed with 25 ml charge that varies with varying pH. At low pH, the surface 2+ of Zn solution (2M NH4Cl, 0.1073M ZnCl2), stirred with a of diatomite is protonated and the surface becomes posi- mechanical shaker for 24 h, and then, allowed to stand over- tively charged: 2+ night. The supernatant solution was analyzed for Zn con- + + H +MOH  MOH2 (4) tent using atomic absorption spectroscopy. while at high pH, the surface hydroxides loose their 2.6 Adsorption isotherms protons, and the surface becomes anionic. Adsorption isotherms were obtained by adding into 100- MOH  H++MO- (5) ml Erlenmeyer flasks a variable mass of sorbent into a fixed + volume of chromium(III) ions synthetic solution containing When the concentration of cationic MOH2 sites - a constant metal-ion concentration (1.5g L-1, pH 3.5). The equals the concentration of anionic MO sites, the average contact time for equilibrium is 2 h. A thermostatic agitator surface charge is neutral, at the zero point of charge (ZPC). was used for mixing the mixtures at room temperature. A The corresponding pH is named as pHZPC. Whatman membrane filter (pore size -2.7 mm) was for filtering the samples, and the filtrate was analyzed by atomic 2.8 Surface charge density adsorption spectroscopy. The Langmuir isotherm was used Owing to existence of ionizable functional groups, in this paper. The quantity of adsorbed heavy metal ions on such as -OH, -COOH, -SH and -NH3, the surface charges of MTD was calculated from the difference between initial and diatomite can be increased by chemical reaction at the sur- the final concentration at equilibrium using Eq. (1). The re- face. Electric charges are caused by the ionization of such lationship between the adsorption capacity of MTD and con- groups [11]. centration of heavy metal ions can be calculated from the For example, in the case of oxides and hydroxides, the Langmuir equation as Eq.(2) [8-10]. + surface charge is formed by the ionization of [MOH2 ] and [MO-]: ()CCVie qe  1) + + W [MOH2 ]=[MOH]+[H ] (6)

3775 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[MOH]=[MO-]+[H+] (7) 65 and 90 m2 g−1 in the same solution acidity based on the 2+ The surface charge density caused by the above sur- Zn adsorption method. Fourier transform infrared (FT-IR) face chemical reactions is called as . It can be concluded spectra were used to show the effects of the treated method. from Eq.(6) and Eq.(7) that  is dependent on the degree Figure 1 shows the FT-IR spectra of MTD and RD. of ionization and pH of the system, and the H+ and OH- It is exhibited in Fig. 1 that the trend of spectra curves ions are potential determining ions. Therefore, the total are similar at pH 4 and 5.2. The adsorption bands, such as positive and negative charge is determined by the quantity 486, 991, 1383 and 3602 cm-1 disappear in the MTD spec- of total base and acid consumed, respectively. So, the sur- trum at pH 4, and the corresponding bands are 455, 1015, face charge density  can be calculated from Eq. (8): 1495 and 3611 cm-1 in the MTD spectrum at pH 5.2, respectively. This disappearance can be attributed to the chemical C - C +[OH-] -[H+]=[MOH +]-[MO-] (8) A B 2 interactions between the manganese oxide and the silanol where CA and CB represent the concentration of the groups on the diatomite surface. The broad band at 3630 cm-1 - + base and acid added, respectively, while [OH ] and [H ] are in the MTD spectrum at pH 4 and at 3750 cm-1 in that at pH the equilibrium concentrations of the base and acid. The 5.2 represent the O-H stretching of interlayer molecules - + concentration of [OH ] and [H ] are measured potentiomet- and framework hydroxyl groups. Compared with the spec- rically with a glass electrode, and the final values are esti- trum of RD and MTD in Fig.1(a), the most important peak mated by using the following equation: of RD spectrum (the band at 821 cm-1) transfers in the pH=-log[H+] and Kw=[OH-] [H+]=10-14 (9) MTD spectrum (the band at 753 cm-1), and the similar con- The mean surface charge, Q (mol/g), can then be cal- clusion can be drawn from Fig. 1(b). The cause is the Mn- culated as a function of pH from the difference between the O stretching [6, 7]. total added base or acid, and the equilibrium OH- and H+ ion concentrations for a given quantity of the adsorbent a : 3.2 The surface charges of MTD and RD - + The surface charge density of the diatomite as a func- (CA- CB+[OH ] -[H ])/ a = Q (10) tion of pH is shown in Fig. 2. Function relation curve be- 2 If the surface area S (m /g) of the adsorbent is known, tween the surface charge of RD and MTD and pH value are 2 the surface charges (C/m ) can be calculated from Eq.(11). shown in Fig. 2. The intersection of the curve with the x- QF axis at surface charge density equaling zero, defines the   (11) zero point of charge (the intersection of the curve with x- S axis at  equaling zero, gives the zero point of charge where F is the Faraday’s constant (96,500 C/mol). (pHZPC)). The total charge from the cations and anions at the adsorbent surface are equal to zero, and this zero point of charge occurs at pH 4.3 for MTD and 4.6 for RD under 3. RESULTS AND DISCUSSION the same conditions. Fig. 2 also shows that the surface charge density of both MTD and RD decreases with in- 3.1 Properties of MTD and RD creasing pH. The hydroxyl groups present on the surface of The surface areas of the MTD and RD depend on the the diatomite can gain or loose a proton, which results in a solution acidity and the duration of the treatment. In this surface charge varying with pH. study, their surface areas were estimated to be approximately

(a) pH=4 (b) pH=5.2

FIGURE 1 - FT-IR spectra of RD and MTD (experimental conditions: sample of adsorbent = 0.05g, temperature = 23 ºC, particle size = 300– 500 μm at different pH values).

3776 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

creasing chromium ion(III) concentration until a maximum adsorption capacity (Q) is reached. The values of Q in Table 1 demonstrate that the maximum adsorption capacity of RD for chromium(III) ion removal is lower than that of MTD. Obviously, this can be benefited from the increment in the surface area of the diatomite by this treatment way. And the resultant surface charge is offered by the formation of the manganese oxides on the diatomite surface. Manganese oxide surface charge in solution is normally higher than that of other oxides, e.g. SiO2, TiO2 and Al2O3, due to the large acidity constant of the manganese oxide surface [3-5]. This means that the surface ionizes at a low pH carry a net negative charge greater than that of other oxides [15, 16].

FIGURE 2 - Surface charge of RD and MTD with different pH values (experimental variables: particle size = 300–500 μm, sample of adsorbent = 0.05g, volume of solution = 25 cm3, equilibrium time = 24 h, shaking speed = 125 rpm.

3.3 Adsorption capacity for chromium(III) ions: adsorption isotherms As known, alkaline solution conditions will readily lead to heavy metals precipitating. In addition, high pH will cause diatomite powder to be slightly unstable but strong acidic pH (pH≤4), thus improving its capability of filtration based on a previous study [3, 4]. Therefore, batch isotherm studies are commonly conducted under the acidic con- ≤ ditions (pH 4). The equilibrium measurements focus on FIGURE 3 - Adsorption isotherm of chromium(III) ions on RD and the determination of the adsorption isotherms. The relation- MTD (experimental conditions: sample of adsorbent = 0.05 g; volume ship between the quantities of chromium(III) ions adsorbed of solution = 50 cm3; T = 23 ºC; equilibrium time = 24 h; pH = 4, per unit mass of RD and MTD and the equilibrium concen- shaking speed = 125 rpm). tration at the same temperature (23 ºC) are shown in Fig. 3. According to the values of correlation coefficients (R2) for 3.4 Effect of pH on the removal rate of chromium(III) ions the Langmuir equation (R2>0.99), both of the chromium(III) ions adsorption by RD and MTD could be described as an The effect of pH on the percentage of heavy metal ions isotherm of Langmuir type, which supports the theory that removal in aqueous solution by RD and MTD are illus- the number of sites on the diatomite surface is limited and trated in Fig. showing that the maximum percentage of the chromium(III) ions form a mono-molecular layer on the sur- heavy metal ions removal by both RD and MTD takes face at maximum capacity for RD and MTD. place at nearly pH 4 (The corresponding pH values of MTD The Langmuir values of qs are calculated by the slope and RD are 3.7 and 3.9, respectively), and the removal rates of each isotherm, and b is calculated based on the ordinate decline when pH >4. Figure 4 displays that the heavy metal at the origin of each isotherm line, respectively, as shown ions removal by MTD is higher than that by RD for the in Table 1. The results of Rl in Table 1 indicate that both of same heavy metal at the same pH, and at the same pH, the RD and MTD adsorption behaviors are extremely fa- within the range of pH (pH <3), the difference of the re- vorable (Rl<1), and tend to be weakly irreversible (Rl = 0) moval for the same heavy metal by between MTD and RD [13,14] . Due to its lower value of Rl, the heavy metal ad- is obviously less than that when 3

TABLE 1 - Langmuir isotherm constants for adsorption of chromium (III) ions on RD, MTD.

-1 3 -1 3 -1 -3 2 Type of diatomite Q (mgg ) A (dm mg ) KL (dm mg ) R1 (  10 ) R

RD 18.18 0.23 3.66 2.1 0.993 MTD 35.47 0.16 7.82 0.46 0.996

3777 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

in pH of the solution, the electrostatic gravitation between 3.5 Comparison of chromium(III) ions solution concentration chromium(III) and the surface of RD and MTD (negative The variation of the heavy metal ions removal by RD charge) is both lowered, which leads to a decrement of the and MTD in different equilibrium times (pH 3.8) is illus- adsorption rate. Owing to less decline in the surface charge trated in Fig. 5. The result shows that the heavy metal ions density, that the electrostatic gravitation of MTD is more removal by both RD and MTD increases with the increas- than that of RD, which lead to the higher sorption of chro- ing synthetic solution concentration, and 48 h for RD and mium(III) ions by MTD than that by RD [19, 20]. 36 h for MTD are optimal equilibrium times for a maximum removal, respectively.

4. CONCLUSION

Compared with raw diatomite (RD), the surface areas of diatomite treated with manganese oxide (MTD) is increased, and its capacity for the sorption of chromium(III) ions are improved owing to the formation of the manganese oxides on the diatomite surface. Due to the different electrostatic forces between chromium(III) and the surfaces of the adsorbent, the heavy metal ion removal by MTD is higher than that by RD at the same pH. In addition, the increased heavy metal synthetic solution concentrations leads to the improvement of the removal by both RD and FIGURE 4 - pH-dependency of adsorption isotherms of RD and MTD MTD. (experimental variables: chromium solution concentration = 150 mg/ dm, particle size =300–500 μm, sample of adsorbent = 0.05 g, volume of solution = 25 cm3, equilibrium time = 24 h, shaking speed =125 rpm).

FIGURE 5 - Influence of chromium synthetic solution on chromium(III) ions removal at different equilibrium times for RD and MTD.

The authors have declared no conflict of interest. [2] Chen, J.P. and Lin, M .(2001). Equilibrium and kinetic of metal ion adsorption onto a commercial H-type granular activated carbon: experimental and modelling studies. Water Re- search 35 (10), 2385-2394. 5. REFERENCES [3] Akin, S., Schembre, J.M., Bhat, S.K. and Kovscek, A.R. [1] Bailey, S.E., Olin, T.J., Brica, R.M. and Adrian, D.D.(1999). (2000). Spontaneous imbibition characteristics of diatomite. A review of the potentially low cost sorbents for heavy metals. Journal of Petroleum Science and Engineering 25 (2), 149- Water Research 33 (11), 2469-2479. 165.

3778 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[4] Al-Degs, Y., Khraisheh, M.A. and Tutunji, M.F. (2001). Sorp- tion of lead ions on diatomite and manganese oxides modified diatomite. Water Research 35 (15), 3724-3728.

[5] Tsai, W.T., Hsien, K.J. and Yang, J.M. (2004). Silica adsorbent prepared from spent diatomaceous earth and its application to removal of dye from aqueous solution. Journal of Col- loid and Interface Science 275 (2), 428-433.

[6] Khokhotva, O. and Waara, S. (2010). The influence of dissolved organic carbon on sorption of heavy metals on urea- treated pine bark. Journal of Hazardous Materials 173 (1), 689–696. [7] Lourie, E. and Gjengedal, E. (2011).Metal sorption by peat and algae treated peat: kinetics and factors affecting the process. Chemosphere 85 (5), 759–764. [8] Tsai, W.T., Hsien, K.J. and Yang, J.M. (2004). Silica adsorbent prepared from spent diatomaceous earth and its application to removal of dye from aqueous solution. Journal of Col- loid and Interface Science 275 (2), 428-433. [9] Nurchi, V.N. and Villaescusa, I. (2012). Sorption of toxic metal ions by solid sorbents: A predictive speciation approach based on complex formation constants in aqueous solution. Coordination Chemistry Reviews 256(2), 212-221.

[10] Vafakhah, S., Bahrololoom, M.E., Bazarganlari, N. and Saeedikhani, M. (2014). Removal of copper ions from electroplating effluent solutions with native corn cob and corn stalk and chemically modified corn stalk. Journal of Environmental Chemical Engineering 2(1), 256-361. [11] Vijayaraghavan, K. and Raja, F.D. (2014). Experimental characterisation and evaluation of perlite as a sorbent for heavy metal ions in single and quaternary. Journal of Water Process Engineering 4(12), 179-184.

[12] Rajender, K., Divya, B., Rajesh, S. and Narsi, R.B. (2012). Metal tolerance and sequestration of Ni(II), Zn(II) and Cr(VI) ions from simulated and electroplating wastewater in batch process: Kinetics and equilibrium study. International Bio- deterioration and Biodegradation 66(1), 82-90. [13] Aivalioti,M., Vamvasakis,I., Gidarakos, E. (2010).BTEX and MTBE adsorption onto raw and thermally modified diatomite. Journal of Hazardous Materials 178(2),136-143.

[14] Sarı,A., Çıtak,D.,Tuzen,M.(2010).Equilibrium, thermody- namic and kinetic studies on adsorption of Sb(III) from aqueous solution using low-cost natural diatomite. Chemical Engi- neering Journal 162(2), 521-527. [15] Yuan, P., He, H.P., Wu, D.Q., Wang, D.Q. and Chen, L.J. (2004). Characterization of diatomaceous silica by Raman spectroscopy. Spectrochimica Acta Part A 60 (3) 2941-2945. Received: March 19, 2015 [16] Reyad, A., Shawabkeh, M.F. and Tutunji, M. (2003). Experi- Accepted: April 27, 2015 mental study and modeling of basic dye sorption by diatomaceous clay. Applied Clay Science 24 (2), 111-120. [17] Dantas,T.N., Dantas,N.A., Moura,M.C. (1999) Use of im- CORRESPONDING AUTHOR pregnated natural clay with microemulsion in the treatment of effluents containing toxic metals, Enviroment Management 27 Xiangying Zeng (2),2256-2268. Wuhan Municipal Engineering Design and Research [18] Ediz,N., Bentli,İ., Tatar,İ.(2010).Improvement in filtration Institute characteristics of diatomite by calcination.International Jour- Changqing Road 1037 nal of Mineral Processing 94(3),129-134. Jianghan District, Wuhan 430015 [19] Beltran-Heredia, J. and Sanchez-Martín, J. (2008) Heavy met- P.R. CHINA als removal from surface with Moringa oleifera seed extract as flocculant agent. Fresenius Environmental Bulletin 17(12), 425-435. Phone: +86+13720153406 [20] Tengku, I., Tengku, H., Ahmad, Z.A. and Shantini, S (2011). Fax: +86+27+85629124 Potential removal of Pb, Cu and Zn by Phylidrum lanugino- E-mail: [email protected] sum in aquatic environment. Fresenius Environmental Bulletin 20(5), 230-240. FEB/ Vol 24/ No 11a/ 2015 – pages 3774 - 3779

3779 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

LOSS CHARACTERISTICS OF NITROGEN AND PHOSPHORUS IN SURFACE RUNOFF WITH DIFFERENT LAND USE TYPES OF A SMALL WATERSHED IN FREEZE-THAW AGRICULTURAL AREA

Xuehui Lai1,*, Fanghua Hao2 and Xiaoli Ren1

1Department of Environment and Safety Engineering, Taiyuan Institute of Technology, Taiyuan, 030008, China 2School of Environment, State Key Laboratory of Water Environment Simulation, Beijing Normal University, Beijing, 100875, China

ABSTRACT 1. INTRODUCTION

This study was carried out in the Abujiaohe watershed With increasing population and rapid economic devel- in Sanjiang Plain, northeast China. This study investigated opment, surface water is heavily polluted due to large the characteristics of nitrogen (N) and phosphorus (P) losses amount of municipal sewage, industrial wastewater and by surface runoff during natural rainfall events. The surface non-point source pollution. Non-point source pollution ac- runoff on the Abujiaohe watershed was estimated based on counts for a large proportion of total pollution contribution the Soil Conservation Service Curve Number (SCS-CN) as reported in various countries [1,2]. Non-point source model. The results showed that N loss in runoff of paddy pollution of surface waters by nitrogen (N) and phosphorus land was higher than that of dry land, wetland and forest (P) can accelerate eutrophication and threaten drinking wa- land. The average event mean concentrations in runoff of ter safety [3]. Agricultural sources of N and P often con- paddy land were 18.39mg/L for TN, 2.81mg/L for ammo- tribute the largest share of non-point source pollution. For + nium nitrogen (NH4 -N), 8.64mg/L for nitrate nitrogen instance, agricultural non-point pollution contamination of - (NO3 -N). The minimum N loss occurred on the forest land, surface waters in the mid United States is regarded as a ma- the average concentrations in surface runoff of which were jor threat to drinking water, recreational resources, and + 11.45mg/L for TN, 1.64 mg/L for NH4 -N, 4.04mg/L for aquatic ecosystems by the U.S. Environmental Protection - - NO3 -N, respectively. In the six rainfall events, NO3 -N was Agency [4]. Thereby, the eutrophication of water bodies the major form of N loss, which was 2.11, 3.07, 2.46 and has aroused international concern, wherein N and P are key + 3.45 times higher than NH4 -N loss for dry land, paddy land, determinants for eutrophicated process. In arable lands, forest land and wetland. However, the average TP concen- due to excessive application of chemical fertilizers, nutri- trations in runoff measured for four land use types showed ent were largely transferred by runoff into surface [5], and followed the order: paddy land (0.35mg/L)>dry land they could raise N and P concentrations in runoff waters to (0.27mg/L)>wetland (0.26mg/L)>forest land (0.08mg/L). In levels above values generally proposed as guidelines for addition, the loads of N and P (kg·ha-1·a-1) in runoff were degraded waters [6]. Therefore, to carry out forecasting N determined by concentrations and runoff volume. The esti- and P pollution load of the watershed in rural area is of mated results using SCS-CN model demonstrated that N great significance for the protection of the water environ- loads were higher than TP loads in runoff for four land use ment safety and for the prevention of the agricultural non- types. Meanwhile, the peak value of TN load in surface point source pollution. runoff was 21.78 kg·ha-1·a-1 for paddy land and the maxi- - -1 -1 A variety of factors can influence nutrient losses from mum NO3 -N load was 17.82kg·ha ·a for wetland. This soils, among which are soil properties, rainfall intensity, phenomenon was due to the fact that the nitrate in soil is topographical features, vegetation coverage [7,8]. The con- easily lost by water movement, especially when runoff dis- tributing factors of agricultural non-point source nutrient charge is low. Dry land, which accounted for 24.1% of the loss by rainfall runoff from land soils have been widely total watershed area, contributed to 29.9% TN loss studied [9] and rainfall intensity is considered as the most amounts and 49.2% TP losses, and however, paddy land important factor leading to soil nutrient loss [10]. As an accounting for 32.0% the total area contributed to 42.5% example, it is found that the losses of N and P from farm- of TN losses and 29.0% of TP losses. These indicated that land are positively correlated with the rainfall intensity and agricultural activity had an obvious impact on the N and P negatively correlated with the vegetation cover in the North losses in the Abujiaohe watershed. China Plain [11]. Accordingly, N and P loads from farmland soil may significantly impact the quality of the receiving KEYWORDS: Land use types; Surface runoff; Non-point source water body. Understanding the variations of nutrient concen- pollution; Nitrogen and phosphorus losses tration in runoff during the rainfall and the observation of

3780 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

various rainfall events are very helpful in controlling the runoff estimation for single rainfall event in small agricul- potential pollution hazards in a watershed [12]. In order to tural watershed [24,25]. improve the water quality and to reduce pollutants, a con- In the current study area, the so-called freeze-thaw ag- siderable amount of studies have focused on laws of water- ricultural region, Sanjiang Plain, has experienced several shed non-point pollutions [13], the estimation of non-point agricultural reclamations to increase the arable area since source pollution [14], simulation of total pollution load the 1950s. It is recognized by many studies that a large por- [15] and a variety of models to simulate hydrological pro- tion of wetland and forest land has been converted for ara- cesses [16,17]. ble. Whilst the N and P loss processes, the generation Nutrient loss is heavily influenced by farming prac- mechanism of non-point source pollution and the influence tices, vegetation coverage and cropping system under agri- factors have been extensively investigated [26]. However, cultural land use. Buffers zones and year-round vegetated little is known about agricultural non-point source pollu- fields have therefore been introduced as methods to reduce tion in intensive agricultural area, especially for watershed soil erosion and the P loss in surface runoff waters [18]. In in the freeze-thaw agricultural area, because it is generally addition, the duration between fertilizer application and the believed that the load and concentration of pollutants from first rainfall-runoff event has also been shown to be an im- agricultural area is relatively low and therefore possesses portant factor in controlling N and P losses. It was found less of a threat to the environment. This study aims at in- that when an irrigation event occurred immediately after vestigating the N and P losses during typical rainfall fertilizer application, the P concentration occurred as the events, and exploring the nitrogen and phosphorous load water deficit reaching 50mm [19]. Runoff induced by in- patterns by surface runoff under different land use types. appropriate tillage practices and land use can cause severe Beyond that, this study also aims to provide a basis for the soil erosion and nutrient loss [20]. Land use change can estimation of overall contribution of nutrient loss by sur- potentially lead to changes in future rainfall runoff and can face runoff in freeze-thaw region. significantly impact the N and P losses [21]. Agricultural non-point source pollution takes place in the form of N and P losses induced by surface runoff [22]. The simulation ac- 2. MATERIALS AND METHODS curacy of surface runoff in various land use types is primarily dependent on rainfall, temperature, topography, geomor- 2.1 Study area phology, land use type, and other parameters. At present, the The watershed studied, Abujiaohe, is located in one of the Soil Conservation Service (SCS) curve number (CN) model branches of the Wusuli River, Sanjiang Plain of Northeast developed by the U.S. Department of Agriculture National China (Fig.1). This area belongs to a continental monsoon cli- Resources Conversion Service is the most popular and mate zone and has a mean annual temperature of 2.94 and widely applied model for direct runoff estimation [23]. The precipitation of 583.2mm. Rainfall distribution during the pe- SCS-CN model is a simple, empirical model with clearly riod of April through September accounts for above 80% of stated assumptions and few data requirements. Therefore, it the total, while the period of November through the following has been widely used for water resource management and March accounts for only about 10% of the total.

FIGURE 1 - Location of the study area

3781 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

The total area of the watershed is 142.5 km2, with main NaOH diffusion method [29]. Soil available phosphorus land uses as paddy land, dry land, forest land and wetland (SAP) was extracted with 0.03 M NH4F, 0.025M HCl and in the proportion of 31.99%, 24.13%, 25.41% and 11.78%, then measured using the molybdate spectrophotometric respectively. The soil property database was directly re- method [30]. The soil properties of the four land use types ferred to the national database, and there were six types of are provided in Table 1. soils in this watershed, which were dark-brown earths, albic soil, meadow albic soil, gleyed albic soil, meadow soil 2.2 Runoff samplings sites and bog soil [27]. With net grids, surface soils (0-20cm Water sampling sites were chosen using information depth) from 40 sites (16 dry land sites, 12 paddy land sites, obtained from land use data. The distribution of land type 7 existing wetland sites and 5 forest land sites ) of different was extracted from the cloud-free LANDSAT TM image land use types were collected in late April 2010 prior to on May 20th in 2009. Surface runoff samples from 7 sites fertilization, plowing and seeding. The sampling sites are (2 dry land sites, 2 paddy land sites, 2 existing forest land in the middle of each field to avoid field ridges, roadsides, sties and 1 wetland site) of four land use types were se- farmhouses and soils that have been obviously disturbed lected along Abujiaohe watershed to study N and P loss after the last harvest. The forested mountain area (eleva- characteristics. As was shown in Fig.1, 2 forest land sites tion≥100m) and the riparian flooding area (elevation≤43m) are located in the headstream and southwest of watershed, in the watershed were not considered for soil sampling and where the runoff samples are represented the background analysis. On each sample site, three soil cores (0-20cm) value without agricultural interference and the pollution were taken in a 100cm×100cm area. For each site, 3 soil value influenced by agricultural activity of dry land, re- samples were homogenized by hand mixing and sealed in spectively. The runoff sampling sites of arable land are in plastic bags. The sampler was rinsed with distilled water the middle of watershed to avoid livestock industry pollu- between soil samples to avoid cross-contamination. tion and sanitary sewage. Whilst wetland land site is cho- All soil samples were air-dried at 25℃, freed of plant sen in the estuary of Wusuli River because Abujiaohe fi- residues, ground to pass through 2mm nylon sieves, and nally flows through the wetland into the Wusuli River. The then stored in plastic bags. With a soil water ration of 1:5, detailed information of surface runoff sampling sites is pro- the soil pH was measured using a pH meter. Soil total ni- vided in Table 2. trogen (STN) was measured with a CHN Elemental Ana- lyzer (Eurovector EA 3000; EuroVector SpA, Milan, Italy; 2.3 Runoff collection and laboratory analysis dry combustion temperature, 900℃). Soil total phosphorus The rain season occurs mainly in the period from July (STP) were digested with a mixed acid containing HF, to September, sometimes in June. All rainfall events were HNO3, and HClO4 [28] and then measured using the induc- recorded from 1964 to 2012 (Fig.2). At the current study, tively coupled plasma optical emission spectrometry (IRIS runoff samples of six rainfall events were collected during Intrepid II XSP, Thermo Electron, Madison WI, USA). the period of April through August in consideration of crop Soil available nitrogen (SAN) was measured using the growing season.

TABLE 1 - Soil properties of four land use types in the study area

Land use Area Ration STN SAN STP SAP pH type (km2) (%) (g.kg-1) (mg.kg-1) (g.kg-1) (mg.kg-1) Dry land Mean 5.73 2.58 247.22 0.91 28.77 34.38 24.13 (n=16) S.D. 0.26 0.60 60.30 0.27 14.03 Paddy land Mean 6.00 2.15 223.39 0.88 12.01 45.58 31.99 (n=12) S.D. 0.22 0.49 42.43 0.10 4.29 Forest land Mean 5.72 3.18 274.52 0.84 9.08 36.21 25.41 (n=5) S.D. 0.20 0.50 26.49 0.37 2.39 Wetland Mean 5.76 6.54 381.72 1.36 17.00 16.78 11.78 (n=7) S.D. 0.67 4.98 205.27 0.78 9.06 n, the number of soil sites for each land use type; S.D., standard deviation

TABLE 2 - Description of runoff sampling sites

Number Geographical position Land use types Vegetation composition 1 47°24.317′N 134°02.219′E Forest land Deciduous broadleaved forest 2 47°24.676′N 134°02.569′E Dry land Corn 3 47°25.037′N 134°03.567′E Forest land Deciduous broadleaved forest 4 47°25.326′N 134°03.688′E Dry land Corn 5 47°25.800′N 134°05.672′E Paddy land Rice 6 47°26.242′N 134°09.619′E Paddy land Rice 7 47°26.561′N 134°19.953′E Wetland Deyeuxia angustifolia

3782 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

FIGURE 2(a) - The annual precipitation change of the study area from 1964 to 2012

FIGURE 2(b) - The daily precipitation of study area in 2012

Surface runoff samples from 7 sites (Table 2) of dry through 0.45μm filter paper. To determine total phosphorus land, paddy land, forest land and wetland were collected (TP), the runoff samples were digested in a H2SO4 and H2O2 after the surface runoff started. For each site, samples were mixture as catalyst followed by the colorimetric measure- collected at specified intervals during the rainfall. Specifi- ment of the molybdenum-blue complex [32]. cally, one sample was collected every 5 min within 30 min after the surface runoff started, every 10min during a pe- 2.4 Rainfall runoff estimation riod of 30 to 60 min, and thereafter every 30 min till the The most commonly used and popular method for direct runoff disappeared or became gradually stable. All samples surface runoff estimation of small agricultural watershed is at specified intervals were mixed together and three 250 mL the SCS-CN model designed by American soil conservation parallel runoff samples were collected in the polyvinyl service [33]. The theory underlying the SCS-CN model is chloride bottles. Before sampling, all bottles were washed that runoff can be related to soil-cover complexes (soil prop- clean to avoid contamination. erties, vegetation coverage, topography, soil types, etc) and In order to maintain sample integrity and prevent the rainfall through a curve number. The SCS-CN model can be degradation of samples, the runoff water samples were expressed in its general form as follows [34]: sealed in plastic bottles, cooled at <4 and immediately 2 taken to the laboratory for analysis. Runoff total nitrogen (P  0.2S) Q  P  0.2S (TN) was measured according to the alkaline potassium (P  0.8S) (1) persulfate digestion-ultraviolet spectrophotometric method - P  0.2S [31]. Nitrate nitrogen (NO3 -N) was determined according to Q  0 the phenol disulfonic acid ultra-spectrophotometric after the samples were filtered through 0.45μm filter paper. Ammo- Where Q is the direct surface runoff (mm), P is the to- + nia nitrogen (NH4 -N) was measured by the method of sa- tal rainfall (mm); S is the potential maximum retention licylate-hypochlorous acid spectrophotometric after filtering (mm) after runoff begins. S is expressed in the form of a

3783 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

dimensionless runoff curve number (CN) which represents CN  23CN /(10  0.13CN ) (4) the runoff potential of the land cover-soil complex charac- III II II teristics determined by soil type, soil antecedent moisture To apply this SCS-CN model to Abujiaohe watershed, condition (AMC), land use and treatment [33]: the CN values were obtained by land use and soil type data. In this study, the primary soil in the watershed was light or 25400 S   254 (2) heavy clay, thus the HSG was classification D. According CN to the previous researches [39], the CNII of dry land, paddy Where CN is a dimensionless catchment parameter land, forest land and wetland were estimated to be 90, 86, ranging from 0 to 100 [35]. A CN of 100 conceptually rep- 79 and 89, respectively. Since there were different five-day antecedent rainfall in six rainfall events, CN needed to be resents a limiting condition of a perfectly impermeable II catch with zero retention, in which all rainfall becomes run- corrected to match CNI and CNIII. Using Eq. (3) and (4), the off. A CN of zero represents the other extreme, with the different CN values of the Abujiaohe watershed for three AMCs are shown in Table 5. catchment abstracting all rainfall and with no runoff regardless of the rainfall amount. However, the CN ranges 2.6 Calculation of N and P loads from 30 to 100 in most realistic cases. Typically, the load of a pollutant in surface runoff is 2.5 Determining the CN value described based on the loss amounts during rainfall events. The CN can be determined from empirical infor- A pollutant load was calculated by runoff volume multi- mation. The SCS has developed standard tables of curve plied by the pollutant concentration. The N and P loads can number values as functions of watershed land use condi- be defined as the total pollutant load of four land use types during rainfall events according to the formula (5) [40]: tions and HSG. The HSG reflects the soil permeability and T surface runoff potential. And it refers to the standard soil i Li  Ct,iQt,i dt (5) classifications ranging from A, which corresponds to sand 0 and aggregated silts with high infiltration rates, to classifi- Where, Li is the event total concentration (g), Ct,i is the cation D, which refers to swell soils when wet and have time variable concentration (mg/L); Qt,i is the time variable low infiltration rates [36]. The HSG characteristics are pre- runoff depth (mm/s) and Ti refers to the discrete time inter- sented in Table 3. val during rainfall event (s). Three AMCs were defined as dry (lower limit of mois- The formula (5) can also be expressed as the formula ture), moderate (normal or average soil moisture condition) (6) due to the surface runoff process being difficult to be and wet (upper limit of moisture), which were denoted as monitored continuously. The pollutant loads in the study AMCI, AMCII and AMCIII, respectively [37]. Table 4 were calculated by the formula (6). listed the classification of antecedent moisture condition. n The CN value of AMCII (CN ) was provided by the II Li  C j,iQ j,i (6) SCS-CN manual and the CN value of AMCI (CNI) and CN j1 value of AMCIII (CNIII) can be calculated by the following Where n is the time section number during a rainfall equations [38]: event. (3) CN I  4.2CN II /(10  0.058CN II )

TABLE 3 - HSG characteristics summary

Final infiltration rate Permeability rate HSG Surface runoff potential Soil texture (mm/h) (mm/h) A Low 25 >7.6 Sand, Loamy sand, or sandy loam B Moderately low 13 3.8~7.6 Silt or loam C Moderately high 6 1.3~3.8 Sandy clay loam D High 3 <1.3 Clay loam, silty clay loam, sandy clay, silty clay, or clay

TABLE 4 - Grade classification of soil antecedent moisture TABLE 5 - CN values of different land use types in Abujioahe watershed Antecedent Five-day antecedent rainfall (mm) moisture condition Growing season Telogen Dry land Paddy land Forest land Wetland AMCI <13 >28 CNI 79 72 61 77 AMCII 13~28 36~53 CNII 90 86 79 89 AMCIII >28 >53 CNIII 95 93 90 95

3784 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

+ 2.7Statistical analysis 3.1.2 Ammonia nitrogen (NH4 -N) loss + The data were analyzed using SPSS 13.0 for Windows NH4 -N losses for different land use types by surface (SPSS Inc., IL). Significant correlation coefficients between runoff in six rainfall events were shown in Fig.4. In the + nitrogen or phosphorus concentrations and the values of im- current study, the mean NH4 -N concentration order in run- pact factors were determined at ɑ=0.05.To explore whether off from high to low was same as TN concentration for four + nitrogen or phosphorus contents of surface runoff are signif- land use types. Peak of NH4 -N losses (3.71mg/L) in the icantly different in different lands, an analysis of variance six rainfall events occurred on June 11 and it varied from (ANOVA) was used in this study. If p≤0.05, differences be- 2.39mg/L to 3.04mg/L in a gentle change during other five + tween land groups were considered significant, and if p>0.05 events. The SD of the NH4 -N losses for the paddy land, there was no significant difference. dry land, forest land and wetland were 0.37, 0.64, 0.25 and 0.03 respectively. The variability of wetland was relatively low, which meant that there were no significant differences + + 3. RESULTS among NH4 -N loss in six rainfall events. NH4 -N concentration of forest land in runoff was lower than the other land 3.1 Characteristics of N loss use types and varied from 0.44 to 4.14mg/L, with a median 3.1.1 TN loss value of 1.64mg/L. It was found that there was significant + The TN concentrations in surface runoff of four land difference between paddy land and forest land for NH4 -N use types are shown in Fig. 3. For surface runoff, the TN loss by surface runoff (p<0.05). This phenomenon can be + concentration of paddy land was higher than the other land explained by the previous research result that NH4 -N con- use types in every rainfall event. The mean TN concentra- centration in runoff had positive correlation with vegetation in surface runoff measured for four land use type was tion coverage. Various studies proposed that the higher the paddy land (18.39mg/L)>wetland (15.05mg/L)>dry land vegetation coverage rate of forest and dry land was, the + (12.61mg/L)>forest land (11.45mg/L). It could be found smaller the NH4 -N concentration in surface runoff [6]. + that the TN concentration of paddy land by surface runoff The amount of runoff NH4 -N loss was determined by the was more than those from dry land and forest land amount of runoff and the amount of soil nitrogen in the soil (p<0.05). The peak TN loss of paddy land (24.23mg/L) water [41].Because of the large forest vegetation coverage, was found in the rainfall event on August 29, which was the canopy and litters can mitigate the precipitation and rain- 1.5 times higher than the foot loss on April 26 (16.3mg/L). fall intensity arrive the ground and decrease rainfall splash Whilst the maximum TN loss of forest land appeared in the effect on surface to minimize runoff depth [42]. Thus, the + rainfall event of June 16 (23.93mg/L) and the minimum NH4 -N output is relatively lower with the small surface run- loss was 5.93 mg/L on April 26. The statistical analysis on off volume of forestland (Table 6). The vegetation coverage the standard deviation (SD) of the TN losses for the paddy of farmland is considerably below forestland, and the agri- land, dry land, forest land and wetland were 2.51, 2.26, cultural production started later in freeze-thaw agricultural 1.18 and 1.13 respectively, which meant there were huge area. This means that the evapotranspiration decreases and differences among the TN loss in different rainfall events. the rainfall runoff erosion increases, which will accelerate

FIGURE 3 - TN loss of four land use types by surface runoff in Abujiaohe watershed

3785 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

+ FIGURE 4 - NH4 -N loss of four land use types by surface runoff in Abujiaohe watershed

- FIGURE 5 - NO3 -N loss of four land use types by surface runoff in Abujiaohe watershed

+ - NH4 -N accumulation in runoff. Simultaneously, the run- components in which had been converted into NO3 -N be- + - off NH4 -N concentration values of different land use types fore adsorbed by vegetation [44]. Meanwhile, the NO3 -N were also determined by livestock and poultry and rural loss in runoff was higher than the other land use types, and residents according to the relevant literatures [43]. this phenomenon can be explained by the fact that wetland in this study is located in the downstream and slope also - 3.1.3Nitrate nitrogen (NO3 -N) loss existed. In additional to soil characteristics, the terrain also - NO3 -N losses of four land use types by surface runoff affects the N loss. The N loss due to surface runoff was were presented in Fig.5. In six rainfall events, the average higher in sloped land compared with flat land [45]. To - - NO3 -N concentration order in runoff for different land use identify whether NO3 -N concentration in runoff is signifi- types from high to low was wetland (8.89 mg/L)>paddy land cantly different in four land use types, ANOVA was ap- (8.64 mg/L)>dry land (5.26 mg/L)>forest land (4.04 mg/L). plied to analyze. It could be found that differences among - The primary reason of high NO3 -N loss in arable was the paddy land, dry land and forest land were considered sig- - excessive application of chemical fertilizers, a majority of nificantly (p<0.05). The maximum NO3 -N loss of wetland

3786 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

(15.58mg/L) occurred on May 8, which was 6.06 times In the present study, the concentration of TN is higher higher than the minimum loss (2.57mg/L) on July 11. The than TP transported by runoff. This can be explained by that - SD values of the NO3 -N loss for the paddy land, dry land, the amount of P fertilizer applied in the farmland was less. forest land and wetland were 2.98, 1.22, 0.57 and 1.12 re- The amount fertilizer used in this study was 236.5kg/ha for + spectively, which meant NH4 -N losses varied slightly paddy land (N:P:K=2:1:1.5) and 279.0kg/ha for dry land - changes in six rainfall events than NO3 -N. In the six rainfall (N:P:K=2:1:0.5) in each year. After application of P ferti- - events, NO3 -N was the major form of N loss, which was lizer to the soil, phosphorus is quickly fixed through a series + 2.11, 3.07, 2.46 and 3.45 times higher than NH4 -N loss for of precipitation and adsorption processes of phosphate ions dry land, paddy land, forest land and wetland. These results with organic matter or metals such as aluminum and iron were similar to those of another three adjacent agricultural [48]. However, in the long-term studies, the difference in watersheds in Knox County, MO, USA [46]. However, surface soil nutrient contents and soil properties should be some research has shown that suspended particulate nitro- considered. The result showed TP losses by surface runoff gen (SN) was the major form of N loss in China with dif- were positively affected by soil TP content of different land ferent ecological settings [47], which can be explained by use type (Table 2). Soil properties and soil adsorption that the variation of rainfall intensity and soil erodibility as might also contribute to the difference between N and P well as nutrient behavior can vary N loss substantially. losses. Soil types of dark-brown earths, albic soil, meadow albic soil, gleyed albic soil, meadow soil and bog soil in 3.2 Characteristic of P loss this watershed were the clay, and the percentage of parti- Compared with the TP concentrations during rainfall cles smaller than 0.02mm were 48.8% for dry land, 43.6% events, the trends of mean TP and TN concentrations in for paddy land, 51.0% for forest land and 42.0% for wet- surface runoff of the land uses in this study were almost the land, respectively. Phosphorus enters the runoff and open same (e.g., the TN and TP concentrations of paddy land were water bodies mainly through transport with clay and fine the highest, followed dry land and wetland, while the forest silt fractions [47]. Some research shows that P is absorbed land was the lowest (Fig.6). Average TP concentration in by micro-aggregates (0.02-0.002 mm) and particles smaller runoff measured for four land use types was paddy land than 0.002 mm [49]. (0.35 mg/L) > dry land (0.27 mg/L) > wetland (0.26 mg/L) > forest land (0.08 mg/L). The peak TP loss of paddy land The primary growth limiting nutrient in most of fresh- (0.81 mg/L) was found in the rainfall event on June 11, water systems is not N but P [50] and consequently the which was 27 times higher than the foot loss (0.03 mg/L) on framework of most eutrophication management has been May 5. Meanwhile, the maximum TP loss (0.15 mg/L) of focused on the control of P concentration [51]. Although forest land occurred on May 8 and the foot loss appeared on there is a lack of consensus concerning the concentration July 11 and August 29 (0.04 mg/L). The ANOVA analysis at which P in runoff leads to eutrophication in agricultural showed that there was no significant difference between production area, it is generally accepted that runoff water four land use types (p>0.05). All the SD values of the TP is degraded if it contains more than 0.1mg/L [47]. In this loss for four land use types were below the 0.03 level, study, TP exceeded the guideline value of 0.1mg/L in 50% which meant TP losses varied smoothly changes in six of the tested samples (Fig.6). The present study showed rainfall events. that N and P concentrations in surface runoff were high and

FIGURE 6 - TP losses of four land use types by surface runoff in Abujiaohe watershed

3787 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

their losses to watershed would stimulate eutrophication in Surface runoff samples from 7 sites (2 dried lands, 2 those ecosystems and worsen the pollution in Abujiaohe paddy lands, 2 forest lands and 1 wetland) were collected watershed. after the surface runoff started. All samples at specified intervals were mixed together and three 250mL parallel runoff 3.3 Estimation of N and P loads and losses in surface runoff samples were collected in the polyvinyl chloride bottles. The loads of N and P (kg·ha-1·a-1) in runoff were de- Therefore, data of N and P concentration for dry land, paddy termined by concentrations and runoff volume. In this land and forest land during the rainfall were the mean value study, six rainfall events were chosen and characteristics of of six parallel samples, and the data of N and P concentration these events are summarized in Table 6. It is apparent that for wetland was the mean value of three parallel samples. dry conditions prevailed for the six events, since measured The mean values with standard deviations of N and P con- five-day antecedent rainfall of May 28 event was more than centrations in these four types are given in Table 7. 28mm and it was less than 13mm for other five events (Ta- According to the normal mean rainfall (583.2mm) of ble 4). Therefore, CNI for AMCI and CNIII for AMCIII (Ta- Abujiaohe watershed, the average annual pollutant load ble 5) were used in assessing the predicted runoff depth for and loss per unit area by surface runoff could be calculated the six events as shown Table 6. based on the weighted average of different land type areas (Table 8).

TABLE 6 - Estimated runoff depths of six rainfall events by SCS-CN method

Date Amount of Five-day antecedent Surface runoff depth (mm) Estimated mean run- (Month-Day) rainfall (mm) rainfall (mm) Dry land Paddy land Forest land Wetland off depth(mm) 04-26 16.0 11.1 2.8 1.2 0.1 2.3 1.5 05-05 19.6 9.9 9.5 7.1 4.6 9.4 7.3 05-08 23.6 41.1 12.8 10.1 7.0 12.8 9.6 06-11 9.4 0 0.4 0.1 0 0.3 0.2 07-11 18.5 7.3 4.0 2.0 0.3 3.4 2.3 08-29 13.0 9.0 1.5 0.5 0 1.2 0.7

TABLE 7 - N and P concentrations in surface runoff of four land use types

N and P concentrations on different sampling date(Year-Month-Day) (mg/L) 2012-4-26 2012-5-5 2012-5-8 2012-6-11 2012-7-11 2012-8-29 Dry land TN 8.60±1.76 10.50±2.83 11.58±2.06 20.17±1.67 12.57±2.21 12.25±2.96 + NH4 -N 1.66±0.44 2.29±0.73 1.83±0.88 1.28±0.22 3.35±0.83 4.46±1.53 - NO3 -N 4.80±0.95 5.27±1.03 5.21±1.50 5.57±0.88 3.25±1.13 7.43±3.21 TP 0.011±0.002 0.032±0.012 0.240±0.029 0.166±0.038 0.492±0.074 0.698±0.064 Paddy land TN 16.30±2.71 17.45±2.24 18.85±2.88 20.57±3.03 12.92±2.95 24.23±5.05 + NH4 -N 2.68±0.30 3.04±0.49 2.52±0.11 3.71±0.91 2.54±0.24 2.39±1.13 - NO3 -N 9.27±3.97 9.87±3.94 11.83±4.57 9.63±4.12 2.86±0.75 8.35±2.05 TP 0.035±0.008 0.032±0.012 0.078±0.010 0.806±0.055 0.795±0.035 0.343±0.025 Forest land TN 5.93±1.07 6.88±0.87 8.03±0.66 23.93±2.71 11.23±1.08 12.71±1.54 + NH4 -N 0.44±0.34 1.82±0.55 1.49±0.04 4.14±0.72 0.53±0.07 1.39±0.09 - NO3 -N 1.77±0.21 3.16±0.83 1.93±0.93 10.07±1.26 2.59±0.36 4.72±0.25 TP 0.055±0.008 0.076±0.016 0.147±0.013 0.114±0.015 0.040±0.014 0.038±0.007 Wetland TN 8.40±0.66 10.90±0.56 21.30±3.00 19.17±0.68 10.43±1.37 20.10±2.72 + NH4 -N 0.87±0.01 1.16±0.02 3.02±0.01 2.88±0.12 2.76±0.03 4.80±0.10 - NO3 -N 4.40±0.32 7.63±0.24 15.58±2.99 13.20±1.83 2.57±1.03 9.93±1.42 TP 0.070±0.002 0.085±0.002 0.062±0.005 0.856±0.017 0.123±0.020 0.364±0.011

TABLE 8 - The average annual N and P loads and losses by surface runoff for four land use types Pollutant loads (kg·ha-1·a-1) Loss amounts (t·a-1) Land use type + - + - TN NH4 -N NO3 -N TP TN NH4 -N NO3 -N TP Dry land 20.32 4.10 9.12 0.38 69.86 14.10 31.35 1.31 Paddy land 21.78 3.31 12.32 0.17 99.27 15.09 56.15 0.77 Forest land 5.35 1.11 1.69 0.08 19.37 4.02 6.12 0.29 Wetland 26.79 3.94 17.82 0.17 44.95 6.61 29.90 0.29

3788 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

FIGURE 7 - Spatial distribution of N and P loads in surface runoff of Abujiaohe watershed

It can be seen in Table 8 that N loads in surface runoff have shown that there is large fluctuation in the N concen- - for four land use types were higher TP loads. Whilst NO3 tration in bare soil because of the lag effect of the extrac- -N load was the major form of N load in Abujiaohe water- tion process between the surface and the internal N con- shed. It was found that the peak TN load in surface runoff centrations [57]. In addition, fertilization also affects the was 21.78 kg·ha-1·a-1 for paddy land and the maximum losses of N and P, and the use of fertilizer can obviously - -1 -1 NO3 -N load was 17.82kg·ha ·a for wetland. This phe- increase the outputs of N and P [58]. In the present study, nomenon was due to the fact that the nitrate in soil is easily N and P loads in rainfall runoff would significantly in- lost by water movement, especially when runoff discharge crease after topdressing during a period from June to July. is low [52]. Ammonium is difficult to be released or trans- The majority of wetland was located in the downstream of ported by runoff, but it can be strongly adsorbed by soil Abujiaohe watershed and most of wetland soil was contam- particles [53]. The soil physical characteristics affect the N inated by farmland in the transformation process from wet- loss, which includes the soil type, the saturated soil mois- land to paddy land. In the agricultural pollution, dissolved ture content, the organic matter, and the particle distribu- nitrogen was the predominant form of N loss, but for phos- tion [11]. In sands dry area, particulate nitrogen carried by phorus, because phosphorus was less mobile than N and surface runoff sands were the primary forms of nitrogen less soluble, the predominant loss form was influenced by losses under typical rainfall conditions [54]. In addition to the rainfall intensity, and the vegetation cover [59]. Non- soil characteristics, the watershed topography is also im- point source pollution load would be caused by land use portant factor that influence the hydrological process and change as a result of agricultural reclamation. Therefore, degree of erosion [55]. The TN loss due to runoff was reducing N fertilizer application to decrease N loss in run- higher in sloped land compared with flat land. The rainfall off would be a good measure for reducing N and P entering intensity, vegetation fraction and soil moisture content into watershed. were three factors influencing the runoff depth [56]. Be- yond that, the rainfall intensity, vegetation fraction and soil The losses of N (t·a-1) in runoff for land use types were moisture also had a significant influence on the loss of N. determined by concentrations, runoff depth and the land The higher rainfall intensity and the soil moisture content area. In this study, the N losses of nitrogen pollutants like + - are conductive to the surface runoff formation and the N TN, NH4 -N and NO3 -N from paddy land were signifi- output [8]. For the vegetation fraction, some researchers cantly higher than those from dry land, wetland and forest

3789 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

+ - land. Peaks of TN, NH4 -N and NO3 -N losses amounts watershed studied is characterized by the apparent freeze- were 99.27 t·a-1, 15.09 t·a-1 and 56.15 t·a-1 for paddy land thaw action, and it is a typical agricultural development (Table 8). The TP losses amounts by surface runoff meas- area and some of wetlands have been converted into paddy ured for four land use types was dry land (1.31t·a-1)>paddy rice since the 1950s. In this study, soil has been frozen be- land (0.77t·a-1)> forest land and wetland (0.29t·a-1). The fore the rainfall is discharged in the cold season. Whilst reason for such a phenomenon might be the relatively large ever frozen ground prevent water from infiltrating to deep area. In the current study, dry land accounting for 24.1% of layer, which increases the depth of runoff. Beyond that, re- the total watershed area contributed to 29.9% of TN losses gional flow process due to the flat terrain and long flood - and 49.2% TP losses. Whilst for paddy land, which ac- season can accelerate pollutants transportation. NO3 -N counted for 32% of the total area, contributed to 42.5% of was the major form of the N loss by surface runoff in this TN losses and 28.95% of TP losses. The estimated results study. This was in accordance with the previous result that - showed that agricultural production had significant im- NO3 -N loss accounted for 79-88% of the total inorganic pacts on the N and P losses after the surface runoff started, nitrogen losses [61]. This phenomenon can be explained by - even the whole freeze-thaw agricultural area. the NO3 -N characteristics of dissolving in water easily and difficulty adsorbed by the sand [62]. 3.4 Spatial pattern of N and P loads However, little is known about the quantitative rela- The spatial distributions of N and P loads in surface tionship between surface runoff volume and pollutants runoff were explored based on the analysis results of land losses because of the limited data obtained in this study. use distribution carried out with GIS software ArcGIS The existing data is hard to meet the request of agricultural (Version 9.3). The land use datasets of watershed were ex- hydrological model, which further resulting in the eco-hy- tracted from the cloud-free LANDSAT TM image obtained drological process is difficult to be obtained. Additionally, in 2012. The TM image was geometry-corrected and man- the estimated N and P losses by surface runoff may be machine interactive interpreted with the assistance of the higher than the average annual values because rainfall 1:250000 national standard topographic databases and the amounts in 2012 is relatively abundant compared with the EDRAS remote sensing software; next, the land use data annual mean precipitation. Therefore, it is necessary to ac- was acquired by ArcGIS software (Fig.1). After the N and cumulate more hydrological data to acquire the regression P loads of different land use types in surface runoff were equation between pollutants losses and runoff volume of calculated, the distributions of pollutant loads were then Abujiaohe watershed, which would provide a scientific ba- analyzed according to the land use distribution (Fig.7). sis to estimate the annual average N and P losses more ac- - The spatial pattern of the TN and NO3 -N loads in run- curately. off of the land uses in this study were almost the same (e.g., the highest pollutant loads was located on the east of Abu- jiaohe watershed, while the lowest loads appeared on the 5. CONCLUSION west of watershed). However, the spatial distribution of + NH4 -N loads was more complicated. Besides for the ob- This study demonstrated the N and P losses by surface + viously higher value of NH4 -N loads in the eastern water- runoff from different land use types under different rainfall + shed, the NH4 -N loads also had obvious higher level in the events. Soil properties, vegetation coverage, rainfall inten- central area of watershed. Compared with the distribution sity and fertilizer application may lead to serious N and P of N losses, the TP loads in runoff for four land use types loss, and then cause water eutrophication. It was proven were lower. Maximum TP load (0.38 kg·ha-1·a-1) occurred that N loss by surface runoff of paddy land was higher than on the sampling sites of dry land, which was near the center other land use types, and the forest land possessed the min- of the watershed. imum N loss. It emerged that the vegetation coverage was an effective method of preventing runoff and N loss. For N - loss, NO3 -N was the main loss, the concentration of which + 4. DISCUSSION was about three times higher than NH4 -N in runoff. The loss of TP in runoff was less than that of nitrogen, but the The contributing factors of N and P losses by surface content of TP in runoff also had surpassed the threshold runoff from land use types have been widely studied and level to prevent eutrophication of water bodies. There were soil properties, rainfall amounts, agricultural activity and obvious differences on N and P losses by surface runoff, land use change are considered as the most important fac- the major reason of which is that dissolved nitrogen is the tors leading to nutrient loss [60]. This study was conducted predominant form of N loss. But for phosphorus, it is less in the Abujiaohe watershed, located in the typical farm area mobile than N and less soluble. Meanwhile, the applicabil- of Heilongjiang Province, China. N and P losses resulting ity of the SCS-CN model to the Abujiaohe watershed was from high intensity farming activity is one of the major evaluated in this study. The TN loads of wetland and forest causes of the quality of receiving waters. Therefore, the land were the maximum with 26.79 kg·ha-1·a-1 and the min- quantification and characterization of N and P losses from imum with 5.35 kg·ha-1·a-1 respectively. Dry land, which rainfall runoff is necessary for the assessment of their con- accounted for 24.1% of the total watershed area, contrib- tribution of agricultural non-point source pollution. The uted to 29.9% N loss amounts and 49.2% P loss amounts,

3790 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

and paddy land accounting for 32% the total area contrib- [11] Liu, R.M., Wang, J.W., Shi, J.H., Chen, Y.X., Sun, C.C., uted to 42.5% of N loss amounts and 28.95% of TP losses. Zhang, P.P. and Shen, Z.Y. (2014) Runoff characteristics and nutrient loss mechanism from plain farmland under simulated These showed that agricultural production had great influ- rainfall conditions. Science of the Total Environment 468-469, ence on the N and P losses of Abujiaohe watershed and 1069-1077. even the whole freeze-thaw agricultural area. [12] Owens, L.B., Van Keuren, R.W. and Edward, W.M. (2003) Non-nitrogen nutrient inputs and outputs for fertilized pastures in silt loam soils in four small Ohio watersheds. Agriculture Ecosystems & Environment 97, 117-130. ACKNOWLEDGEMENTS [13] Arheimer, B. and Liden, R. (2000) Nitrogen and phosphorus concentrations from agricultural catchments-influence of spa- This research benefited from financial support from the tial and temporal variables. Journal of Hydrology 227, 140- University Technology Innovation Project of Shanxi Prov- 159. ince (2014151) and sponsored by the Fund for Shanxi Key [14] Ding, X.W., Shen, Z.Y., Hong, Q., Yang, Z.F., Wu, X. and Subjects Construction (FSKSC). Liu, R.M. (2010) Development and test of the export coefficient model in the upper reach of the Yangtze River. Journal The authors have declared no conflict of interest. of Hydrology 383, 233-244. [15] Srinivasan, R. and Engel, B.A. (1998) A spatial decision on support system for assessing agricultural non-point source pollution. Water Resources Bulletin 30, 441-452. REFERENCES [16] Whitehead, P.G., Wilson, E.J. and Butterfield, D. (1998) A semi-distributed integrated nitrogen model for multiple source [1] Heckrath, G., Brookes, P.C., Poulton, P.R. and Goulding, assessment in catchments (INCA): part I-model structure and K.W.T. (1995) Phosphorus leaching from soils containing dif- process equations. Science of The Total Environment 210, ferent phosphorus concentrations in the Broadbalk experi- 547-558. ment. Journal of Environmental Quality 24, 904-910. [17] Viney, N.R., Sivapalan, M. and Deeley, D. (2000) A concep- [2] Stutter, M.I., Langan, S.J. and Cooper, R.J. (2008) Spatial and tual model of nutrient mobilisation and transport applicable at temporal dynamics of stream water particulate and dissolved large catchment scales. Journal of Hydrology 240, 23-44. N, P and C forms along a catchment transect, NE Scotland. Journal of Hydrology 350, 187-202. [18] Han, J.G., Li, Z.B., Li, P. and Tian, J.L. (2010) Nitrogen and phosphorous concentrations in runoff from a purple soil in an [3] Cameira, M.R., Fernando, R.M. and Pereira, L.S. (2003) Mon- agricultural watershed. Agricultural Water Management 97, itoring water and NO3-N in irrigated maize fields in the Sorraia 757-762. Watershed, Portugal. Agricultural Water Management 60, 199-216. [19] Bush, B.J. and Austin, N.R. (2001) Timing of phosphorus fertilizer application within an irrigation cycle for perennial pas- [4] Sweeney, D.W., Pierzynski, G.M. and Barnes, P.L. (2012) Nu- ture. Journal of Environmental Quality 30, 939-946. trient losses in field-scale surface runoff from claypan soil receiving turkey litter and fertilizer. Agriculture, Ecosystems & [20] Wilson, G.V., McGregor, K.C. and Boykin, D. (2008) Residue Environment 150, 19-26. impacts on runoff and soil erosion for different corn plant populations. Soil and Tillage Research 99, 300-307. [5] Owens, L.B., Van Keuren, R.W. and Edwards, W.M. (1998) Budgets of non-nitrogen nutrients in a high fertility pasture [21] Wu, L., Long, T.Y., Liu, X. and Guo, J.S. (2012) Impacts of system. Agriculture Ecosystems & Environment 70, 7-18. climate and land-use changes on the migration of non-point source nitrogen and phosphorus during rainfall-runoff in the [6] Kwong, K.F.N.K., Bholah, A., Volcy, L. and Pynee, K. (2002) jialing River Watershed, China. Journal of Hydrology 475, 26- Nitrogen and phosphorus transport by surface runoff from a 41. silty clay loam soil under sugarcane in the humid tropical environment of Mauritius. Agriculture, Ecosystems & Environ- [22] Novotny, V. (1999) Diffuse pollution from agriculture-a ment 91, 147-157. worldwide outlook. Water Science and Technology 39, 1-13.

[7] Santhi, C., Srinivasan, R., Arnold, J.G. and Williams, J.R. [23] Ajmal, M., Moon, G.W., Ahn, J.H. and Kim, T.W. (2015) In- (2006) A modeling approach to evaluate the impacts of water vestigation of SCS-CN and its inspired modified models for quality management plans implemented in a watershed in runoff estimation in South Korean watersheds. Journal of Hy- Texas. Environmental Modelling & Software 21, 1141-1157. dro-environment Research xx, 1-2.

[8] Shigaki, F., Sharpley, A. and Prochnow, L.I. (2007) Rainfall [24] Lewis, D., Singer, M.J. and Tate, K.W. (2000) Applicability intensity and phosphorus source effects on phosphorus of SCS curve number method for a California oak woodlands transport in surface runoff from soil trays. Science of the Total watershed. Journal of Soil and Water Conservation 55, 226- Environment 373, 334-343. 230.

[9] Kong, L.L., Zhang, Z.Y. and Zhu, L. (2010) Review of the [25] Sahu, R.K., Mishra, S.K. and Eldho, T.I. (2010) Comparative trends in the study of non-point source nitrogen in hydrologi- evaluation of SCS-CN-inspired models in applications to clas- cal processes on farmland over irrigation district. Advances in sified datasets. Agricultural Water Management 97, 749-756. Water Science 21, 853-860. [26] Chen, Z.L., Chen, J., Liu, P., Peng, X.C., Xu, Z.C. and Wu, [10] Ganlin, Z., Jinling, Y. and Yuguo, Z. (2003) Nutrient dis- Z.F. (2008) An experiment on influence of storm on nitrogen charges from a typical watershed in the hilly area of subtropi- loss and phosphorus loss under different land use in river ba- cal China. Pedosphere 13, 23-30. sin. Journal of Soil and Water Conservation 22, 30-33.

3791 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[27] Ouyang, W., Huang, H.B., Hao, F.H., Shan, Y.S. and Guo, [44] Zeng, S.C., Su, Z.Y., Chen, B.G., Wu, Q.T. and Ouyang, Y. B.B. (2012) Evaluating spatial interaction of soil property with (2008) Nitrogen and phosphorus runoff losses from orchard non-point source pollution at watershed scale: The phosphorus soils in south China as affected by fertilization depths and indicator in Northeast China. Science of the Total Environ- rates. Pedosphere 18, 45-53. ment 432, 412-421. [45] Zhao, G.J., Hörmann, G., Fohrer, N., Li, H.P., Gao, J.F. and [28] Kara, D., Özsavasci, C. and Alkan, M. (1997) Investigation of Tian, K. (2011) Development and application of a nitrogen suitable digestion methods for the determination of total phos- simulation model in a data scarce catchment in South China. phorus in soils. Talanta 44, 2027-2032. Agricultural Water Management 98, 619-631.

[29] Stanford, G. (1982) Assessment of soil nitrogen availability. [46] Udawatta, R.P., Motavalli, P.P., Garrett, H.E. and Krstansky, Nitrogen in agricultural soils , (nitrogeninagrics) 22, 651-688. J.J. (2006) Nitrogen losses in runoff from three adjacent agricultural watersheds with claypan soils. Agricultural, Ecosys- [30] Bray, R.H. and Kurtz, L.T. (1945) Determination of total, or- tems & Environment 117, 39-48. ganic, and available forms of phosphorus in soils. Soil Science 59, 39-46. [47] Yang, J.L., Zhang, G.L., Shi, X.Z., Wang, H.J., Gao, Z.H. and Ritsema, C.J. (2009) Dynamic changes of nitrogen and phos- [31] National Environmental Protection Agency/Editor Board of phorus losses in ephemeral runoff processes by typical storm Water And Waste Water Monitoring and Analyzing Methods. events in Sichuan Basin, Southwest China. Soil and Tillage (2002) Water And Waste Monitoring and Analyzing Methods, Research 105, 292-299. fourth ed. China Environmental Science Press, Beijing, China. [48] Nyakatawa, E.Z., Mays, D.A., Tolbert, V.R., Green, T.H. and [32] Cabrera, F., Diaz, E., Toca, C.G. and de Arambarri, P. (1982) Bingham, L. (2006) Runoff, sediment, nitrogen, and phospho- A modification of the hydrogen peroxide method for determi- rus losses from agricultural land converted to sweetgum and nation of total phosphorus in natural waters. Water Research switchgrass bioenergy feedstock. Biomass and Bioenergy 30, 16, 1061-1064. 655-664. [49] Cai, Q.G. and Wu, S.A. (1998) Effect of different land use on [33] USDA-Soil Conservation Service (USDA SCS), 1985. Na- soil and water loss processes on purple steep slopeland. Bulle- tional Engineering Handbook. Section 4, Hydrology. USDA- tin of Soil and Water Conservation 18, 1-8. SCS, Washington, D.G. [50] Conley, D.J. (2000) Biogeochemical nutrient cycles and nutri- [34] Bosznay, M. (1989) Generalization of SCS curve number ent management strategies. Hydrobiologia 410, 87-96. method. Journal of Irrigation and Drainage Engineering 115, [51] Smith, V.H., Tilman, G.D. and Nekola, J.C. (1999) Eutrophi- 139-144. cation: impacts of excess nutrient inputs in freshewater, marine and terrestrial ecosystems. Environmental Pollution 100, [35] Ponce, V.M. and Hawkins, R.H. (1996) Runoff curve number: 179-196. Has it reached maturity. Journal of Hydrologic Engineering 1, 11-19. [52] Luo C.Y., Gao, Y., Zhu, B. and Wang, T. (2013) Sprinkler- based rainfall simulation experiments to assess nitrogen and [36] Shadeed, S. and Almasri, M. (2010) Application of GIS-based phosphorus losses from a hillslope cropland of purple soil in SCS-CN method in west bank catchments, Palestine. Water China. Sustainability of Water Quality and Ecology 1, 40-47. Science and Engineering 3, 1-13. [53] Zhu, Z.L. and Wen, Q.X. (1992) Nitrogen of Chinese Soil. [37] Xu, Y. and Pan, W.B. (2006) SCS model for watershed runoff Jiangsu Science and Technology Press, Nanjing. calculation in ArcView. Research of Soil and Water Conser- [54] Qian, X.Y., Shen, G.X., Huang, L.H. and Wang, S.B. (2010) vation13, 176-182. Characteristics of nitrogen and phosphorus losses with rainfall-runoff from sandy dry field in Chongming Dongtan. Jour- [38] Bhaduri, B., Harbor, J., Engel, B. and Grove, M. (2000) As- nal of Soil and Water Conservation 24, 11-14. sessing watershed-scale, long-term hydrologic impacts of land use change using a GIS-NPS model. Environmental Manage- [55] Scanlon, T.M., Kiely, G. and Xie, Q.S. (2004) A nested catchment, 26, 643-658. ment approach for defining the hydrological controls on non- point phosphorus transport. Journal of Hydrology 29, 218-231. [39] Liu, X.Z., Kang, S.Z., Liu, D.L. and Zhang, X.P. (2005) SCS [56] Bouldin, J.L., Farris, J.L., Moore, M.T. and Cooper, C.M. model based bon geographic information and its application to (2004) Vegetative and structural characteristics of agricultural simulate rainfall-runoff relationship at typical small watershed drainages in the Mississippi Delta landscapes. Environmental level in Loess Plateau. Transactions of the CSAE 21, 93-97. Pollution 132, 403-411. [40] Zhao, J.Q. (2002) The pollution control on urban surface run- [57] Dan, B.Q., Yin, C.Q. and Yu, J. (2001) Study on phosphorus off. Beijing: China Environmental Science Press. transport in the surface layer of soil with rainfall simulation method. Acta Science Circumstance 21, 7-12. [41] Xia, L.Z., Hoermann, G., Ma, L. and Yang, L.Z. (2013) Re- [58] Wang, J., Wang, E., Yang, X.G., Zhang, F.S. and Yin, H. ducing nitrogen and phosphorus losses from arable slope land (2012) Increased yield potential of wheat-maize cropping sys- with contour hedgerows and perennial alfalfa mulching in tem in the North China Plain by climate change adaptation. Three Gorges Area, China. Catena 110, 86-94. Climate Change 113, 825-840. [42] Song, Z.F., Wang, K.Q., Sun X.L., He, J.Q., Li, H.B. and Yi, [59] De Jager, N.R. and Houser, J.N. (2012) Variation in water- J.C. (2008) Phosphorous and nitrogen loss characteristics with mediated connectivity influences patch distributions of total runoff on different lands use pattern in small watersheds in N, total P, and TN, TP rations in the Upper Mississippi River, Jianshan River, Chengjiang. Research of Environmental Sci- USA. Freshwater Science 31,1254-1272. ences 21, 109-113. [60] Sun, Z.B., Chen, Z.J., Liao, X.Y. and Wang, H.M. (2011) [43] Yoon, S.W., Chung, S.W., Oh, D.G. and Lee, J.W. (2010) Characteristics of agricultural non-point source nitrogen and Monitoring of non-point source pollutants loads from a mixed phosphorus losses in a typical small watershed in Three forest land use. Journal of Environmental Sciences 22, 801- Georges Reservoir area. Chinese Journal of Ecology 30, 1720- 805. 1725.

3792 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[61] Han, J.G., Li, Z.B. and Qian, C. (2010) Nitrogen, phosphorous and soil losses by runoff in a farmland-dominated watershed in purple soil region. Ecology and Environmental Sciences 19, 423-427.

[62] Zhu, B., Wang, T., Xu, T.P., Kuang, F.B., Luo, Z.X. and Gao, Y. (2006) Non-point source nitrogen movement and its environmental effects in a small watershed in hilly area of purple soil. Journal of Mountain Science 24, 601-606.

Received: March 24, 2015 Revised: June 16, 2015; July 13, 2015 Accepted: August 25, 2015

CORRESPONDING AUTHOR

Xuehui Lai Department of Environment and Safety Engineering Taiyuan Institute of Technology Taiyuan, 030008 P.R. CHINA

Phone: (+86) 351 3566125 E-mail: [email protected]

FEB/ Vol 24/ No 11a/ 2015 – pages 3780 - 3793

3793 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

FeCl3/NaNO2-CATALYZED OXIDATIVE DEGRADATION OF AQUATIC STEROID ESTROGENS WITH MOLECULAR OXYGEN

Lianzhi Wang1,*, Zhengchao Duan1, Taoli Qu1, Dongmei Fu2 and Xinmiao Liang2,*

1 School of Chemical and Environmental Engineering, Hubei University for Nationalities, Enshi 445000, China. 2 Dalian Institute of Chemical Physics, Chinese Academy of Sciences, Dalian 116023, China.

ABSTRACT the growing knowledge about their adverse effects have resulted in an urgent need for eradication of these pollutants The oxidative degradation of natural estrogens 17β-es- from the environment [13-21]. tradiol, estriol and synthetic estrogen 17-ethynylestradiol Although there is significant importance in removing in water was carried out with FeCl3/NaNO2 as catalyst and these estrogens from the environment, only a few reports dioxygen as terminal oxidizing agent. Under a representa- on this particular subject have been found. Most reports in- o tive reaction conditions (150 C and 0.5 MPa of O2, 50 mol volved biological degradation [9,22-25] and advanced ox- % of FeCl3/NaNO2, 4 h reaction), the reaction achieved the idation processes [26, 27] of estrogens. Only recently, a removals of 17β-estradiol, estriol and 17-ethynylestradiol few examples involving degradations of estrogens with being 99%, 99%, and 86%, respectively. By GC/MS anal- photo-catalysis methods have been reported [28-34]. Typ- ysis, the main products were small organic acids. The es- ically, the photocatalytic degradations of estrogens were trogenic activity of the intermediate products was negligi- carried out with TiO2-based catalysts or photo-Fenton un- ble by evaluation of estrogenic activities. der UV/O3/sunlight irradiation. [26-28,30,33-41]. How-

ever, for a number of reasons, these degradation methods

KEYWORDS: for estrogens have a limited degree of success in view of Oxidative degradation, estrogens, iron chloride, sodium nitrite their efficiency, complexity, and industrial viability. On the other hand, wet catalytic air oxidation as a versatile

strategy is increasingly used in the destruction of the envi- 1. INTRODUCTION ronmental pollutants and the reduction of COD level. How- ever, rare examples have been reported that the use of such Estrogens are an important class of steroidal com- wet catalytic air oxidation methods treat estrogens. Here pounds that have found widespread use in hormone re- we report that FeCl3 in combination with NaNO2 is a com- placement therapeutic agents, contraceptive and farm ani- petent catalyst for the oxidative degradation of estrogens, mal meat or milk production regulators. They may also E2, E3, and EE2 using molecular oxygen, and the interme- elicit a variety of adverse effects on both humans and ani- diate products have no estrogenic activity. mals by intervening the normal functions of their endocrine Formerly, we presented that FeCl3/NaNO2 was capa- pathways [1-4]. Documented impacts on wildlife include ble of degrading E2, E3, and EE2 under the mimic natural hermaphrodite polar bears and decreasing fertility in birds, conditions.[42] However, the degradation method suffered fish, frogs and mammals [5]. Human health effects include from a major shortage, namely the degradation rate was reproductive abnormalities, decrease in the sperm count slow. Attempting to obtain a highly efficient degradation and increase in testicular, breast, uterine, ovarian and pros- method that not only can remove estrogens with rapid deg- tate cancers, hypospadias and cryptorchidism [6-8]. New radation rates but also can treat high concentrations of the research suggests that the endocrine disrupting chemicals pollutants, such as the waste from the industrial effluents (EDCs) can rob plants of key nutrients [5]. of contraceptive pill production, we use FeCl3/NaNO2 as The presence of steroid estrogenic compounds in aque- the catalyst and dioxygen as the oxidizing agent for the ex- ous environments can be attributed to both natural occur- ploration of the degradation of estrogenic pollutants with ring estrogens such as 17β-estradiol (E2), estriol (E3), es- the method of catalytic wet air oxidation (CWAO). trone (E1), and anthropogenic activities, such as those ex- creted from humans as a result of their use for contracep- tion and hormone replacement therapy, 17-ethynylestra- 2. MATERIALS AND METHODS diol (EE2), etc. The significant quantities of these estrogen pollutants found in the global environment [4, 9-12] and 2.1 The degradation of estrogens: E2, E3 and EE2 were purchased from Sigma Chemical * Corresponding author Company Ltd (GC grade >98 %). FeCl3 and NaNO2 were

3794 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

purchased from KeMiOu reagent plant (China). E2 was dis- The details of the method were similar to those reported in solved in water (Milli-Q) by ultrasonication at 50 oC. The our previous paper [43]. catalytic oxidation of estrogens was carried out in a 50 mL The test solution was concentrated 20-fold by evapo- Teflon-lined, stainless steel autoclave. The autoclave was ration under vacuum and eluted with dimethyl sulphoxide -1 added with 20 mL of E2 solution (5 mg·L ) and the desired (DMSO, 1 mL), then diluted by 7 orders of magnitude. In amounts of FeCl3 and NaNO2 and then was closed and the control experiment, the original E2 solution was treated charged with pure oxygen to 0.5 MPa. Subsequently it was by the same method. heated to the desired temperature with a stirring speed of 300 rpm. After reaction (heating time included), the autoclave was cooled to room temperature in a water bath, care- 3. RESULTS AND DISCUSSION fully depressurized and sampled for further analysis. The degradation of other estrogens was carried out in a similar 3.1 Oxidative degradation of E2 in the presence of Fe(III) and way. NaNO2 Initial experiments were carried out with 17β-estradiol 2.2 Analysis of the treated solution: (E2) as a prototypical estrogenic pollutant. When 10 mol % Residual estrogens were determined at 30 C using a of FeCl3 and 50 mol % of NaNO2 were added to the reaction high performance liquid chromatography (HPLC) equipped system, the removal of E2 was higher than 74 % after 4 h with an ODS column (250 mm × 4.6 mm) and a Waters 474 under 0.5 MPa of oxygen at 150 oC (Table 1, entry 4). To fluorescence detector (excitation wavelength, 280 nm; understand the role of FeCl3 and NaNO2 in this degradation emission wavelength, 320 nm). The mobile phase was an system, some control experiments were designed. In the acetonitrile /water mixture (4/6, v/v). The flow rate was set first control experiment, the solution of E2 was carried out to 1.0 mL·min-1 for E2 and EE2, and 0.8 mL·min-1 for E3. without catalyst, which indicated the removal of E2 was The injection volume was 8 L. negligible (Table 1, entry 1). The second experiment was With GS-MS analysis, the concentration of E2 was performed with sole NaNO2 as the catalyst. Unfortunately, 100 mg·L-1 (containing 40 % acetonitrile) before reaction. E2 remained recalcitrant toward the degradation method. GC-MS analysis was performed using an Agilent 6890 GC- For example, when 50 mol % NaNO2 (corresponding to MS system with a DB-5 capillary column (30 m × 0.25 mm) 1:2 mol ratio of NaNO2 to E2) was used with E2 for 4 h in o and helium was used as the carrier gas. The injection volume the presence of 0.5 MPa dioxygen at 150 C, the disappear- was 1 L, the split ratio was 50:1, and the oven program was ance of E2 was also negligible (Table 1, entry 2). This re- held at 50 C for 5 minutes, increased by 3 C·min-1 to sult clearly indicated that NaNO2 did not destroy E2 in the system. In the third control experiment, the reaction was 200 C, then 10 C·min-1 to 300 C, and held at 300 C for performed with sole FeCl (10 mol %) as the catalyst, the 30 min. The voltage of electron impact ionization was set 3 removal ratio of E2 was 65 % (Table 1, entry 3). at 70 eV. The scanning range of m/z was from 30 to 500. Before GC-MS analysis, the solution was concentrated 20- The above results showed that the combination of fold by evaporation under vacuum and eluted with acetoni- FeCl3 and NaNO2 in the system exhibited a robust catalytic trile (2 mL), and then was dried under a nitrogen stream. activity for the degradation of E2. The efficiency of E2 The sample of intermediate products analysis was eluted degradation in the combination of FeCl3 and NaNO2 sys- with pyridine (100 L) and derivatized by adding 100 L tem was higher than that in the presence of FeCl3 or NaNO2 of the derivatization reagent (N-methyl-N-trimethylsilyl alone. In view of this promising result, we systematically trifluoroacetamide, MSTFA), and the sample was incu- investigated and optimized the reaction conditions (tem- bated for 30 min (160 rpm) at 37 C , and then stored for perature, reaction time, oxygen pressure, and the amounts 2 h at ambient temperature. The termination products were of NaNO2 and FeCl3). + - derivatized by ((CH3)3S OH ) before analysis. 3.2 Effects of temperature on the degradation of E2 2.3 Evaluation of estrogenic activities Temperature is an extremely important factor in the The estrogenic activities of the test solutions were in- oxidative degradation process. The oxidative degradation vestigated by a recombinant yeast screening assay (YES). of E2 has been studied at temperatures ranging from 90 oC

a TABLE 1 - The removal of E2 in the absence of NaNO2 or FeCl3

Entry NaNO2 (mol%) FeCl3 (mol %) Removal of E2(%) 1 0 0 0 2 50 0 0 3 0 10 65 4 50 10 74 a. The reaction were carried out under 150 oC, 0.5 MPa of oxygen pressure and the reaction time was 4 h.

3795 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

120

100

40 Removal of E2(%) of Removal 20

0 80 100 120 140 160 180 200

Temperature(℃)

FIGURE 1 - Influence of temperature on the removals of E2 -1 Experimental conditions: CE2 = 5 mg L , PO2 = 0.5 MPa, the mol ratio of E2/NaNO2

o to 180 C (the oil bath temperature) for 4 h under 0.5 MPa of O2 and in the presence of 50 mol % of FeCl3/NaNO2 o of oxygen pressure with FeCl3/NaNO2 catalyst. As shown catalyst system at 150 C. As shown in Fig. 2, the removal in Fig. 1, the removal of E2 increases with the reaction tem- of E2 increased with the increase of reaction time from 1 h perature from 90 oC to 150 oC. For example, when the re- to 4 h. For example, there was only 53% removal of E2 in o action temperature was lowered to 90 C, the removal of the presence of 50 mol% FeCl3 and 50 mol% NaNO2 after E2 was only 47 %. As the temperature reached 150 oC, the 1 h reaction. For further prolonging the reaction time to 2 h, removal of E2 was higher than 99 %. 88% of removal of E2 was obtained. When the reaction time was 4 h, the removal of E2 was higher than 99 %. 3.3 Effect of reaction time on the degradation of E2 However, for prolonging the reaction time from 4 h to 8 h, Under the ideal reaction temperature condition, the re- there was no further improvement in E2 removal. So the action time from 1 h to 8 h has been studied under 0.5 MPa ideal reaction time was 4 h in the system.

120

100

40 Removal of E2(%) of Removal 20

0 0246810 Reaction time(h)

FIGURE 2 - Influence of reaction time on the removal of E2 -1 Experimental conditions: CE2=5 mg L , PO2=0.5 MPa, the mol ratio of E2/NaNO2 /FeCl3=1:0.5:0.5

3796 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

o 3.4 Effect of oxygen pressure on the degradation of E2 NaNO2 at 150 C under 0.5 MPa oxygen pressure on the The oxygen pressure is also an important factor in the degradation of E2. As shown in Fig. 3a, the increase of oxidative degradation system for the environmental pollu- FeCl3 amount leads to more efficient degradation of E2 tants. The effect might be related to the oxygen concentra- when the amount of NaNO2 was fixed at 50 mol % and the tion in the liquid phase. Therefore an increase in the oxy- reaction time was set at 4 h. For example, the removal of gen pressure in the gas phase will increase the dissolved E2 was negligible in the presence of 50 mol % NaNO2 oxygen concentration in the liquid phase which is benefi- alone. However, adding of 1 mol % of FeCl3, there was a cial to the pollutant oxidation. To investigate the effect of significant improvement in E2 removal from 0 to 52 %. In- oxygen pressure on oxidative degradation of E2, three ex- creasing the FeCl3 amount to 20 mol % led to 82 % E2 periments were conducted under pressures ranging from removal. More than 99 % of E2 removal was obtained in 0.1 to 0.5 MPa. As shown in Table 2, the removal of E2 the experiment with 50 mol % of FeCl3. However, when increases with the increase of the oxygen pressure. the FeCl3 amount was increased to 100 mol %, there was The control experiment was carried out under 0.5 MPa no further improvement in E2 removal. The results indi- of air instead of oxygen. Surprisingly, 77% of E2 was re- cated that 50 mol % of FeCl3 may be the ideal dosage using moved under this condition. This degradation process was in the current degradation system for E2. of great interest from its potential low cost of industrial ap- Fig. 3b shows the effect of NaNO2 amount on the deg- plications. radation of E2 in the presence of 50 mol % molar ratio of o FeCl3 at 150 C under 0.5 MPa of oxygen pressure. It can 3.5 Influence of the amount of the catalyst on the degradation be observed from Fig. 3b that the larger the NaNO2 amount, of E2 the higher the E2 removal efficiency will be obtained. For The amount of catalyst has significant influence on example, with the addition of 1, 20 and 50 mol % of NaNO2 the process of oxidative degradation. In the study, we in the system, the removal of E2 increased from 97 % to have explored the effect of the amounts of FeCl3 and more than 99 %. It has been demonstrated that steroid estro-

TABLE 2 - The removal of E2 in the different oxygen partial pressure a

Entry Pressure (MPa) NaNO2 (mol%) FeCl3 (mol %) Removal of E2(%) 1 0.1 50 50 70 2 0.3 50 50 94 3 0.5 50 50 99 4b 0.5 50 50 77 a. The reaction were carried out under 150 oC and the reaction time was 4 h. b. Under 0.5 MPa of air.

120

100

40 Removal of E2(%) of Removal 20

0 -101030507090110

The mole ratio of FeCl3(%)

FIGURE 3a - Influence of mol ratio of FeCl3 on the removal of E2. -1 Experimental conditions: CE2=5 mg L , PO2=0.5 Mpa; the mol ratio of E2/ NaNO2=1: 0.5

3797 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

99,5

98,5

97,5

Removal of E2(%) of Removal 97

96,5 -10 0 10 20 30 40 50 60

The mole ratio of NaNO2(%)

FIGURE 3b - Influence of mol ratio of NaNO2 on the removal of E2. -1 Experimental conditions: CE2=5 mg L , PO2=0.5 Mpa; the mol ratio of E2/ FeCl3=1: 0.5

gens, such as E2, can provoke reproductive disturbances in hydroxyl radical, and Fe(III) might act as the most im- riverine fish even at nanogram per liter concentration lev- portant active oxidants for the degradation of pollutants in- els (10-100 ng·L-1) [3,6]. Therefore, a thorough removal of cluding E2. E2 is very important. Hence, the best amount of NaNO2 is 50 mol % in the present system. 3.6 GC/MS analysis of intermediate products and termination products Recently, FeCl3/NaNO2 was found to be highly efficient for the wet oxidative degradation of dye pollutants To further probe the fate of estrogen E2 in the model [44, 45]. We postulate that NO2/NO, Fe(III)/Fe(II) and process, we utilized gas chromatography-mass spectrome- ONOO- are most likely to be involved in the present cata- try (GC-MS) to identify the intermediate products in the lytic system. The same mechanism was supposed in this reaction process. When 50 mol% of NaNO2 and 50 mol% study. Peroxynitrite, which has oxidative activity similar to of FeCl3 were mixed with E2 solution under the standard

Abundance

TIC: 060125-5.D 2.6e+07

2.4e+07 3

2.2e+07

2e+07

1.8e+07

1.6e+07

1.4e+07

1.2e+07 5 1e+07 4

8000000

6000000

4000000 1 2

2000000

10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 Time--> FIGURE 4a -. The Total ion current (TIC) of GC-MS analysis for residual organic compounds of E2 degradation. (reaction time: 1 hour, the samples were derivatized by the derivatization (N-methyl-N-trimethylsilyl trifluoroacetamide, MSTFA).1: silanamine, N,1,1,1-tetramethyl-N- [trimethylsilyl] which produced by derivatization reagents. 2: Butanedioic acid, bis(trimethylsilyl) ester. 3: acetic acid, bis(trimethylsilyl) ester. 4 and 5: bis(trimethylsilyl) 2 or 4-nitroestradiol.)

3798 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

FIGURE 4b - The Total ion current (TIC) of GC-MS analysis for residual organic compounds of E2 degradation. (reaction time: 4 h, the + - samples were derivatized by the derivatization ((CH3)3S OH ) 1 and 5: Cyclotrisiloxane, hexamethyl which produced by column losing. 2: Ethane, 1,2-bis(methylthio-) which produced by derivatization reagents. 3 and 4: electronic response.)

120 100 a 80 b 60 40 20 Relative(%) response 0 -20 1.00E-06 1.00E-04 1.00E-02 1.00E+00

E2 concentration (ng L-1)

FIGURE 5 - Estrogenic activity of aqueous solutions containing various amounts of E2. (a) E2 original solution; (b) E2 solution after reacted 1 h, and the mol ratio of E2/ NaNO2/FeCl3 were 1:0.5:0.5.

reaction conditions for 1 h, mass spectrometric analysis in- assay (YES) was selected to determine the estrogenic activ- dicated that the presence of low molecular weight organic ity of E2 to chart a time course. We prepared an E2 solution acids (e.g. butanedioic acid and acetic acid) and nitroestro- which has been reacted for 1 h. The removal of E2 was gens (i.e. 2-nitroestradiol and 4-nitroestradiol) as the major about 53 %. The solution was concentrated 20-fold by chemical offspring (Fig. 4 a). Of particular interests are the evaporation under vacuum and eluted with dimethyl observations for low molecular weight organic acids and the sulphoxide (DMSO, 1 mL). The treated solution, which nitrified E2 compounds. These compounds were further de- contained about 47 μg·mL-1 of E2, was diluted by about graded with prolonged reaction time to 4 h (Fig. 4 b). 7 orders of magnitude, and the estrogenic activities of the diluted solutions were evaluated and plotted in Fig. 5, curve 3.7 Estrogenic activities b against the E2 concentrations in the diluted solutions. In Estrogenic activity is also one of the critical factors for the control experiment, the original E2 solution was treated evaluating the effectiveness of a new degradation system for to contain the same concentration of E2 with the reaction so- estrogens. For this reason, the recombinant yeast screening lution. The activity-concentration relationship (dose-response

3799 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

curve) for E2 solution without reaction treatment was investi- REFERENCES gated (Fig. 5, curve a). The two curves were in good agreement. This result exhibited that the estrogenic activities of the [1] Jobling, S., Beresford, N., Nolan, M., Rodgers-Gray, T., Brighty, G. C., Sumpter, J. P., Tyler, C. R. (2002) Altered sexual maturation intermediate products were negligible and that there was no and gamete production in wild roach (Rutilus rutilus) living in riv- secondary risk associated with the increased estrogenic activ- ers that receive treated sewage effluents. Biol. Reprod. 66, 272– ity as a result of the model degradation of E2 in water under 281. the present experimental conditions. If the intermediate prod- [2] Rodgers-Gray, T. P., Jobling, S., Morris, S., Kelly, C., Kirby, S., ucts in the treated solution have appreciable estrogenic activi- Janbakhsh, A., Harries, J. E., Waldock, M. J., Sumpter, J. P., Tyler, C. R. (2000) Long-term temporal changes in the estrogenic com- ties, the plot should exhibit higher values than the original position of treated sewage effluent and its biological effects on dose-response curve in a certain concentration region [6]. fish. Environ. Sci. Technol. 34, 1521–1528. [3] Routledge, E. J., Sheahan, D., Desbrow, C., Brighty, G. C., 3.8 The degradation of other estrogens Waldock, M., Sumpter, J. P. (1998) Identification of estrogenic chemicals in STW effluent. 2. In vivo responses in Trout and The degradation model was also applied to estrogen E3 Roach. Environ. Sci. Technol. 32, 1559–1565. and EE2. A similar degradation effect was observed when [4] Liu, J., Du, X., Zhang, C., Lei, Y., Gao, Q., Li, Y. (2011) Deter- they were treated with NaNO2 and FeCl3 under the stand- mination of Estrohenic Hormones in Water Samples Using High ard conditions. For example, with 50 mol% of NaNO2 and Performance Liquid Chromatography Combined With Dispersive o Liquid-Liquid Micro-Extraction Method Based on Solidification 50 mol% of FeCl3 under 0.5 MPa of O2 and 150 C, the -1 of floating organic drop and Pre-Column Derivatization. Fresenius removal of E3 (10 mg·L ) reached 99.2 % after reacting Environmental Bulletin 20, 1075-1083. 4 h and 86.2 % of EE2 (4 mg·L-1) was degraded. Espe- [5] Pennisi, E. (2004) Society for integrative and comparative biology cially, it is noteworthy that EE2, which is generally be- meeting: Estrogen may disrupt nitrogen fixation. Science 303, 950. lieved to be a recalcitrant substrate for degradation, was ef- [6] Hess, R. A., Bunick, D., Lee, K. H., Bahr, J., Taylor, J. A., Korach, ficiently removed in the present degradation system. K. S., Lubahn, B. B. (1997) A role for oestrogens in the male reproductive system. Nature 390, 509-512. The FeCl3/NaNO2/O2 system can rapidly and efficiently degrade estrogens with no secondary pollution. [7] Service, R. F. (1998) Cancer: New role for estrogen in cancer? Sci- Furthermore, the use of dioxygen as the oxidizing agent ence 279, 1631-1633. and FeCl3/NaNO2 as the catalyst renders the reaction inter- [8] Spearow, J. L., Doemeny, P., Sera, R., Leffler, R., Barkley, M. esting from both economical and environmental points of (1999) Genetic variation in susceptibility to endocrine disruption by estrogen in mice. Science 285, 1259-1261. view. Therefore, this inexpensive, simple and efficient catalytic degradation system facilitates its industrial applica- [9] Huang, B., Li, X., Sun, W., Ren, D., Li, X., Li, X., Liu, Y., Li, Q., Pan, X. (2014) Occurrence, removal, and fate of progestogens, an- tions. drogens, estrogens, and phenols in six sewage treatment plants around Dianchi Lake in China. Environ Sci Pollut Res Int. 21, 12898-12908. 4. CONCLUSIONS [10] Grzybowski, W. and Szydlowski, J. (2014) The impact of chromo- phoric dissolved organic matter on the photodegradation of 17α- ethinylestradiol (EE2) in natural waters. Chemosphere 111, 13-17. The FeCl3/NaNO2/O2 system can rapidly and effi- [11] Sarkar, S., Ali, S., Rehmann, L., Nakhla, G., Ray, M. (2014) Deg- ciently degrade natural estrogens 17β-estradiol, estriol and radation of estrone in water and wastewater by various advanced the synthetic estrogen 17-ethynylestradiol. After reacting oxidation processes. J. Hazard. Mater. 278, 16-24. o for 4 h at 150 C and 0.5 MPa of O2, 50 mol% of FeCl3 and [12] Padhye, L.; Tezel, U. (2014) Fate of Environmental Pollutants. 50 mol% NaNO2 as the catalyst, the removals of 17β-estra- Water Environ. Res. 86, 1714-1773. diol, estriol and 17-ethynylestradiol were 99%, 99%, and [13] Raloff, J. (2002) Hormones: Here's the Beef. Environmental con- 86%, respectively. By GC/MS analysis, the main products cerns reemerge over steroids given to livestock. Science News 161, were small organic acids and the estrogenic activities of the 10. intermediate products were negligible by evaluation of es- [14] Baronti, C., Curini, R., D’Assenzo, G.., Corcia, A.D., Gentili, A., trogenic activities. Samperi, R. (2000) Monitoring natural and synthetic estrogens at activated sludge sewage treatment plants and in a receiving river water. Environ. Sci. Technol. 34, 5059-5066. [15] Casey, X.M.; Larsen, G..L., Hakk, H., Šimůnek, J. (2003) Fate and transport of 17β-estradiol in soil-water systems. Environ. Sci. ACKNOWLEDGEMENTS Technol. 37, 2400-2409. [16] Hanselman, T.A., Graetz, D.A., Wilkle, A.C. (2003) Manure- We gratefully acknowledge the financial support from Borne Estrogens as Potential Environmental Contaminants: A Re- the National Natural Science Foundation of China (No. view. Environ. Sci. Technol. 37, 5471-5477. 21267009, 21107108). [17] Nasu, M., Goto, M., Kato, H., Oshima, Y., Tanaka, H. (2001) Study on endocrine disrupting chemicals in wastewater treatment plants. Water Sci. Technol. 43, 101-108. The authors have declared no conflict of interest. [18] Soto, A.M., Calabro, J.M., Prechtl, N.V., Yau, A.Y., Orlando, E.F., Daxenberger, A., Lolok, A.S., Guillette, L.J., Bizec. B.L., Lange, I.G.., Sonnenschein, C. (2004) Androgenic and estrogenic activity in water bodies receiving cattle feedlot effluent in eastern Ne- braska, USA. Environ. Health Perspect. 112, 346-352.

3800 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[19] Ternes, T.A., Stumpf, M., Mueller, J., Haberer, K., Wilken, R.D. [38] Ohko, Y., Iuchi, K., Tatsuma, T. (2002) 17β-Estradiol degradation (1999) Behavior and occurrence of estrogens in municipal sewage by TiO2 photocatalysis as a means of reducing estrogenic activity. treatment plants - I. Investigations in Germany, Canada and Brazil. Environ. Sci. Technol. 36, 4175–4181. Sci.Total Environ. 225, 81-90. [39] Segmuller, B.E., Armstrong, B.L., Dunphy, R. Oyler,. A.R. (2000) [20] Ying, G.G., Kookana, R.S. (2003) Degradation of five selected en- Identification of autoxidation and photodegradation products of docrine-disrupting chemicals in seawater and marine sediment. ethynylestradiol by on-line HPLC-NMR and HPLC-MS. J. Pharm. Environ. Sci. Technol. 37, 1256-1260. Biomed. Anal. 23, 927–937. [21] Yu, C.P., Roh, H., Chu, K.H. (2007) 17β-Estradiol-degrading bac- [40] Tanizaki, T., Kadokami, K., Shinohara, R. (2002) Catalytic photo- teria isolated from activated sludge. Environ. Sci. Technol. 41, degradation of endocrine disrupting chemicals using titanium di- 486-492. oxide photosemiconductor thin films. Bull. Environ. Contam. 68, 732–739. [22] Fujii, K., Kikuchi, S., Satomi, M., Ushio-Sata, N., Morita, N. (2002) Degradation of 17 beta-estradiol by a gram-negative bacterium iso- [41] Zhou, D., Wu, F., Deng, N. (2004) Fe(III)-oxalate complexes in- lated from activated sludge in a sewage treatment plant in Tokyo, duced photooxidation of diethylstilbestrol in water. Chemosphere Japan. Appl. Environ. Microbiol. 68, 2057-2060. 57, 283–291. [23] Shi, J., Fujisawa, S., Nakai, S., Hosomi, M. (2004) Biodegradation [42] Wang, L.Z., Zhang, F.F., Liu, R.H., Zhang, T.Y., Xue, X.Y., Xu, of natural and synthetic estrogens by nitrifying activated sludge Q., Liang, X.M. (2007) FeCl3/NaNO2: An efficient photocatalyst and ammonia-oxidizing bacterium Nitrosomonas europaea. Water for the degradation of aquatic steroid estrogens under natural light Res. 38, 2323-2330. irradiation. Environ. Sci. Technol. 41, 3747-3751. [24] Weber, S., Leuschner, P., Kampfer, P., Dott, W., Hollender, J. [43] Zhang, Y., Chen, J., Zhang, C., Wu, W., Liang, X. (2005) Analysis (2005) Degradation of estradiol and ethinyl estradiol by activated of the estrogenic components in kudzu root by bioassay and high sludge and by a defined mixed culture. Appl. Microbiol. Biotech- performance liquid chromatography. J. Steroid Biochem. 94, 375– nol. 67, 106-112. 381 [25] Yoshimoto, T., Nagai, F., Fujimoto, J., Watanabe, K., Mizukoshi, [44] Peng, Y., Fu, D., Liu R., Zhang F., Xue, X., Xu, Q., Liang, X. H., Makino, T., Kimura, K., Saino, H., Sawada, H., Omura, H. (2007) NaNO2/FeCl3 dioxygen recyclable activator: an efficient (2004) Degradation of estrogens by Rhodococcus zopfii and Rho- approach to active oxygen species for degradation of a broad range dococcus equi isolates from activated sludge in wastewater treat- of organic dye pollutants in water. Appl. Catal. B: Environ. 79, ment plants. Appl. Environ. Microbiol. 70, 5283-5289. 163-170.

[26] Zhang, A., Li, Y. (2014) Removal of phenolic endocrine disrupting [45] Peng, Y., Fu, D., Liu R., Zhang F., Liang, X. (2007) NaNO2/FeCl3 compounds from waste activated sludge using UV, H2O2, and catalyzed wet oxidation of the azo dye acid orange 7. Chemosphere UV/H2O2 oxidation processes: Effects of reaction conditions and 71, 990-997. sludge matrix. Sci. Total Enviro. 493, 307-323.

[27] Li, Y., Zhang, A. (2014) Removal of steroid estrogens from waste activated sludge using Fenton oxidation: Influencing factors and degradation intermediates. Chemosphere 105, 24-30. [28] Koutantou, V., Kostadima, M., Chatzisymeon, E. (2013) Solar photocatalytic decomposition of estrogens over immobilized zinc oxide. Catal. Today 209, 66-73. [29] Mao, K., Li, Y., Zhang, H., Zhang, W., Yan, W. (2013) Photocata- lytic Degradation of 17 alpha-Ethinylestradiol and Inactivation of Escherichia coli Using Ag-Modified TiO2 Nanotube Arrays. Clean- Soil Air Water 41, 455-462. Received: March 25, 2015 [30] Marinho, B., de Liz, M., Lopes Tiburtius, E., Noemi, N., Peralta-Za- Revised: May 07, 2015 mora, P. (2013) TiO2 and ZnO mediated photocatalytic degradation Accepted: June 03, 2015 of E2 and EE2 estrogens. Photoch. Photobio. Sci. 12, 678-683. [31] Yang, J., Wen, G., Zhao, J., Shao, X., Ma, J. (2013) Oxidation of bisphenol A by permanganate: reaction kinetics and removal of es- CORRESPONDING AUTHOR trogenic activity. Water Sci. Technol. 67, 1867-1872. [32] Jiang, L., Zhang, L., Chen, J., Ji, H. (2013) Degradation of 17 beta- Lianzhi Wang estradiol in aqueous solution by ozonation in the presence of manganese(II) and oxalic acid. Environ. Technol. 34, 131-138. School of Chemical and Environmental Engineering Hubei University for Nationalities [33] Han, J., Liu, Y., Singhal, N., Wang, L., Gao, W. (2012) Compara- tive photocatalytic degradation of estrone in water by ZnO and Enshi 445000 TiO2 under artificial UVA and solar irradiation. Chem. Eng. J. P.R. CHINA 213, 150-162. Phone: +86 718 8437531) [34] Liu, L., Li,Y., Zhang, H., Zhang, W. (2012) Photocatalytic Degra- E-mail: [email protected] dation of 17 alpha-Ethinylestradiol Using ZnO self-assembly Mi- crospheres. Fresenius Environmental Bulletin 21, 2232-2237. Xinmiao Liang [35] Coleman, H.M., Eggins, B.R., Byrne, J.A., Palmer, F.L., King, E. Dalian Institute of Chemical Physics (2000) Photocatalytic degradation of 17-β-oestradiol on immobi- Chinese Academy of Sciences lised TiO2. Appl. Catal. B: Environ. 24, L1–L5. Dalian 116023 [36] Feng, X., Tu, J., Ding, S., Wu, F., Deng, N. (2005) Photodegrada- P.R. CHINA tion of 17β-estradiol in water by UV–vis/Fe(III)/H2O2 system. J. Hazard. Mater. B. 127, 129-133. Phone: + 86 411 84379519), [37] Liu, B., Wu, F., Deng, N. S. (2003) UV-light induced photodegra- E-mail: [email protected] dation of 17α-ethynylestradiol in aqueous solutions. J. Hazard. Mater. 98, 311–316. FEB/ Vol 24/ No 11a/ 2015 – pages 3794 - 3801

3801 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

EVOLUTION OF THE MICROBIAL COMMUNITY RESPONSE TO DIFFERENT INDUSTRIAL WASTEWATERS

Qing-Xiang Yang1,2,*, Rui-Fei Wang1,2, Ying Xu1, Kai-Yue Zhuo1, Hong-Li Zhao1, Fei Zhang1 and Liu Yang1

1College of Life Sciences, Henan Normal University, Xinxiang 453007, China 2Key Laboratory for Microorganisms and Functional Molecules (Henan Normal University), University of Henan Province, Xinxiang 453007, China

ABSTRACT 1. INTRODUCTION

The adaptation process of a microbial community to Biological processes are conventionally used for treat- industrial wastewater plays key roles in the successful es- ment of municipal and industrial wastewaters with various tablishment and start-up of wastewater treatment plants. physicochemical characteristics. Performance of a biolog- Here, real-time PCR and 454 high-throughput pyrosequenc- ical wastewater treatment process is primarily dependent on ing were used to analyze the evolutionary process of a mi- microbial activities, which underlies the importance of the crobial community in a municipal wastewater treatment sys- structure and dynamics of microbial communities. With the tem, as compared to printing and dyeing wastewater, and pa- introduction of various molecular techniques, the evolution permaking wastewater system, respectively. During the do- and dynamics of microbial communities in many wastewater mestication processes, the numbers and compositions of treatment processes have been extensively investigated, in- the dominant protozoa largely varied, in which Rotifera cluding anaerobic digestion processes [1], submerged mem- species were generally observed, while Epistylidae, Flag- brane bioreactors [2], and nitrifying-denitrifying sludge sys- ellate, and Vorticella species were more sensitive to vari- tems [3], among others. Several important factors, such as ous industrial wastewaters. Contrary to the total number of the characteristics of wastewater, operation parameters, and microorganisms, the ratio of fungi to bacteria increased at process controlling, separately or jointly influence the struc- the beginning of the domestication processes and then de- ture and composition of microbial communities in waste- creased to initial levels. With the domestication process and water treatment systems [2, 4]. However, most current stud- as compared to seed sludge and control samples, the richness ies have primarily focused on the effects of operation param- of the genera Proteiniclasticum, Arcobacter, and Nitrospira, eters, while relatively few have investigated the impacts of etc. largely decreased or disappeared altogether in both in- wastewater characteristics on microbial communities, es- dustrial wastewater systems. However, species from the pecially the evolution of microbial communities for treat- genera Caldilinea, Phenylobacterium, Brevundimonas, etc. ment of different industrial wastewaters. were largely enhanced in the printing and dyeing wastewater In most current industrial wastewater treatment plants, domestication system. Members of the order Bacillales seed sludge inoculated in the system at the establishment (phylum Firmicutes), especially the genera Bacillus, Exigu- stage is normally collected from municipal wastewater (MW) obacterium, Caryophanon, etc. were largely enhanced in the treatment plants. This sludge undergoes a dramatic variation paper-making wastewater domestication system. Rhodobac- in microbial community structure during acclimation to the ter, Caldilinea, unclassified Rhodobacteraceae, and other new wastewater environment, which sometimes results in unclassified Bacteria were present in all the tested systems, failure of the system start-up, especially for bio-recalcitrant indicating their possible roles in the maintenance of acti- wastewater. Therefore, elucidation of the evolutionary pro- vated sludge structures. The hydraulic retention time was cess of the same microbial community in different indus- very important for sustaining microbial function during the trial wastewater treatment systems will be helpful for sys- domestication process of the printing and dyeing wastewater tem establishment and start-up. system. A short hydraulic retention time resulted in some Treatment of printing and dyeing wastewater (PDW) or important genera, such as Caldilinea, Phenylobacterium, straw pulp papermaking wastewater (PW) is difficult due to and Brevundimonas, being washed away. the low ratio of biological oxygen demand to chemical oxy-

gen demand (BOD5/COD ratio) and toxic content. The establishment of systems to treat these two types of wastewater KEYWORDS: printing and dyeing wastewater; papermaking is very difficult and requires an extended period of time for wastewater; microbial community; 454 high-throughput pyrosequencing; domestication microbial domestication. In this study, with the help of 454 high-throughput pyrosequencing, we investigated the effects of these two types of wastewater on the structure of * Corresponding author microbial communities.

3802 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 1 - Characteristics of wastewater samples.

Samples COD BOD5 Colority pH TN TP Lignin (mg l-1) (mg l-1) (dilutes) (mg l-1) (mg l-1) (mg l-1) PDW 2360-3510 590-1053 128-200 6.8-8.1 5.2-21.3 0.8-8.7 - PW 914-1140 365-479 80-128 5.4-7.2 9.8-16 3.5-6.7 664 MW 177-448 78-169 32-64 4.3-6.7 51-64 42-58 - PDW, collected from printing and dyeing wastewater plant; PW, collected from papermaking wastewater plant; MW, collected from municipal wastewater plant; TN, total nitrogen; TP, total phosphate

2. MATERIALS AND METHODS V3 region of the 16S rRNA gene [8]. The PCR mixture contained 4 μl of 5×FastPfu Buffer, 2 μl of deoxy-ribonu- 2.1 Characteristics of wastewater and sludge samples cleoside triphosphate (2.5 mM each), 0.8 μl of forward pri- Seed sludge samples were collected from the secondary mer (5 μM), 0.8 μl of reverse primer (5 μM), 0.4 μl of sedimentation tank of an MW treatment plant. Industrial FastPfu polymerase and 10 ng of template DNA. PCR am- wastewater (sample PW) was collected from a papermaking plifications were carried out in a total volume of 20 μl in a plant that produces paper from straw pulp, which contained GeneAmp® 9700 thermocycler (Applied Biosystems, Inc., copious amounts of toxic and recalcitrant COD, such as re- Foster City, CA, USA) under the following cycling condi- sidual lignin and brighteners, although most of the lignin tions: initial denaturation for 2 min at 95°C followed by content was recovered via an alkali method [5]. Industrial 25 cycles at 95°C for 30 s, 55°C for 30 s, 72°C for 30 s, wastewater (sample PDW), which contained high concen- and a final extension at 72°C for 5 min. PCR products were trations of COD, different types of dyes (mainly activated sequenced by Beijing Genomics Institute (Shenzhen, China) dyes) and dyestuffs, and salt, was collected from a printing using a 454/Roche GS-FLX Titanium instrument (Roche and dyeing plant. COD, BOD5, pH, and colority were as- Molecular Systems, Inc., Branchburg, NJ, USA). sayed using standard methods [6]. MW was collected from Pyrosequencing data were processed using mother, an the influents of the same plant as the seed sludge and used for open-source, platform-independent, community-supported comparison. The characteristics of the collected wastewater software for describing and comparing microbial commu- samples are listed in Table 1. nities [9], based on the 454 standard operating procedure (http://www.mothur.org/wiki/454_SOP as of November 2.2 Domestication of activated sludge in industrial wastewater 2014). Sequences shorter than 200 bp were excluded. Chi- Seed sludge (5 g wet weight) was inoculated into meric sequences were removed using the UCHIME algo- 1000 mL Erlenmeyer flasks containing 500 mL of differ- rithm (http://drive5.com/uchime). Archaeal, mitochondrial, ent wastewater samples, respectively. Three parallel flasks and chloroplast sequences were also removed. Datasets were prepared for each type of wastewater. The flasks were were normalized to reach the same sequencing depth for cultivated at 25oC while shaking at 150 rpm. Every 12 h, each sample. Sequences were classified using the Riboso- the contents of the flasks were allowed to settle for 20 min mal Database Project reference version 9 and binned into and then the supernatants (400 mL) were replaced with operational taxonomic units (OTUs, 97% similarity) based fresh wastewater, which resulted in a hydraulic retention on taxonomic classifications. Rarefaction curves were ob- time (HRT) of at least 15 h. The wastewater treatment ef- tained at the genus level. The dominant genera in each sam- ficiencies were monitored daily. Sludge samples from the ple (more than 100 sequences) were extracted and plotted domestication flasks were periodically collected on days in a figure. 10, 25, 41, and 86 for microbial community analysis. Principal component analysis (PCA) of seed sludge samples was conducted according to genera proportion us- 2.3 Total DNA extraction ing SPSS 19.0 data analysis and processing software (IBM- For total bacterial community analysis, sludge samples SPSS, Inc., Chicago, IL, USA). In the PCA plots, samples were washed three times with phosphate-buffered saline positioned closely together along an axis were considered (pH 8.0) and centrifuged at 4oC and 10,000 rpm for 15 min. to have similar microbial community compositions. Total genomic DNA of the microbial community was extracted using the cetyltrimethylammonium bromide method 2.5 Quantification of microbial populations [7]. The yield and fragmentation of the crude or purified Quantification of the total microbial community was DNA were determined by agarose gel electrophoresis (1% performed by enumerating cells stained with 4, 6-diamidino- w/v agarose) and Ultraviolet Rays visualization after eth- 2-phenylindole (DAPI) and fixed with 4% paraformalde- idium bromide staining. hyde (m/v) solution [10]. After the samples were concentrated onto 0.2-pore-size black polycarbonate filters (EMD 2.4 454 high-throughput 16S rRNA gene pyrosequencing Millipore, Billerica, MA, USA), the total number of cells The bacterial universal primer pair, 27F (5'-AGAG- was counted under an epifluorescence microscope (Ti-U; TTTGATCCTGGCTCAG-3') and 534R (5'-ATTAC- Nikon Corporation, Tokyo, Japan). At least 10 fields of CGCGGCTGCTGG-3'), was used for analysis of the V1- view per sample were enumerated using oil immersion at a

3803 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

magnification of 1000×. The average number of cells in 3. RESULTS AND DISCUSSION each wastewater treatment period was obtained from three parallel samples. 3.1 Domestication process of seed sludge in different industrial wastewater samples Bacterial and fungal populations were also quantified through a real-time PCR technique using the primer pairs Efficient removal of COD and colority is a key chal- 338F (5′-ACTCCTACGGGAGGCAGCAG) and 518R lenge in the treatment of PW and PDW. Therefore, these (5′-ATTACC-GCGGCTGCTGG) for bacteria, and two parameters were monitored daily during the entire do- FF390 (5′-CGATAACGAAC-GAGACCT) and FR1 (5′- mestication process (Fig. 1). When the seed sludge was in- AICCATTCAATCGGTAIT) for fungi [11, 12]. The PCR oculated into PW, there seemed to be no domestication pe- parameters are reported elsewhere [13]. Amplified 16S or riod in COD removal efficiency, which reached 90% on the 18S rRNA sequences were inserted into plasmids that were first test day and was later maintained at around 77%– transformed into competent Escherichia coli DH5-α cells. 89.8% under the conditions of feeding two batches of The standards were prepared using serially diluted plasmid wastewater. However, for color removal, the seed sludge DNA with 103–108 gene copies µL-1. The specificity of the underwent a significant and long-term domestication pro- amplified products was confirmed by both melting curve cess (Fig. 1A’). The color removal efficiencies increased analysis and agarose gel electrophoresis. Two independent sharply from 60% to 80% after treatment for 21 days, real-time PCR assays were performed for each of the three thereby indicating successful establishment of a domesti- replicate DNA samples. Standard curves were generated cation system. by plotting the threshold cycle values versus log10-transformed gene copy numbers [14]. Compared to PW, it was much more difficult for the seed sludge to grow in PDW. During the first 50 treatment 2.6 Enumeration and observation of protozoa days, the COD and color removal efficiencies were very Identification and enumeration of protozoa were per- low (only 20%–30% for COD and 10%–35% for colority) formed by bright-field microscopy within 5 h after sample (Figs. 1B and B’) and the sludge grew very slowly, as collection, as described in a previous report [15]. Protozoa shown in Figure 2. In consideration of high inhibition po- were identified mostly to the genus level, mainly based on tential, the HRT was extended from 15 to 30 h from day 51 morphology and movement during live observations, and to provide sufficient time for the growth of the minor according to several key factors [16-19]. adaptable microorganisms. We found that the COD and

A (PW)

A’ (PW)

3804 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

B (PDW)

B’ (PDW)

C (MW)

C’ (MW)

FIGURE 1 - Removal efficiencies of COD (A, B, C) and colority (A’, B’, C’) in papermaking (PW), printing and dyeing (PDW), and municipal wastewater (MW), respectively. (■ influent; ◆ effluent; ▲ removal efficiency).

3805 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

FIGURE 2 - Growth of sludge during the domestication process of municipal wastewater, papermaking wastewater, and printing and dyeing wastewater domestication systems (MW, PW, and PDW, respectively).

FIGURE 3 - Richness of protozoa during different wastewater treatment processes. The number of each protozoan species was calculated from an average of 10 microscopic fields. ASS, seed sludge; MW-10, MW-25, and MW-41 indicate municipal wastewater samples obtained on days 10, 25, and 41, respectively; PW-10, PW-25, and PW-41, indicate papermaking wastewater samples collected on days 10, 25, and 41, respectively; PDW-10, PDW-25, and PDW-85 indicate printing and dyeing wastewater samples collected on days 10, 25, and 85, respectively.

color removal efficiencies gradually increased and reached specific protozoa populations in different industrial waste- 91% and 75%, respectively (Figs. 1B and B’). Correspond- waters, even though the same seed sludge was used for all ing to this process, the amount of sludge in the flasks in- inoculations (Fig. 3). In the control system (MW), there creased significantly and the sludge color changed from was no significant variation in protozoan number or species brown to black, which suggested successful establishment during the entire test period. In the PW and PDW systems, of an enrichment system for PDW treatment. the numbers of protozoa significantly decreased on day 10, but recovered to the same or higher level than in the control As compared to the industrial wastewaters, the seed system (MW) at the end of the domestication process (day sludge in the control samples (MW) underwent no signifi- 25 for PW and day 85 for PDW). cant domestication period (Figs. 1C and C’). There was no variation in protozoan populations in 3.2 Dynamic evolution of protozoa population during the do- MW during the entire cultivation period. These popula- mestication process of seed sludge in different industrial tions were mainly composed of Rotifera, Vorticella, Nem- wastewaters atoda, Epistylidae, Flagellate, and Arcellinida with the Microscopic observations indicated changes in both same species diversity as those present in the seed sludge. the number and dominance of protozoan species in the seed However, when the seed sludge was inoculated into PW sludge during the domestication process, which then formed flasks, the environment was obviously unsuitable for growth

3806 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

of Flagellate and Vorticella, as indicated by the absence of protozoans are widely employed as indicators of activated these organisms in the samples before day 21. As the do- sludge performance and effluent quality [20]. In this study, mestication process proceeded, especially after feeding the Ciliatea were only observed in well-domesticated sludge, second batch of PW, Flagellate and Vorticella spp. reap- while Rotifera were generally observed in the two indus- peared. trial wastewater systems and during the entire domestica- In the case of the PDW domestication process, more tion process, which suggested strong suitability to these en- dramatic changes occurred in the protozoan population. Ep- vironments. However, Epistylidae, Flagellate, and Vorti- istylidae spp. completely disappeared from PDW system cella spp. were obviously more sensitive to the various in- shortly after the seed sludge was inoculated. Vorticella spp. dustrial wastewater environments. A study by Hu et al. [15] disappeared during the first 25 days, but reappeared at the reported similar results, in which the abundances of Epi- end of the domestication period, especially after the HRT stylis plicatilis and Vorticella striata largely changed in the was extended. Moreover, at the end of the domestication activated sludge in response to the operation parameters. period, populations of Ciliatea and Paramecium spp. were 3.3 Dynamic evolution of microbial populations during the do- especially prevalent, as indicated in Fig. 3. mestication process of seed sludge in different industrial According to Dubber and Gray, ciliate protozoa play a wastewaters vital role in sludge activation process by reducing the den- The total number of microorganisms was counted us- sity of dispersed bacteria through predation, thereby con- ing the DAPI staining technique. As shown in Fig. 4A, as tributing to the flocculation process, and improving efflu- compared to the seed sludge, there was no significant ent quality by reducing BOD and concentrations of sus- change in microbial concentrations in the control flasks pended solids in effluents. The diversity and abundance of (MW flasks) (1.08 × 1014–2.15 × 1014 g-1 dried activated

s 16

15 ) -1

14 (log 10 g (log 13

Total number of microorganism of number Total 12 ASS MW-10 MW-25 MW-41 PW-10 PW-25 PW-41 PDW-10 PDW-25 PDW-85

Samples

A 20 ) -1 15

0 ASS MW-10 MW-25 MW-41 PW-10 PW-25 PW-41 PDW-10 PDW-25 PDW-85 -5 Microbial concentrations (log10 g (log10 concentrations Microbial -10 Samples

FIGURE 4 - Enumeration results of total microbes (bacteria and fungi) in the activated sludge during the domestication processes in different wastewaters using the DAPI method (A) and real-time PCR (B) ( , bacteria; , fungi; , fungi to bacteria; ASS, seed sludge). MW-10, MW-25, and MW-41 indicate municipal wastewater samples collected on operation days 10, 25, and 41, respectively; PW-10, PW-25, and PW- 41 indicate papermaking wastewater samples collected on operation days 10, 25, and 41, respectively; PDW-10, PDW-25 and PDW-85 indicate printing and dyeing wastewater samples collected on operation days 10, 25, and 85, respectively.

3807 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

sludge). However, during the first 10 days of the domestica- abundance of fungi in the wastewater systems. The results tion period of industrial wastewater, the total number of mi- of our previous study of full-scale printing and dyeing croorganisms significantly decreased (8.8 × 1013 g-1 for PW wastewater treatment systems were similar [22]. However, and 3.4 × 1013 g-1 for PDW), but then recovered to the initial the large abundance of bacteria in the activated sludge was values after domestication (25 days for PW and 85 days for endowed with great functional redundancy [4], thus once PDW). This phenomenon possibly resulted from the toxicity the resistant bacterial population was established, the and lethality of PW and PDW to some microorganisms. growth of bacteria exceeded that of fungi. However, with the progress of the domestication process, the microbial communities changed or became resistant to 3.4 Bacterial compositions in different wastewater treatment the toxicity, resulting in an increased number of microbes systems at the end of the domestication period in the two industrial wastewater systems. The PDW was To explore how microbial communities change from obviously more toxic to the microorganisms. The sludge seed sludge during the domestication process in industrial concentrations and number of microbes recovered very wastewater systems, we performed high-throughput se- slowly in PDW system, which actually occurred some time quencing of the 16S rRNA genes, which yielded between after the HRT was extended (after 41 days). 5125 and 7703 sequences per sample. The sequences were Variations in the abundance of bacteria and fungi were binned into OTUs based on taxonomical attributes and rar- also monitored by real-time PCR. As shown in Fig. 4B, the efaction curves were plotted at the genus level (Fig. 5). concentrations of bacteria and fungi were in the range of With the OTUs defined at the genus level, the curves of all 1013–1015 and 107–109 g-1 of dried activated sludge, respec- samples nearly plateaued at 5125 sequences sampled. tively. These levels were similar to or slightly higher than Therefore, we could normalize the dataset to reach similar the number of microbes determined by the DAPI method sequencing depth by randomly selecting 5125 sequences used for total microbial counts in this study. Compared to from each sample. Mothur analysis indicated that these the control, the ratio of fungi to bacteria increased in both sequences were clustered in 224, 221, 138, 130, and 144 industrial domestication systems during the first 10 days OTUs for the sample of seed sludge (ASS), a sample col- and then gradually decreased throughout the rest of the do- lected on day 25 from the control system (MW-25), a PW mestication period, which suggested that the bacteria were sample collected at the end of the domestication period more sensitive to changes or toxicity of wastewaters than (PW-25), and PDW samples collected during the middle the fungi. Although occupying a small proportion in the and end of the analysis (PDW-25 and PDW-85), respec- microbial community in wastewater systems, some fungi, tively. The number of OTUs in sample MW-25 was very especially yeasts, produce extracellular enzymes, such as close to that in the seed sludge, indicating that lab condi- manganese dependent peroxidase or lignin peroxidase, as tions did not impact the bacterial diversity in the municipal a stress response to bio-recalcitrant or toxic substances [10, wastewater system. However, the OTUs in the two indus- 21]. After the seed sludge from the municipal wastewater trial wastewater systems decreased by 35.7%–47.3%, indi- treatment system was inoculated into the industrial cating that the domestication process significantly de- wastewater, the microbes experienced toxic shock, while creased the bacterial diversity in the industrial wastewater. the fungi were more tolerant, as indicated by the relative The number of OTUs from sample PDW-85 was higher

FIGURE 5 - Rarefaction curves constructed from the 16S sequencing results (sequences were binned to OTUs based on taxonomic characteristics at the genus level); MW-25, PW-25, PDW-25, and PDW-85 indicate sludge samples collected from the control, papermaking wastewater domestication system on day 25, and the printing and dyeing wastewater system on days 25 and 85, respectively.

3808 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

FIGURE 6 - Bacterial compositions of sludge samples based on mothur analysis of 16S rRNA gene sequences detected by the 454 pyrosequencing method (A, phylum level; B, family level; C, genus level); MW-25, PW-25, PDW-25, and PDW-85 indicate sludge samples collected from the control, papermaking wastewater domestication system on day 25, and the printing and dyeing wastewater system on days 25 and 85, respectively.

than that from sample PDW-25, indicating that extension clustered together, unlike the industrial domestication sam- of the HRT was helpful to maintain a greater diversity of ples. bacterial species. With the domestication process and as compared to the Bacterial sequences with a mean relative abundance of seed sludge and control samples, the richness of Protein- > 1% used to identify the dominant bacterial compositions. iclasticum, Arcobacter, unclassified Rhizobiales, Nitro- As shown in Fig. 6, Proteobacteria, Firmicutes, Bac- spira, unclassified Actinobacteria and unclassified Bac- teroidetes, Actinobacteria, and Chloroflexi were generally teroidetes genera largely decreased or disappeared alto- dominant in all samples, although the relative richness of gether in both industrial wastewater systems. PW and each differed among samples (Fig. 6A). At the end of do- PDW are considered to be very difficult to treat because of mestication process, as compared to the seed sludge, there the large amount of toxic and bio-recalcitrant pollutants, were no significant changes in the richness of composition such as various synthetic dyes and dye additives in PDW, of bacterial communities in the control flask (MW) at each and lignin and various brighteners in PW. Nitrospira are classification level (Fig. 6). However, large variation in nitrite-oxidizing bacteria that are important for nitrogen re- bacterial communities occurred in the two industrial moval from municipal wastewater; however, they fail to wastewater flasks (Figs. 6B and C). This phenomenon was grow in PW and PDW. Species of the genus Arcobacter clearly indicated by the PCA analysis that was based on are generally found in both animal and environmental genus-level bacterial community data (Fig. 7), in which the sources. However, they were also sensitive to the two in- seed sludge and the control sample (MW-25) were closely dustrial wastewater conditions in this study. Reportedly,

3809 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

specific in the PW system. For example, the abundance of members of the order Bacillales (phylum Firmicutes), especially genera of Bacillus, Exiguobacterium, Caryopha- non, Lysinibacillus, Planococcaceae_incertae_sedis, Soli- bacillus, Sporosarcina, unclassified Planococcaceae, as well as unclassified Bacillales, was very low in the seed sludge, the control samples and the PDW system, but largely enhanced in the paper-making wastewater domestication system becoming the dominant population. These bacteria are possibly able to degrade bio-recalcitrant substances, such as lignin or their metabolic intermediates.

Some genera of bacteria, which included Rhodobacter, Caldilinea, unclassified Rhodobacteraceae, unclassified Alphaproteobacteria, and unclassified Bacteria, were dominant in all systems, although the prevalence of each differed among the wastewater systems (Fig. 6C). These bacteria appear to be the core populations of the activated sludge and are possibly important to sustain the structure of the sludge particles. FIGURE 7 - PCA analysis based on the 16S sequencing results (sequences were binned to OTUs based on taxonomic characteristics at Together, these results suggest that developing a quick the genus level); MW-25, PW-25, PDW-25, and PDW-85 indicate domestication method for a specific wastewater system by sludge samples collected from the control, papermaking wastewater screening specific bacterial consortiums and adding these domestication system on day 25, and the printing and dyeing wastewater system on days 25 and 85, respectively. The closer the two species to the seed sludge at system establishment and samples were in the graph, the more similar the genera compositions. start-up is practical. Notably, extending the HRT was very important for sustaining the functional microbes during the some species of genus Proteiniclasticum are both anaero- domestication process. In the PDW system, samples PDW- bic and proteolytic [23]. We detected a high abundance of 25 and PDW-85 were chosen as representative samples be- such species in the seed sludge (about 10%), but these con- fore and after extension of the HRT, respectively. In PDW- centrations decreased during the domestication process in 25 (HRT of 15 h), a genus of unclassified Rhodobacter- both systems, as well as in the control flasks. The aerobic aceae was the predominant population, accounting for domestication conditions employed in this study were ob- 46.8% of all bacteria, and the diversity of bacteria was sig- viously deleterious to this genus of bacteria. nificantly reduced, as compared to the seed sludge, control samples, and sample PDW-85 (HRT of 30 h) (Fig. 6C). Contrary to the above bacterial populations, quantities The prevalence of some possibly important populations, of some genera of bacteria were significantly enhanced in such as Caldilinea, Phenylobacterium, Brevundimonas, the industrial wastewater systems. For examples, the rich- Xanthobacter, Stappia, unclassified Sphingobacteriales, ness of Caldilinea, Phenylobacterium, Brevundimonas, unclassified Xanthobacteraceae and unclassified Alphap- Sphingobium, Xanthobacter, Stappia, unclassified Sphin- roteobacteria species, was high in sample PDW-85, but gobacteriales, unclassified Cytophagaceae and unclassi- were detected or very low in sample PDW-25. This phe- fied Xanthobacteraceae species was very low in the seed nomenon suggests that these bacteria grew slowly and were sludge and MW-25, but significantly higher in PDW-85. washed out during the domestication system under the con- According to Kragelund et al., members of the genus Cald- dition of a short HRT (15 h), resulting in low wastewater ilinea (phylum Chloroflexi) were generally detected in treatment efficiency. Extension of the HRT (i.e., 30 h after wastewater treatment plants, and the total Chloroflexi pop- day 51) significantly increased bacterial diversity (as indi- ulation constituted 12% of the entire microbial population cated by sample PDW-85 in Fig. 6C) and improved the and seems to play an important structural role in the sludge COD and color removal of the wastewater (Figs. 1B and floc formation. Some Caldilinea species showed high po- B’). Hu et al. reported that different treatment processes tential for macromolecule conversion, as evidenced by the with the same wastewater may affect the abundance and high expression levels of exoenzymes [24]. Another genus, diversity of microbial communities, and longer sludge re- Sphingobium, is believed to degrade a variety of environ- tention time or lower food/microorganism ratios provided mental chemical pollutants, such as aromatic and chloroar- more time for the bacteria to adapt to pollutants [4]. There- omatic compounds, as well as phenols and polycyclic aro- fore, before starting up a wastewater treatment system, matic hydrocarbons. The high prevalence of the above gen- comprehensive research on the operation parameters, espe- era in sample PDW-85 suggested that these bacteria might cially HRT, is very important for sustaining a greater di- be very efficient in the degradation of various synthetic versity of microbial species. dyes and dye additives. The bacterial community was very

3810 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

4. CONCLUSIONS [7] Zhou, J., Bruns, M. and Tiedje, J. (1996) DNA recovery from soils of diverse composition. Appl. Environ. Microbiol. 62, 316-322. During the domestication process of the same seed [8] Guo, H., Liu, R., Yu, Z., Zhang, H., Yun, J., Li, Y., Liu, X. sludge from an MW treatment plant in the PDW and PW and Pan, J. (2012) Pyrosequencing reveals the dominance of wastewater systems, the following general observations methylotrophic methanogenesis in a coal bed methane reser- were noted: the numbers and species of protozoa dramati- voir associated with Eastern Ordos Basin in China. Int. J. Coal. cally varied; the ratio of fungi to bacteria slightly fluctuated Geol. 93, 56-61. at first and then recovered; the abundance of the dominant [9] Schloss, P.D., Westcott, S.L., Ryabin, T., Hall, J.R., Hart- bacterial genera changed, forming specific bacterial com- mann, M., Hollister, E.B., Lesniewski, R.A., Oakley, B.B., Parks, D.H., Robinson, C.J., Sahl, J.W., Stres, B., Thallinger, munities in response to the components of different G.G., Horn, D.J.V. and Weber, C.F. (2009) Introducing wastewaters; operation parameters (i.e., HRT) signifi- mothur: open-source, platform-independent, community-sup- cantly influenced the persistence of some important genera ported software for describing and comparing microbial com- of bacteria during domestication of the PDW system, munities. Appl. Environ. Microbiol. 75, 7537-7541. which subsequently influenced the establishment of an ef- [10] Yang, Q., Angly, F.E., Wang, Z. and Zhang, H. (2011) ficient wastewater treatment system. Wastewater treatment systems harbor specific and diverse yeast communities. Biochem. Engin. J. 58-59, 168-176. [11] Overas, L., Forney, L. and Dae, F.L. (1997) Distribution of bacterial plankton in meromictic lake saelevanner, as determined by denaturing gradient gel electrophoresis of PCR am- ACKNOWLEDGEMENTS plified gene fragments coding for16S rRNA. Appl. Environ. Microbiol. 63, 3367-3373. The authors would like to acknowledge the financial [12] Valášková, V. and Baldrian, P. (2009) Denaturing gradient gel support from the National Natural Science Foundation of electrophoresis as a fingerprinting method for the analysis of China (NSFC 21277041 and NSFC 21477035), Program for soil microbial communities. Plant soil environ. 55, 413-423. Innovative Research Team (in Science and Technology) in [13] Yang, Q., Zhang, W., Zhang, H., Li, Y. and Li, C. (2011) University of Henan Province (IRTSTHN, 13IRTSTHN009), Wastewater treatment by alkali bacteria and dynamics of microbial communities in two bioreactors. Bioresour. Technol. The Outstanding Talented Persons Foundation of Henan 102, 3790-3798. Province (144200510007), the Specialized Research [14] Gutiérrez, R.A., Green, P.J., Keegstra, K. and Ohlrogge, J.B. Fund for the Doctoral Program of Higher Education (2004) Phylogenetic profiling of the Arabidopsis thaliana pro- (20134104110006) and Doctor Initiative Foundation of teome: what proteins distinguish plants from other organisms. Henan Normal University (qd14166). Genome Biol. 5, R53. [15] Hu, B., Qi, R., An, W., Xu, M., Zhang, Y., Bai, X., Bao, H., Wen, Y., Gu, J. and Yang, M. (2013) Dynamics of the micro- The authors have declared no conflict of interest. fauna community in a full-scale municipal wastewater treatment plant experiencing sludge bulking. Eur. J. Protistol. 49, 491-499. [16] Foissner, W., Berger, H. and Schaumburg, J. (1999) Identifi- REFERENCES cation and Ecology of Limnetic Plankton Ciliates. Bavarian State Office for Water Management, Munich. [1] Kim, J., Lee, S. and Lee, C. (2013) Comparative study of [17] Kudo, R. (1966) Protozology. Springfield, Illinois. changes in reaction profile and microbial community structure [18] Patterson, D.J. (1996) Free-Living Freshwater Protozoa: A in two anaerobic repeated-batch reactors started up with dif- ColourGuide. Manson Publishing Ltd, London. ferent seed sludge. Bioresour. Technol. 129, 495-505. [19] Shen, Y.F., Zhang, Z.S., Gong, X.J., Gu, M.R., Shi, Z.X. and [2] Ma, J., Wang, Z., Yang, Y., Mei, X. and Wu, Z. (2013) Corre- Wei, Y.X. (1990) Modern biomonitoring techniques using lating microbial community structure and composition with freshwater microbiota. China Architecture & Building Press, aeration intensity in submerged membrane bioreactors by 454 Beijing. high-throughput pyrosequencing. Water Res. 47, 859-869. [20] Dubber, D. and Gray, N.F. (2009) Enumeration of protozoan [3] He, S.Y., Li, J.X., Peng, Z.G., Zhang, Q.Z., Xu, Y.T. and ciliates in activated sludge: Determination of replicate number Yang, H.Z. (2011) Nitrobacteria community structure during using probability. Water Res. 43, 3443-3452. the startup in a submerged biofilters for in-situ remediation of contaminated stream. Fresen. Environ. Bull. 20, 2111-2118. [21] Yang, Q., Zhang, H., Li, X., Wang, Z., Xu, Y., Ren, S., Chen, X., Xu, Y., Hao, H. and Wang, H. (2013) Extracellular enzyme [4] Hu, M., Wang, X., Wen, X. and Xia, Y. (2012) Microbial com- production and phylogenetic distribution of yeasts in munity structures in different wastewater treatment plants as wastewater treatment systems. Bioresour. Technol. 129, 264- revealed by 454-pyrosequencing analysis. Bioresour. Technol. 273. 117, 72-79. [22] Yang, Q., Wang, J., Wang, H., Chen, X., Ren, S., Li. X, Xu, [5] Ouyang, H., Song B. (2005) Techniques of pulping and pa- Y., Zhang, H and Li, X. (2012) Evolution of the microbial permaking wastewater treatment and its progress. China Pulp community in a full-scale printing and dyeing wastewater & Paper 24, 48-52. (Chinese Journal) treatment system. Bioresour. Technol. 117, 155-163. [6] Taras, M. (1971) Standard methods for the examination of wa- [23] Zhang, K., Song, L. and Dong, X. (2010) Proteiniclasticum ru- ter and wastewater. American Public Health Association: New minis gen. nov., sp. nov., a strictly anaerobic proteolytic bac- York, N Y, the 13 edition, 874 P. terium isolated from yak rumen. Int. J. Syst. Evol. Microbiol. 60, 2221-2225.

3811 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

[24] Kragelund, C., Thomsen, T.R., Mielczarek, A.T. and Nielsen, P.H. (2011) Eikelboom's morphotype 0803 in activated sludge belongs to the genus Caldilinea in the phylum Chloroflexi. FEMS Microbiol. Ecol. 76, 451–462.

Received: March 27, 2015 Revised: July 07, 2015 Accepted: July 09, 2015

CORRESPONDING AUTHOR

Qingxiang Yang

College of Life Sciences

Henan Normal University

Xinxiang 453007

P.R. CHINA

Phone: 86-373 3325528

Fax: 86-373 3325528

E-mail: [email protected]

FEB/ Vol 24/ No 11a/ 2015 – pages 3802 - 3812

3812 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

OPTIMIZATION OF DERIVATIZATION OF ACIDIC DRUGS FOR ANALYSIS BY GC-MS AND ITS APPLICATION FOR DETERMINATION OF DRUG RESIDUES IN WASTEWATER

Richard Sýkora1,*, Milada Vávrová1, Petr Lacina1 and Ludmila Mravcova1

1 Institute of Chemistry and Technology of Environmental Protection; Brno University of Technology – Faculty of Chemistry, Purkyňova 118, 612 00 Brno, Czech Republic

ABSTRACT the process of cleaning. Generally in wastewater treatment plants (WWTPs) drugs are removed from 60-90% [2-4]. This article is focused on optimization of derivatization From the WWTP these substances enter surface waters, of acidic drugs from the group of non-steroidal anti-inflam- where they can lead to intoxication of aquatic organisms matory drugs (NSAID): salicylic acid, ibuprofen, naproxen, [3, 5]. ketoprofen and diclofenac. Derivatization is a necessary step Due to the high consumption of drugs its input to the en- for analysis of pharmaceuticals by GC. MSTFA (N-methyl- vironment is practically continual. In waters trace amounts of N-(trimethylsilyl)trifluoroacetamide) and BSTFA (N,O- drugs are found, which may not cause acute risks, but can bis(trimethylsilyl)trifluoroacetamide) were used as a deri- become dangerous in the long-term exposure [6]. Negative vatization reagents and tested under different conditions: effect on aquatic organisms has been demonstrated for hor- volume, temperature and time of derivatization. Results of monal and antibacterial drugs. Largely also analgesics and optimization were compared and optimized conditions nonsteroidal anti-inflammatory drugs (NSAIDs) enter the were successfully applied for analysis of real samples of environment, which are one of the world's most widely wastewater. Wastewater samples were collected from influ- used groups of drugs. Those do not have adverse effects as ent and effluent of wastewater treatment plant (WWTP). The much as hormone or antibacterial drugs, but their increas- whole analytical procedure involved solid phase extraction ing presence in the aquatic environment should be con- (SPE), optimized derivatization and analysis by GC-TOF MS. trolled as well as other xenobiotics. It requires sufficiently Limits of detection (LOD) ranged between 8.9 - 71.6 ng.L-1 sensitive analytical methods that can qualitatively and and limits of quantification (LOQ) ranged between 29.5 – quantitatively determine the amount of these substances in 238.7 ng.L-1 depending on the specific compound. Concen- waste and surface waters. tration of selected drugs in real samples of wastewater varied -1 -1 LC-MS is the most common analytical method used from 0.1 to 30 g.L in influent and from 0.05 to 2 g.L in for drug analysis from water due to the fact, that pharma- effluent, depending on the compound. ceuticals are generally non-volatile compounds. Neverthe- less GC-MS offers substantially better separation effi-

KEYWORDS: ciency based on fused silica capillary column. Mass spec- Non-steroidal anti-inflammatory drugs, derivatization, silylation, trometric detection with electron ionization enables relia- derivatization reagents, gas chromatography, wastewater ble identification of separated compounds on the base of library searches. However use of chromatography for de-

termination of drugs requires the necessity of derivatiza- 1. INTRODUCTION tion by suitable derivatization reagents. Derivatization provides increasing of volatility and thermal stability. Deriva- Worldwide production and consumption of drugs char- tives with these properties are more suitable for analysis by acterized by different structure increases every year. Due gas chromatography. The most common used derivati- to the fact that these substances are not always completely zation reagents for acidic drugs are MSTFA (N-methyl- eliminated in the human or animal body, subsequently get N-(trimethylsilyl)trifluoroacetamide) and BSTFA (N,O- into the environment [1], which leads to a negative effect bis(trimethylsilyl)trifluoroacetamide) [3, 7, 8-13]. of certain constituents of natural ecosystems [2]. Their big- This study is focused on the optimization of derivati- gest representation is in the waste water. However, it was zation necessary for GC-MS analysis of selected drugs. shown that most of them are very difficult to decompose in Five of the most used acidic drugs (salicylic acid, ibuprofen, naproxen, ketoprofen diclofenac) from the group of * Corresponding author non-steroidal anti-inflammatory drugs were selected.

3813 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

2. MATERIALS AND METHODS

2.1 Chemicals The following chemicals have been purchased from these establishments: methanole for HPLC from Lach-Ner, s.r.o., Czech Republic, Pyridin from Lach-Ner, s.r.o., Czech Republic, salicylic acid from Sigma-Aldrich, Germany, ibu- FIGURE 2 - Chemical structure of tested silylation reagents profen sodium salt from Sigma-Aldrich, Germany, naproxen from Sigma-Aldrich, Germany, ketoprofen from Sigma- Aldrich, Germany, diclofenac sodium salt from Sigma- Aldrich, Germany, N-methyl-N-(trimethylsilyl)trifluoroacetamid (MSTFA) from Sigma-Aldrich, Germany, N,O- bis(trimethylsilyl)trifluoroacetamid (BSTFA) from Sigma- FIGURE 3 - Example of derivatization of ibuprofen Aldrich, Germany.

Due to the fact that the selected drugs (salicylic acid, 2.2 GC-MS instrumentation ibuprofen, naproxen, ketoprofen diclofenac) are all carbox- Gas chromatographic analysis was carried out with a ylic acids (Fig. 1), they were identified and quantified by Pegasus IV D from LECO, USA equipped with column HT- GC-MS as silyl esters [13]. Trimethylsilyl derivatives 5 (30 m x 0,25 mm; d.f. = 0,25 m). Samples were injected (TMS) were prepared with two silylating reagents: N-me- (1 l) into the GC in splitless mode at 270 °C. The carrier thyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA) and gas was helium set at the constant flow mode (1 ml/min). For N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) (Fig. chromatographic separation of all derivatives, the tempera- 2), under different conditions. Optimization of derivatiza- ture was programmed as follows: at 80 °C for 1 min, from tion includes following: volume of derivatization reagents, 80 °C to 300 °C at 20 °C/min and 300 °C at 2 min. The gas time of derivatization and temperature of derivatization. chromatograph was connected to mass spectrometry with Optimized derivatization method then was verified and analysator time-of-flight. MS was operated under elec- used for determination of selected drugs in real samples. tronic impact (EI) mass spectra mode at 70 eV ionization Whole analytical method including solid-phase-extraction energy. The MS transfer line temperature was maintained (SPE), optimized derivatization and analysis by gas chro- at 280 °C, whereas the ion source temperature was 220 °C. matography with „time-of-flight“ mass spectrometric de- Range of observed molecular weights was from 50 to 600. tection (GC-TOF MS) was successfully applied for deter- Scanning velocity was programmed on 5 spectrum.s-1. mination of selected drugs from wastewater on influent and effluent of wastewater treatment plant Brno – Modřice in 2.3 Assessment of Retention times, monitored mass and mass Czech Republic. spectrum The stock solution of selected drugs in methanol with a concentration of 80 g/mL was prepared with analytical pre- cision for assessment of retention times, monitored mass and mass spectrum. Then from the stock solution 500 l were taken into 1.5ml vial and evaporated to dryness under a gentle stream of nitrogen and redissolved in mix of 100 l pyridine (as solvent) and 300 l derivatization reagent (MSTFA or BSTFA). The incubation of prepared samples processed at 70 °C for 90 min. This values were used and modified from [13] for primary assessment and in the next steps this values were optimized for our specific analysis. After incubation, samples were analyzed by GC-TOF MS.

2.4 Optimization of derivatization

For optimization of derivatization the following conditions were tested: different volume ratios of solvent (pyridine) and derivatization reagent (MSTFA or BSTFA); different temperatures of incubation and different time of incubation. The stock solution of disolved drugs in methanol with a concentration of 80 ng/mL was prepared. From this stock solution were taken 500 l into 1,5ml vial in every case of tests and evaporated to dryness under a gentle

FIGURE 1 - Chemical structure of selected NSAID stream of nitrogen. Then followed dissolving in different

3814 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

mixes of pyridine and derivatization reagent and incuba- used for this extraction method were previously optimized tion at different temperatures for a different times. and published in [14, 15]. Before SPE method, 400 ml sam- For optimal volume ratio of pyridine and MSTFA the ples of wastewater were filtered on a glass microfiber paper following three different ratios were tested: 100 l of and adjusted to pH 2 with HCl. SPE cartridges were condi- MSTFA + 300 l of pyridine; 150 l of MSTFA + 250 l tioned with 6 ml of methanol and 3 ml of Milli-Q water adjusted to pH 2. Then 400 ml of prepared wastewater sam- of pyridine; 200 l of MSTFA + 200 l of pyridine. ple were loaded. After sample loading, cartridges were For optimal volume ratio of pyridine and BSTFA the washed with 3 ml of Milli-Q water and dried under vacuum following four different ratios were tested: 100 l of BSTFA for 20 min. After elution with 8 ml of methanol, 500 l of + 300 l of pyridine; 150 l of BSTFA + 250 l of pyridine; eluate were taken into 1,5ml vial and evaporated to dryness 200 l of BSTFA + 200 l of pyridine and 300 l of under a gentle stream of nitrogen and redissolved in opti- BSTFA + 100 l of pyridine. mized mix of 200 l pyridine and 200 l of MSTFA and For optimal time of incubation the following different incubated at optimized conditions: 70 °C, 90 min. After in- times were tested: 30 min, 60 min, 90 min, 120 min; and cubation, samples were analyzed by GC-TOF MS with the for optimal temperature of incubation were tested follow- same adjustment as for optimization. ing different temperatures: 30 °C, 50 °C, 70 °C, 90 °C.

2.5 Wastewater sampling 3. RESULTS AND DISCUSSION Wastewater samples were collected from the WWTP in Brno – Modřice in one-day intervals for 10 days. The 3.1 Analytical data amber glass sample bottles were used for this collection. Fig. 4 shows chromatograms of trimethylsilyl deriva- Samples were collected from influent and effluent of tives of selected drugs for MSTFA and BSTFA. In Table 1 WWTP. The samples were processed and analyzed imme- values of retention times and monitored mass for MSTFA diately after sampling and transportation into laboratory. and BSTFA derivatives of selected drugs are given. Reten- tion times, mass spectrum and unique monitored mass are 2.6 Sample preparation and analysis the same for MSTFA and BSTFA due to the fact, that For the extraction was used solid phase extraction MSTFA even BSTFA create the same derivatives of se- (SPE) with Oasis HLB cartridges from Waters. Conditions lected drugs.

FIGURE 4 - Chromatogram of model sample with MSTFA (a), BSTFA (b); 1 – salicylic acid, 2 – ibuprofen; 3 – naproxen; 4 – ketoprofen; 5 – diclofenac

3815 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

TABLE 1 - Analytical data (monitored mass, retention time)

MSTFA BSTFA TMS of compound monitored mass retention time (s) monitored mass retention time (s) salicylic acid 267 495,3 267 498,7 ibuprofen 160 531,5 160 534,3 naproxen 185 692,7 185 692,9 ketoprofen 282 727,7 282 727,7 diclofenac 214 749,5 214 749,5

3.2 Optimization of derivatization derivatization reagent are 200 l of MSTFA + 200 l of Results of optimal volume of derivatization reagents pyridine (Fig. 5) and 300 l of BSTFA + 100 l of pyridine are shown on folowing pictures of chromatograms. Pic- (Fig. 6). tures show, that the best ratio of volume of pyridine and

FIGURE 5 - Chromatograms – different volumes of MSTFA; (a) 100 l of MSTFA + 300 l of pyridine, (b) 150 l of MSTFA + 250 l of pyridine, (c) 200 l of MSTFA + 200 l of pyridine; 1 – salicylic acid, 2 – ibuprofen; 3 – naproxen; 4 – ketoprofen; 5 - diclofenac

3816 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

FIGURE 6 - Chromatograms – different volumes of BSTFA; (a) 100 l of BSTFA + 300 l of pyridine, (b) 150 l of BSTFA + 250 l of pyridine, (c) 200 l of BSTFA + 200 l of pyridine, (d) 300 l of BSTFA + 100 l of pyridine; 1 – salicylic acid, 2 – ibuprofen; 3 – naproxen; 4 – ketoprofen; 5 – diclofenac

Because of similarity of using MSTFA and BSTFA, ysis of real samples. Fig. 7 shows dependence of signal in- only one derivatization reagent was chosen. Its use was op- tensity of each drug derivative on different derivatization timized under different conditions, and then used for anal- reagents. Higher intensity is reached with use of MSTFA,

3817 © by PSP Volume 24 – No 11a. 2015 Fresenius Environmental Bulletin

1,6E+07

1,4E+07

1,2E+07

1,0E+07 MSTFA 8,0E+06 BSTFA 6,0E+06

4,0E+06

2,0E+06 Intensity of signal (area of peak) of (area signal of Intensity 0,0E+00

d en ac ox pr clofen Ibuprofen Na i Ketoprofen D Salicylic aci TMS of drug

FIGURE 7 - Dependence of signal intensity each drug derivative on different derivatization reagent

(a) 1,8E+07

1,6E+07

1,4E+07

1,2E+07

1,0E+07 30°C 50°C 8,0E+06 70°C 6,0E+06 90°C

4,0E+06

2,0E+06 Intensity of signal (area of of peak) signal Intensity

0,0E+00

c a cid ofen r roxen p ylic a Ibup Na lic Ketoprofen Diclofen Sa TMS of drug

1,8E+07 (b) 1,6E+07 1,4E+07 1,2E+07 30 min 1,0E+07 60 min 8,0E+06 90 min 6,0E+06 120 min 4,0E+06 2,0E+06

Intensity of signal of peak) (area 0,0E+00

n n id ac n rofe rofe p p Ibu Naproxen iclofe licylic ac Keto D a S TMS of drug

FIGURE 8 - Dependence of signal intensity of each drug derivative on different conditions: temperature (a), time (b)

therefore this reagent was chosen for another tests and op- of derivatization. Fig. 8(a) shows depending of signal in- timizations. tensity on different temperature of incubation and Fig. 8(b) shows depending of signal intensity on different time of in- Following figures show dependence of signal intensity cubation. From the results is obvious, that the best temper- of each drug derivative (created by MSTFA) on conditions ature for incubation is 70 °C and the best time is 90 min.

Therefore this conditions were used for analysis of real TABLE 2 - Linearity, limits of detection and quantification samples. Linearity LOD LOQ TMS of compound (R) (ng.L-1) (ng.L-1) 3.3 Real samples of wastewater salicylic acid 0,9987 8,9 29,5 Analytical method including SPE, optimized derivatization and GC/MS-TOF analysis were successfully applied ibuprofen 0,9986 47,0 156,8 on real samples of wastewater from influent and effluent of naproxen 0,9981 20,0 66,7 WWTP Brno – Modřice. Calibration curves of each drug ketoprofen 0,9966 56,3 187,8 were constructed for evaluation of their concentrations in diclofenac 0,9963 71,6 238,7 waste water. In Table 2 are linearity (R) limits of detection (LOD) and quantification (LOQ).

Salicylic acid

0,30

0,25

0,20

c (μg.L-1) 0,15

0,10

0,05

0,00 23.III. 24.III. 25.III. 26.III. 27.III. 30.III. 31.III. 1.IV. 2.IV. 3.IV.

date influent effluent

Ibuprofen

35,0

30,0

25,0

20,0 c (μg.L-1) 15,0

10,0

5,0

0,0 23.III. 24.III. 25.III. 26.III. 27.III. 30.III. 31.III. 1.IV. 2.IV. 3.IV.

date influent effluent Naproxen

1,40

1,20

1,00

0,80 c (μg.L-1) 0,60

0,40

0,20

0,00 23.III. 24.III. 25.III. 26.III. 27.III. 30.III. 31.III. 1.IV. 2.IV. 3.IV.

date influent effluent

Ketoprofen

5,00 4,50 4,00 3,50 3,00 c (μg.L-1) 2,50 2,00 1,50 1,00 0,50 0,00 23.III. 24.III. 25.III. 26.III. 27.III. 30.III. 31.III. 1.IV. 2.IV. 3.IV. date influent effluent

Diclofenac

15,0

12,0

9,0 c (μg.L-1) 6,0

3,0

0,0 23.III. 24.III. 25.III. 26.III. 27.III. 30.III. 31.III. 1.IV . 2.IV . 3.IV .

date influent effluent

FIGURE 9 - Concentrations of selected drugs in influent and effluent of WWTP Brno – Modřice and their comparing

Fig. 9 shows graphs of concentrations of each drug on MSTFA was chosen as a very suitable derivatization rea- influent and effluent of WWTP and their comparing. The gent. Derivatization method optimalized and performed in ranges of concentrations of selected drugs in influent and this paper is convenient not only for derivatization of per- effluent were as follows: salicylic acid 0.15 – 0.27 / 0,09 – formed drugs from group of non-steroidal anti-inflamma- 0,15 g/L; ibuprofen 18.75 – 29,11 / 0.06 – 0.11 g/L; tory drugs, but even for derivatization of another acidic naproxen 0.91 – 1,31 / 0.17 – 0.74 g/L; ketoprofen 1.73 – drugs. This method thus enables using of advantages of gas 4.32 / 0.47 – 1.74 g/L; diclofenac 1.29 – 14.27 / 0.53 – chromatography for trace analysis of wide group of phar- 1.54 g/L.Average removal efficiency of the WWTP for maceuticals. Performed derivatization method was suc- each drug was as follows: salicylic acid 35.4 %; ibuprofen cessfully applied for determination of selected drugs in real 99.3 %; naproxen 60.1 %; ketoprofen 63.1 % and diclo- samples of wastewater. Results of this paper prove, that fenac 62.4 %. non-steroidal anti-inflammatory drugs are relatively significant contaminants of aquatic environment due to their presence in effluent of WWTP. 4. CONCLUSION

GC-MS is very suitable analytical method for determination of organic compounds, such as non-steroidal anti- ACKNOWLEDGEMENT inflammatory drugs, in environmental samples due to high separation efficiency of GC and high selectivity and sensi- Financial support from the MŠMT project Czech Re- tivity of MS. Only disadvantage is necessity of derivatiza- public; research plan No. FCH-S-15-2869. tion with suitable derivatization reagent. However this study proves, that use of silylation reagents is not difficult. The authors have declared no conflict of interest.

REFERENCES [15] Čapka, L., Lacina, P., Vávrová, M. (2012) Development and application of SPE/CZE method for detection and determination of selected nonsteroidal anti-inflammatory drugs in [1] Cooper, E. R., Siewicki, T. C., Phillips, K. (2008) Preliminary wastewater. Fresenius Environmental Bulletin 21, (11a), risk assessment database and risk ranking of pharmaceuticals 3312-3317. in the environment. Science of Total Environment. 398 (1-3), 26-33. [2] Kotyza, J., Soudek, P., Kafka, Z., Vaněk, T. (2009) Léčiva – „nový“ environmentální polutant. Chemické listy. 103, 540- 547. [3] Lacina, P., Ženatová, P., Vávrová, M. (2012) The assesment of contamination of selected river streams in the Czech republic by human and veterinary drug residues with liquid and gas chromatography. Fresenius Environmental Bulletin. 21 (11a), 3318-3324.

[4] Cunningham, V. L., Binks, S. P., Olson, M. J. (2009) Human health risk assessment from the presence of human pharmaceuticals in the aquatic environment. Regulatory Toxicology and Pharmacology. 53 (1), 39-45. [5] Hernando, M.D., Mezcua, M., Fernández-Alba, A. R. and Bar- celó, D. (2006) Environmental risk assesment of pharmaceutical residuem in wastewater effluents, surface waters and sediments. Talanta 69, 334-342. [6] Gómez, M. J., Petrović, M., Fernández-Alba, A. R., Barceló, D. (2006) Determination of pharmaceuticals of various therapeutic classes by solid-phase extraction and liquid chromatography–tandem mass spectrometry analysis in hospital effluent wastewaters. Journal of Chromatography A 1114 (2), 224-233.

[7] Ahrer, W., Scherwenk, E., Buchenberger, W. (2000) Determi- nation of drug residues in water by the combination of liquid chromatography or capillary electrophoresis with electrospray mass spectrometry. Journal of Chromatography A 910 (1), 69- 78. [8] Abu-Qare, A. W., Abou-Donia, M. B. (2001) A validated HPLC method for the determination of pyridostigmine bromide, acetaminophen, acetylsalicylic acid and caffeine in rat plasma and urine. Journal of Pharmaceutical and Biomedical Analysis 26 (5-6), 939-947.

[9] Lacey, C., McMahon, G., Bones, J., Barron, L. P., Morrissey, A., Tobin, J. M. (2008) An LC–MS method for the determination of pharmaceutical compounds in wastewater treatment plant influent and effluent samples. Talanta 75 (4), 1089-1097. [10] Kostranoj, A. V., Golubitski, G. B., Basova, E. M., Budko, E. V., Ivanov, V. M. (2008) High-Performance Liquid Chroma- tography in the Analysis of Multicomponent Pharmaceutical Preparations. Journal of Analytical Chemistry 63 (6), 516-529. [11] Ahrer, W., Scherwenk, E., Buchberger, W. (2001) Determina- Received: April 01, 2015 tion of drug residues in water by the combination of liquid Accepted: June 18, 2015 chromatography or capillary electrophoresis with electrospray mass spectrometry. Journal of Chromatography A.. 910, 69- 78. CORRESPONDING AUTHOR [12] Nebot, C., Gibb, S. W., Boyd, K. G. (2007) Quantification of human pharmaceuticals in water samples by high performance Richard Sýkora liquid chromatography-tandem mass spectrometry. Analytical Chemica Acta 598, 87-94. Brno University of Technology Faculty of Chemistry – Institute of Chemistry and [13] Sebök, Á., Vasanits-Zsigrai, A., Palkó, G., Záray, G., Molnár- Technology of Environmental Protection Perl, I. (2008) Identification and quantification of ibuprofen, naproxen, ketoprofen and diclofenac present in waste-waters, Purkyňova 118 as their trimethylsilyl derivatives, by gas chromatography 612 00 Brno mass spectrometry. Talanta 76, 642-650. CZECH REPUBLIC [14] Čapka, L., Lacina, P. and Vávrová, M. (2012) Optimisation of solid phase extraction of selected non-steroidal anti-inflamma- E-mail: [email protected] tory drugs by capillary zone electrophoresis. Chemické listy 106, 30-35. FEB/ Vol 24/ No 11a/ 2015 – pages 3813 - 3821

ADSORPTION BEHAVIOR OF VICTORIA BLUE DYE ON HALLOYSITE NANOTUBES FROM AQUEOUS SOLUTIONS

Pınar Turan Beyli

Balikesir University Faculty of Science and Literture Department of Chemistry 10145 Çağış, Balıkesir, Turkey

ABSTRACT Several methods like precipitation, coagulation, membrane separation, photo-degradation, biosorption, ion ex- This study examines the adsorption and electrokinetic change, solvent extraction, filtration, electrochemical treat- properties of victoria blue dye (VB) onto halloysite nano- ment, and advanced oxidation processes have been exten- tubes from aqueous solutions. The adsorption of VB was sively exploited for the removal of dyes-contaminated studied with batch experiments. Equilibrium was rapidly wastewater treatment [5-7]. But most of the methods men- attained after 60 min of contact time. The halloysite nano- tioned above have some disadvantages, such as consump- tubes used as adsorbent in this work were initially charac- tion of large quantity of chemicals, generation of secondary terized by BET, XRD, FTIR-ATR, SEM/EDX and Zetasizer pollutants, being time consuming, having high cost, nonbi- Nano ZS. The zeta potential of the halloysite was negative odegradable and incomplete removal of dyes, and are not in the wide pH range (>2.2) and indicated that the material able to effectively remove the dyes. In contrast to the tech- had a potential for binding of cationic dyes. The effects of niques mentioned earlier, adsorption is a most versatile and pH, ionic strength and temperature on adsorption process convenient method for the removal of dyes due to high ef- were investigated. Adsorption increased with increasing ficiency, environmental friendliness, low cost, easy opera- initial pH, ionic strength and temperature. The Langmuir tion, insensitivity to toxic substances and the possibility of and Freundlich models were applied to describe the equi- the materials recycling [6,8]. Generally, cost effectiveness, librium isotherms and the Langmuir model agreed very locally availability and adsorptive property are the main well with experimental data. The isosteric heat of adsorp- criteria for choosing an adsorbent. Currently, activated car- tion (H°) indicated that the adsorption process was spon- bon, the most widely used adsorbent for dyes removal, has taneous and endothermic. The obtained results indicated the disadvantages of high costs and problems associated that the halloysite nanotubes had the potential to be utilized with its subsequent treatment and regeneration. These lim- as low-cost and effective alternative for dye removal in itations have encouraged the searches for inexpensive and wastewater. readily available materials as adsorbents for dyes removal

[9]. Many low-cost adsorbents including zeolites, kaolinite, perlite, sepiolite and montmorillonite clays, banana KEYWORDS: Halloysite nanotube, victoria blue, adsorption, zeta peel, resin, saw dust, fly ash, and other materials have been potential, characterization. tested as adsorbents to remove dyes from colored wastewater

[6]. Among them, clays are being considered as promising alternative adsorbents due to their cheap and abundant re- 1. INTRODUCTION sources, simply operation, large specific surface area and high adsorption capacity [9]. Synthetic dyes and pigments are widely used in various fields, such as leather, paper, printing, plastics, textile, Recently, halloysite nanotubes, economically viable clay dye synthesis, pulp mills, etc., and their discharge into nat- mineral, attracted a great interest due to their hollow tubular ural waters causes environmental problems, related to their structure, high surface area and unique surface properties [6, carcinogenicity, toxicity to aquatic life and undesirable 10]. Besides, in contrast to other nanomaterials, naturally oc- aesthetic aspect [1,2]. Besides, they also interfere with the curring halloysite is easily obtained and an inexpensive na- transmission of light and upset the biological metabolism noscale container. Halloysite is an aluminosilicate which processes which cause the destruction of aquatic commu- has two different basal faces. The first one consists of a nities present in the ecosystem [3,4]. To meet environmen- tetrahedral silicate surface Si-O-Si while the other basal tal regulations and reduce the harm to human, it is impera- surface has gibbsite octahedral layer (Al(OH)3) [11]. Hal- tive for industries to decrease the dyes in their effluents to loysite nanotubes are widely used in the areas of electron- an acceptable level before discharging into the environ- ics, catalysis, biological and functional materials due to ment. Therefore, increasing attention is paid to the removal their submicron hollow tubular structure, rich reactive of dyes from aqueous systems in the last few years. groups, and large surface/volume ratio [12,13]. Compared

with other tubular nanomaterials (such as carbon nanotubes 2.2 Characterization of halloysite nanotubes (CNTs)), natural halloysite nanotubes are cheap and envi- Nitrogen adsorption–desorption isotherms were meas- ronmentally friendly [14]. Due to these excellent proper- ured with NOVA2200e, Quanto Chrome Instruments at 77 K ties, halloysite nanotubes can be used as cheap and efficient for the specific surface area estimations with the BET adsorbents for dyes wastewater. In the literature, only a (Brunauer-Emmett-Teller) technique. X-ray diffraction meas- limited number of studies on the use of halloysite nanotube urement of halloysite nanotubes was performed using an An- as an adsorbent have been found. Luo et al. [9] investigated alytical Philips X’Pert-Pro X-ray diffractometer equipped the removal of methylene blue from aqueous solutions by with a back monochromator operating at 40 kV and a cop- adsorption onto chemically activated halloysite nanotubes. per cathode as the X-ray source (λ = 1.54 Å). Fourier trans- Peng et al. [14] investigated the adsorption of methylene formed infrared spectroscopy was performed with Perkin- blue and malachite green dyes in aqueous solutions by chi- Elmer Spectrum 100 FTIR-ATR spectrometer in the range tosan-halloysite nanotubes composite hydrogel beads. of 600–4000 cm-1 to observe halloysite nanotube surface Kiani et al. [2] investigated the removal of malachite green functional groups. The size and morphology, and chemical from aqueous solutions onto halloysite nanotubes; Liu et analysis of halloysite nanotubes were characterized with a al. [1] investigated the adsorption of methyl violet from Zeiss EVO LS 10 scanning electron microscope (SEM/EDX). aqueous solution by halloysite nanotubes. Zhao and Liu [15] investigated the adsorption behavior of methylene blue on 2.3 Zeta potential analysis halloysite nanotubes. Luo et al [16] investigated the adsorp- Zeta potential measurements of dilute halloysite sus- tion of neutral red from aqueous solution onto halloysite pensions (1 mg/mL) in an aqueous solution with pH adjust- nanotubes. Therefore, the aim of the present study was to ment using 0.1 M HCl or 0.1 M NaOH, were performed investigate the ability of halloysite nanotubes as an adsor- using a Zetasizer Nano ZS, Malvern. The operating condi- bent forthe removal of victoria blue (VB) from aqueous so- tions were checked and adjusted using a calibrated latex lutions using batch method.The halloysite nanotubes were solution supplied by the instrument manufacturer [17]. characterized using BET, XRD, FTIR-ATR and SEM/EDX. Theelectrokinetic properties of halloysite were studied by 2.4 Adsorption studies Zetasizer Nano ZS instrument. The effects of pH, ionic Adsorption studies were carried out by the batch tech- strength and temperature on adsorption process were evalu- nique to obtain equilibrium data. Batch adsorption studies ated and the equilibrium adsorption isotherms were reported. were performed at different pHs, ionic strengths and temperatures. For isotherm studies, adsorption experiments were carried out by shaking 0.1 g halloysite nanotubes with 50 mL 2. MATERIAL AND METHODS aqueous solution in a series of 100 mL polyethylene flasks. Each polyethylene flask was filled with 50 mL of a dye so- 2.1 Material lution of varying concentrations (20-140 ppm) and adjusted Halloysite and victoria blue (C33H32ClN3) were pur- to the desired pH, ionic strength and temperature. A known chased from Sigma-Aldrich and Fluka, respectively. The amount of adsorbent was added to each polyethylene flask dye stock solution was prepared from an exact and pre- and agitated for the desired time periods, up to a maximum weighed quantity of VB dye (4-[[4-anilino-1-naphthyl] of about 60 min. Preliminary experiments demonstrated [4(dimethylamino)-phenyl]-methylene]-cyclohexa-2,5-dien- that the equilibrium was established in 60 min. Equilibra- 1-ylidene]dimethylammonium chloride) having a molecular tion for longer times gave practically the same uptake. formula of C33H32ClN3 and a formula weight of 506.10 g/mol Therefore, a contact period of 60 min was finally selected that was dissolved in 1000 mL of double distilled water and for all of the equilibrium tests. At the end of the adsorption used throughout the experiment. The experiments were period, the solution was centrifuged for 15 min at 5000 rpm carried out by diluting the bulk solution with the desired and then the concentrations of the residual dye, Ce, were concentration. Double distilled water was used for all the determined with the aid of a PerkinElmer Lambda 25 UV– experiments. Figure 1 shows the structure of VB. vis spectrophotometer. The measurements were made at the wavelengths= 620 nm, which corresponds to maximum absorbance. Blanks containing no dye were used for each series of experiments. The effect of pH was observed by studying the adsorption of dye over a pH range of 2–6 with NaOH or HCl solution by using an Orion 920A pH-meter equipped with a combined pH electrode. pH-meter was standardized with NBS buffers before every measurement. The sorption studies were also carried out at different temperatures, i.e., 30, 40, 50 and 60 0C, to determine the effect of temperature and to evaluate the isostreic heat of sorption. A thermostated shaker bath was used to keep the temperature constant. The sorption studies were also carried FIGURE 1 - The structures of victoria blue dye out at different ionic strengths, i.e.,1x10-1, 1x10-2, 1x10-3

and 1x10-4 mol/L NaCl concentrations. The amounts of in the adsorbent material is identified with Fourier Trans- dyes adsorbed were calculated from the concentrations in form Infrared Spectroscopy (FTIR) in the range of 4000– solutions before and after adsorption. All the experiments 600 cm−1. The FTIR-ATR spectrum of halloysite nanotubes is were carried out in duplicate. The amount of dyes adsorbed depicted in Figure 3. The peaks at 3691 and 3622 cm−1 are (mg/g), (qe), onto halloysite nanotubes was calculated from attributed to the stretching vibrations of inner-surface hy- the mass balance equation as follows: droxyl groups, whose deformation vibration is at 907 cm−1. The band at 998 cm−1 is caused by the stretching vibrations V of Si-O-Si. Interlayer water is indicated by the deformation q  (C - C ) (1) e 0 e W vibration at 1652 cm−1 [8]. The VB dye adsorbed on hal- −1 −1 loysite nanotubes shows peaks at 1584 cm and 1369 cm . where C0 and Ce are the initial and equilibrium liquid- These peaks of VB on halloysite nanotubes imply that VB phase concentrations of dye solution (mg/L), respectively; dye was physically adsorbed on the surface of halloysite Vthe volume of dye solution (L), and Wthe mass of hal- nanotubes because it was not observed the shift in the FTIR loysite sample used (g) [18]. peaks of halloysite. Figure 4 presents the scanning electron

100 3. RESULTS AND DISCUSSION 80 3.1 Characterization Halloysite nanotubes were characterized using BET, 60 XRD, FTIR-ATR and SEM/EDX. The specific surface area of the halloysite was found as 32 m2/g. Figure 2 shows Counts 40 the XRD pattern of halloysite nanotubes. X-ray diffraction (XRD) peaks of the nanotubes are consistent with those of 20 halloysite-7A (Al2Si2O5(OH)4, JCPDS Card 29-1487). The significant broadening of the diffraction peaks is ascribed 0 to the very small crystallite size [19]. The plot for the hal- 5 15253545 loysite in Figure 2 shows the basal spacing reflections in- Position [°2Th.] dicating a peak at 120 (2) which translate to a (001) basal spacing of 7 A. The presence of various functional groups FIGURE 2 - XRD pattern of halloysite nanotube

a 1652

3691 1115 3621 792 749

%T b 998 907 1652 1584 1369

3691 1116 906 792 3621 995 748

4000.0 3000 2000 1500 1000 650.0 cm-1 FIGURE 3 - FTIR-ATR spectrum of (a) halloysite nanotube and (b) VB adsorbed halloysite nanotube.

cps/eV 18

8 O Al Si

0 1 2 3 4 5 6 7 8 9 10 keV FIGURE 4 - SEM/EDX pattern of halloysite nanotube

microscopy studies of halloysite nanotubes. Scanning elec- the solid–liquid interface increase with increase in solid tron microscopy reveals that majority of sample consists of concentration and that using inadequate solids concentra- cylindrical tubes of 50 – 100 nm diameter and length of tion can lead to erroneous conclusions in the interpretation 0.5-2 μm. Previous works [19,20] stated that the halloysite of zeta potential measurements. Therefore, in the subse- nanotubes were geometrically similar to multiwall carbon quent zeta potentials measurements the solid-to-liquid ratio nanotubes. But, different to multiwall carbon nanotubes, has been kept constant as 1 g/L. the halloysite nanotubes observed in Figure 4 are straight without entanglement. Energy-dispersive X-ray spectros- -15 copy (EDX) analysis is an analytical technique used for the 0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 elemental analysis or chemical characterization of a sam- -16 ple. It relies on an interaction of some source of X-ray excitation and a sample. Its characterization capabilities are -17 due in large part to the fundamental principle that each ele- -18 ment has a unique atomic structure allowing unique set of peaks on its X-ray spectrum [21]. As seen from Figure 4, the -19 SEM/EDX analysis has shown the presence of Si, Al and O

atoms in the structure of halloysite nanotubes. Halloysite has (mV) potentialZeta -20 (Al2Si2O5(OH)4.2H2O formula unit. The weight percentages of Al, Si, O and H are 18.3, 19.1, 59.8 and 2.8, respec- -21 tively. The weight percentages from EDX analysis are 18, g/100mL 19 and 60 for Al, Si and O, respectively. In this case, it can -22 be said that the results have quite a good fit. FIGURE 5 - The effect of solid concentration on the zeta potential of 3.2 Electrokinetic properties halloysite nanotube 3.2.1 Effect of solid concentration The solid concentration has an important effect in the 3.2.2 Effect of pH measurements of zeta potential of solids [17]. As shown in The zeta potential curve for halloysite was determined Figure 5, the zeta potential change with the solid concen- in the pH ranges of 2 to 10. Figure 6 illustrates the effect tration in solution, indicating that solid concentration in so- of pH on the variation of zeta potential of halloysite nano- lution is a major parameter governing the surface charge tubes. Zeta potential of halloysite nanotube suspensions at generation. This means that the ionic species produced at pH 2.2–10 was negative and isoelectrical ponit (pHiep) of

Silica, being an acidic oxide, causes the reaction on the 10 right-hand of the equation to be predominant over a large

0 range of pH values, while alumina shows an amphoteric be- 024681012havior. Since the surface is mainly occupied by silica, the -10 surface charge will be negatively charged over a wide range

-20 of pH values. As a result, halloysite nanotubes tend to readily bind cationic compounds in aqueous solution [16]. -30

-40 70 (mV) potentialZeta 60 -50 50 -60 pH 40 FIGURE 6 - The effect of pH on the zeta potential of halloysite nano- 30 tube qe (mg/g) 20 halloysite nanotubes was at pH = 2.2 (Figure 6). The adsorbent outer surface has a net positive charge below 10 pHiep, while at pH > pHiep, the surface has a net negative 0 charge. Similar results were reported by Luo et al. [16], 0 5 10 15 Mellouk et al. [22] and Levis and Deasy [23]. At pH 2.2, Ce (mg/L) the number of negatively charged Al–OH sites equals the number of positively charged Si–OH sites and the surface FIGURE 7 - Adsorption isotherm of VB on halloysite nanotube. (Ad- is electrically neutral [6].The reason for the negative sur- sorbent dose: 2 g/L, shaking rate: 150 rpm, contact time: 1 h, pH: 3.5, 0 face charge with increasing pH is that in its curved struc- temperature: 30 C) ture, the silica is mainly positioned on the outer surfaces of 3.3 Adsorption properties the tubule, whereas the alumina is present mainly on the inner surface and edges of the tubules. Exposure of both 3.3.1 Adsorption isotherms oxides to water caused the formation of the surface hy- Figure 7 shows the adsorption isotherm of VB on the droxyl group. Since the surface is mainly silica, the surface halloysite nanotubes. Again, Figure 8 shows the photo of charge will be negative over a wide range of pH [24].The the dye solution before and after the adsorption of VB dye surface of halloysite nanotubes bears abundant Si–OH and on halloysite nanotube. An increase in initial dye concen- Al–OH, which can ionize in the following form: trations resulted in increased dye uptake. The adsorption capacity of the adsorbent at equilibrium increased from 9.9 Si⁄ Al OH Si⁄ Al OH Si⁄ Al O (2) to 56.1 mg/g with an increase in the initial VB concentra-

a) b)

FIGURE 8 - The photo of the dye solution before adsorption of VB dye on halloysite nanotube a), and after adsorption of VB dye on halloysite nanotube b). (Adsorbent dose: 2 g/L, shaking rate: 150 rpm, contact time: 1 h, temperature: 30 0C, VB concentrations: 20-100 ppm.)

tions from 20 to 140 mg/L. The initial dye concentration favorability for VB adsorption at higher ionic strength provides the necessary driving force to overcome the re- could be explained by two effects: (1) dimerization of dyes sistance to the mass transfer of VB between the aqueous in the presence of salts, which leads to the adsorption of phase and the solid phase. The increase in initial dye con- more molecules per adsorption site [27, 28] and (2) com- centration also enhances the interaction between VB and pression of the diffuse double layer on the adsorbent sur- halloysite nanotubes. Therefore, an increase in initial con- face with the increase of ionic strength, which facilitates centration of VB enhances the adsorption uptake of VB. the electrostatic attraction and contributes to the adsorption This is due to the increase in the driving force of the con- [6,29]. Similar results were found for Rhodamine 6G on centration gradient, as an increase in the initial dye concen- hallosiyte nanotube [6] and methylene blue and methyl vi- tration. A similar trend was reported for the adsorption of olet on sepiolite [30], respectively. dyes such as reactive blue 221 onto kaolinite [25] and max- ilon blue 5G on sepiolite [26]. Dyes can be removed from 50 adsorbet filters by burning at high temperatures. Second generation reusable adsorbed filters were made and their efficiency was demonstrated using VB. Filtrate is almost 45 completely transparent at 20 ppm concentration of VB dye. This result shows that halloysite almost completely preserved the dye removal efficiency after first reuse. Further- 40 more, halloysite filters have been effectively regenerated qe (mg/g) by burning the adsorbed dye. 35

3.3.2 Effect of pH The pH of the solution always affects the nature and 30 23456 surface charge of the adsorbent and is responsible for the Equilibrium pH interaction of the dye molecules with the adsorbent material may be due to protonated or deprotonated surface charges in solution at different pH values [26]. Generally FIGURE 9 - The effect of pH of the solution on the adsorption of victoria blue on halloysite nanotube (Adsorbent dose: 2 g/L, shaking in aqueous medium, the ionization of basic dyes acquires a rate: 150 rpm, contact time: 1 h, temperature: 30 0C, VB initial con- net positive charge and the adsorbents containing organic centration: 100 ppm.) functional groups on their surfaces also develop charges on their own surfaces. As can be seen in Figure 9, the amount 50 of VB adsorbed increased with increasing pH of the solution. The zeta-potential of halloysite nanotubes is negative at pH 2.2–10 due to the surface potential of SiO2 with a small contribution from the positive Al2O3 inner surface. 49 The surface of halloysite nanotube is positively charged below pH 2.2 while it acquires a negative charge above this pH. The zeta potential of halloysite nanotube is more negatively charged with increase in pH in the pH range of 2.2 qe (mg/g) 48 to 10.0. Halloysite nanotubes contain significant amount of hydroxyl groups as evidenced by FTIR-ATR spectrum. Therefore, increasing electrostatic attractions between negatively charged adsorption sites and positively charged VB 47 cations causes an increase in VB removal. Lower adsorp- -4 -3 -2 -1 + log [NaCl] tion at lower pH is due to the presence of excess of H ions competing with the dye cations for adsorption sites. FIGURE 10 - The effect of ionic strength of the solution on the adsorption of victoria blue on halloysite nanotube (Adsorbent dose: 2 3.3.3 Effect of ionic strength g/L, shaking rate: 150 rpm, contact time: 1 h, temperature: 30 0C, There are a number of studies showing the changing of pH: 3.5, VB initial concentration: 100 ppm.) the removal order of dye with the concentration of various electrolyte types in dye medium [25,26]. In order to inves- 3.3.4 Effect of temperature tigate the effect of ionic strength, halloysite nanotubes as Temperature also has important impact on the adsorp- adsorbent were added to VB solution, in which concentration process. As seen from Figure 11, the adsorbed amount tion of NaCl ranges from 0 to 0.1 mol/L. The results, shown of VB on halloysite nanotubes has increased with increas- in Figure 10, indicate that the inorganic electrolyte sup- ing temperature from 30 to 60 0C that corresponds to the presses distinctly the adsorption of dye. The chloride ion endothermic heat of adsorption. This increment is attributed concentration increases from 0 to 0.1 mol/L, the VB ad- to the increasing rate of diffusion of the adsorbate molecules sorption capacity of halloysite nanotubes increased. The across the external boundary layer and in the internal pores

of the adsorbent particles, owing to the decrease in the vis- where KF(mg/g) (L/g) is an indicator of the adsorption cosity of the solution, and change the equilibrium capacity capacity, 1/n is the adsorption intensity. KF and n can be of the adsorbent for a particular adsorbate. In addition, a determined from the linear plot of lnqe versus lnCe [34]. change of the temperature modifies the equilibrium capac- Langmuir and Freundlich models were applied to de- ity of the adsorbent for a particular adsorbate [31,32]. scribe the adsorption process. The isotherm constants of Langmuir and Freundlich models were given in Table 1. The 70 isotherm data were calculated from the least square method 2 60 and the related correlation coefficients (R values) were given in the same table. As seen from Table 1, the Lang- 50 muir equation represents the adsorption process very well; the R2 values were all higher than 0.99, indicating a very 40 30 C good mathematical fit. Figure 12 shows the Langmuir plots 30 40 C for the adsorption of VB onto halloysite nanotubes at 30, qe (mg/g) 50 C 40, 50 and 60 °C. The fact that the Langmuir isotherm fits 20 60 C the experimental data very well may be due to homogenous 10 distribution of active sites on hallosiyte nanotube surface; since the Langmuir equation assumes that the surface is ho- 0 mogenous, adsorption sites are energetically equivalent 0 5 10 15 20 25 30 and only monolayer adsorption occurs in the process [33]. Ce (mg/L) 0,3 FIGURE 11 - The effect of temperature of the solution on the adsorption of victoria blue on halloysite nanotube (Adsorbent dose: 2 g/L, 30 C shaking rate: 150 rpm, contact time: 1 h, pH: 3.5.) 0,25 40 C 3.4 Adsorption isotherms 0,2 The adsorption isotherms predict whether the adsorb- 50 C ate molecules combine with the adsorbent and also provide 0,15

information about the nature of combinations. Several Ce/qe (g/L) types of isotherms have been developed for the determina- 0,1 tion of adsorption equilibrium capacity of the adsorbent. In 0,05 this study both Langmuir and Freundlich models were used for the fitting experimental data. The linear form of Lang- 0 muir equation is given: 0 5 10 15 20 Ce (mg/L) Ce 1 Ce   (3) FIGURE 12 - Langmuir isotherm plots qe qmK qm The shape of the isotherm may also be considered with where qe (mg/g) is the equilibrium amount of VB ad- a view to predicting if an adsorption system is ‘favourable’ sorption by halloysite nanotubes, Ce(mg/L) is the equilib- or ‘unfavourable’. The essential characteristics of Langmuir rium VB concentration in the solution, qm (mg/g) is the maximum adsorption of dye and K (L/mg) is the Langmuir isotherm can be expressed by a dimensionless constant called equilibrium parameter, RL [35], which is defined by constant related to the enthalpy of the process. qm and K can be calculated from the slope and intercept of the plot 1 R  (5) of Ce/qe versus Ce, respectively [33]. L 1 KCe Freundlich isotherm model is based on the assumption The value of R indicates the type of the isotherm to of exponentially decaying adsorption site energy distribu- L be either unfavorable (R >1), linear (R = 1), favorable tion. It is used for heterogeneous surface energy systems. L L (0

lnqe  lnKF  1 nlnCe (4) was favorable (0

TABLE 1 - Isotherm constants for victoria blue adsorption onto halloysite nanotube Langmuir isotherm Freundlich isotherm 0 2 2 Temp.( C) pH [NaCl] qm K R RL n KF R (M) (mg/g) (L/mg) 30 5.5 --- 58.8 1.8 0.9980 0.030-0.92 3.2 30.6 0.9770 40 5.5 --- 62.5 1.7 0.9990 0.034-0.82 3.0 30.5 0.8520 50 5.5 --- 66.6 3.0 0.9990 0.022-0.89 3.6 35.4 0.8310 60 5.5 --- 66.6 1.9 0.9920 0.044-0.83 3.3 34.1 0.7990

3.5 Heat of adsorption The authors have declared no conflict of interest. The isosteric heat of adsorption, H0, from the adsorption data at various temperatures as a function of coverage fraction (=qe/qm) can be estimated from the following equation [35]: REFERENCES

0 [1] Liu R., Zhang B., Mei D., Zhang H., Liu J. (2011). Adsorption ∆H (lnC e )     (6) of methyl violet from aqueous solution by halloysite nano- R g  1/T  θ0.5 tubes. Desalination. 268, 111–116. [2] Kiani G., Dostali M., Rostami A., Khataee A.R. (2011). Ad- where Rgis the gas constant. Figure 13 shows the plots sorptionstudies on theremoval of Malachite Green from aque- 0 of –lnCe against 1/T. The value of H was calculated at a ous solutions onto halloysite nanotubes. Applied Clay Science. specific coverage fraction of 0.5 as 25 kJ/mol. The result 54, 34–39. shows that the interactions between surface and adsorbate [3] Hameed B.H., Ahmad A.A. (2009). Batch adsorption of meth- molecules are a physical interaction. Since adsorption is an ylene blue from aqueous solution by garlic peel, an agricultural endothermic process, it would be expected that an increase wastebiomass. J. Hazard. Mater. 164, 870–875. in solution temperature would result in an increase in ad- [4] Royer B., Cardoso N.F., Lima E.C., Macedo T.R., Airoldi C. sorption capacity [36]. Similar result was found for the ad- (2010). Sodicandacidic crystalline lamel armagadiite adsor- sorption of MV on perlite [37]. bents for the removal of methylene blue from aqueous solutions: Kinetic and equilibrium studies. Sep. Sci. Technol. 45, 129–141. 11 [5] Kumar M. And Tamilarasan R. (2013). Kinetics and Equilib- rium Studies on the Removal of Victoria Blue Using Prosopisjuliflora-Modified Carbon/Zn/Alginate Polymer Composite Beads. J. Chem. Eng. Data. 58, 517−527. 10,5 [6] Zhao Y., Abdullayev E., Vasiliev A., Lvov Y. (2013). Hal- loysite nanotubule clay for efficient water purification. Journal of Colloid and Interface Science. 406, 121–129. -ln Ce 10 [7] Karaoglu, M.H., Doğan, M., Alkan, M., Ugurlu, M. (2012). Photo oxidative degradation of cationic dyes using UV/H2O2 and UV/H2O2/TiO2 process. Fresenius Environmental Bulle- tin. 21(7), 1758-1763. 9,5 0,0029 0,003 0,0031 0,0032 0,0033 0,0034 [8] Kumar M., Tamilarasan R., Sivakumar V. (2013). Adsorption of Victoria blue by carbon/Ba/alginate beads: Kinetics, ther- 1/T (K-1) modynamics and isotherm studies. Carbohydrate Polymers. 98, 505– 513. FIGURE 13 - Plots of -lnCe versus 1/T for adsorption of VB dye on halloysite nanotube [9] Luo P., Zhang B., Zhao Y., Wang J., Zhang H., and Liu J. (2011). Removal of methylene blue from aqueous solutions by adsorption onto chemically activated halloysite nanotubes. Korean J. Chem. Eng..28(3), 800-807. 4. CONCLUSIONS [10] Liu M., Jia Z., Jia D., Zhou C. (2014). Recent advance in research on halloysite nanotubes-polymernanocomposite. Pro- From the experimental data obtained in the present gress in Polymer Science. 39, 1498–1525. study, it is concluded that the halloysite nanotubes may be [11] Duarte H.A., Lourenço M.P., Heine T. and Guimarães L. used as an effective nano-adsorbent for the removal of VB (2012). Clay Mineral Nanotubes: Stability, Structure and via adsorption from aqueous solution. Halloysite nano- Properties. Stoichiometry and Materials Science – When tubes were characterized by BET, XRD, FTIR-ATR and Numbers Matter. Editedby Dr. AlessioInnocenti, Publisher SEM/EDX. The morphology of halloysite was observed to InTech. have tubular. Over a wide range of pH, halloysite has a [12] Lvov Y., Aerov A., Fakhrullin R. (2014). Clay nanotube en- negative zeta potential. It has an isoelectrical point at pH capsulation for functional biocomposites. Advances in Colloid 2.2. At a pH above 2.2, halloysite can bind cationic dyes. and Interface Science. 207, 189–198. The batch mode study was conducted by varying pH, ionic [13] Zhai R., Zhang B., Wan Y., Li C., Wang J., Liu J. (2013). Chi- strength and temperature. The adsorption data were evalu- tosan–halloysite hybrid-nanotubes: Horseradish peroxidase ated by Freundlich and Langmuir isotherm equations and immobilization and applications in phenol removal. Chemical Engineering Journal. 214, 304–309. the best-fit adsorption isotherm was achieved with the Langmuir model, indicating that homogeneous adsorption [14] Peng Q., Liu M., Zheng J., Zhou C. (2015). Adsorption of dyes in aqueous solutions by chitosan–halloysite nanotubes occurred. The dimensionless separation factor (RL) was composite hydrogel beads. Microporous and Mesoporous Ma- found to lie between 0 and 1, revealing that the adsorption terials. 201, 190–201. process was favorable for the removal of VB dye. The iso- [15] Zhao M. and Liu P. (2008). Adsorption behavior of methylene 0 steric heat of adsorption (H ) shows a positive value, indi- blue on halloysite nanotubes. Microporous and Mesoporous cating that the adsorption process was endothermic in nature. Materials. 112, 419–424.

[16] Luo P., Zhao Y., Zhang B., Liu J., Yang Y., Liu J..(2010). [35] Doğan, M., Alkan, M. and Onganer, Y. (2000). Adsorption of Study on the adsorption of Neutral Redfrom aqueous solution methylene blue on perlite from aqueous solutions. Water, Air onto halloysite nanotubes. Water research. 44, 1489–1497. and Soil Pollution. 120, 229-248 [17] Doğan, M., Alkan, M. and Çakır, Ü. (1997). Electrokinetic [36] Alkan M., Dogan M. (2001). Adsorption of copper (II) onto properties of perlite. Journal of Colloid and Interface Science. perlite. J. Colloid Interface Sci. 243, 280–291. 192, 114-118 [37] Dogan M., Alkan M. (2003). Removal of methyl violet from [18] Alkan, M. and Doğan, M. (2003). Adsorption kinetics of Vic- aqueous solution by perlite. J. Colloid Interface Sci. 267, 32– toria blue onto perlite. Fresenius Environmental Bulletin. 41. 12(5), 418-425.

[19] Xing B., Yin X.B.. Novel Poly-Dopamine Adhesive for a Hal- 2+ loysite Nanotube-Ru(bpy)3 Electrochemilum inescent Sen- sor. 4(7), e6451. [20] Ye Y.P., Chen H.B., Wu J.S., Ye L. (2007). High impact strength epoxy nanocomposites with natural nanotubes. Poly- mer. 48, 6426–6433. [21] Vindhya C.R, HafeezKhan, N.S Kumar. (2014). Experimental investigation on halloysite nanotube sand clay an infilled composite steel tube. International Journal of Research in Engi- neering and Technology. 3(6), 195-204. [22] Mellouk S., Cherifi S., Sassi M., Marouf-Khelifa K., Bengueddach A., Schott J., Khelifa A. (2009). Intercalation of halloysite from Djebel Debagh (Algeria) and adsorption of copper ions. Appl. Clay Sci. 44, 230-236. [23] Levis, S.R., Deasy, P.B. (2002). Characterisation of halloysite for use as microtubular drug delivery system. International Journal of Pharmaceutics. 243 (1–2), 125–134. [24] Bordeepong S., Bhongsuwan D., Pungrassami T. and Bhong- suwan T. (2011). Characterization of halloysite from Thung Yai District, Nakhon Si Thammarat Province, in Southern- Thailand. Songklanakarin J. Sci. Technol. 33 (5), 599-607. [25] Karaoğlu M.H., Doğan M., Alkan M. (2010). Kinetic analysis of reactive blue 221 adsorption on kaolinite. Desalination. 256, 154–165.

[26] Alkan M., Doğan M., Turhan Y., Demirbas O., Turan P. (2008). Adsorption kinetic sand mechanism of maxilonblue 5 G dye on sepiolite from aqueous solutions. Chem. Eng. J. 139, 213–223. [27] Li Q., Yue Q., Sun H., Y. Su, B. Gao, (2010). A comparative study on the properties, mechanisms and process designs for the adsorption of non-ionic or anionic dyes onto cationic-polymer/bentonite. J. Environ. Manage. 91 1601-11. [28] Prasad A.L, Santhi T. (2012). Adsorption of hazardous cationic dyes from aqueous solution onto Acacia nilotica leaves as an eco friendly adsorbent. Sustain. Environ. Res. 22, 113–122.

[29] Tsai W., Chang C., Ing C.C., Chang, J. (2004). Adsorption of acid dyes from aqueous solution on activated bleaching earth. Colloid Interface Sci. 275, 72-78. Received: April 13, 2015 Revised: May 26, 2015 [30] Özdemir, Y., Doğan, M. and Alkan, M. (2006). ‘‘Adsorption Accepted: June 15, 2015 of cationic dyes from aqueous solutions by sepiolite’’. Mi- croporous and Mesoporous Materials. 96(1-3), 419-427.

[31] Doğan, M., Alkan, M., Türkyılmaz, A. and Özdemir, Y. CORRESPONDING AUTHOR (2004). Kineticsandmechanism of removal of methyleneblue- byadsorptıononto perlite. Journal of Hazardous Materials. B109, 141-148. Pinar Turan Beyli [32] Alkan, M., Demirbaş, Ö. and Doğan, M. (2004). Removal of Balikesir University acidyellow 49 from aqueous solution by adsorption. Fresenius Faculty of Science and Literture Environmental Bulletin. 13(11a), 1112-1121 Department of Chemistry [33] Langmuir, I. (1918). The adsorption of gases on plane surfaces Çağış, Balıkesir, 10145 of glass, mica and platinum. J. Am. Chem. Soc., 40 (9), TURKEY 1361−1403. [34] Freundlich H. (1932). Of the adsorption of gases. Section II. E-mail: [email protected] Kinetics and energetics of gas adsorption. Introductory paper to section II. Trans. Faraday Soc. 28, 195–201. FEB/ Vol 24/ No 11a/ 2015 – pages 3822 - 3830

EFFECTS OF POTASSIUM SUPPLY ON PHYSIOLOGICAL PROPERTIES IN PEANUT UNDER ENHANCED ULTRAVIOLET-B RADIATION

Yan Han1, 2, Yunsheng Lou1,*, Chenguang Han2, Meng Li1 and Chengda Hu1

1 Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters / Jiangsu Key Laboratory of Agricultural Meteorology, Nanjing University of Information Science and Technology, Nanjing 210044, China 2 College of Environment and Planning, Henan University, Kaifeng 475004, China

ABSTRACT 2% while the ozone in the stratosphere reduces 1%. UV-B has a direct effect on earth’s life in the view of biology [2, A field experiment was conducted to investigate the 3]. The influence of enhanced UV-B radiation on crops influence of potassium fertilization on net photosynthetic has received more attentions in recent years. Some studies rate (Pn), transpiration rate (Tr), stomatal conductance (Gs) showed that enhanced UV-B radiation led to dwarfing of and water use efficiency (WUE) of peanut at different plant height, decreasing leaf area, stunting growth, reduc- growth stages under enhanced ultraviolet-B (UV-B) radia- ing leaf stomatal conductance, photosynthetic rate and tion. The experiment was designed with three levels of UV- transpiration rate [4-6]. Enhanced UV-B radiation also de- B radiation, i.e. ambient UV-B radiation (CK, natural light, pressed crop root, stem, vegetative and reproductive or- 1.5kJ/m2), enhanced UV-B by 20% (E1, 1.8kJ/m2), en- gans, thus reduced the biomass and yield [7-11]. hanced UV-B by 40% (E2, 2.1kJ/m2) and two levels of po- Plant growth is influenced by many factors. Recently, tassium fertilization, i.e. low (K1, 75 kg K O/hm2) and 2 more attentions have been paid to the effects of UV-B and high (K2, 200kg K O/hm2). The results showed that, com- 2 other abiotic factors on plant growth, development and re- pared with control, enhanced UV-B radiation decreased the production, such as elevated CO , global warming, net photosynthetic rate, transpiration rate, stomatal con- 2 drought, mineral nutrition deficiency and air pollution [12- ductance, and leaf water use efficiency, and with increas- 14]. Potassium is an essential nutrient for crop. Potassium ing intensity of UV-B radiation, the above decreasing ten- application can increase plant resistance against drought, dency was enhanced. Potassium supply could increase the cold, disease, salt and lodge [15, 16]. Peanut is the fourth net photosynthetic rate, transpiration rate, stomatal con- major oil crops. So far few studies have been available re- ductance and leaf water use efficiency, and with increasing garding whether potassium application can mitigate the de- potassium supply the above parameters were also in- pressive effect of enhanced UV-B on peanut physiological creased. This indicated that potassium supply could miti- properties. The purposes were to investigate the influences gate the depressive effects of enhanced UV-B on photosyn- of potassium fertilization on photosynthetic and transpira- thesis and transpiration in peanut to some degree. This tion parameters in peanut through a field experiment under finding is helpful in further investigating the effect of po- enhanced UV-B radiation. tassium fertilization on yield and quality in peanut under enhanced UV-B radiation.

2. MATERIAL AND METHODS

KEYWORDS: Enhanced UV-B; Potassium Supply; Peanut; Photo- 2.1 Materials synthesis; Transpiration The tested peanut variety is Kainong 64, from Kaifeng Agriculture and Forestry Science Research Institute, He- nan Province. The tested potassium fertilizer is potassium 1. INTRODUCTION sulphate.

The ozone in the stratosphere has been reduced by the 2.2 Experimental design release of artificial CFCs and nitrogen oxide gases [1]. Ul- A field experiment was conducted from June 6, 2013 traviolet radiation reaching the earth's surface increases to October 13, 2013 at the Experimental Station of Agro-

meteorology, Zhengzhou (34°43′N, 113°39′E). Peanut was * Corresponding author sowed at the rate of 180,000 holes hm-2 in the filed after

cultivation and fertilization. The experiment was designed increasing potassium supply could obviously alleviate the with 3 levels of UV-B radiation intensity, and 2 levels of inhibitory effect of enhanced UV-B on net photosynthetic potassium fertilization. The tested field was divided into rate in the leaf of peanut. three zones, i.e., one natural light zone (A, ambient at Compared with control, enhanced UV-B by 20% sig- 1.5kJ·m-2·h-1), and two UV-B treated zones (E1 and E2, at nificantly reduced Pn by 9.09% (low potassium), and 1.8 kJ•m-2•h- and 2.1 kJ•m-2•h-1, elevated by 20% and 40%, 9.77% (high potassium), respectively. The difference was respectively). The UV-B lamps were fixed on a liftable significant at p < 0.05 at flowering stage and fruiting stage, steel frame over the canopy to simulate the enhanced UV- but not at full fruit stage. B radiation. The UV-B radiation intensity was monitored with a UV meter (BNU297, Beijing). The potassium ferti- Compared with the control, enhanced UV-B 40% (E2) lization was set at 2 levels, i.e. low potassium (K1, 75 kg sharply decreased Pn by 13.65% (low potassium) and -2 -2 K2O hm ), and high potassium (K2, 200 kg K2O hm ). The 13.99% (high potassium), in contrast, E1 reduced Pn by experiment was designed with a complete random design 4.56% (low potassium), 4.22% (high potassium). The dif- method. The treatments were listed as follows, (1) ambient ference in Pn among the treatments was significant at the UV-B and low potassium (A+K1); (2) ambient UV-B and three growth stages (p<0.05). Net photosynthetic rate un- high potassium (A+K2); (3) enhanced UV-B by 20% and der high potassium was generally higher than low potas- low potassium (E1+K1); (4) enhanced UV-B by 20% and sium under the same UV-B intensity, indicating that in- high potassium (E1+K2); (5) enhanced UV-B by 40% and creasing potassium supply improved Pn in peanut leaf, and low potassium (E2+K1); and (6) enhanced UV-B by 40% alleviated the depressive effect of UV-B radiation. and high potassium (E2+K2). Compound fertilizer and potassium fertilizer were basally applied into the field before TABLE 1 - The influences of potassium fertilization on Pn in peanut peanut sowing. Each treatment was replicated 3 times. There at different growth stages under enhanced UV-B radiation (μmol CO2 m-2 s-1) were totally 18 plots. Each plot area was 3m×3m =9 m2. The distance between the lamps and the canopy of the Treatment Time peanut was frequently adjusted to keep the distance at 0.8m UV-B K Flowering Fruiting Full fruit mean level supply stage stage stage during the experimental duration. The UV-B radiation was CK K1 24.16b 23.53b 15.37ab 21.02 daily supplied for 8h from 8:00 to 16:00 except cloudy and K2 25.64a 25.47a 16.42a 22.51 rainy days. Field routine management such as irrigation E1 K1 21.93c 21.35c 14.04b 19.11 and disease control was conducted according to actual re- K2 23.02bc 22.97b 14.93a 20.31 E2 K1 20.89d 20.89d 12.76b 18.15 quirement. K2 22.29c 21.94c 13.85b 19.36 Note: Values with different letter in the same column were significantly 2.3 Testing Items and methods different at p<0.05 level. The same below. The physiological parameters were measured with a portable photosynthesis system (LI-6400, USA) at flower- 3.2 Transpiration rate (Tr) ing stage (July 17, 2013), fruiting stage (August 13, 2013), Table 2 showed that the highest of transpiration rate full fruit stage (September 26, 2013). The main parameters (Tr) occurred at flowering stage, and lowest at full fruit included net photosynthetic rate (Pn), transpiration rate stage. In general, the highest value of Tr occurred in the (Tr), stomatal conductance (Gs), leaf water use efficiency treatment with ambient UV-B and high potassium (A+K2), (WUE) and intercellular CO2 concentration. The third leaf in contrast, the lowest in the treatment with enhanced UV- below the stem was selected to measure the parameters B 40% and low potassium (E2+K1), indicating that in- from 9:00 to 11:00 on a sunny day. creasing potassium supply might alleviate the depressive effect of UV-B radiation on transpiration rate. Compared 2.4 Statistic analysis with control, enhanced UV-B 20% (E1) reduced Tr by Statistic analysis of the data was performed with SPSS 7.58% (low potassium) and 7.47% (high potassium), but software. The differences among different treatments and the differences between treatments and growth stages were growth stages were subjected to multi-way ANOVA. not significant at p＞ 0.05.

TABLE 2 - The influences of potassium fertilization on transpiration 3. RESULTS rate in peanut at different growth stages under enhanced UV-B radi- -2 -1 ation (mmol H2O m s ) 3.1 Net photosynthetic rate (Pn) Treatment Time Pn peaked at flowering stage and bottomed at full fruit UV-B K Flowering Fruiting Full fruit mean stage (Table 1). At the same growth stage, obvious differ- level supply stage stage stage ences among different treatments were observed in Pn. CK K1 13.10ab 10.28ab 8.69ab 10.69 Generally, the highest Pn occurred in the treatment with K2 13.31a 10.79a 9.22a 11.11 E1 K1 12.04b 9.43b 8.16b 9.88 ambient UV-B and high potassium treatment (A+K2), and K2 12.38ab 9.89b 8.58ab 10.28 the lowest in the treatments with enhanced UV-B by 40% E2 K1 11.62b 9.32b 7.63b 9.52 and low potassium treatment (E2+K1). This indicated that K2 12.38ab 9.73b 7.84b 9.98

In comparison with the control, UV-B enhanced 40% ducive to photosynthesis and the accumulation of pho- (E2) decreased Tr by 10.94% (low potassium), 10.17% toassimilate. (high potassium), In contrast, E1 reduced Tr by 3.36% (low potassium) and 2.70% (high potassium). The differences in Intercellular CO2 concentration significantly increased transpiration rate between low and high potassium treat- by 3.22% (low potassium), 4.77% (high potassium) when ment was not significant at the same UV-B level, implying UV-B enhanced 20% (E1) compared with control. The dif- that increasing potassium supply could not mitigate the in- ference was not significant (p＞ 0.05) at different stages. hibitory effect of enhanced UV-B on transpiration rate in Compared with the control, enhanced UV-B 40% (E2) in- peanut leaf. creased intercellular CO2 concentration by 4.94% (low potassium) and 6.96% (high potassium). The difference was 3.3 Stomatal conductance (Gs) significant (p < 0.05) at different stages. Intercellular CO2 The stomata functioned as modulating water vapor and concentration under high potassium was generally lower CO2 exchange and controlling photosynthesis and transpi- than low potassium under the same UV-B intensity, indi- ration. Stomatal conductance peaked at fruiting stage and cating the decrease of the intercellular CO2 concentration bottomed at full fruit stage (Table 3). Generally, the highest might be associated with the improvement of the potassium Gs occurred in the treatment with ambient UV-B and high nutrition. potassium (A+K2), and the lowest in the treatment with enhanced UV-B 40% and low potassium (E2+K1), indicating TABLE 4 - The influences of potassium fertilization on intercellular CO2 concentration at different growth stages in peanut under en- -1 that elevated UV-B inhibited stomatal conductance, but in- hanced UV-B radiation (μmol CO2 mol ) creasing potassium supply could alleviate the depressive effect of elevated UV-B on Gs. Treatment Time UV-B K Flowering Fruiting Full mean Compared with control, enhanced UV-B 20% (E1) sig- level supply stage stage fruit stage nificantly reduced Gs by 16.28% (low potassium) and CK K1 229.30b 259.04ab 325.73b 271.36 K2 217.52c 237.51c 308.87c 254.63 12.00% (high potassium). The differences in Gs among E1 K1 237.24ab 266.76ab 336.30a 280.10 treatments and growth stages were significant at p<0.05. K2 227.05b 257.94b 315.36c 266.78 Stomatal conductance sharply decreased by 20.93% (low E2 K1 240.67a 270.32a 343.28a 284.76 potassium), 28.00% (high potassium) under enhanced UV- K2 231.82b 262.84ab 322.38b 272.35 B 40% (E2), compared to the control; in contrast, it reduced by 4.65% (low potassium) and 16.00% (high potassium) 3.5 Water use efficiency (WUE) under E1. Under the same UV-B, Stomatal conductance in Leaf water use efficiency referred to the amount of as- the treatment with high potassium was generally higher similated CO2 per unit of consumed water, usually ex- than that with low potassium. According to stomatal dy- pressed as the ratio of net photosynthetic rate to transpira- namics, the increased stomatal conductance might be associated with the improvement of potassium supply [17]. tion rate. The formula is WUE= Pn / Tr , WUE for water Pn Tr TABLE 3 - The influences of potassium fertilization on stomatal con- use efficiency, , for the average of the Pn and Tr, ductance in peanut at different growth stages under enhanced UV-B respectively. radiation (mol m-2 s-1) TABLE 5 - The influences of potassium fertilization on leaf water use Treatment Time efficiency in peanut at different growth stages under enhanced UV-B -1 UV-B K Flowering Fruiting Full fruit mean radiation (μmol CO2 · mmol H2O) level supply stage stage stage CK K1 0.41b 0.49b 0.40a 0.43 Treatment Time K2 0.51a 0.57a 0.43a 0.50 UV-B K Flowering Fruiting Full mean E1 K1 0.36b 0.41c 0.32bc 0.36 level supply stage stage fruit stage K2 0.48a 0.50b 0.34b 0.44 CK K1 1.84a 2.29a 1.77a 1.97 E2 K1 0.35c 0.42c 0.24d 0.34 K2 1.93a 2.36a 1.78a 2.03 K2 0.38b 0.43c 0.27c 0.36 E1 K1 1.82a 2.26a 1.72a 1.93 K2 1.86a 2.32a 1.74a 1.97 E2 K1 1.79a 2.24a 1.67a 1.91 3.4 Intercellular CO2 concentration (Ci) K2 1.80a 2.25a 1.77a 1.94 Intercellular CO2 concentration gradually ascended during the whole growth stages (Table 4). Generally, the The highest water use efficiency (WUE) in peanut leaf Ci peaked in the treatment with enhanced UV-B 40% and was found at fruiting stage and the lowest at full fruit stage low potassium (E2+K1) and was lowest in the treatment (Table 5). In general, the leaf water use efficiency maxim- with ambient UV-B and high potassium (A+K2), indicat- ized in the treatment with ambient UV-B and high potas- ing that enhanced UV-B increased the intercellular CO2 sium (A+K2), and minimized in the treatment with en- concentration, but increasing potassium supply decreased hanced UV-B 40% and low potassium (E2+K1), but the it. The result showed that potassium significantly improved differences were not significant at p<0.05) at different the intercellular CO2 assimilation ability, which was con- stages, indicating that enhanced UV-B inhibited water use

efficiency, and potassium application alleviated the depres- vided strong transpiration pull for nutrient delivery. Potas- sive effect of enhanced UV-B on WUE. sium was needed in the formation of high-energy compounds (ATP) for photosynthesis and respiration. Compared with control, UV-B enhanced 20% (E1) significantly decreased water use efficiency by 2.03% (low According to previous research, plants mainly ab- potassium) and 2.96% (high potassium). However, the dif- sorbed K+ in the soil solution and soil colloid as nutrition. ference was not significant (p<0.05) at different stages. En- Potassium existed in plant body by ions or soluble salt, par- hanced UV-B 40% (E2) obviously decreased water use ef- ticipated in the photosynthesis, metabolism of sugar, pro- ficiency by 3.05% (low potassium) and 4.43% (high potas- tein and fat synthesis and transportation process, and ad- sium).. Under the same UV-B intensity, water use effi- justed the osmotic pressure and stomata opening and clos- ciency in high potassium treatment was generally higher ing, improved plant drought resistance, cold resistance, than that in low potassium treatment, indicating that the in- disease resistance, salt resistance, lodging resistance [24]. crease of the leaf water use efficiency was associated with Enhanced UV-B radiation significantly increased leaf in- the improvement of the potassium nutrition, and potassium tercellular CO2 concentration in peanut. Potassium appli- application alleviated the depressive effect of enhanced cation could decrease intercellular CO2 concentration, UV-B radiation on leaf water use efficiency. which indicated that potassium might improve the intercellular CO2 assimilation ability, conducive to photosynthesis and the accumulation of photo assimilate. The photosyn- 4. DISCUSSION AND CONCLUSION thetic rate decrease was mainly caused by low stomatal conductance when the intercellular CO2 concentration and Enhanced UV-B radiation obviously decreased net stomatal limitation increased. If photosynthetic rate de- photosynthetic rate, stomatal conductance, transpiration creased, but the intercellular CO2 concentration increased, rate and water use efficiency at different peanut stages (Ta- the limit of photosynthesis might be non-stomatal factors, bles 1-3, Table5). The reasons might be related to that en- namely the decline in crop photosynthetic activity of mes- hanced UV-B hurt the chloroplasts and obstructed the syn- ophyll cells. Furthermore, UV-B enhancement influenced thesis of chlorophyll, chlorophyll and carotenoid were in- the photosynthetic reaction process directly. duced to the non-enzymatic photooxidation, lowered hill reaction activity, increased stomata resistance or closed Potassium regulating effect on related photosynthesis stomata, thus inhibited the photosynthesis of plants [18]. parameters was remarkable with increasing potassium lev- UV-B enhancement also impaired photosynthesis function, els. The stomatal conductance of function leaf increased characterized by optical system II reaction center inactiva- with increasing potassium supply, implying that providing tion, function inhibition of electron transport chain, low potent guarantee for CO2 exchange, and abundant supply saturation value of electronic light transmission rate, defor photosynthesis [25]. In this study, net photosynthetic crease in light energy utilization rate, decline of unsatu- rate, transpiration rate, stomatal conductance and water use rated fatty acid (linolenic acid), decline of membrane flu- efficiency were higher in the treatment with high potassium idity caused by the increase of saturated fatty acids (pal- than low potassium, indicating that potassium application mitic acid) and stearic acid etc [19]. Under enhanced UV- levels alleviated the effect of UV-B enhancement on net B radiation, the net photosynthetic rate in peanut leaf de- photosynthetic rate, transpiration rate, stomatal conduct- creased through the way of lowering transpiration rate and ance and intercellular CO2 concentration and water use ef- stomatal conductance to improve water use efficiency, to ficiency. Although potassium was not a part of the plant resist against the damage of enhanced UV-B [20]. Thus, organic compounds, it was indispensable to sustain plant enhanced UV-B radiation had a significant effect on pho- life. Enough potassium accelerated the transport rate in tosynthetic and transpiration. The net photosynthetic rate, vessels and sieve tubes to promote a variety of metabolic transpiration rate and water use efficiency decreased with processes and also played a unique physiological role in increasing UV-B levels, leading to decrease of the peanut plant growth and development. yield, and increase of water demand [21, 22].

Potassium application improved the net photosynthetic rate, transpiration rate, stomatal conductance and water use ACKNOWLEDGEMENT efficiency, to a certain extent, and alleviated the inhibitive effect of enhanced UV-B on net photosynthetic rate and This research was conducted under the financial sup- transpiration rate. In this experiment, the regulating effect ports from the National Natural Science Foundation of of potassium supply on related photosynthetic parameters China (51309093; 41375159), the Science and Technolog- was remarkable, especially stomatal conductance, because ical Fund of Henan Provincial Education Department K+ directly influenced stomata opening and effectively (14A170001), the Natural Science Foundation of Jiangsu regulated stomata movement [23]. Potassium could also Province (BK20131430) and the 333 High-level Talent regulate CO2 permeating rate and water transpiration rate Training Program of Jiangsu Province. by influencing stomata opening and closing. In our experiment, potassium increased the transpiration rate and pro- The authors have declared no conflict of interest.

REFERENCES [17] Gao F., Zhai Z.X. and Wang M.L.(2011) Effects of potassium application rate on photosynthetic characteristics and yield of summer-planted peanut. Journal of Peanut Scienc 40( 1):13- [1] Ji M.J, Feng H.Y., An L.Z. and Wang X.L. (2002) Present sta- 17. tus and prospects in research on effect of enhanced UV-B radiation on plants. Chinese Journal of Applied Ecology 13(3): [18] Zheng Y.F., Jian W.M. and Li X.F. (1998) Effect of enhanced 359-364. solar ultraviolet radiation on soybean. Acta Scientiae Circum- stantiae 18(5):549-552. [2] Kerr J.B. and Mcelroy C.T. (1993) Evidence for large upward trends of ultraviolet-B radiation linked to ozone depletion. Sci- [19] Zhou Q. and Huang X.H.(2001) The Survival stress-ecophys- ence 262 (5136): 1032-1034. iological effect of enhanced ultraviolet-B radiation on plant. Chinese Journal of Nature 23(4):199-203. [3] Wang S.B., Su W.H. and Wei D.W. (1993) Boiologically effective radiation of solar ultraviolet radiation and the depletion [20] Yan J.Y., Yang Z.M. and Zheng Z.M. (1995) Study on the bi- of ozone layer. Acta Scientiae Circumstantiae 13(1): 114-119. ological effect of soybean to ultraviolet radiation. Agro-environmental Protection 14(4): 154-157. [4] Wu X.C., Lin W.X. and Huang Z.L. (2007) Influence of enhanced ultraviolet-B radiation on photosynthetic physiologies [21] Lou Y.S., Wu J. and Yu J.Q. (2012) Effects of nitrogen ferti- and ultrastructure of leaves in two different resistivity rice cul- lization on physiological characteristics of barley leaves at tivars. Acta Ecologica Sinica 27(2):554-564. booting stage under enhanced UV-B radiation. Chinese Jour- nal of Agrometeorology 33(2):202-206. [5] Fedlna I.S., Grlgorova I.D. and Georgleva K.M. (2003) Re- sponse of barley seedlings to UV-B radiation as affected by [22] Lou Y.S., Han Y. and Liu Z.Y. (2013) Effects of silicon ferti- NaCl. Plant Physiol 160(2): 205-208. lization on diurnal variations of photosynthesis and transpiration at barley heading stage under elevated UV-B radiation. [6] Xu K. and Qiu B.S. (2007) Response of superhigh-yield hy- Chinese Journal of Agrometeorology 6: 668-672. brid rice Liangyoupeijiu to enhancement of ultraviolet-B radi- [23] Zhao T., Zheng X.L., and Xu G.Z. (2011) Effects of fertiliza- ation. Plant Science 172(1): 139-149. tion on peanut physiology and quality. Fujian. Journal of Ag- [7] Liu B., Wang C., and Jin J. (2009) Responses of soybean and ricultural Sciences 26(3): 490-497. other plants to enhanced UV-B radiation. Soybean Science [24] Wang Z. (2000) Plant physiology. Beijing: China Agriculture 28(6): 1097-1100. Press. [8] Guo W., Yin H., and Mao X.Y. (2008) Effect of enhancing [25] Zhang F.S. and Liu J.Q. (1993) The recommendation of main UV-B radiation combined with nitrogen addition on the phys- crop potassium supplied technology. Beijing: Beijing Agricul- iological response of rice. Hubei Agricultural Sciences 47(5): tural University Press. 521-524.

[9] Luo N.S. and Zhong Z.C. (2006) Effects of enhanced ultraviolet-B radiation on photosynthesis of maize. Chinese Journal of Ecology 25(4): 369-373.

[10] Chen J.J., Zu Y.Q., Chen H.Y. and Li Y. (2004) Influence of enhanced UV-B radiation on growth and biomass allocation of twenty soybean cultivars. Journal of Agro-Environment Sci- ence 23(1): 29-33. [11] Zi X.N., Chen Z.Y. and Guo S.C. (2006) Influence of enhanced UV-B radiation on crop morphology and physiological function. Chinese Journal of Agrometeorology 27(2): 102- 106. [12] Van de Leun J.C., Tang X.Y. and Tevini M. (1995) Environ- mental effects of ozone depletion: 1994 assessment. (UNEP Received: April 13, 2015 1994). Ambio 24:102-224. Accepted: June 15, 2015

[13] Caldwell, M.M., Björn, L.O., Bornman, J.F., Flint, S.D., Ku- landaivelu, G., Teramura, A.H. and Teramura M.T. (1998) Ef- fects of increased solar ultraviolet radiation on terrestrial eco- CORRESPONDING AUTHOR systems. Journal of Photochemistry & Photobiology B Biol- ogy 46, 98, 40–52. Yunsheng Lou [14] Kang Y.J., Wang Y.F., Zhao C.X., Wang M.L., Xu L. and College of Applied Meteorology Cheng X. (2010) Effects of different potassium application on Nanjing University of Information Science and Tech- senescence and yield in peanut. Chinese Agricultural Science nology Bulletin 26(4): 117-122. No 219 Ningliu Road, Pukou District [15] He C.M., Wang F., Li Q.H., Lin C., Li Y. and Lin X.J. (2009) Nanjing 210044 Effect of the different ratios of potassium, magnesium and sul- P.R. CHINA fur elements applications on the peanut's nutrient absorption, accumulation and assignment. Chinese Journal of Soil Science 40(6):1385-1389. Phone: +86-25-58699904 Fax: +86-25-58731193 [16] Zhou L.Y., Li X.D. and Wang L.L. (2006) Effects of different application rates of N, P, K, Ca fertilizer on photosynthesis E-mail: [email protected] properties, yield and kernel quality of peanut. Journal of Pea- nut Science 35(2):11-16. FEB/ Vol 24/ No 11a/ 2015 – pages 3831 - 3835

CHEMICAL, MICROBIAL AND PHYSICAL EVALUATION OF COMMERCIAL BOTTLED DRINKING WATER AVAILABLE IN BUSHEHR CITY, IRAN

Elham Shabankareh Fard1, Sina Dobaradaran1,2,3,* and Roughayeh Hayati1

1Department of Environmental Health Engineering, Faculty of Health, Bushehr University of Medical Sciences, Bushehr, Iran 2The Persian Gulf Marine Biotechnology Research Center, Bushehr University of Medical Sciences, Bushehr, Iran 3Systems Environmental Health, Oil, Gas and Energy Research Center, Bushehr University of Medical Sciences, Bushehr, Iran

ABSTRACT 1. INTRODUCTION

Due to many reasons people had been interested in Availability of healthy drinking water sources is a consuming bottled waters. Therefore, this study was done main concern in many countries around the world [1]. to give a clear view of physical, chemical and microbial About 80% of worldwide mortalities among children arise quality of bottled water available in Bushehr. Ten commer- due to gastrointestinal diseases such as diarrhea following cial brands of bottled water were analyzed to determine drinking of contaminated water [2]. Accordingly, aware- physical, chemical and microbial parameters. Multiple ness about drinking water quality has been continually in- tube fermentation method was used to determine fecal and creased during recent years [3]. total coliform bacteria and spread plate method was used to measure heterotrophic bacteria. Results were compared Physical, chemical and microbial properties of drink- with Iranian National Regulation (INR), WHO and EPA ing water may affect its safety and consumer's satisfaction guidelines. The mean values of physical, chemical and mi- [4]. Hence, drinking water guidelines have been developed crobial parameters of bottled drinking water were electrical by national and international organizations to define limits conductivity (452.38 µS/cm), turbidity (0 NTU), pH (7.7), for physical, chemical and microbial contents to ensure the alkalinity (172.4), total hardness (170.18), calcium hard- required quality [5]. ness (113.92), magnesium hardness (56.26) mg/L CaCO3, calcium (45.56), magnesium (14.02), residual chlorine (0), Among different sources of drinking water, bottled chloride (16.08), TDS (226.19), iron (0.04), fluoride drinking water has been increasingly paid attention by the (0.18), phosphate (0.04), nitrate (3.68), nitrite (0.14) and consumers worldwide. The main reasons for this rapid ten- sulphate (39.88) mg/L ,total coliform (0), fecal coliform (0) dency are the lack of sufficiently safe drinking water, un- MPN/100ml and HPC (260.8 CFU/mL). Except HPC con- favorable taste of tap water and the availability and porta- centrations in one bottled drinking water brand all men- bility of bottled drinking water [5-9]. Accordingly, people tioned results met the INR, WHO and EPA guidelines. in Iran have shown considerable appeal to consumption of Physicochemical parameters in some brands had a statisti- bottled drinking water in the last decade [7]. Although con- cally significant difference with labeled values. Physical, sumers mainly consider bottled drinking water as pure and chemical and microbial quality of bottled water in Bushehr safe [10], there are growing concerns regarding the chemi- were not problematical from health point of view. As the cal and microbiological characteristics of such waters. A residual chlorine level of bottled water was zero, there is a number of studies have shown that microbial quality of risk of secondary contamination of bottled water if stored bottled drinking water were at levels much higher than the in unsuitable conditions. national and international standards set for drinking water [11-13]. In addition, levels of some chemical parameters (e.g. pH and calcium) were reported to be higher than guidelines for drinking waters [7].

KEYWORDS: Bottled drinking water; physical, chemical and microbial quality; Bushehr Therefore, the aims of this study were to investigate the physical, chemical and microbial properties of different commercial brands of bottled drinking water available in Bushehr city, Iran and to compare their properties with the Ira- nian National Regulation (INR), US Environmental Protec- tion Agency (EPA) and World Health Organization (WHO) * Corresponding author guidelines as well as with the values on the labels.

TABLE 1 - Regulation and guidelines for drinking water Drinking water quality guidelines Parameter Unit EPAa WHOb INRc Ec µS/cm * * 1500 pH - 6.5-8.5 * 6.5-9 Turbidity NTU 5 * 5 Residual-Chlorine mg/L * * * TDS mg/L 500 500 1500 Ca2+ mg/L * * * Mg2+ mg/L 30# * * Cl- mg/L 250 * 400 Fe mg/L 0.3 * 0.3 F- mg/L 4 1.5 1.5 - No3 mg/L 45 50 50 - No2 mg/L 0.3 3 3 2- So4 mg/L 250 * * 2- Po4 mg/L * * * Alkalinity mg/L as CaCO3 * * * Bicarbonate mg/L as CaCO3 * * * Total hardness mg/L as CaCO3 * * 500 Ca Hardness mg/L as CaCO3 * * 250 Mg Hardness mg/L as CaCO3 * * 50 Total Coliform MPN/100mL 0 0 0 Fecal Coliform MPN/100mL 0 0 0 HPC CFU/ml 500 500 500 * Not determined # desirable admissible concentration aUS Environmental Protection Agency guideline for drinking water (maximum admissible concentration) bWorld Health Organization guideline for drinking water (maximum admissible concentration) cIranian National Regulation guideline for drinking water (maximum admissible concentration)

2. MATERIALS AND METHODS In order to determine heterotrophic bacteria, 0.1 ml water samples were spread onto R2A agar plates and incu- In this study, 10 commercial brands of 1.5 liter bottled bated at 35°C for 48 hours and colonies were counted by drinking water were analyzed for their physical, chemical using a Scan 100 Interscience colony counter and results and microbial properties. Five bottles of each brand, some reported as CFU/mL. packaged in different locations, were purchased during March 2012 to February 2013 from supermarkets, grocery Statistical analyses were performed using Microsoft stores, and health shops in the Bushehr, Iran. Bottled drink- Excel 2010 and Statistical Package for Social Science ing water samples were immediately transferred to the la- (SPSS) software (Version 16). The results were expressed boratory and were stored in a dark place at room tempera- as mean ± SD and mean value of each parameter was com- ture in their original sealed plastic containers until the anal- pared to national and international guidelines. Kolmogo- yses were made. Residual chlorine level and pH were rov-Smirnov statistical test was done and result showed measured using DPD colorimetric kit test and pH meter re- that all data had a normal distribution (P-value > 0.05). spectively. Hardness and chloride were measured by stand- Therefore, parametric one sample t-test was performed to ard method [14]. Spectrophotometric method was used for compare mean of each measured physicochemical param- - 2+ - analyses of F (570 nm), Fe (510 nm), NO3 (400 nm), eter for every brand with its labeled value (P-value < 0.05). - 2- 2- NO2 (507nm), PO4 (890nm), SO4 (450nm) using a DR/2000 spectrophotometer (HACH Company, USA). Electric conductivity (EC) and turbidity were measured by 3. RESULTS AND DISCUSSION using Greisinger-GLM020 electrical conductivity meter and LUTRON-2016 turbidity meter (nephlometric The EPA, WHO and INR guideline values for physico- method), respectively. chemical and bacteriological parameters of drinking water are shown in table1. The measured physicochemical param- For total coliform (TC) and fecal coliform (FC) 100ml eters of each brand were compared to labeled values of the samples (10 tubes/10ml) were analyzed according to the samples (Tables 2 and 3). The level of turbidity and residual multiple-tube fermentation method [15]. For total coliform chlorine were 0 mg/L in all bottled drinking water samples. determination, Lactose Broth (incubated at 35°C for 48 hours) and Brilliant Green Bile Lactose (incubated at 35°C The results of FC, TC and HPC measurements of all for 48 hours) and for fecal (FC), Brilliant Green Bile Lac- bottled drinking water samples are presented in Table 4. tose (incubated at 44.5°C for 24 hours) were used and the The results of HPC measurements compared with the EPA, results reported as Most Probable Number (MPN) per WHO and INR guidelines in Figure 1. Table 5 shows the 100mL. distribution of HPC in bottled drinking water brands.

The majority of physicochemical and microbiological lack of guideline values for some parameters such as cal- parameters of bottled drinking water samples were in ac- cium etc. in drinking water [16]. The total hardness of wa- cordance with the EPA, WHO and INR guidelines. How- ter may range from trace amounts to hundreds of milli- ever, there were deviations from the labeled values for a grams per liter [17]. EPA and WHO have not set guideline number of physicochemical parameters in some brands of value for total hardness, but the INR has set a guideline value bottled drinking water samples. of 500 mg/L for total hardness in drinking water [18]. In some instances, consumers tolerate water hardness higher Bottled mineralized water can offer a significant than 500 mg/L [16]. The minimum and maximum levels of amount of required nutrients and minerals [10]. There are TH were 58 (sample 22, brand 5) and 316 (sample 31, brand insufficient data to suggest either minimum or maximum 7) mg/L. Calcium and magnesium salts make water hard. concentrations of minerals currently, as adequate intake However, there are evidences that have suggested magnesium will depend on a range of other factors. This is reason for and calcium hardness may play a role in protection against

TABLE 2 - Measured physicochemical parameters of bottled drinking water in conjunction with label values

2+ 2+ 2– ─ − − − Ca (mg/L) Mg (mg/L) SO4 (mg/L) Cl (mg/L) F (mg/L) Fe (mg/L) NO3 (mg/L) NO2 (mg/L) Brand Source Ma Lb M L M L M L M L M L M L M L 1 spring 44.8±3.7 NL 15.96±1.45 NL 30.57±29.64 NL 1.69±1.25 NL 0.19±0.1 NL 0.033±0.006 NL 3.08±0.79 NL 0.2±0.44 NL 2 spring 40.8±3.34* 49 13.96±2.36* 19 46.25±18.87 NL 32.68±1.64 NL 0.16±0.06* 0.26 0.058±0.041 0.02 3.91±0.59* 2 0.0018±0.001* 0.008 3 spring 40.16±2.67 40 22.24±1.6* 11 44.37±44.003 6 3.49±0.99* 6 0.31±0.04 0.3 0.043±0.021 NL 5.066±0.59* 4 0.4±0.89 0.001 4 spring 40.16±1.53* 45 13.56±1.19* 10 45.71±26.75 NL 32.19±3.27* 15 0.19±0.02 0.2 0.046±0.01 NL 3.28±0.7* 5 0.002±0.0018* 0.01 ground 5 3.68±3.81 5 14.16±0.906* 13 117.59±26.53* 51 20.69±0.44 NL 0.13±0.08 NL 0.034±0.008 NL 0.004±0.01* 0.1 0.0028±0.0019 NL water 6 spring 47.84±3.16 48.6 8.37±4.36 12 8.98±2.59* 4.8 1.29±1.3 NL 0.072±0.05* 0.23 0.04±0.008 NL 3.77±0.57* 0.8 0.005±0.0057 NL 7 spring 83.36±3.01* 50 22.34±2.42* 12 36.43±22.61 16 31.39±1.38* 18 0.22±0.08 NL 0.036±0.01 NL 7.57±3.008* 1.8 0.202±0.445 NL 8 spring 51.84±4.21 48 8.37±2.57 8 23.94±15.64 5 1.59±1.13* 3.7 0.12±0.06* 0.24 0.036±0.005 NL 3.89±0.91 3.7 0.203±0.445 0.001 9 spring 60.32±1.75 62 10.92±1.27* 8.4 16.75±3.89 NL 4.59±0.22* 8 0.27±0.04* 0.48 0.035±0.009 NL 3.06±0.68* 5 0.203±0.445 0.005 10 spring 42.72±3.17 45.6 10.37±4.59 10 28.18±11.26 NL 31.19±0.56* 12 0.12±0.07* 0.024 0.042±0.01 NL 3.16±0.49* 4.8 0.202±0.446 0.007

a Measured value b labeled value NL: Not Labeled * (P < 0.05) statistically significant compared to the labeled value

TABLE 3 - Measured physicochemical parameters of bottled drinking water in conjunction with label values

─ Total Hardness HCO3 Ca Hardness Mg Hardness 3─ TDS (mg/L) pH EC (µS/cm) PO4 (mg/L) Brand Source (mg/L as CaCO3) (mg/L as CaCO3) (mg/L as CaCO3) (mg/L as CaCO3) Ma Lb M L M L M L M L M L M L M L 1 spring 189.3±26.97 NL 176±9.89 NL 189±8.94 NL 7.72±0.1 NL 378.6±53.9 NL 112±9.27 NL 3.08±0.79 NL 0.042 ±0.02 NL 2 spring 251±42.58 NL 158±8.12 NL 159±11.4* 190 7.8±0.0 NL 502±85.16 NL 102±8.36 NL 3.91±0.59* NL 0.049 ±0.01 NL 3 spring 239.4±32.39 214 189.6±2.19* 160 208±9.08 NL 7.68±0.1 7.6 478.8±64.79 NL 100.4±6.69 NL 5.066±0.59* NL 0.044 ±0.01 NL 4 spring 237.2±30.69* 180 154.8±6.57 150 159±7.41* 170 7.8±0.0 7.6 474.4±61.39 NL 100.4±3.84 NL 3.28±0.7* NL 0.034 ±0.01 NL ground 5 140.9±24.63* 110 66±10.39* 53 26±5.47 NL 7.44±0.43 7 281.8±49.27 NL 9.2±9.54 NL 0.004±0.01* NL 0.002±0.00 NL water 6 spring 170.7±23.6 172.8 153.2±13.38 142.6 164±11.4 170.8 7.72±0.17* 7 341.4±47.21 NL 119.6±7.92 NL 3.77±0.57* NL 0.05± 0.01 NL 7 spring 337.8±35* 235 298±14.69 NL 257±51.67 NL 7.64±0.26* 7.06 675.6±70.01 NL 208.4±7.53 NL 7.57±3.008* NL 0.08± 0.01 NL 8 spring 200.7±34.85 185 163.2±7.82* 132 187±47.64 NL 7.8±0.0 7.5 401.4±69.71 NL 129.6±10.52 NL 3.89±0.91 NL 0.043± 0.01 NL 9 spring 253±44.3 230 194.6±7.056 190 219±13.41 NL 7.72±0.1 7.7 506±88.61 NL 150.8±4.38 NL 3.06±0.68* NL 0.062± 0.02 NL 10 spring 241.9±38.47 265 148.4±17.68 156 156±5.47 NL 7.76±0.08 7.8 483.8±76.95 NL 106.8±7.94 NL 3.16±0.49* NL 0.029±0.01 NL

a Measured value b labeled value NL: Not Labeled * (P < 0.05) statistically significant compared to the labeled value

TABLE 4 - Microbiological quality (Mean±SD) of bottled drinking water samples

Microbiological properties of bottled drinking water samples Brands TC FC HPC ( MPN/100mL) ( MPN/100mL) (CFU/mL) 1 0±0.0 0±0.0 224±59.41 2 0±0.0 0±0.0 326±142.23 3 0±0.0 0±0.0 184±174.44 4 0±0.0 0±0.0 508±394.42 5 0±0.0 0±0.0 224±124.81 6 0±0.0 0±0.0 278±78.23 7 0±0.0 0±0.0 116±63.48 8 0±0.0 0±0.0 288±84.08 9 0±0.0 0±0.0 204±105.02 10 0±0.0 0±0.0 236±177.42 Mean ± SD 0±0.0 0±0.0 260.8±182.6

FIGURE 1 - Mean of the HPC results compared to HPC guidelines for drinking water

TABLE 5 - Distribution of HPC in bottled drinking water brands

Number of positive sample HPC (CFU/mL) (Percent of positive sample) < 1 0 (0.0) 1 ─ 500 47 (94) > 500 3 (6) Range 40-940 Mean ± SD 260±182.6

cardiovascular diseases and prostate cancer [16,19-21], in Most brand of bottled waters survey in our study were nat- addition calcium is an essential element for nervous system ural mineral water. As regards, natural mineral water orig- as well as formation of the human bones [22]. Our results inated from an underground aquifer tapped via one or more showed that the minimum level of measured calcium was natural or drilled wells [25], so the reason for high hardness 1.6 mg/L for brand 5 and the maximum level was 86.4 content of water, may because of their underground origin. mg/L for brand7. The mean concentration of calcium was The results of this study are in agreement with Loloei et al. found to vary widely between 3.68 and 83.36 mg/L. This [26] and Van der Aa et al. [25] findings who reported most range met INR value for calcium (MCLG: 30 mg/L) [18]. of the bottled water samples in their study as hard water. Our result in case of calcium is in agreement with study The range of pH was found to vary between 7.44 and 7.8. results of Samadi et al. [7] and Moazeni et al. [10] reported All the bottled water samples had pH value in the range of the measured calcium values in some other Iranian bottled EPA and INR guidelines which is generally consistent with water between 0-202 mg/L and 11.22–49.70 mg/L, respec- Loloi et al. [26]. Analyses of turbidity and residual chlorine tively. The Mg hardness level in our study was between showed that amount of these two parameters in all samples 33.6 - 89.6 mg/L. The magnesium level ranged from 2.99 were equal to zero (0 mg/L). The guideline value for tur- (sample 46, brand 10) to 24.93 (sample 31, brand 7) mg/L. bidity is 5 NTU according to WHO and INR [16,18], so all The brand 7 (22.34 mg/L) had the highest mean level of Mg, the bottled water brands met the turbidity regulations. Ni- whereas the brands 6 and 8 had the lowest levels (8.37 mg/L). trate and nitrite in water can affect the health of consumers So all the bottled drinking water brands met the EPA guideline specially in the case of infants [10]. The maximum measured (50 mg/L) [23]. In a similar study on bottled drinking water concentrations of nitrate and nitrite were 7.57 (brand 7) and samples in Iran, Eslami et al. [24] reported the Mg level in 0.4 (brand 3) mg/L, respectively. So nitrate and nitrite of all the range of standard. According to the classification of bottled water brands in this study met EPA, WHO and INR drinking water hardness [25], 2% of the bottled drinking guidelines. EPA and INR set a 0.3 mg/L guideline value for water samples in our study were considered as soft, 10% as iron in drinking water. Iron can be found in natural water moderately hard, 54% as hard and 34% as very hard water. especially groundwater. Minimum and maximum levels of

iron in bottled drinking water in our study were found vary reported that from 7 brands of bottled water, 3 brands were widely between 0.025 and 0.128 mg/l. All bottled drinking positive with a HPC levels of < 350 CFU/mL and 4 brands water samples met the EPA and INR guidelines. The mean were negative in the case of HPC. Also, they found that the level of fluoride ion in our study ranged from 0.072 (brand quality of all bottled drinking water samples were con- 6) to 0.32 (brand 3) mg/L which is generally consistent sistent with the WHO and Iranian drinking water guide- with results reported by Shams et al. [27], Dobaradaran et lines [37]. But in another study on evaluation of microbio- al. [28] and Nabipour et al. [29]. Some studies were done logical quality of bottled water in Bangladesh, the results about excessive intake of fluoride and association with an indicated that most of bottled water samples had HPC level adverse health effect such as higher risk of hypertension, higher than their safety guidelines. dental caries and dental fluorosis [30-32]. In this regard, different studies have reported the occurrence of elevated fluoride levels in drinking water, air, fish as well as in con- 4. CONCLUSIONS nection with its removal from aqueous solution [33-36]. Maximum mean values of sulfate and chloride ions were Results of the present study showed that all of 10 ex- 117.59 mg/L and 32.68 mg/L, respectively. All these re- amined bottled water brands distributed in the Bushehr ex- sults met the maximum admissible concentration set by cept HPC for one brand generally complied with current EPA, INR and WHO guidelines [16, 18, 23]. drinking water regulations regarding their bacterial and physiochemical properties. The measured levels of Ca2+, Among nineteen physicochemical parameters that 2+ 2------Mg , SO4 , Cl , F , NO3 , NO2 , pH, HCO3 , TDS and TH in were measured in the present study, only eleven values of some brands had a statistically significant difference with these parameters including calcium, magnesium, sulfate, manufacturer's labeled values. Finally, it should be noted chloride, fluoride, nitrate, nitrite, TDS, TH, bicarbonate that the quality of bottled drinking water distributed in Bu- and pH have been labeled on the bottled. As shown in table shehr was not problematical from health point of view. 2 and 3, results of one-sample t-test indicated that amount of analytical examinations of these parameters in some brands had a significant statistical difference with manufacturer's labeling values (P< 0.05). As Moazeni et al. [10] ACKNOWLEDGMENTS have reported that the amount of pH and ions including F−, 2- 2+ 2+ + - SO4 , Ca , Mg , K and Cl in 21 samples of bottled min- The authors are grateful to the Bushehr University of eral waters in Isfahan have differences with amounts of Medical Sciences for financial support. their labels on the bottles. Samadi et al. [7] have also reported that except for magnesium and pH values there were The authors have declared no conflict of interest. statistically significant difference between measured parameters with manufacturer's labeled values (P> 0.05). Also, Saberi Bidgoli et al. [8] have done a study on nitrate concentration of bottled drinking water in Qom and have REFERENCES reported that 66.7 % of the bottled samples had measured nitrate concentration less than the labeled value whereas [1] Mara, D. D. (2003) Water, sanitation and hygiene for the health of 33.3% of the bottled samples had measured nitrate concen- developing nations. Public health 117, 452-456. tration higher than the labeled value . [2] Balbus, J. M. and Lang, M. E. (2001) Is the water safe for my baby? Pediatric Clinics of North America 48, 1129-1152. The results of all FC and TC measurements were neg- [3] de França Doria, M. (2010) Factors influencing public perception ative. In other words, all the bottled drinking water brands of drinking water quality. Water policy 12, 1-19. were found free from coliforms and met the INR, EPA and [4] Ngari, M. S. Wangui, W. T. Ngoci, N. S. and Mavura, W. J. (2013) WHO guidelines (0 MPN/100mL) [15, 16, 23]. In agree- Physico-Chemical Properties of Spring Water in Kabare and Baragwi Locations, Gichugu Division Kirinyaga County of Kenya. ment with our study, Jahed khaniki et al. [13] reported that International Journal of Science and Research 2, 280-285. none of the bottled water samples in their study were posi- [5] Güler, C. and Alpaslan, M. (2009) Mineral content of 70 bottled tive for fecal coliform. In present study, the results showed water brands sold on the Turkish market: assessment of their com- that 94% and 6% of the bottled drinking water samples pliance with current regulations. Journal of Food composition and contained heterotrophic plate counts (HPC) within a range Analysis 22, 728-737. of 1-500 CFU/mL and greater than 500 CFU /mL, respec- [6] Kermanshahi, K. Y. Tabaraki, R. Karimi, H. Nikorazm, M. and Abbasi, S. (2010) Classification of Iranian bottled waters as indi- tively. Brand 7 and brand 4 had the lowest and highest mean cated by manufacturer’s labellings. Food chemistry 120, 1218- levels of HPC with 116 and 508 CFU/mL values respec- 1223. tively. A total of 3 from 50 examined bottled water showed [7] Samadi, M. Rahmani, A. Sedehi, M. and Sonboli, N. (2009) Evalua- heterotrophic plate counts (HPC) greater than 500 CFU/ml tion of chemical quality in 17 brands of Iranian bottled drinking wa- and as shown in Figure 1, except brand 4, other brands of ters. Journal of research in health sciences 9, 25-31. our samples met the INR, EPA and WHO guidelines [15, [8] Bidgoli, M. S. Ahmadi, E. Yari, A. R. Hashemi, S. Majidi, G. Naz- ari, S. Jadidiyan, M. and Azari, A. (2013) Concentration of Nitrate 16, 23]. In a study, Zazouli et al. surveyed bacterial quality in Bottled Drinking Water in Qom, Iran. Archives of Hygiene Sci- of bottled drinking water in Semnan City, Iran. They have ences 2, 121-125.

[9] Güler, C. (2007) Evaluation of maximum contaminant levels in [28] Dobaradaran, S. Mahvi, A. H. and Dehdashti, S. (2008) Fluoride Turkish bottled drinking waters utilizing parameters reported on content of bottled drinking water available in Iran. Fluoride 41, 93. manufacturer's labeling and government-issued production li- [29] Nabipour, I. and Dobaradaran, S. (2013) Fluoride concentrations censes. Journal of Food Composition and Analysis 20, 262-272. of bottled drinking water available in Bushehr, Iran. Fluoride 46, [10] Moazeni, M. Atefi, M. Ebrahimi, A. Razmjoo, P. and Vahid 63-64. Dastjerdi, M. (2013) Evaluation of chemical and microbiological [30] Spittle, B. (2008) Dyspepsia associated with fluoridated water. quality in 21 brands of Iranian bottled drinking waters in 2012: a Fluoride 41, 89. comparison study on label and real contents. Journal of environmental and public health 2013, 1-4. [31] Ostovar, A. Dobaradaran, S. Ravanipour, M. and Khajeian, A. (2013) Correlation between Fluoride Level in Drinking Water and [11] Majumder, A. K. Islam, K. N. Nite, R. N. and Noor, R. (2011) the Prevalence of Hypertension: an Ecological Correlation Study. Evaluation of microbiological quality of commercially available The international journal of occupational and environmental med- bottled water in the city of Dhaka, Bangladesh. Stamford Journal icine 4, 216-217. of Microbiology 1, 24-30. [32] Rahmani, A. Rahmani, K. Dobaradaran, S. Mahvi, A. H. Moham- [12] Lalumandier, J. A. and Ayers, L. W. (2000) Fluoride and bacterial adjani, R. and Rahmani, H. (2010) Child dental caries in relation content of bottled water vs tap water. Archives of family medicine to fluoride and some inorganic constituents in drinking water in 9, 246. Arsanjan, Iran. Fluoride 43, 179-186. [13] Jahed Kh, G. R. Zarei, A. Kamkar, A. Fazlzadeh, D. M. [33] Dobaradaran, S. Fazelinia, F. Mahvi, A. H. and Hosseini, S. S. Ghaderpouri, M. and Zarei, A. (2010) Paper: Bactriological eval- (2009) Particulate airborne fluoride from an aluminium production uation of bottled water from domestic brands in Tehran, Iran. plant in Arak, Iran. Fluoride 42, 228. World Applied Science Journal 8, 274-278. [34] Dobaradaran, S. Kakuee, M. Nabipour, I. Pazira, A. Zazouli, M. [14] APHA - American Public Health Association (2005) Standard A. Keshtkar, M. and Khorsand, M. (2015) Fluoride removal from Methods for the Examination of Water and Wastewater. 21th Ed. aqueous solutions using Moringa oleifera seed ash as an environ- APHA, AWWA and WEF, Washington, DC. mental freindly and cheap biosorbent. Fresenius Environmental [15] Detection and enumeration of coliform organisms in water by mul- Bulletin 24, 1269-1274. tiple tube method. ISIRI No. 3759 Institute of standards and in- [35] Dobaradaran, S. Vakil Abadi, D. R. Hossein Mahvi, A. and Javid, dustrial Research of Iran; 2008. Available from: A. (2011) Fluoride in skin and muscle of two commercial species http://www.isiri.org/portal/files/std/3759.htm. (Persian). of fish harvested off the Bushehr shores of the Persian Gulf. Fluo- [16] WHO, 2011. Guidelines for Drinking Water Quality. 4th Ed: World ride 44, 143. Health Organization Geneva. ISBN: 978 92 4 154815 1. [36] Shams, M. Qasemi, M. Dobaradaran, S. and Mahvi, A. H. (2013) [17] Saleh, M. A. Ewane, E. Jones, J. and Wilson, B. L. (2001) Chem- Evaluation of waste aluminum filing in removal of fluoride from ical evaluation of commercial bottled drinking water from Egypt. aqueous solution. Fresenius Environmental Bulletin 22, 2604- Journal of Food Composition and Analysis 14, 127-152. 2609. [18] Drinking water - Physical and Chemical specifications. ISIRI No. [37] Zazouli, M. Safarpour Ghadi, M. Veisi, A. and Habibkhani, P. 1053 Institute of standards and industrial Research of Iran; 2008. (2013) Bacterial contamination in bottled water and drinking water Available from: http://www.isiri.org/Portal/Home/De- distribution network in Semnan, 2012. J Mazandaran Univ Med fault.aspx?CategoryID=cab3ebf1-95d7-4813-b28c- Sci 22, 151-159. c5fc205c35b3.

[19] Derry, C. Bourne, D. and Sayed, A. (1990) The relationship be-

tween the hardness of treated water and cardiovascular disease mortality in South African urban areas. South African medical journal= Suid-Afrikaanse tydskrif vir geneeskunde 77, 522-524. [20] Yang, C.-Y. Chiu, H.-F. Tsai, S.-S. Cheng, M.-F. Lin, M.-C. and Sung, F.-C. (2000) Calcium and magnesium in drinking water and risk of death from prostate cancer. Journal of Toxicology and En- vironmental Health Part A 60, 17-26. [21] Sauvant, M.-P. and Pepin, D. (2002) Drinking water and cardio- Received: April 13, 2015 vascular disease. Food and chemical toxicology 40, 1311-1325. Revised: July 07, 2015 [22] Nkamare, M. B. Ofili, A. N. and Adeleke, A. J. (2012) Physico- Accepted: July 07, 2015 chemical and microbiological assessment of borehole water in Okutukutu, Bayelsa State, Nigeria. Advances in Applied Science Research 3, 2549-2552. CORRESPONDING AUTHOR [23] EPA 2012. Drinking Water Standards and Health Advisors. Wash- ington, DC: Office of Drinking Water, US Environmental Protec- tion Agency, USA. Sina Dobaradaran [24] Eslami, A. Jamali, H. and Naderi, S. (2013) Comparative Study of Bushehr University of Medical Sciences Bottled Water Microbial and Physicochemical Quality with Na- Faculty of Health tional Standards and its label (A Case Study in Qazvin City, Iran). Iranian Journal of Health and Environment 6, 155-166. The Persian Gulf Marine Biotechnology Research Center [25] van der Aa, M. (2003) Classification of mineral water types and comparison with drinking water standards. Environmental Geol- Department of Environmental Health Engineering ogy 44, 554-563. Bushehr [26] Loloei, M. and Zolala, F. (2011) Survey on the quality of mineral IRAN bottled waters in Kerman city in 2009. Journal of Rafsenjan Uni- versity of Medical Sciences 10, 183-192. Phone: +987714550134 [27] Shams, M. Dobaradaran, S. Mazloomi, S. Afsharnia, M. Ghasemi, M. and Bahreinie, M. (2012) Drinking water in gonabad iran fluo- Fax: +987714550134 ride levels in bottled distribution network point of use desalinator E-mail: [email protected] and decentralized municipal desalination plant water. Fluoride 45, 138-141. FEB/ Vol 24/ No 11a/ 2015 – pages 3836 - 3841

EFFECTS OF CHLORPYRIFOS INSECTICIDE ON CYP1 FAMILY GENES EXPRESSION AND ACETYLCHOLINESTERASE ENZYME ACTIVITY IN ORYZIAS JAVANICUS

Trieu Tuan*, Yoshio Kaminishi, Aki Funahashi, Takao Itakura

Laboratory of Marine Biotechnology, Faculty of Fisheries, Kagoshima University, 4-5-20 Shimoarata, Kagoshima 890-0056, Japan.

ABSTRACT 1. INTRODUCTION

Effects of chlorpyrifos (0.01-0.5 mg/L) on differential Chlorpyrifos insecticide is widely used for pest control expression of CYP1 genes in exposed-fish tissues for 5 days in agricultural areas. Unfortunately, the chemical has been and characteristics of the intrinsic variations of acetylcholin- detected in environmental water areas and unintended ef- esterase (AChE) activity and recovery in medaka fish tis- fects on aquatic animals by aerial overspray or run-off have sues were evaluated for 20 days. Javanese medaka CYP1 occurred [1]. Chlorpyrifos is a broad-spectrum insecticide, genes transcript expression was detected in all tissues, in- killing insects upon contact by affecting their nervous sys- cluding the liver, gill and intestine of the fish exposed to tem normal function by preventing the breaking down of different concentrations of the insecticide for 5 days. The the acetylcholine (ACh) in the synaptic cleft [2]. In aquatic CYP1A gene induction level in each tissue increased with organisms, ACh accumulation in the neuronal cells affects higher concentrations of chlorpyrifos. CYP1B1 transcrip- neurotoxicity, abnormal respiration, swimming, feeding, tion was detected in all of observed tissues. The CYP1C1 muscle spasms and convulsions [3]. However, the specific transcription in all evaluated tissues increased 2- to 5-fold tissue acetylcholinesterase (AChE) activities of fish may at all chlorpyrifos levels after 1 day of exposure. Significant be affected by many factors. For example, differences in effects and interactions of the insecticide concentrations AChE activity were found in relation to fish species [4], and exposure time were found in fish tissues AChE activ- fish organs [5] and fish populations [6], and environmental ity. A trend occurred in which AChE inhibition increased factors, such as temperature [7]. with higher insecticide concentrations and exposure duration. Tissue AChE activity was significantly inhibited after Marine Javanese medaka Oryzias javanicus is distrib- 1 day of exposure to the chemical, but liver AChE activity uted in a wide range of salinity in the estuarine regions of showed the lowest concentration. Recovery was occurred Southeast Asia [8]. Therefore, the marine medaka has been after holding the fish in clean water for 5 days, but the ac- extensively used in toxicity test for toxic pollutants in dif- tivity in all exposure groups was still significantly lower ferent environments [9]. Cytochrome P450 is largely used than that in the control group. Full recovery was only found as an indicator of exposure to environmental contaminants in liver AChE activity after keeping the fish in clean water [10] because CYP proteins play a critical role in the oxida- for 15 days at the lowest concentration. The findings of the tive metabolism of endogenous and xenobiotic (exoge- study can serve as sensitive multi-biomarkers to character- nous) compounds, including pharmaceuticals and environ- ize toxicological impacts and monitor aquatic environmental toxins [11]. Fish CYP1 genes expression patterns ments. could become useful biomarkers for measuring the response of aquatic organisms when assessing exposure to contaminants in environmental systems [12].

In the present study, we evaluated the time duration KEYWORDS: Chlorpyrifos, Oryzias javanicus, acetylcholinester- changes in the expression of the cytochrome P450 CYP1A ase, cytochrome P450, CYP1 (GenkBank accession number, KJ689303), CYP1B1

(GenkBank accession number, KJ689304) and CYP1C1 (GenkBank accession number, KJ689305) gene of the marine medaka fish exposed to low chemical contamination levels for 5 days. In addition, the study also investigated the time period chlorpyrifos inhibition and recovery on tissues AChE activity in the fish exposed to environ- * Corresponding author mentally-realistic concentrations of chlorpyrifos for 5 days

before being transferred to clean water to be kept for an- the chemical. Total RNA was isolated from medaka fish other 5 to 15 days. liver, gill and intestine using QuickGene RNA Tissue Kit S II (RT-2) (Fujifilm, Japan) according to manufacturer in- structions. mRNA reverse transcription was conducted with 2. MATERIALS AND METHODS a PrimeScript™ 1st strand cDNA Synthesis Kit (Takara Bio, Japan) following supplier protocols. Gene-specific primers 2.1 Test animal and cultivation conditions for the CYP1 genes and β-actin (GenBank accession number, Javanese medaka Oryzias javanicus was cultured in JQ905607), as an internal control gene, were designed by the the aquarium facility (Faculty of Fisheries, Kagoshima web-based software, Primer3Plus [13], with a product size o University, Japan). Before the experiment, the fish were between 50-150 bp, Tm ranging from 57-63 C and all the acclimatized to the experimental conditions for one week. default parameters (Table 1). Real-time PCR was analyzed The fish were fed a medaka commercial diet (Kyorin, Ja- using a FastStart Essential DNA Green Master Kit and a pan) once daily at 5% of their biomass during the acclima- LightCycler® Nano system (Roche Applied Science). For tization. The aquariums were cleaned everyday by renew- each sample, gene expression was analyzed in triplicate ing 80% of the water. The culture tanks were filled with with the following protocol: initial holding at 95oC for 10 min, natural seawater (salinity 33±1 ppt) and maintained at pH 3-step amplification in 45 cycles (95oC for 10 sec, 60oC for (7.4±0.5), and temperature (25.1±0.4oC) throughout the 10 sec, 72oC for 15 sec), holding stage at 95oC for 30 sec, raising period. The raising systems were continuously aer- melting at 60oC for 20 sec and 95oC for 20 sec. Melt curve ated to maintain adequate levels of the dissolved oxygen analysis was performed after each PCR run to assure that levels. During the culturing cycle, the aquariums were kept the single product was amplified. The 2-Ct method [14] in a 12-hour dark and 12-hour light photoperiod regime. was used to compare the different CYP1 expression levels within a given organ and using β-actin as a reference gene 2.2 Chlorpyrifos to calculate changes in fold-induction in response to exper- The chlorpyrifos (CAS No. 2921-88-2) insecticide was imental conditions. obtained from Wako Pure Chemical Industries (Japan). The insecticide was of 99.5% purity. Its stock solution was TABLE 1 - Oligonucleotide primers used for real-time PCR analysis prepared by dissolving in 0.02% acetone in water inside a Primer name Location Primer sequences (5'-3') standard volumetric flask to obtain the desired concentra- Java_CYP1A 1F 1163-1182 CATTCACAATCCCACACTGC tion. Although the stock pesticide solution was prepared in Java_CYP1A 2R 1269-1288 ATGGATCCTGCCACAGTTTC acetone, the final solvent amount in the experimental sys- Java_CYP1B 1F 1188-1207 GAGCTACACCATCCCCAAGA tem, considering the total system volume, was negligible, Java_CYP1B 2R 1301-1320 CTTGTTCAGCTTCCCCTCTG Java_CYP1C 1F 1244-1263 ACCAGTTCTCCGTCAACCAC as recommended by OECD guidelines for testing chemi- Java_CYP1C 2R 1362-1381 GCCGTTTACCTGTGGAGAAA cals with some modifications. Java_actin 1F 346-365 AGGGAGAAGATGACCCAGAT Java_actin 2R 447-466 CAGAGTCCATGACGATACCA 2.3 Experimental design The exposure regimen of the experiment was carried 2.5 Effects of chlorpyrifos on tissue AChE activity in the out in 3 L beakers with net lids containing 2 L of natural medaka fish seawater at the stocking density of five adult fish per Four fish were randomly collected from beakers of beaker while aerated using glass Pasteur pipettes fitted onto each treatment group after 1, 3 and 5 days of exposure. each lid. Four fish groups including three experimental Their brain, muscle and liver were removed to be used for groups and one control group, were prepared. The insecti- determining AChE activity in these tissues. The fish re- cide stock solution was distributed to each beaker, reaching maining after 5 days of exposure were moved to new beak- nominal concentrations of 0.02, 0.1 and 0.5 mg/L chlorpyr- ers containing clean seawater. The water in each beaker ifos. No mortality was observed under the experimental was changed with approximate 80% clean natural seawater conditions employed. The control group was kept in natu- every day. The fish were kept for another 5 to 15 days, at ral seawater with only 0.02% acetone. The insecticide con- which point four fish from each group, were collected as centrations used in the study were based on results from the described above. The tissues AChE activity was deter- concentrations of preliminary tests (0.01-10 mg/L) (data mined using an Acetylcholinesterase Activity Assay Kit not shown). The experiments were carried out in the semi- (Sigma-Aldrich, USA), according to the manufacturer in- static systems with a beaker solution renewal of 80% and structions. The tissues were homogenized in 0.1 M phos- an appropriate volume of stock solution was added to phate buffer, pH 7.5 (Sigma-Aldrich, USA), at concentra- maintain constant concentrations of active chemicals every tion of approximately 25 mg fresh tissues per milliliter. The 24 hours after measuring water quality. The experimental homogenized solution was centrifuged at 14,000 rpm at conditions were maintained as described above. 4oC for 10 minutes. Seven hundred micro-litters (700 µl) of the supernatant were kept on ice for tissue AChE analy- 2.4 Real-time PCR and RNA preparation sis within 6 hours. The analyses were carried out in an air- Four fish were randomly collected from beakers of conditioned room (25oC). This assay is an optimized ver- each treatment group after 1, 3 and 5 days of exposure to sion of the Ellman method [15] in which thiocholine, pro-

duced by AChE, reacts with 5,5’-dithiobis (2-nitrobenzoic tively. These ranges are considered within optimal values acid) to form a colorimetric product proportional to the for Javanese medaka fish [9]. AChE activity present. One unit of AChE equals the amount of enzyme that catalyzes the production of 1.0 µmole of thi- 3.1 CYP1A expression changes induced by chlorpyrifos expo- ocholine per minute at pH 7.5 (at room temperature). En- sure zyme AChE activity was determined using a spectrophoto- Expression of CYP1A in chlorpyrifos-exposed fish tis- metric microplate reader (Immuno Mini NJ-2300, USA) at sues after 1-, 3- and 5-day exposure was illustrated in Fig. a 412 nm wavelength. The results of these measurements 1A, 1B and 1C, respectively. A significant increase in were expressed as units/L and the AChE activity calculated CYP1A transcription in liver, gill and intestine was found as follows: after 1 day of exposure to the insecticide at the highest concentration (p<0.05) (Fig. 1A). A significant change in AChE Activity CYP1A mRNA induction in the liver was observed only in A412final – A412initial units/L= x n x 200 the 0.5 mg/L chlorpyrifos-exposed group after 5 days of A412calibrator – A412blank exposure. In the gill, CYP1A transcription peaked at 0.02 where, 200 = equivalent activity (units/L) of the cali- mg/L chlorpyrifos-exposed group after 3-day exposure. In brator when assayed is read at 2 minutes and 10 minutes the intestine, CYP1A transcription reached the maximal n = dilution factor level at 0.5 mg/L of chlorpyrifos-exposed group after 3-day (A ) = Absorbance of the sample at 10 minutes exposure (Fig. 1B). CYP1A expression in the liver peaked 412 final at 0.1 mg/L chlorpyrifos-exposed group after 5-day expo- (A ) = Absorbance of the sample at 2 minutes 412 initial sure. In the gill, CYP1A expression peaked at the highest (A412)calibrator = Absorbance of the calibrator at 10 minutes concentration group after 5 days of exposure. In the intes- (A412)blank = Absorbance of the blank at 10 minutes tine, significant changes in CYP1A expression were observed at all exposure levels after 5-day exposure com- 2.6 Statistical analysis pared with the control group (Fig. 1C). The statistical analysis was carried out by using Super- ANOVA (Abacus Concepts, Berkeley, CA). All data were 3.2 CYP1B1 expression changes induced by chlorpyrifos ex- presented as mean±standard deviation (S.D.) of triplicate posure studies. Two-way analysis of variance (ANOVA) was per- Chlorpyrifos induced CYP1B1 mRNA expression in formed to test the effects of insecticide concentrations and the medaka fish tissues after 1-, 3- and 5-day exposure, as time course of exposure. If interaction between the two fac- presented in Fig. 2A, 2B and 2C, respectively. CYP1B1 in- tors was found, the difference in means was tested within duction level in the liver increased with higher chlorpyrifos each factor. The means of groups were compared using one- concentrations after 1 day of exposure. Significant CYP1B1 way ANOVA followed by Duncan’s multiple comparison expression changes in the gill were only observed in the 0.5 tests. A value of p<0.05 indicated statistical significance. mg/L group compared with the control after 1-day exposure. No significant CYP1B1 expression changes were observed in the intestine after incubation with chlorpyrifos for 3. RESULTS 1-day exposure compared with the control group (Fig. 2A). CYP1B1 expression in medaka fish liver and intestine were Environmental parameters, temperature, pH and salin- not significantly altered at any chlorpyrifos concentrations ity varied during the study period. Temperature, pH and after 3 days of exposure. In the gill, significant differences salinity were 24.3±0.6oC, 7.5±0.4, and 31±1 ppt, respec- in CYP1B1 expression were only observed in the 0.1 and

20 A d 16 c 12 b 8 b

Fold change Fold a a a 4 a aaaa 0 LIVER GILL INTESTINE

60 B 50 e d 40

20 Fold chnage Fold c c b 10 a a aaaa a 0 LIVER GILL INTESTINE

8 C f f 6 e

4 d cd Fold chnage Fold bc ab ab 2 a aaa

0 LIVER GILL INTESTINE

FIGURE 1 - Relative mRNA levels of CYP1A gene in liver, gill and intestine of the medaka fish after exposure to chlorpyrifos insecticide for 5 days. A, expression of CYP1A in these tissues after 1-day exposure; B, expression of CYP1A in these tissues after 3-day exposure; C, expression of CYP1A in the tissues after 5-day exposure. Black, 0 mg/L; white, 0.02 mg/L; upward diagonal, 0.1 mg/L; horizontal brick, 0.5 mg/L. Different letters indicate a statistically significant difference (p<0.05).

25 A e

20 d 15 c 10 Fold change Fold b b ab 5 ab ab ab aaa 0 LIVER GILL INTESTINE

50 B b 40 b

20 Fold chnage Fold

10 a a a aaaa a a a 0 LIVER GILL INTESTINE

6 C b 5 b 4

a 2

Fold chnage Fold a a a a a aaaa 1

0 LIVER GILL INTESTINE

FIGURE 2 - Relative mRNA levels of CYP1B1 gene in liver, gill and intestine of the medaka fish after exposure to chlorpyrifos insecticide for 5 days. A, expression of CYP1B1 in these tissues after 1-day exposure; B, expression of CYP1B1 in these tissues after 3-day exposure; C, expression of CYP1B1 in the tissues after 5-day exposure. Black, 0 mg/L; white, 0.02 mg/L; upward diagonal, 0.1 mg/L; horizontal brick, 0.5 mg/L. Different letters indicate a statistically significant difference (p<0.05).

0.5 mg/L groups after 3-day exposure (Fig. 2B). CYP1B1 changes in CYP1C1 expression in the liver were only ob- expression in medaka fish liver was not significantly al- served in the 0.1 mg/L group after 3-day exposure. No sig- tered at any chlorpyrifos concentration after 5 days of ex- nificant changes in CYP1C1 induction were observed in posure. CYP1B1 transcription in gill and intestine changed the intestine after 3-day exposure to all levels of the insec- significantly only after exposure to 0.1 mg/L of the insecticide. CYP1C1 transcription in the gill peaked at 0.1 mg/L ticide for 5 days (Fig. 2C). chlorpyrifos-exposed group after 3-day exposure (Fig. 3B). Five days of exposure to various concentrations of chlorpyr- 3.3 CYP1C1 expression changes induced by chlorpyrifos ex- ifos indicated that the CYP1C1 mRNA levels in the liver, posure gill and intestine reached the maximal level at 0.1 mg/L The CYP1C1 gene expression in medaka fish tissues chlorpyrifos-exposed group (Fig. 3C). after 1-, 3- and 5-day exposure are depicted in Fig. 3A, 3B and 3C, respectively. The transcription of CYP1C1 in the 3.4 Response of tissues AChE activity to chlorpyrifos liver and gill reached the maximal level at 0.1 mg/L AChE activities in various tissues, such as brain, mus- chlorpyrifos-exposed group after 1 day of exposure. Signif- cle and liver of the marine medaka are presented in Fig. icant changes in CYP1C1 expression in the intestine tissue 4A, 4B and 4C, respectively. In the brain, significant inter- was observed in all chlorpyrifos levels, compared with the actions (p<0.05) between insecticide concentrations and control group after 1 day of exposure (Fig. 3A). Significant exposure time were found in AChE enzyme activity. Brain

8 A f ef 7 def 6 cde cde bc 5 bc bcd 4 b 3 Fold chnage Fold 2 aaa 1

0 LIVER GILL INTESTINE

25 B e 20 d 15

Fold change Fold c 5 b b b ab aba ab ab ab 0 LIVER GILL INTESTINE

6 f C 5 e 4

2 Fold chnage Fold d d d c cccb c 1 a

0 LIVER GILL INTESTINE

FIGURE 3 - Relative mRNA levels of CYP1C1 gene in liver, gill and intestine of the medaka fish after exposure to chlorpyrifos insecticide for 5 days. A, expression of CYP1C1 in these tissues after 1-day exposure; B, expression of CYP1C1 in these tissues after 3-days exposure; C, expression of CYP1C1 in the tissues after 5-day exposure. Black, 0 mg/L; white, 0.02 mg/L; upward diagonal, 0.1 mg/L; horizontal brick, 0.5 mg/L. Different letters indicate a statistically significant difference (p<0.05).

A j 120,00 j j j j i i 100,00 h fg gh f 80,00 e e 60,00 d d d c 40,00 b a a 20,00

Brain AChE activity (units/L) 0,00 DAY 1 DAY 3 DAY 5 DAY +5 DAY +15

B 160,00 140,00 i i i i i h 120,00 g fg g g e ef 100,00 e d d 80,00 c 60,00 40,00 b ab a a 20,00

Muscle AChE activity (units/L) activity Muscle AChE 0,00 DAY 1 DAY 3 DAY 5 DAY +5 DAY +15

C 90,00 g fg fg fg fg fg 80,00 f e 70,00 60,00 50,00 d d 40,00 c c 30,00 b a a a 20,00 a a a a 10,00 0,00 Liver AChE activity (units/L) activity Liver AChE DAY 1 DAY 3 DAY 5 DAY +5 DAY +15

FIGURE 4 - AChE activity in brain (A), muscle (B) and liver (C) of the medaka fish exposed to chlorpyrifos insecticide. Black, 0 mg/L; white, 0.02 mg/L; upward diagonal, 0.1 mg/L; horizontal brick, 0.5 mg/L. Day +5, +15, the fish were kept in clean water after exposure for 5 days or 15 days. Two-way ANOVA data where P values are pesticide concentration 0.0001, time of exposure 0.001 and interaction 0.001. Different letters indicate a statistically significant difference (p<0.05).

AChE activity was significantly inhibited by all chlorpyri- at the highest concentration (0.5 mg/L). Recovery was ob- fos treatments after 1 day of exposure to the chemical. Per- served after replacing the water, but the AChE enzyme ac- centage inhibition increased with higher concentrations tivity in all groups was still significantly lower than that in and reached a maximum (82%) after five days of exposure the control group after 15 days holding the fish in clean

water (Fig. 4A). In the muscle, two-way ANOVA showed 73 to over 700 µg/L [22, 23]. In India, chlorpyrifos is clas- significant effects (p<0.05) of insecticide concentrations sified as an extremely hazardous pesticide. It has been and exposure time on the AChE enzyme activity. Inhibition found 4.8 ppb in soft drinks, which was 47 times higher levels ranged from 19-22%, 21-47%, and 27-86% of the than the permissible limit recommended by Bureau of In- control values in the 0.02, 0.1 and 0.5 mg/L chlorpyrifos dian Standards [19]. Therefore, the nominal concentrations treatment, respectively. No sign of recovery was found in (0.02-0.5 mg/L) we used in the study are realistic for estu- AChE activity, and the effects remained significant when arine regions where Javanese medaka is likely to develop. compared to the control group until the end of the exposure The biological responses of organisms to exterior stresses, period. Muscle AChE activity recovery was slow, and at identified at the gene transcript level and enzyme activities, the end of the experiment, was still significantly lower than provide a basis for assessing physiological and biochemi- the AChE enzyme activity in the control group at all cal responses, and afford an early warning of ecological chlorpyrifos insecticide levels (Fig. 4B). In the liver, two- risks at the individual and community levels. way ANOVA shows significant effects and interactions (p<0.05) of insecticide concentrations and exposure time The present study reported the organs distribution of on the AChE enzyme activity. Liver AChE activity was CYP1 genes induction in the medaka fish exposed to vari- significantly inhibited after 1 day of exposure to the chem- ous concentrations of chlorpyrifos insecticide for 5 days. ical, except in the lowest concentration (0.02 mg/L), which In aquatic organisms, CYP1A is one of the major P450 was not significantly inhibited compared to the control. In subfamilies that are responsive to toxic organic chemicals. general, the higher concentration of chlorpyrifos, the lower CYP1A gene has been used as a biomarker to monitor the the liver AChE activity during 5-day exposure. Recovery presence and effects of contaminants in a wide ranges of occurred after changing the water for 5 days, but the activ- aquatic environments [24]. In the study, we evaluated dif- ity in all exposure groups was still significantly lower than ferential expression of the biomarker gene in exposed-fish that in the control group (p<0.05). Full recovery was found responsed to chlorpyrifos insecticide exposure for 5 days. after 15 days holding the fish in clean water at the lowest The expression of the CYP1A gene in chlorpyrifos-ex- concentration (0.02 mg/L) group (Fig. 4C). posed medaka varied in the organs studied, suggesting that the gene has different functions in the tissues. The results also revealed that the CYP1A gene induction level in each 4. DISCUSSION tissue increased with higher chlorpyrifos concentrations. Specially, the CYP1A gene was predominantly expressed Rice farming is the main agricultural industry world- in the gill during each sampling point. Gill can be stimu- wide. The amount of used organophosphorus insecticide lated to produce oxidative stress through exposure to envi- increases every year. A large number of agricultural and ronmental pollution [25]. However, this result is opposite industrial chemicals have been developed and overuse of to previous study on the effect of chlorpyrifos on the ex- these chemicals causes concern regarding their toxicity or pression of CYP1A in gill of common carp [26]. The persistence in the environment. Living organisms in the CYP1A induced expression can be served as a biomarker contaminated environment take such chemicals into their for assessing animal exposure to pollutants [27]. The pre- bodies and respond to exterior stresses through defense and sent study shows CYP1A gene involvement in other phys- adaptation. Despite the wide use of organophosphorus in- iological functions of the Javanese medaka. secticide in agriculture, there are few studies on their hazardous effects on aquatic organisms or the risks arising In this study, we used stress-related gene, CYP1B1 as from the exposure of aquatic animals. Therefore, the study a molecular biomarker in the medaka fish for assessing the was conducted to provide information on the physiological marine environment and the fish health status. Javanese or genetic changes induced by such environmental stresses. medaka CYP1B1 transcripts were detected in all tissues ex- In addition, the result of the study may serve as useful bi- amined, including the liver, gill and intestine. The highest omarkers for chlorpyrifos insecticide pollution. Since there CYP1B1 gene level expressions were found in the liver af- is the development of sensitive biomarkers for detection of ter 1 day of exposure to the chemical. The result of this genotoxic effects in aquatic organisms has gained im- study is in line with previous findings in killifish, zebrafish portance because of growing concern over the presence of and mice, in which the CYP1 gene expression patterns genotoxins in the aquatic environment [16]. We chose to were high in the liver tissue [28-30]. The CYP1B1 induc- examine chlorpyrifos due to its relatively common agricul- tion in the liver may consider functions associated with the tural chemical use in many countries such as the United role of the organ in endogenous metabolite [31]. In addi- States [17], Australia [18], India [19] and United Kingdom tion, CYP1B1 mRNA level changes were observed in the [20]. Direct application of chlorpyrifos to water bodies to gill of the fish exposed to the high concentration groups control mosquitoes or agricultural run-off from treated ar- after 3 and 5 days of exposure to the insecticide. Similarly, eas can result in contamination of chlorpyrifos up to 4.3 CYP1B1 transcripts were induced in the gill of zebrafish mg/L in streams and lakes [21]. In addition, previous works exposed to βNF and PCB126 [29]. However, the result of examined chlorpyrifos concentrations in small water bod- this study is in contrast with previous study on the effect of ies adjacent to cropland report concentrations ranging from chlorpyrifos on the CYP1B expression in gill of common

carp [26]. The present study shows that CYP1B1 transcrip- were not previously exposed to any inhibitory agent. It is tion appears to be an appropriate biomarker for insecticide worth mentioning that AChE decrease is not necessarily exposure in aquatic animals. fatal. In the study, no mortality was found in any treatment at any levels of chlorpyrifos insecticide, even when the in- The study have shown that CYP1C1 transcription was hibition rate reached over 80% of the control values. The expressed in all observed tissues increased 2- to 5-fold at relationship between AChE inhibition and mortality in fish all chlorpyrifos levels after 1 day of exposure. The highest has not been known because some species survival has CYP1C1 mRNA expression was observed in the gill at been observed, even in brain AChE activity inhibition of high chlorpyrifos concentration groups after 3 days of ex- 90-95%, though some have died [43]. These observations posure to the chemical. This result is consistent with previ- strongly support the importance of marine medaka tissue ous study on the effect of chlorpyrifos on the expression of AChE activity as a biomarker for assessing patho-physio- CYP1C in gill of common carp [26]. Specifically, the high- logical changes in fish caused by chlorpyrifos insecticide est CYP1C1 gene expression was found in all exposed-fish pollution and its mitigation by depuration. tissues at 0.1 mg/L chlorpyrifos group after 5 days of exposure. The markedly similar patterns for basal expression AChE inhibition increased rapidly within 1 day after suggested conserved physiological functions of CYP1C exposure and remained significant until the end of the ex- gene in fish [12]. The CYP1Cs genes were recently isolated posure period, even in the lowest chemical treatment con- and its roles in animals are little understood. CYP1C genes centration. Five days after the medaka fish were transferred have been described in fish, frogs and birds [32, 33]. Stud- to chlorpyrifos-free water, the percentage of AChE activity ies on expression of CYP1C1 and CYP1C2 in different tis- inhibition decreased. Full recovery was seen in the liver sues of carp have showed CYP1Cs induced expression AChE activity after 15 days in the lowest insecticide con- only in the gill and kidney, respectively [34, 35]. In addi- centration group. The current results show that the AChE tion, CYP1Cs subfamily genes expression in many tissues enzyme activity responds to the insecticide, which is cur- by exposure to xenobiotics was also reported in zebrafish rently regulated in different kind of tissues. Long-term in- [29] and killifish [30]. The result of the study indicate that hibition of tissue AChE activity exposed to organophos- the CYP1C1 of Javanese medaka is involved largely in the phate insecticides followed by recovery in insecticide-free fish tissues, and may play an important role in the meta- water has previously been presented in the snakehead fish bolic detoxification systems of the fish, especially in re- Channa striata [39], gold fish Carassius auratus [43], hy- sponse to the insecticide. brid catfish Clarias macrocephalus x Clarias gariepinus [44], tilapia Oreochromis mossambicus [45] and rainbow Stress indicators at cellular and tissue levels have been trout Oncorhynchus mykiss [5]. The long-term AChE en- presented in fish and other aquatic organisms to monitor zyme activity inhibition found in the study may be ex- environmental pollution for recent years [36]. Recently, plained by slow rate of spontaneous reactivation of the en- studies have been carried out to investigate AChE activity zyme [46]. In addition, the organophosphates insecticide in fish tissues as early-warning biomarkers for insecticide with a thiol group (P=S bond, as chlorpyrifos in the present pollution assessment in tissue-specific variations in AChE study) may have a slow metabolism than the oxon form responses [36]. Tissues AChE and non-protein reduced (P=0 bond) in fish [45]. Moreover, the AChE activity re- glutathione, which protects cells against oxidative injury covery depends on multiple factors such as the tissue being and detoxicates xenobiotics and/or their metabolites, have evaluated, the species and class of pesticide [47]. In this been validated as pollution biomarkers in fish and other study, full liver AChE activity recovery was only found at aquatic organisms [37]. In the present study, AChE enzyme the lowest insecticide concentration after keeping the fish activity in all tissues was rapidly inhibited by chlorpyrifos in clean water for 15 days. The different rate of AChE re- insecticide during 5 days of exposure. The duration of the covery may be explained by various protein syntheses in effect following chlorpyrifos exposure was not only con- the tissue. In the current experiment, the AChE enzyme ac- centration-dependent but also exposure-time dependent. tivity recovery was observed after keeping the fish in clean All chlorpyrifos concentrations tested in the experiment water for 5 days and increased after 15 days. However, caused significant inhibition of tissues AChE activity at some extra time was needed for total enzyme activity re- any sampling point during the study. Low tissue AChE ac- covery in the high concentration groups. tivity, such as brain, liver, gill, muscle and plasma, was observed in snakehead fish Channa striata exposed to the or- ganophosphate insecticide, diazinon [38, 39], channel cat- 5. CONCLUSION fish Ictalurus punctatus exposed to chlorpyrifos [40], and mosquitofish Gambusia affinis exposed to chlorpyrifos, The regulated CYP1 gene expression in the marine parathion and methyl parathion [41]. AChE activity inhibi- medaka fish can be used as potential biomarkers in control- tion above 20% has been suggested as indicative of expo- ling aquatic environments. The findings of the study pro- sure to anticholinesterase agents [42]. However, the exper- vide a better understanding of the genetic responses of the imental activity of AChE for the marine medaka fish in this medaka fish to chlorpyrifos insecticide exposure. In addi- work indicate variations of over 20%, even though fish tion, this study demonstrates the effect of chlorpyrifos on

different tissue AChE activities in the marine medaka fish. [9] Koyama, J., Miki, K., Imai, S., Fukunaga, M., Uno, S., AChE inhibition in the tissues evaluated appears as an Kakuno, A. (2007) Java Medaka: a proposed new marine test fish for ecotoxicology. Environmental Toxicology, 487-491. early process in response to sublethal exposures and could be a sensitive biomarker to characterize toxicological im- [10] Hahn, M.E., B. R Woodin, J. J. Stegeman and D. E. Tillitt (1998) Aryl hydrocarbon receptor function in early verte- pacts. The duration of the effects of chlorpyrifos exposure brates: inducibility of cytochrome P450 lA in agnathan and was not only concentration-dependent but also exposure- elasmobranch fish. Comp Biochem Physiol C Pharmacol Tox- time dependent. AChE inhibition increased with higher icol EndocrinoI, 120, 67-75. chlorpyrifos concentrations and exposure duration. The [11] Diotel, N., Le Page, Y., Mouriec, K., Tong, S., Pellegrini, E., quick and long-term inhibition of AChE activity indicates Vaillant, C., Anglade, I., Brion, F., Pakdel, F., Chung, B., Kah, that the AChE inhibition may not be the main cause of O. (2010) Aromatase in the brain of teleost fish: expression, regulation and putative functions. Frontiers in Neuroendo- acute toxicity in fish. crinolog, 31, 172-192.

[12] Jönsson, M.E., Gao, K., Olsson, J.A., Goldstone, J.V., Brandt, I. (2010) Induction patterns of new CYP1 genes in environ- ACKNOWLEDGEMENTS mentally exposed rainbow trout. Aquatic Toxicology, 98, 311- 321. We thank Prof. Dr. Jiro Koyama (Education and Re- [13] Andreas, U., Harm, N., Xiangyu, R., Ton, B., René, G., Jack, search Center for Marine Resources and Environment, A. (2007) Leunissen: Primer3Plus, an enhanced web interface Faculty of Fisheries, Kagoshima University, Japan), who to Primer3. Nucleic Acids Research, 35, 71-74. provided Javanese medaka for the studies and his sugges- [14] Livak, K.J., Schmittgen, T.D. (2001) Analysis of Relative tions for using the insecticide on the fish. The authors are Gene Expression Data Using Real-Time Quantitative PCR and also thankful to Prof. Dr. Manabu Ishikawa and Ms. Tru- the 2−ΔΔCT Method. METHODS, 25, 402-408. ong Huynh Nhu (Laboratory of Aquatic Animal Nutrition, [15] Ellman, G., Courtney, D., Anderdres, V., Featherstone, R. Faculty of Fisheries, Kagoshima University, Japan) for (1961) A new and rapid colorimetric determination of acetyl- their support in measuring the AChE enzyme activity and cholinesterase activity. Biochemical Pharmacology, 7, 88-95. statistical analyses. [16] Hayashi, M., Ueda, T., Uyeno, K., Wada, K., Kinae, N., Sao- tome, K., Tanaka, N., Takai, A., Sasaki, Y., Asano, N., Sofuni, The authors have declared no conflict of interest. T., Ojima, Y. (1998) Development of genotoxicity assay systems that use aquatic organisms. Mutation Research/Funda- mental and Molecular Mechanisms of Mutagenesis, 399, 125- 133. REFERENCES [17] Cutler, G., Purdy, J., Giesy, J., Solomon, K. (2014) Risk to pollinators from the use of chlorpyrifos in the United States. [1] Somnuek, C., Boonphakdee, C., Cheevaporn, V., Tanaka, K. Reviews of Environmental Contamination and Toxicology, (2009) Gene expression of acetylcholinesterase in hybrid cat- 231, 219-265. fish (Clarias gariepinus X Clarias macrocephalus) exposed to chlorpyrifos and carbaryl. Journal of Environmental of Biol- [18] Baskaran, S., Kookana, R., Naidu, R. (1999) Degradation of ogy, 30, 83-88. bifenthrin, chlorpyrifos and imidacloprid in soil and bedding materials at termiticidal application rates. Pesticide Science [2] Karanth, S., Pope, C. (2000) Carboxylesterase and A-esterase 55, 1222-1228. activities during maturation and aging: relationship to the toxicity of chlorpyrifos and parathion in rats. Toxicological Sci- [19] Ali, D., Nagpure, N., Kumar, S., Kumar, R., Kushwaha, B. ences, 58, 282-289. (2008) Genotoxicity assessment of acute exposure of chlorpyrifos to freshwater fish Channa punctatus (Bloch) using micro- [3] Peakall, D. (1992) Animal biomarkers as pollution indicators. nucleus assay and alkaline single-cell gel electrophoresis. Chapman & Hall, London, UK, 1-291. Chemosphere, 71, 1823-1831. [4] Chuiko, G., Podgornaya, V., Zhelnin, Y. (2003) Acetylcholin- [20] Brajesh, K., Walker, A., Morgan, J., Wright, D. (2003) Effects esterase and butyrylcholinesterase activities in brain and of soil pH on the biodegradation of chlorpyrifos and isolation plasma of freshwater teleosts: cross-species and cross-family of a chlorpyrifos-degrading bacterium. Applied and Environ- differences. Comparative Biochemistry and Physiology, mental Microbiology, 69, 5198-5206. 135B, 55-61. [21] Wood, B., Stark, J. (2002) Acute toxicity of drainage ditch wa- [5] Ferrari, A., Andrés, V., de D'Angelo, A. (2007) Muscular and ter from a Washington State cranberry-growing region to brain cholinesterase sensitivities to azinphos methyl and car- Daphnia pulex in laboratory bioassays. Ecotoxicology and En- baryl in the juvenile rainbow trout Oncorhynchus mykiss. Com- vironmental Safety, 53, 273-280. parative Biochemistry and Physiology, Part C 146, 308-313. [22] Moore, M., Schulz, R., Cooper, C., Smith, J., Rodgers, J. [6] Gibson, J., Ludke, J., Frerguson, D. (1969) Sources of error in (2002) Mitagation of chlorpyrifos runoff using constructed the use of fishbrain acetylcholinesterase activity as a monitor wetlands. Chemosphere, 46, 827-835. for pollution. Bulletin of Environmental Contamination and [23] Mazanti, L., Rice, C., Bialek, K., Sparling, D., Stevenson, C., Toxicology, 17-23. Johnson, W., Kangas, P., Rheinstein, J. (2003) Aqueous-phase [7] Hogan, J. (1970) Water temperature as a source of variation in disappearance of atrazine, metolachlor, and chlorpyrifos in la- specific activity of brain acetylcholinesterase of bluegills. Bul- boratory aquaria and outdoor macrocosms. Archives of Envi- letin of Environmental Contamination and Toxicology, 5, 347- ronmental Contamination and Toxicology, 44, 67-76. 353. [24] Woo, S.e.a. (2009) Effects of heavy metals on antioxidants and [8] Kinoshita, M., Murata, K., Naruse, K., Tanaka, M. (2009) stress-responsive gene expression in Javanese medaka Medaka: Biology, management, and experimental protocols. (Oryzias javanicus). Comp Biochem Physiol C Toxicol Phar- 1st, Wiley-Blackwell, Iowa, USA, 1-444. macol 149, 289-299.

[25] Tiedke, J., Cubuk, C., Burmester, T. (2013) Environmental [39] Cong, N., Phuong, N., Mark, B. (2008) Brain cholinesterase acidification triggers oxidative stress and enhances globin ex- response in the snakehead fish (Channa striata) after field expression in zebrafish gills. Biochemical and Biophysical Re- posure to diazinon. Ecotoxicology and Environmental Safety search Communications, 441, 624-629. 71, 314-318. [26] Fu, Y., Li, M., Liu, C., Qu, J.P., Zhu, W.J., Xing, H.J., Xu, [40] Straus, D., Chambers, J. (1995) Inhibition of acetylcholinester- S.W., Li, S. (2013) Effect of atrazine and chlorpyrifos expo- ase and aliesterases of fingerling channel catfish by chlorpyri- sure on cytochrome P450 contents and enzyme activities in fos, parathion, and S,S,S-tributyl phosphorotrithioate (DEF). common carp gills. Ecotoxicology and Environmental Safety, Aquatic Toxicology, 33, 311-324. 94, 28-36. [41] Scott, J., Chambers, J. (1996) Time course of inhibition of cho- [27] Kim, R., Kim, B., Hwang, D., Au, D., Jung, J., Shim, W., linesterase and aliesterase activities, and nonprotein sulfhydryl Leung, K., Wu, R., Rhee, J., Lee, J. (2013) Evaluation of bi- levels following exposure to organophosphorus insecticides in omarker potential of cytochrome P450 1A (CYP1A) gene in mosquitofish (Gambusia affinis). Fundamental and Applied the marine medaka, Oryzias melastigma exposed to water-ac- Toxicology, 29, 202-207. commodated fractions (WAFs) of Iranian crude oil. Compara- tive Biochemistry and Physiology Part C: Toxicology & Phar- [42] Ludke, J., Hill, E., Dieter, M. (1975) Cholinesterase (ChE) re- macology, 157, 172-182. sponse and related mortality among birds fed ChE inhibitors. Archives of Environmental Contamination and Toxicology, 3, [28] Choudhary, D., Jansson, I., Stoilov, I., Sarfarazi, M., 1-21. Schenkman, J.B. (2005) Expression patterns of mouse and human CYP orthologs (families 1–4) during development and in [43] Ferrari, A., Venturino, A., de D'Angelo, A. (2004) Time course different adult tissues. Archives of Biochemistry and Biophys- of brain cholinesterase inhibition and recovery following acute ics, 436, 50-61. and subacute azinphosmethyl, parathion and carbaryl exposure in the goldfish (Carassius auratus). Ecotoxicology and Envi- [29] Jonsson, M.E., R. Orrego, B. R. Woodin, J. V. Goldstone and ronmental Safety, 57, 420-425. J. J. Stegeman (2007) Basal and 3,3',4,4',5-pentachloro-bi- phenyl- induced expression of cytochrome P450 lA, 1B and [44] Chawanrat, S., Cheevaporn, V., Saengkul, C., Beamish, W. 1C genes in zebrafish. Toxicol Appl PharmacoI, 221, 29-41. (2007) Variability in acetylcholinesterase upon exposure to chlorpyrifos and carbaryl in hybrid catfish. ScienceAsia, 33, [30] Zanette, J., Jenny, M., Goldstone, J., Woodin, B., Watka, L., 301-305. Bainy, A., Stegeman, J. (2009) New cytochrome P450 1B1, 1C2 and 1D1 genes in the killifish Fundulus heteroclitus: Ba- [45] Rao, J. (2004) Effects of monocrotophos and its analogs in ac- sal expression and response of five killifish CYP1s to the AHR etylcholinesterase activity’s inhibition and its pattern of recov- agonist PCB126. Aquatic Toxicology, 93, 234-243. ery on euryhaline fish, Oreochromis mossambicus. Ecotoxi- [31] Nebert, D.W., Roe, A.L., Dieter, M.Z., Solis, W.A., Yang, Y., cology and Environmental Safety, 59, 217-222. Dalton, T.P. (2000) Role of the aromatic hydrocarbon receptor [46] Costa, L. (1997) Basic toxicology of pesticides. In: Keifer, M. and [Ah] gene battery in the oxidative stress response, cell cycle Human Health Effects of Pesticides, Occupational Medicine control, and apoptosis. Biochemical Pharmacology, 59, 65-85. State of Art Review. 12, Hanley & Belfus, Inc., Philadelphia, [32] Goldstone, J.V., Andrew, G.M., Akira, K., Juliano, Z., Thiago, PA, 251-268. P., Maria, E.J., Nelson, D.R., Stegeman, J.J. (2010) Identifica- [47] Fulton, M., Key, P. (2001) Acetylcholinesterase inhibition in tion and developmental expression of the full complement of estuarine fish and invertebrates as an indicator of organophos- Cytochrome P450 genes in Zebrafish. BMC Genomics, 11, 1- phorus insecticide exposure and effects. Environmental Toxi- 21. cology and Chemistry, 20, 37-45. [33] Goldstone, J., Goldstone, H., Morrison, A., Tarrant, A., Kern,

S., Woodin, B., Stegeman, J. (2007) Cytochrome P450 1 genes in early deuterostomes (tunicates and sea urchins) and verte- brates (chicken and frog): origin and diversification of the CVPl gene family. Mol Biol Evol, 24, 2619-2631. [34] Itakura, T., El-kady, M., Mitsuo, R. (2005) Complementary DNA cloning and constitutive expression of Cytochrome P450 1C1 in the gills of carp (Cyprinus carpio). Environmental Sci- Received: April 30, 2015 ences, 12, 111-120. Revised: July 09, 2015 Accepted: August 03, 2015 [35] Kaminishi, Y., El-Kady, M., Mitsuo, R., Itakura, T. (2007) Complementary DNA cloning and organ expression of cytochrome P450 1C2 in carp (Cyprinus carpio). Environmental Sciences, 14, 23-33. CORRESPONDING AUTHOR [36] Mdegela, R., Braathen, M., Mosha, R., Skaare, J., Sandvik, M. (2010) Assessment of pollution in sewage ponds using bi- Trieu Tuan omarker responses in wild African sharptooth catfish (Clarias Kagoshima University gariepinus) in Tanzania. Ecotoxicology, 19, 722-734. Faculty of Fisheries [37] Stefano, B., Ilaria, C., Silvano, F. (2008) Cholinesterase activ- Laboratory of Marine Biotechnology ities in the scallop Pectin Jacobeans: characterization and effects of exposure to aquatic contaminants. Science of the Total 4-5-20 Shimoarata, Kagoshima, 890-0056 Environment, 392, 99-109. JAPAN [38] Cong, N., Phuong, N., Mark, B. (2006) Sensitivity of brain cholinesterase activity to diazinon (BASUDIN 50EC) and fe- Tel/Fax: +81-99-286-4220 nobucarb (BASSA 50EC) insecticides in the air-breathing fish E-mail: [email protected]. Channa striata (Bloch, 1793). Environmental Toxicology and Chemistry, 25, 1418-1425. FEB/ Vol 24/ No 11a/ 2015 – pages 3842 - 3852

CHEMICAL CHARACTERISTICS, SOURCES AND HEALTH

IMPACTS OF URBAN AMBIENT PM2.5 IN WUHAN DURING THE CHINESE SPRING FESTIVAL POLLUTION EPISODE

Li Liu1, Hui Hu1,2,*, Yiping Tian3, Nan Chen3, Jihong Quan3, Hong Liu3

1School of Environmental Science and Engineering, Huazhong University of Science and Technology, Wuhan 430000, China 2Hubei Key Laboratory of Industrial Fume & Dust Pollution Control, Jianghan University, Wuhan 430000, China 3Hubei Environmental Monitoring Central Station, Wuhan 430000, China

ABSTRACT crease in particulate matter exposure and increase in mortality and morbidity [1-4]. Moreover, there are better cor- A real-time air pollution monitoring campaign was relations in short-term air quality degradation events be- conducted in Wuhan, the largest city in mid China, during tween particle pollution and human diseases [5-8], and the the 2013 Spring Festival pollution episode. Analysis of festival-type air pollution episode has become an important chemical characteristics indicated that eight water soluble type of short-term air pollution events around the world [9- inorganic ions (WSIIs) and seventeen elements, respec- 11]. Furthermore, the Chinese Spring Festival will be the tively, accounted for 42.1% and 4.2% of PM2.5 mass. The biggest one due to the time-concentrated and wide-range study period was divided into three phases: the pre-holiday enormous burning firework and vehicle exhaust. pollution phase (phase 1), the holiday pollution phase (phase Although there were already some research reports 2) and the post-holiday pollution phase (phase 3). The ratio - + about PM2.5 pollution during the Spring Festival pollution of Cl /K in phase 2 was 0.6, which was exactly equal to the episode [12-15], all of them just focused on the impacts of value of firework-burning aerosol reported by Shen et al. 2- firework burning on the chemical and physical characteris- (2009). Phase average concentrations of PM2.5, SO2, SO4 + tics of PM2.5, and there were rather limited studies provid- and K reached the highest in phase 2, whereas those of NO2, - 2+ 2+ ing information about emission sources and health risk as- NO3 , Ca , and Mg peaked in phase 3. K-mean cluster anal- sessment of PM during the period. ysis and PMF were performed successfully to resolve seven 2.5 sources of PM2.5 during the pollution period. Motor vehicle In this study we selected the positive matrix factoriza- exhaust (25.0%) and coal combustion (21.6%) were the mation (PMF) model to do the source apportionment of PM2.5 jor contributors to PM2.5 mass followed by steel works in Wuhan, the largest city in mid China. PMF is one of the (15.5%), secondary inorganic source (14.8%), firework burn- most widely used methods for source apportionment [16, ing (10.9%), glass/cement plants (6.5%) and road dust 17], and it has been successfully applied in many cities all (5.7%). Additionally coal burning was the primary source of over the world [18-23]. Due to the fact that source appor- aerosol Pb in Wuhan during the study period. And the re- tionment with PMF required plenty of samples [17], a high solved sources of steel works and glass/cement plants mainly time-resolution monitoring campaign with online instru- referred to the local sources in Wuhan. The results of health ments was conducted to provide sufficient samples and cap- risk assessment from heavy metal indicated that As, Pb and ture the whole pollution process of the Chinese Spring Fes- Cd were the three main metals in PM2.5, which could cause tival pollution episode (1 to 25 February, 2013). K-mean non-carcinogenic effects. cluster analysis was innovatively performed to assist the PMF with the identification of source type and dig out more

information about sources. Moreover health risk assessment KEYWORDS: PM2.5; water soluble ion; element; source apportion- from heavy metal in PM2.5 during the pollution episode was ment; PMF; K-mean cluster analysis; health risk assessment also discussed.

1. INTRODUCTION 2. MATERIALS AND METHODS

Over the past decades, loads of epidemiological studies 2.1 Study region have already demonstrated the good correlations between in- As the most populous city in central China, Wuhan had a population of 10,120,000 people with 8,217,170 residents * Corresponding author in its urban area as of 2012. And Wuhan is also a major

transportation hub, with dozens of railways, roads and ex- pressways passing through the city. Because of its key role in domestic transportation, Wuhan was sometimes referred to as the "Chicago of China”. Moreover, it not only has a huge amount of migrant residents, but also has the largest number of enrolled college students (about 1,040,000 college students in 2012) in the world. Therefore there will be extremely high traffic flow as numerous college students and migrant residents come back to their hometowns to re- unite with their families and celebrate the traditional Chi- nese New Year. The transport system including trains, air- planes and buses experiences tremendous challenges. In 2013, the total number of journeys during the “spring travel rush” reached ~3.4 billion, which exceeded the entire population of China. It was estimated that about 120 million people traveled via the Wuhan transport system. Obviously Wuhan plays a very significant role in the Chinese New Year exodus which was regarded as the world’s biggest annual migration on Earth. Apart from tremendous vehicle exhaust, there will be a citywide fireworks burning during the Spring Festival. Furthermore Wuhan is also a typical heavy industry city with numerous complex emission FIGURE 1 - Location of the sampling site in Wuhan sources including steel works, petrochemical plants, glass plants, cement mills, and power stations. Hourly PM2.5 concentration was measured by the TEOM™ Series 1405 Ambient Particulate Monitors 2.2 Measurement site (Thermo Fisher Scientific Inc., Franklin, MA) and its WSIIs The monitoring station (30°31′N, 114°21′E; see Fig. 1) and elements species were measured every 60 min by a WSII is situated in the Hubei Environmental Monitoring Centre and elements analyzer (Marga 1S/Metrohm, Switzerland; (HEMC), which is located in an urban district of Wuhan. AMMS-100/Focused Photonics Inc., China), respectively. All the instruments in the monitoring campaign are in- The minimum detection limit (MDL) of these equipment was stalled on the top of a 6-floor building. Then the sampling determined by sampling zero air (Parker, USA). Standard de- site is situated in a residential area. Beside it are the Wuhan viation (SD) for each constituents was obtained, and three University and the East Lake, and there is not vital com- times of the SD (99.7% confidence limit) was considered munication traffic line, large scale fireworks burning sites, as the MDL for the respective constituents. The MDL for the railway/coach stations and huge metallurgical steel mills 8 WSIIs ranged from 0.002 to 0.056 μg m-3, while the MDL within 1-km radius. In all, the monitoring site is believed for the 17 elements varied from 0.001 to 0.017 μg m-3. to represent a typical urban pollution environment in Wu- Cleaning with the isopropanol and leakage detection of han during the Spring Festival. the sampling system was conducted prior to sampling. Zero

2.3 Data collection and span checks were done every 5 days to identify possible analyzer malfunction and zero drifts. The multi-point In the intensive atmospheric field campaign, the study calibrations were performed at approximately a 4-day in- period lasted from 1 to 25 February in 2013, covering the terval during the study period. Spring Festival (10-16 February) and the Lantern Festival (24 February). In the campaign, the data to be collected with Meteorological data (WS, WD, T, RH, precipitation) were also measured every 60 min at the sampling site with real-time measurement include gas pollutants (SO2, CO, auto meteorology monitoring station (WS600-UMB/ Swarco- NO2, NO), particulate pollutants (PM2.5), constituents (8 wa- 2- - + - lufft, Austria). However, there were loads of missing val- ter soluble inorganic ions (WSIIs) (SO4 , NO3 , NH4 , Cl , K+, Na+, Ca2+, Mg2+) and 17 elements (Pb, Se, Cr, Cd, Zn, ues among the precipitation data. Thus we collected the daily precipitation of Wuhan from the Weather Under- Cu, Ni, Fe, Mn, Ti, Sb, Sn, V, Ba, As, Mo, Sc)) of PM2.5, and meteorological parameters (wind speed (WS), wind di- ground (http://www.wunderground.com). rection (WD), temperature (T), relative humidity (RH), pre- Finally for all the real-time monitoring items, there cipitation). were 600 samples totally collected during the whole study period. Gas pollutants including sulfur dioxide (SO2), nitrogen oxides (NOX) and carbon monoxide (CO) were measured 2.4 PMF every 60 min by a SO2, NOX and CO analyzer (Thermo Fisher Scientific Inc., Franklin, MA; Model 43i, Model 42i, The PMF model was developed by Paatero and Tapper Model 48i), respectively. The accuracy for SO2 and NOX an- [17]. The Environmental Protection Agency (EPA) PMF is alyzer both were 0.1 µg m-3, while 0.01 mg m-3 for CO. one of the receptor models that the US EPA’s Office of

Research and Development (ORD) has developed. Algo- health impact of pollutant exposure, the average amount of rithms used in the EPA PMF model to compute profiles pollutant exposure per individual’s body weight over a and contributions have been peer-reviewed by leading sci- given time span for the three sensitive groups (aged 2-6 entists in the air quality management community and cer- years and ≥70 years) was calculated [28] as: tified to be scientifically robust. Detailed algorithms were CIFD  introduced elsewhere [16, 17]. In this study, the data in- DE  (4) cluded 8 water soluble inorganic ions, 17 elements. The tW EPA PMF program input data included the concentration where the terms are DE: dose of exposure (mg/kg- of each sample data and the uncertainty of the data which day); C: mean concentrations (mg/m3); I: inhalation rate can be calculated as below: (m3/day); F: exposure frequency (days/year); D: exposure UMDLcMDL2(  ) (1) duration (years); t: average time; and W: body weight (kg). The elemental risk (R) was calculated by Eq. (5). Here RD UPcMDLcMDL()()(22  ) (2) is the reference dose [28]. R  DE (5) Where U is the data uncertainty; P is expressed as a RD percentage of the measurement uncertainty, and P values for all species are set to 10% estimated by experience; c is the concentration of the sample; MDL is the minimum de- 3. RESULTS AND DISCUSSION tection limit of the instruments, calculation method of which has been described above in detail. 3.1 Particle pollution levels

2.5 K-mean cluster analysis The time series of meteorological conditions and various aerosol parameters are shown in Fig. 2. During the The choice of K-mean clustering was made from a se- whole period, concentrations of PM ranged from 7 μg m-3 lection of partitional cluster packages [24]. The mathemat- 2.5 to 672 μg m-3 with an average value of 90 μg m-3. Caused ical details of the method presented are available elsewhere by large-scale firework burning, the maximum PM mass [24, 25]. Briefly, K-means method aims to minimize the 2.5 loading (672 μg m-3) occurred on the Chinese New Year’s sum of squared distances between all points and the cluster Eve (10 February) with a high average 24 h concentration of center. In order to choose the optimum number of cluster 338 μg m-3. For the same reason, extremely high hourly con- the Dunn-Index (DI) was used, which aims to identify centration (285 μg m-3) and daily concentration (177 μg m-3) dense and well-separated clusters. DI is defined as the ratio also happened on the Lunar Lantern Festival. Although the between the minimal intercluster distances to maximal in- peak value on the New Year’s Eve was much higher than tra-cluster distance. In other words, for Dunn’s index we that on the Lunar Lantern Festival, the lasting hour for high wanted to find the clustering which maximizes this index. PM value was pretty shorter on the New Year’s Eve. This The use of cluster analysis was justified in this work using 2.5 phenomenon suggested that the firework-burning pattern a Cluster Tendency test, providing a calculated Hopkins was more concentrated and the duration for large scale Index of 0.2 and implying the presence of structures in the burning was shorter on the New Year’s Eve than that on form of cluster in a dataset [24]. The Dunn-Index for the the Lunar Lantern Festival. Moreover, there was also a peak results of the K-means analysis for different cluster num- value (279 μg m-3) before the holiday, which was mainly ex- bers showed a clear maximum for 13. However, by care- plained by increasing traffic flow to meet the rising demands fully looking at the clusters, these were reduced to 9 as the for transportation resources [29]. Similar high PM concen- difference among some of them was minimal. 2.5 trations during the Spring Festival were also reported in Bei-

jing (224 μg m-3) [13], Shanghai (382 μg m-3) [30], Jinan 2.6 Health risk assessment of heavy metal in PM2.5 (464 μg m-3) [31] and Xi’an (299 μg m-3) [14], especially on In this study, representative heavy metal elemental the New Year’s Eve. All these PM2.5 concentrations far ex- components (Cu, Zn, Ni, Pb, As, Cd, Cr, Co, Mn) of PM2.5 ceeded the China National Ambient Air Quality Standards were applied to calculate elemental risk levels to assess the -3 for 24 h PM2.5 (75 μg m ), which fairly evidences the wide- possible impacts on human health. spread and serious air pollution in China during the Spring Based on the human health evaluation manual of US Festival. As it is nearly impossible to cut human activities, EPA [26] and the experimental data in this study, the fol- such as setting off fireworks and going home to have a lowing equation was used to obtain the 95% upper confi- family reunion through public traffic systems, to a low dence limit (UCL) of the arithmetic mean concentration: level during the Spring Festival, understanding more about tS emission sources during the period is thus crucial to the 95%,n 1 improvement of the air quality in China. UCL195%  X (3) n According to the known human activity habits during where X is the mean concentration, S is the standard the Spring Festival and temporal distribution of air pollu- deviation, t is the Student’s t value, which can be found in tants (Fig. 2), the whole study period could be divided into Gilbert [27], and n is the sample size. To evaluate the three phases: the pre-holiday phase (1-9 February, phase

1), the holiday phase (10-16 February, phase 2) and the According to Table 1, sulfate, nitrate and ammonium post-holiday phase (17-25 February, phase 3). Concentra- (SNA) were the three main ions no matter when in phases 1, tions of PM2.5 and its chemical species in these phases are 2 and 3, and their average concentrations during the whole -3 -3 -3 presented in Table 1. The PM2.5 concentration in these period were 15.3 μg m , 11.5 μg m , and 8.1 μg m , re- -3 three phases followed the order: phase 2 (115 μg m ) > spectively. Percentages of SNA mass in PM2.5 during phase 3 (92 μg m-3) > phase 1 (69 μg m-3). The similar tem- phases 1, 2 and 3 were, respectively, 49.5%, 41.1% and poral distributions of PM2.5 concentration were also re- 47.0%, following a completely opposite order as PM2.5 : ported by Huang et al. [29] and Feng et al. [30] in Shang- phase 1 > phase 3 > phase 2. hai. Obviously, widespread firework-burning and high traf- According to the relevant literatures [29, 36], SO 2- fic volume were the main reasons for the highest PM con- 4 2.5 and NO - are mainly neutralized by NH +. As shown in Fig. centration in phase 2. In addition, low wind speed and stable 3 4 3, all the slopes of linear regression equation between NH + atmosphere (Table 1) in phase 2 favored the accumulation 4 versus SO 2- and NO - in phases 1, 2 and 3 were just a little of air pollutants [32, 33]. There were two main reasons for 4 3 higher or lower than 1.0, indicating that SO 2- and NO - the higher PM concentration in phase 3 than that in phase 4 3 2.5 were just completely neutralized by NH + during the 1: (1) more concentrated high traffic flow in the Spring 4 phases and existed in the form of (NH ) SO and NH NO . Festival return travel rush (phase 3) led to more motor 4 2 4 4 3 - vehicle exhaust in a short period; (2) more rainfall (4 mm Previous studies demonstrated that the ratio of NO3 -1 -1 2- h ) and stronger wind (8.84 m s ) in phase 1 were helpful /SO4 could be reasonably used to evaluate the influence for the diffusion of air pollutants. of mobile versus stationary pollution sources [37, 38]. The - 2- ratios of NO3 /SO4 (0.71~0.91) in Wuhan during the 3.2 WSIIs Spring Festival were fairly close to those in Guangzhou Total WSIIs mass concentrations in PM2.5 varied from (0.79) [39], Shanghai (0.64) [32], and Beijing (0.71) [40], 0.4 μg m-3 to 227.4 μg m-3 with a mean value of 37.9 μg m-3. whose traffic densities were much higher than those in Wu- - 2- Previous studies showed that WSIIs accounted for one-third han. The ratios of NO3 /SO4 in phases 1 and 3 were higher - or more of the fine particles in Chinese urban aerosols [34- than those in phase 2. On the whole, the high ratio of NO3 2- 36]. In this study, WSIIs accounted for 46.3% of PM2.5 mass /SO4 in this study could be explained by the extremely during the whole Spring Festival, implying that WSIIs might high traffic flow and large-scale shutting-down of plants play an important role in PM2.5. during the study period, especially in phases 1 and 3.

FIGURE 2 - Time series of meteorological conditions and various aerosol parameters with time resolution of 1 h from 1 to 25 February 2013 in Wuhan. (a) wind speed and direction (b) relatively humidity (RH), temperature and precipitation (c) PM2.5 mass concentration during the whole study period (d) PM2.5 mass concentration on the Chinese Lunar New Year’s Eve (9 February) and the Chinese Lunar New Year’s Day (10 February)

TABLE 1 - Concentrations of PM2.5 and its compositions in phases 1, 2 and 3

Phase 1 Phase 2 Phase 3 Composition Mean / μg m-3 Range Mean / μg m-3 Range Mean / μg m-3 Range

PM2.5 69 7~279 115 44~672 92 15~285 WS* (m/s) 1.68 0.01~8.84 1.02 0.05~3.26 1.44 0.04~5.32 SO2 22.9 4.38~62.53 46.8 10.67~241.05 36.8 3.61~110.23 NO2 41.1 8.96~105.78 37.4 8.38~95.07 48.5 5.08~116.90 2- SO4 11.43 2.04~50.03 16.68 0.26~26.46 15.74 0.24~51.82 - NO3 8.91 1.59~25.46 11.78 0.23~27.41 12.33 0.22~41.74 - 2+ NO3 / SO4 0.91 0.23~1.74 0.71 0.30~1.26 0.77 0.17~1.39 SOR 0.35 0.13 ~0.69 0.32 0.11~0.61 0.32 0.14~0.80 NOR 0.16 0.02~0.36 0.23 0.02~0.54 0.16 0.02~0.51 + NH4 7.18 1.60~16.45 9.60 0.15~16.60 7.98 0.11~27.32 Cl- 1.93 0.13~13.70 1.79 0.20~82.80 2.21 0.23~20.61 K+ 1.24 0.13~23.02 3.14 0.28~75.90 1.55 0.12~10.48 Ca2+ 0.29 0.06~1.15 0.24 0.10~0.66 0.46 0.09~1.05 Mg2+ 0.14 0.08~0.39 0.28 0.08~0.68 0.37 0.07~2.48 Total ions 31.44 6.55~227.40 43.44 0.96~80.69 40.10 0.35~119.91 Fe 1.566 0.076~6.200 2.364 0.373~8.802 2.612 0.051~8.813 Pb 0.203 0.019~1.871 0.321 0.062~1.497 0.258 0.031~0.923 Zn 0.173 0.009~0.687 0.227 0.041~1.431 0.253 0.006~0.931 Sn 0.158 0.012~0.431 0.127 0.004~0.351 0.112 0.006~0.327 Cu 0.141 0.004~0.572 0.191 0.001~0.796 0.156 0.01~0.658 Ba 0.123 0.002~1.870 0.431 0.018~4.372 0.260 0.009~7.876 Sb 0.105 0.001~0.440 0.081 0.003~0.368 0.060 0.004~0.222 Ti 0.090 0.003~0.290 0.141 0.01~0.928 0.150 0.007~0.449 Cd 0.084 0.001~0.223 0.075 0.003~0.239 0.063 0.001~0.225 Mo 0.078 0.001~0.261 0.071 0.005~0.194 0.064 0.001~0.217 Mn 0.058 0.002~0.214 0.083 0.017~0.616 0.091 0.006~0.309 Sc 0.037 0.002~0.125 0.039 0.004~0.121 0.054 0.001~0.149 Ni 0.025 0.003~0.054 0.029 0.009~0.085 0.030 0.005~0.058 As 0.022 0.001~0.174 0.032 0.001~0.280 0.034 0.001~0.420 Se 0.019 0.001~0.046 0.022 0.001~0.064 0.019 0.001~0.057 Cr 0.018 0.002~0.051 0.024 0.003~0.228 0.024 0.002~0.087 V 0.004 0.001~0.032 0.007 0.001~0.082 0.006 0.001~0.023 Co 0.002 0..001~0.005 0.003 0..001~0.005 0.002 0..001~0.005 Total elements 2.904 0.525~9.426 4.264 1.194~19.623 4.247 0.658~11.817

+ - 2- FIGURE 3 - Correlations between NH4 versus NO3 and SO4 . Linear regression equations are given on the top

Moreover, the nitrogen oxidation ratio (defined as NOR mainly happened from 12:00 to 16:00, when the UV - - NOR = NO3 / (NO3 +NO2)) and the sulfur oxidation ratio intensity reached the highest during a day. When SOR is 2- 2- (defined as SOR = SO4 / (SO4 +SO2)) could be used to higher than 0.1, photochemical reactions of SO2 might estimate secondary transformation processes from NO2 to have occurred in the atmosphere [42]. The SORs of PM2.5 - 2- NO3 and SO2 to SO4 , respectively [41]. Generally, NOR was in the phases were higher than 0.1, suggesting that the for- lower than SOR. In this study, average SOR (0.32~0.35) and mation of secondary pollutants from SO2 occurred in the NOR (0.16~0.23) in the phases were relatively close to whole study period, especially in phase 1. each other, as shown in Table 1. High hourly SOR and

As for K+ and Cl-, their concentrations in the whole study Sn > Cu > Ba > Sb > Ti > Cd > Mo > Mn > Sc > Ni > As period were, respectively, 1.88 μg m-3 and 1.99 μg m-3, which > Se > Cr > V. In this study, Fe is the most abundant ele- were just lower than those of SNA. It was noted that K+ mass ment with an average concentration of 2.17 μg m-3, which in phase 2 was much higher than that in phases 1 and 3, and was much higher than that of other PM2.5 elements. Despite coefficients of variation (106.1% ~ 495.1%) for these two the fact that there were no other crust elements (Si, Al, Ca ions were rather higher than other PM2.5 components, and Mg) measured in this study, it was still easy to know which indicates the huge variations of K+ and Cl- mass dur- crust elements played a comparatively important role in ing the study period. KNO3, KClO3, and KClO4 were often PM2.5 elements [44, 45]. widely used as the oxidizer in order to sustain the burning As for the heavy metal elements whose annual stand- of firecrackers [43]. Thus we could deduce that the large ard was given by the 2012 Chinese ambient air quality scale and time-concentrated firework-burning caused these standard, Pb concentrations (0.203 μg m-3 ~ 0.321 μg m-3) temporal variations for K+ and Cl-. Interestingly, the ratio - + for all these three phases were lower than the standard of Cl /K in phase 2 was 0.6, which was exactly equal to value (0.5 μg m-3). Promoting the use of lead-free gasoline the value of firework-burning aerosol reported by Shen et in China might be one of the reasons for meeting the stand- al. [14]. ard for Pb. As and Cd almost exceeded the yearly standard As for Ca2+ and Mg2+, their mass loadings for the whole value by about 2~16 folds, which should get much atten- period were, respectively, 0.34 μg m-3 and 0.26 μg m-3, tion from the authorities. Therefore, investigation into the slightly lower than those of Cl- and K+. The Ca2+ concentra- measures for figuring out the sources of these heavy metal tion followed the pattern: phase 3 > phase 1 > phase 2, while elements and reducing their concentrations during the for Mg2+, it was phase 3 > phase 2 > phase 1. Usually, Ca2+ Spring Festival seemed particularly urgent. and Mg2+ were the tracer for construction source. But due to the fact that most of construction sites in Wuhan were 3.4 Source apportionment of PM2.5 by PMF and K-mean cluster all shut down during the Spring Festival, the highest Ca2+ analysis 2+ and Mg concentrations in phase 3 should be explained by In this study, 25 species including PM2.5 were used as road dust caused by high traffic volume, which also re- input data for PMF model. PMF was run multiple times, sulted in the highest NO2 concentration in phase 3. In ad- while varying the number of sources (from 1 to 9), boot- dition, Mg is also one of raw materials for fireworks [43], strap (seed: random; minimum correlation R-value: 0.6; and enormous firework burning in phase 2 caused higher block size: 20) and fpeak (0, 0.1, 0.2, 0.3, 0.4, 0.5), which value for Mg2+ than that in phase 1. affects the rotation of matrices for source compositions and contributions. Seven sources were resolved with a resultant 3.3 Elements fpeak of 0. Under the solution, residuals of the majority of All 17 analyzed elements comprised about 4.2% of standardized species were between -3 and +3. G-space plot PM2.5 mass, and all elements together have an average shows data points lying within the source axes. Twenty -3 -3 concentration of 3.77 μg m ranging from 0.53 μg m to runs were made for each factor and the Qrobust is 7383.1 -3 19.62 μg m . Their average concentrations in the whole with the Qrobust / Qtrue ratios of 0.97. Note that Qrobust / Qtrue study period followed a decreasing order: Fe > Pb > Zn > < 2 indicates an overall acceptable fit [46]. A comparison

350 ) -3 300 y = 0.9445x + 1.2005 g m μ 250 R² = 0.9742

200

150

100

0 Predicted concentration ( concentration Predicted 0 50 100 150 200 250 300

Observed concentration (μg m-3)

FIGURE 4 - Predicted and observed PM2.5 mass concentration in Wuhan

FIGURE 5 - Profiles of factors from base run and bootstrap runs

of the daily calculated PM2.5 mass concentration from wear particles from Cd-bearing alloys are also emitted [47, seven sources with the measured mass concentration is 48]. In addition, the highest values of Mo (0.082 μg m-3) shown in Fig. 4. The squared correlation coefficient (R2) of and Cd (0.088 μg m-3) were distributed into cluster 3 with 0.9742, a slope of 0.9445 and an intercept of 1.2005 indi- high values of CO and NOX according to Table 2. The con- cated that the reconstructed mass concentrations were contribution of motor vehicle source (25.0%) (Table 3) during sistent with the measured mass concentrations, and the re- this special period was much higher than the annual contri- solved seven sources profiles were shown in Fig. 5. butions of the same source in Beijing (5.6%~11.5%) [49, In order to accurately figure out source types repre- 50], Zhengzhou (10%) [45] and Jinan (6.06%) [48]. sented by each factor and dig out more information about the Coal combustion source had high concentrations of emission sources, K-means cluster analysis was also per- 2- - - SO4 , NO3 , Cl , Se, Ba, Cr, Ti, Fe, Mn, Pb mixing with formed with the data of the resolved sources’ tracer species, medium concentrations of As. High concentrations of Cl-, gas pollutants and meteorological parameters. Through Se and Cr were probably associated with coal burning [21]. modifying the cluster number from 6 to 15 and checking the According to the cluster analysis, additionally, high con- Dunn Index, we got nine reasonable clusters. The average centrations of Ti (0.096 μg m-3), Ba (0.35 μg m-3) and Pb values associated to each of the nine clusters were pre- (0.51 μg m-3) were put into cluster 1 together with high sented in Table 2. And source contribution estimates of -3 concentration of SO2 (41 μg m ), and low concentration of PM2.5 were calculated for each of the samples from the cor- CO (1783 μg m-3) (Table 2). The concentration of Pb in respondingly modeled species’ absolute mass (Table 3). cluster 1 was higher than that in cluster 3 indicating that Also the results of Bootstrap run are shown in Table 3 with coal burning was the main source of aerosol Pb in Wuhan similar source contribution estimates. (Table 2). Similar results were also reported in other cities Motor vehicle source was characterized by the high [51]. However, Widory et al. [52] argued that in Beijing, - loadings of NO3 , V, Sb, Pb, Zn, Mo and Cd. Zn is emitted metal-refining plants are the dominant sources of aerosol from lubrication oil, brake linings and tyres, whereas metal Pb, followed by thermal power stations and other coal com-

bustion sources. As shown in Table 3, coal combustion was one of the enterprises with the largest air pollutant (SO2, the second major source of PM2.5 with a contribution of NO2, PM2.5, PM10) emission in Wuhan. Therefore it was 21.6%, which could be explained by high ratio of coal easy to understand the high contribution (15.5%) of steel burning in the energy consumption in China [21, 53]. Coal works to PM2.5 mass. Additionally, the good consistency be- consumption in Wuhan was mainly for power generation tween the dominant wind direction of cluster 9 and the di- and steel smelting. Based on Fig. 6, one of the predominant rection of local steel works indicated that the resolved steel wind directions for cluster 1 was north-north-east (NNE), work source was mainly local, not regional. just the direction of the two biggest coal-fired power plants Secondary inorganic source showed not only high load- with more than 1 million-kilowatt of installed capacity 2- + - ings for SO4 , NH4 and NO3 , but also medium loadings for (Fig. 1). Cl-, K+, V, Fe, As, Pb and Zn. All these elements were also Steel works source presented high loading for Fe, Mn, tracers of other sources such as motor vehicle emission, coal As and medium loadings for Sc. Meanwhile, highest load- burning, steel works and firework burning (Fig. 5). This re- -3 -3 ings of Fe (6.21 μg m ) and Sc (0.077 μg m ) were asso- sult suggested that SO2 and NO2 emitted by these sources par- ciated with medium concentration of SO2 in cluster 9 (Ta- ticipated in these secondary reactions. As shown in Table 3, ble 2). Moreover, the absolute dominant winds of the wind secondary inorganic source contributed up to 14.8%. How- rose of cluster 9 (Fig. 6) were north-north-east (NNE) / ever in Beijing, Song et al. [21] reported a higher value (31%) north-east (NE) sides, exactly where a number of steel of secondary inorganic source for the winter in 2000-2001, works belonging to Wuhan Iron and Steel (Group) Corpo- and Zhang et al. [54] gave a lower value (6%) for the winter ration were located. According to the Wuhan City ambient in 2009-2010. The difference could be partly explained by -3 air quality bulletins, Wuhan Iron and Steel (Group) Corp is the different concentration levels of SO2 (2001: 71 μg m ;

TABLE 2 - Average values associated to each of the nine clusters. Bold texts are for particularly high values (Units for all the air pollutants: µg m−3)

Steel Cluster Type Coal combustion Fireworks Motor vehicle Dilution Glass plants Road dust Glass/Cement plants Dilution Plants Cluster number 1 2 3 4 5 6 7 8 9 WS 0.63 0.86 0.60 3.66 1.10 0.73 1.39 3.55 0.23 (m s-1) SO2 41 80 37 14 31 31 23 21 38 CO 1783 2388 3126 1891 2065 2303 2136 1831 2729 NO2 118 62 100 25 36 49 51 25 129 Cl- 1.82 3.33 1.75 1.17 1.83 1.70 1.21 1.25 1.20 K+ 1.49 3.52 1.13 1.28 1.81 1.73 0.76 1.66 0.46 Ca2+ 0.48 0.28 0.42 0.18 0.66 0.68 0.60 0.33 0.56 Mg2+ 0.26 0.35 0.24 0.24 0.28 0.55 0.25 0.20 0.45 Pb 0.51 0.27 0.32 0.10 0.22 0.28 0.23 0.09 0.34 Cr 0.03 0.02 0.05 0.03 0.04 0.02 0.01 0.02 0.04 Cd 0.07 0.08 0.09 0.06 0.07 0.08 0.06 0.06 0.07 Zn 0.14 0.19 0.30 0.09 0.11 0.21 0.10 0.11 0.31 Cu 0.16 0.15 0.57 0.17 0.18 0.17 0.12 0.15 0.28 Ni 0.03 0.03 0.02 0.03 0.04 0.03 0.02 0.03 0.04 Fe 1.74 5.30 1.26 0.20 3.54 2.11 3.69 0.15 6.21 Mn 0.09 0.09 0.08 0.10 0.12 0.07 0.04 0.07 0.12 Ti 0.10 0.12 0.14 0.10 0.11 0.10 0.12 0.09 0.11 Ba 0.35 0.32 0.26 0.22 0.23 0.25 0.19 0.19 0.19 As 0.06 0.04 0.02 0.04 0.03 0.02 0.02 0.02 0.03 Mo 0.07 0.08 0.08 0.07 0.07 0.08 0.06 0.05 0.06 Sc 0.05 0.07 0.06 0.03 0.04 0.04 0.04 0.03 0.08 Co 0.002 0.003 0.002 0.002 0.003 0.002 0.002 0.003 0.002

-3 Table 3 PM2.5 source contribution estimates (SCE) (μg m and %) for PM2.5

Base run Bootstrap run Factor Source name μg m-3 % μg m-3 % Factor 1 Motor vehicle emission 20.7 25.0% 21 26.0% Factor 2 Coal combustion 17.9 21.6% 18.2 22.5% Factor 3 Steel works 12.8 15.5% 11.6 14.4% Factor 4 Secondary inorganic source 12.2 14.8% 11.3 14.0% Factor 5 Firework burning 9 10.9% 8.6 10.6% Factor 6 Glass/Cement plants 5.4 6.5% 5.2 6.4% Factor 7 Road dust 4.7 5.7% 4.9 6.1%

FIGURE 6 - Wind rose maps associated with each cluster

-3 -3 2010: 32 μg m ) and NO2 (2001: 71 μg m ; 2010: 57 μg Sn, Ti, Sc and Pb. CaCO3 was one kind of raw material for m-3) in the atmosphere between 2000 and 2010 [55]. cement production, whereas CaO for glass production. - Moreover, CuO, FeO, Fe O , MnO , CaF , etc. were nec- Factor 5 was composed of high mass fractions of NO3 , 2 3 2 2 K+ and Cl-, mixing with medium loadings for Mg2+, Ba and essary additives for producing colored glass [57]. And rel- -3 Cu. If simply based on the PMF profile, we could make a atively high values of Ca (0.60~0.66 μg m ) and Fe -3 mistake to attribute this factor to biomass burning [56]. Re- (3.54~3.69 μg m ) occurred in clusters 5 and 7 (Table 2). sults of K-mean cluster analysis showed that all the mem- According to Fig. 6, predominant wind directions of clusters + 5 and 7 were west-north-west (WNW) and south-west (SW), bers of cluster 3 with highest concentrations of SO2, K and Cl- occurred on the Chinese Lunar New Year’s Eve and respectively, where plenty of glass/cement plants lie as Lantern Festival when people set off fireworks on the larg- shown in Fig. 1. Therefore this factor was attributed to glass/ cement plants, which contributed 6.5% to PM mass. Fur- est scale. Additionally, KNO3, KClO3, BaNO3, Mg, etc 2.5 were used as raw materials in firework powders. Therefore thermore, like steel works source, the agreement between the this factor should be recognized as firework burning. And dominant wind direction and local glass/cement sources’ direction also suggested that this resolved glass/cement plants firework burning source contributed 10.9% to the PM2.5 mass during the study period (Table 3). were a local source. Factor 6 was characterized by high loadings of Ca2+, Factor 7 had high loadings for Ca2+ and Mg2+, and me- Cu, Fe and Mn mixing with other element species, such as dium loadings for Mn, Fe, Pb and Zn. Ca, Mg and Fe were

elements of soil dust [58]. Ca2+ was usually the tracer ele- vidual element in it. In addition, children were the more ment of construction dust [53]. As almost all the construc- sensitive group to develop non-carcinogenic effects, and tion activities were shut down during the study period, this avoiding possible exposure to these contaminant elements factor should be road dust. The medium loading of those is important. Therefore, it is critical for the government to elements indicated that road dust was mixed with steel adopt effective measures to control PM2.5 pollution and to smelting dust, motor vehicle dust and others. Non-predom- improve urban air quality, especially in the important fes- inant wind direction of cluster 6, of which Ca2+ and Mg2+ tivals. concentrations were the highest, was consistent with the truth that this site was surrounded by roads. Road dust accounted for 5.7% of PM2.5 mass during the study period in 4. CONCLUSIONS Wuhan (Table 3). A real-time air pollution monitoring campaign was 3.5 Assessment of health risk from heavy metal carried out at a semi-residential site in Wuhan from 1 to In this study, nine major metals (Cu, Pb, Cr, Co, Mn, 25 February, 2013, covering the Chinese Spring Festival. Zn, Cd, Ni and As) were used to assess the possible non- During the whole period, concentrations of PM2.5 var- carcinogenic human health risks during the Spring Festival ied from 7 μg m-3 to 672 μg m-3 with an average value of -3 2- - pollution episode. The non-carcinogenic risk assessment 90 μg m . Eight water soluble inorganic ions (SO4 , NO3 + - + + 2+ 2+ results for children (≤6) and elderly (≥70) during the whole , NH4 , Cl , K , Na , Ca , Mg ) and 17 elements (Pb, Se, period, are listed in Table 4. Cr, Cd, Zn, Cu, Ni, Fe, Mn, Ti, Sb, Sn, V, Ba, As, Mo, Sc), respectively, explained 46.3% and 4.2% of PM2.5 mass. TABLE 4 - Health risk estimated from heavy metal elemental com- The whole study period was divided into three phases: ponents of PM2.5 the pre-holiday phase (phase 1), the holiday phase (phase 2) Whole and the post-holiday phase (phase 3). PM , SO , SO 2- and Element 2.5 2 4 Child(≤6) Adult(≥70) K+ presented the highest concentration in phase 2 because Pb 0.7814 0.1197 - 2+ 2+ of firework burning, whereas NO2, NO3 , Ca , and Mg in Zn 0.0076 0.0011 Cu 0.0508 0.0102 phase 3 because of motor vehicle emission. Cd 0.8701 0.1753 The source apportionment by PMF and K-mean anal- Mn 0.0203 0.0049 ysis resolved 7 sources for PM mass: motor vehicle emis- Ni 0.0164 0.0033 2.5 As 1.0042 0.1362 sion (20.7%), coal burning (17.9%), steel works (13.8%), Cr 0.0516 0.0105 secondary inorganic source (13.2%), firework burning Co 0.0013 0.0002 (9.2%), glass/cement plants (9.0%), and road dust (5.4%). Total 2.8037 0.4614 In addition, the coal burning was the main source of aerosol Pb in Wuhan during the study period. And the resolved Based on the study [59], elemental risk level higher steel works and glass/cement plants mainly referred to lo- than 0.1 has adverse health effects on children. As shown cal emission sources. in Table 4, the metals in PM with higher risk value than 2.5 The results of health risk assessment from the heavy 0.1 were As (1.042), Cd (0.8701), and Pb (0.7814) for the metal elements in PM indicated that As, Pb and Cd were whole period in Wuhan. The values of high-risk-level ele- 2.5 the three main metals in PM , which could cause non-car- ments were rather higher than their values in Nanjing [60] 2.5 cinogenic effects. The health impact of the combination of and Jinan [48]. Hu et al. [59] reported that As (0.414), Pb all the heavy metal elements in PM would be more seri- (0.173), Cr (0.109) and Co (1.13) were the metal elements 2.5 ous both for children (≤ 6) and the elderly (≥70). In addi- with high risk value in Nanjing, whereas Yang et al. [48] tion, the children were the more sensitive group than the reported that Cr (0.168), Mn (0.481) and Co (0.463) were elderly to develop non-carcinogenic effects, and avoiding the metal elements in Jinan. Different sampling periods were possible exposure to these contaminant elements is im- the main reason for the higher values. Our sampling period portant. covered almost the most serious air pollution episode, the Chinese Spring Festival pollution episode, but He et al. [60] and Yang et al. [48] collected the annual samples. The elemental risk levels for other metals were much ACKNOWLEDGEMENTS lower than 0.1, especially for Co (0.0002-0.0013). How- ever, the elemental risk according to the impact of heavy This work was supported by Hubei Key Laboratory of metals on health could be cumulative. The total elemental Industrial Fume & Dust Pollution Control, Jianghan Uni- risk for the two groups was much larger than 0.1 with a versity, Wuhan, China (No: HBIK2012-2), and Scientific value of 2.8037 for children (≤6 years), a much lower value Program of Hubei Environmental Protection Bureau (No: of 0.4614 for the elderly (≥70), indicating that the health 2012HB05). impact of the combination of all the heavy metal elements in PM2.5 would be more serious than that of a single indi- The authors have declared no conflict of interest.

REFERENCES [18] Bullock, K.R., Duvall, R.M., Norris, G.A., Mcdow, S.R. and Hays, M.D. (2008) Evaluation of the CMB and PMF models using organic molecular markers in fine particulate matter collected during [1] Guo, Y., Jia, Y., Pan, X., Liu, L. and Wichmann, H. (2009) The as- the Pittsburgh Air Quality Study. Atmospheric Environment. sociation between fine particulate air pollution and hospital emer- 42(29): 6897-6904. gency room visits for cardiovascular diseases in Beijing, China. Sci- ence of the Total Environment. 407(17): 4826-4830. [19] Dutton, S.J., Vedal, S., Piedrahita, R., Milford, J.B., Miller, S.L. and Hannigan, M.P. (2010) Source apportionment using positive [2] Guo, Y., Tong, S., Zhang, Y., Barnett, A.G., Jia, Y. and Pan, X. matrix factorization on daily measurements of inorganic and or- (2010) The relationship between particulate air pollution and ganic speciated PM2.5. Atmospheric Environment. 44(23): 2731- emergency hospital visits for hypertension in Beijing, China. Sci- 2741. ence of the Total Environment. 408(20): 4446-4450. [20] Heo, J., Dulger, M., Olson, M.R., Mcginnis, J.E., Shelton, B.R., [3] Halonen, J.I., Lanki, T., Yli-Tuomi, T., Tiittanen, P., Kulmala, M. Matsunaga, A., Sioutas, C. and Schauer, J.J. (2013) Source appor- and Pekkanen, J. (2009) Particulate air pollution and acute cardi- tionments of PM2.5 organic carbon using molecular marker Posi- orespiratory hospital admissions and mortality among the elderly. tive Matrix Factorization and comparison of results from different Epidemiology. 20(1): 143-153. receptor models. Atmospheric Environment. 73(0): 51-61. [4] Simkhovich, B.Z., Kleinman, M.T. and Kloner, R.A. (2009) Par- [21] Song, Y., Zhang, Y., Xie, S., Zeng, L., Zheng, M., Salmon, L.G., ticulate air pollution and coronary heart disease. Current opinion Shao, M. and Slanina, S. (2006) Source apportionment of PM2.5 in in cardiology. 24(6): 604-609. Beijing by positive matrix factorization. Atmospheric Environ- ment. 40(8): 1526-1537. [5] Le Tertre, A., Medina, S., Samoli, E., Forsberg, B., Michelozzi, P., Boumghar, A., Vonk, J.M., Bellini, A., Atkinson, R. and Ayres, J.G. [22] Tan, J., Duan, J., Ma, Y., Yang, F., Cheng, Y., He, K., Yu, Y. and (2002) Short-term effects of particulate air pollution on cardiovascu- Wang, J. (2014) Source of atmospheric heavy metals in winter in lar diseases in eight European cities. Journal of epidemiology and Foshan, China. Science of the Total Environment. 493(0): 262- community health. 56(10): 773-779. 270. [6] Schwartz, J. (1996) Air pollution and hospital admissions for res- [23] Zhang, Y., Sheesley, R.J., Schauer, J.J., Lewandowski, M., Jaoui, piratory disease. Epidemiology. 7(1): 20-28. M., Offenberg, J.H., Kleindienst, T.E. and Edney, E.O. (2009) Source apportionment of primary and secondary organic aerosols us- [7] Arvanitis, A., Kotzias, D., Kephalopoulos, S., Carrer, P., Cavallo, ing positive matrix factorization (PMF) of molecular markers. At- D., Cesaroni, G., De Brouwere, K., de Oliveira-Fernandes, E., For- mospheric Environment. 43(34): 5567-5574. astiere, F. and Fossati, S. (2010) The INDEX-PM project: Health risks from exposure to indoor particulate matter. Fresenius Envi- [24] Beddows, D.C., Dall Osto, M. and Harrison, R.M. (2009) Cluster ronmental Bulletin. 19(11): 2458-2471. analysis of rural, urban, and curbside atmospheric particle size data. Environmental science & technology. 43(13): 4694-4700. [8] Seleventi, M.K., Saraga, D.E., Helmis, C.G., Bairachtari, K., Va- [25] Dall'Osto, M., Monahan, C., Greaney, R., Beddows, D., Harrison, silakos, C. and Maggos, T. (2012) PM indoor/outdoor relation- 2.5 R.M., Ceburnis, D. and O'Dowd, C.D. (2011) A statistical analysis ship and chemical composition in ions and OC/EC in an apartment of North East Atlantic (submicron) aerosol size distributions. At- in the center of Athens. Fresenius Environmental Bulletin. 21(11): mospheric Chemistry and Physics. 11(24): 12567-12578. 3177-3183. [26] US EPA (United States Environmental Protection Agency). Sup- [9] Tiwari, S., Chate, D.M., Srivastava, M.K., Safai, P.D., Srivastava, plemental guidance for developing soil screening levels for super- A.K., Bisht, D.S. and Padmanabhamurty, B. (2012) Statistical fund sites. OSWER 9355.4-24. Office of Solid Waste and Emer- evaluation of PM10 and distribution of PM1, PM2.5, and PM10 in gency Response; 2001. ambient air due to extreme fireworks episodes (Deepawali festi- vals) in megacity Delhi. Natural hazards. 61(2): 521-531. [27] Gilbert, R.O. Statistical methods for environmental pollution monitoring. John Wiley & Sons, 1987. [10] Tsai, H., Chien, L., Yuan, C., Lin, Y., Jen, Y. and Ie, I. (2012) Influences of fireworks on chemical characteristics of atmospheric [28] US EPA (United States Environmental Protection Agency). Hu- fine and coarse particles during Taiwan's Lantern Festival. Atmos- man health evaluation manual. EPA/540/1-89/002 Risk assess- pheric Environment. 62: 256-264. ment guidance for superfund, vol. I. Office of Solid Waste and Emergency Response; 1989. [11] Licudine, J.A., Yee, H., Chang, W.L. and Whelen, A.C. (2012) Hazardous metals in ambient air due to New Year fireworks during [29] Huang, K., Zhuang, G., Lin, Y., Wang, Q., Fu, J.S., Zhang, R., Li, J., 2004–2011 celebrations in Pearl City, Hawaii. Public Health Re- Deng, C. and Fu, Q. (2012) Impact of anthropogenic emission on air ports. 127(4): 440. quality over a megacity–revealed from an intensive atmospheric campaign during the Chinese Spring Festival. Atmospheric Chemis- [12] Kuo, C., Lee, H. and Lai, J. (2006) Emission of polycyclic aro- try and Physics. 12(23): 11631-11645. matic hydrocarbons and lead during Chinese mid-autumn festival. [30] Feng, J., Sun, P., Hu, X., Zhao, W., Wu, M. and Fu, J. (2012) The Science of the total environment. 366(1): 233-241. chemical composition and sources of PM2.5 during the 2009 Chi- [13] Li, X., Guo, X., Liu, X., Liu, C., Zhang, S. and Wang, Y. (2009) nese New Year's holiday in Shanghai. Atmospheric Research. 118: Distribution and sources of solvent extractable organic compounds 435-444. in PM2.5 during 2007 Chinese Spring Festival in Beijing. Journal [31] Yang, L., Gao, X., Wang, X., Nie, W., Wang, J., Gao, R., Xu, P., of Environmental Sciences. 21(2): 142-149. Shou, Y., Zhang, Q. and Wang, W. (2014) Impacts of firecracker [14] Shen, Z., Cao, J., Arimoto, R., Han, Z., Zhang, R., Han, Y., Liu, burning on aerosol chemical characteristics and human health risk S., Okuda, T., Nakao, S. and Tanaka, S. (2009) Ionic composition levels during the Chinese New Year Celebration in Jinan, China.

of TSP and PM2.5 during dust storms and air pollution episodes at Science of the Total Environment. 476: 57-64. Xi'an, China. Atmospheric Environment. 43(18): 2911-2918. [32] Wang, Y., Zhuang, G., Zhang, X., Huang, K., Xu, C., Tang, A., [15] Shi, Y., Zhang, N., Gao, J., Li, X. and Cai, Y. (2011) Effect of Chen, J. and An, Z. (2006) The ion chemistry, seasonal cycle, and fireworks display on perchlorate in air aerosols during the Spring sources of PM2.5 and TSP aerosol in Shanghai. Atmospheric Envi- Festival. Atmospheric Environment. 45(6): 1323-1327. ronment. 40(16): 2935-2952. [33] Turk, Y.A. and Kavraz, M. (2010) The relation between some me- [16] Paatero, P. (2000) User’s guide for positive matrix factorization teorological factors and air pollutants in Trabzon city. Fresenius programs PMF2 and PMF3. Helsinki: University of Helsinki. Environmental Bulletin. 19(4 A): 721-729. [17] Paatero, P. and Tapper, U. (1994) Positive matrix factorization: A [34] He, K., Yang, F., Ma, Y., Zhang, Q., Yao, X., Chan, C.K., Cadle, non‐negative factor model with optimal utilization of error esti- S., Chan, T. and Mulawa, P. (2001) The characteristics of PM2.5 in mates of data values. Environmetrics. 5(2): 111-126. Beijing, China. Atmospheric Environment. 35(29): 4959-4970.

[35] Wang, G., Huang, L., Gao, S., Gao, S. and Wang, L. (2002) Char- [52] Widory, D., Liu, X. and Dong, S. (2010) Isotopes as tracers of acterization of water-soluble species of PM10 and PM2.5 aerosols sources of lead and strontium in aerosols (TSP & PM2.5) in Beijing. in urban area in Nanjing, China. Atmospheric Environment. 36(8): Atmospheric Environment. 44(30): 3679-3687. 1299-1307. [53] Zheng, M., Salmon, L.G., Schauer, J.J., Zeng, L., Kiang, C.S., [36] Hu, M., He, L., Zhang, Y., Wang, M., Pyo Kim, Y. and Moon, Zhang, Y. and Cass, G.R. (2005) Seasonal trends in PM2. 5 source K.C. (2002) Seasonal variation of ionic species in fine particles at contributions in Beijing, China. Atmospheric Environment. Qingdao, China. Atmospheric Environment. 36(38): 5853-5859. 39(22): 3967-3976. [37] Arimoto, R., Duce, R.A., Savoie, D.L., Prospero, J.M., Talbot, R., [54] Zhang, F., Xu, L., Chen, J., Chen, X., Niu, Z., Lei, T., Li, C. and Cullen, J.D., Tomza, U., Lewis, N.F. and Ray, B.J. (1996) Rela- Zhao, J. (2013) Chemical characteristics of PM2.5 during haze ep- tionships among aerosol constituents from Asia and the North Pa- isodes in the urban of Fuzhou, China. Particuology. 11(3): 264- cific during PEM‐West A. Journal of Geophysical Research: At- 272. mospheres (1984–2012). 101(D1): 2011-2023. [55] Ying, Q., Wu, L. and Zhang, H. (2014) Local and inter-regional [38] Xu, L., Chen, X., Chen, J., Zhang, F., He, C., Zhao, J. and Yin, L. contributions to PM2.5 nitrate and sulfate in China. Atmospheric (2012) Seasonal variations and chemical compositions of PM2.5 Environment. 94(0): 582-592. aerosol in the urban area of Fuzhou, China. Atmospheric Research. [56] Amodio, M., Andriani, E., Dambruoso, P.R., Daresta, B.E., de 104: 264-272. Gennaro, G., Di Gilio, A., Intini, M., Palmisani, J. and Tutino, M. [39] Tan, J., Duan, J., Chen, D., Wang, X., Guo, S., Bi, X., Sheng, G., (2012) Impact of biomass burning on PM10 concentrations. Frese- He, K. and Fu, J. (2009) Chemical characteristics of haze during nius environmental bulletin. 21(11A): 3296-3300. summer and winter in Guangzhou. Atmospheric Research. 94(2): [57] Bamford, C.R. Colour generation and control in glass. Elsevier 238-245. Amsterdam, 1977. [40] Wang, Y., Zhuang, G., Tang, A., Yuan, H., Sun, Y., Chen, S. and [58] Callén, M.S., De La Cruz, M.T., López, J.M., Navarro, M.V. and Zheng, A. (2005) The ion chemistry and the source of PM2.5 aero- Mastral, A.M. (2009) Comparison of receptor models for source sol in Beijing. Atmospheric Environment. 39(21): 3771-3784. apportionment of the PM10 in Zaragoza (Spain). Chemosphere. [41] Kaneyasu, N., Ohta, S. and Murao, N. (1995) Seasonal variation 76(8): 1120-1129. in the chemical composition of atmospheric aerosols and gaseous [59] De Miguel, E., Iribarren, I., Chacón, E., Ordoñez, A. and Charles- species in Sapporo, Japan. Atmospheric Environment. 29(13): worth, S. (2007) Risk-based evaluation of the exposure of children

1559-1568. to trace elements in playgrounds in Madrid (Spain). Chemosphere. [42] Pierson, W.R., Brachaczek, W.W. and Mckee, D.E. (1979) Sulfate 66(3): 505-513. emissions from catalyst-equipped automobiles on the highway. [60] Hu, X., Zhang, Y., Ding, Z., Wang, T., Lian, H., Sun, Y. and Wu, Journal of the Air Pollution Control Association. 29(3): 255-257. J. (2012) Bioaccessibility and health risk of arsenic and heavy met- [43] Wang, Y., Zhuang, G., Xu, C. and An, Z. (2007) The air pollution als (Cd, Co, Cr, Cu, Ni, Pb, Zn and Mn) in TSP and PM2.5 in Nan- caused by the burning of fireworks during the lantern festival in jing, China. Atmospheric Environment. 57(0): 146-152. Beijing. Atmospheric Environment. 41(2): 417-431. [44] Tao, J., Zhang, L., Engling, G., Zhang, R., Yang, Y., Cao, J., Zhu, C., Wang, Q. and Luo, L. (2013) Chemical composition of PM2.5 in an urban environment in Chengdu, China: Importance of spring- time dust storms and biomass burning. Atmospheric Research. 122(0): 270-283. [45] Geng, N., Wang, J., Xu, Y., Zhang, W., Chen, C. and Zhang, R.

(2013) PM2.5 in an industrial district of Zhengzhou, China: Chem- ical composition and source apportionment. Particuology. 11(1): 99-109. [46] Baumann, K., Jayanty, R.K.M., Flanagan, J.B. (2008) Fine particulate matter source apportionment for the chemical speciation Received: May 16, 2015 trends network site at Birmingham, Alabama, using positive ma- Revised: July 13, 2015 trix factorization. Journal of the Air and Waste Management As- Accepted: August 12, 2015 sociation. 2008(1): 27-44. [47] Begum, B.A., Kim, E., Biswas, S.K., Hopke and P.K. (2004) In- vestigation of sources of atmospheric aerosol at urban and semi- CORRESPONDING AUTHOR urban areas in Bangladesh. Atmospheric Environment. 38(19): 3025-3038. Hui Hu [48] Yang, L., Cheng, S., Wang, X., Nie, W., Xu, P., Gao, X., Yuan, C. and Wang, W. (2013) Source identification and health impact of School of Environmental Science and Engineering PM2.5 in a heavily polluted urban atmosphere in China. Atmos- Huazhong University of Science and Technology pheric Environment. 75(0): 265-269. Wuhan 430000 [49] Cheng, S., Lang, J., Zhou, Y., Han, L., Wang, G. and Chen, D. P.R. CHINA (2013) A new monitoring-simulation-source apportionment ap- and proach for investigating the vehicular emission contribution to the PM2.5 pollution in Beijing, China. Atmospheric Environment. Hubei Key Laboratory of Industrial Fume & Dust Pol- 79(0): 308-316. lution Control [50] Li, J., Song, Y., Mao, Y., Mao, Z., Wu, Y., Li, M., Huang, X., He, Jianghan University Q. and Hu, M. (2014) Chemical characteristics and source appor- Wuhan 430000 tionment of PM2.5 during the harvest season in eastern China's ag- P.R. CHINA ricultural regions. Atmospheric Environment. 92(0): 442-448.

[51] Zhang, R., Jing, J., Tao, J., Hsu, S., Wang, G., Cao, J., Lee, C., Zhu, L., Chen, Z. and Zhao, Y. (2013) Chemical characterization E-mail: [email protected] and source apportionment of PM2.5 in Beijing: seasonal perspective. Atmospheric Chemistry and Physics. 13(14): 7053-7074. FEB/ Vol 24/ No 11a/ 2015 – pages 3853 - 3864

APPLICATION OF THE SELF-ORGANIZING MAP FOR AQUEOUS PHOSPHORUS MODELING AND MONITORING IN A NATURAL FRESHWATER WETLAND: A CASE STUDY

Liang Zhang*, Yanhua Zhuang, Yun Du

Key Laboratory of Environment and Disaster Monitoring and Evaluation of Hubei, Institute of Geodesy and Geophysics, Chinese Academy of Sciences, Wuhan 430077, China

ABSTRACT as agricultural activities and large-scale water conservancy projects [6-8]. Water quality simulation is an efficient low-cost tool Water quality modeling and simulation is important of communities investigating approaches for controlling for optimizing the design, operation, and management for eutrophication in water bodies. In this study, the feasibility the ecological treatment and restoration pilot applications of applying self-organizing map (SOM) model to analyze in freshwater [4, 9]. Furthermore, water quality online fore- the water quality data in a large natural freshwater wetland cast is significant for water quality management and im- was assessed. The SOM model and a multivariable linear provement. A fuzzy quality index (FQI) is introduced for regression were used for analyzing water quality in a natu- the environmental assessment in a restored wetland to as- ral freshwater wetland. The relationships between transpar- sess the pollutants removal efficiency [10]. This model is ency, temperature, pH, conductivity, dissolved oxygen simple enough to be immediately grasped by non-technical (DO) and phosphorus were revealed by the multivariable people. Qian and Reckhow developed a predictive model linear regression model and the visualization function of based on nonparametric Bayesian regression for the total SOM model. Furthermore, the SOM model performed sat- phosphorus effluent concentration in an Everglades wet- isfactory in predicting the total phosphorus concentrations land [11]. In another study, an artificial neural network by utilizing those rapidly and easily measurable online var- (ANN) model is applied to establish an interrelationship iables, such as transparency, temperature, pH, conductivity between phosphorus, biochemical oxygen demand and dis- and DO. It could be concluded that the SOM model is a solved oxygen (DO). It shows satisfactory prediction and future modeling and predicting approach for monitoring probably could be used as a future tool for quality monitor- and managing water quality in natural freshwater wetlands. ing [12].

The accuracies of the modeling results are greatly af- KEYWORDS: self-organizing map (SOM); phosphorus; prediction; fected by the quality of the collected data set; it is difficult water quality, Honghu Lake wetland to assemble a robust and multivariate data set [2, 13]. There

are always sometimes resulting in gaps in the data when something goes wrong [14]. Compared with multivariable 1. INTRODUCTION linear regression model, the self-organizing map (SOM), which is not affected by the missing values in the input var- Eutrophication has become generally acknowledged as iables, could process with incomplete input data set and a serious environmental threat for many freshwater bodies perform relatively well in simulation [13, 15]. The model all over the world [1, 2]. Eutrophication leads to a severe is based on an unsupervised neural network algorithm, and reduction in water quality and decrease in the value of wa- has been successfully applied in the field of data analysis ter bodies for recreation, fishing, hunting, and aesthetic en- in complex dimensions [15-17]. The SOM model has been joyment [3, 4]. The overloading of phosphorus is the most well performed in improving water treatment performance, common cause of fresh water eutrophication. A 37-year developing water resource and freshwater ecology man- whole-ecosystem study indicated that eutrophication in agement [18-20]. But it is not often implemented in water lakes was hardly controlled by reducing nitrogen input; it quality forecast and control strategies for large-scale natu- should be focused heavily on phosphorus controlling [5]. ral freshwater wetland in comparison to traditional neural The inputs of phosphorus in freshwater lakes are multiple networks. In our previous research, the SOM model has and significantly determined by the human activities, such been experimentally used as a prediction tool for assessing the nitrogen and phosphorus removal performance in a pi- * Corresponding author lot-scale constructed wetland [9]. However, compared with

previous researches, the feasibility of using SOM model to is based on unsupervised learning, which indicates that no analyze the water quality data in large natural freshwater human intervention is required during the model learning wetland was assessed in this study. Additional aims of this process and that little information needs to be known about work are to find potential approaches for: (1) identifying the characteristics of the input data [24, 25]. The SOM relationships between phosphorus and other water quality models have been widely applied for the visualization of variables in a natural freshwater wetland; and (2) predict- dimensional systems and data mining [15, 26]. The SOM ing the phosphorus concentrations for online, real time consists of two layers (the multi-dimensional input layer monitoring and controlling. and the competitive or output layer) [27, 28]. In the SOM algorithm, the topological relations and the number of the neurons or nodes are fixed from the beginning. Each neu- 2. MATERIALS AND METHODS rons or nodes i (i =1,2,…,M) contains a weight vector Mi = [mi1,…,min] [27]. As the model runs, the weight vectors 2.1 Field site description are initialized to random values at the beginning. During Honghu Lake wetland, located at the north bank of the the training, each weight vector is calculated using a dis- middle reaches of Yangtze River and near the Three Gorges tance measure such as the Euclidian distance, which is de- Dam Project, is the largest natural wetland in Jianghan Plain fined as Equation 1. 2 in central China [2]. It has an area of approximate 344 km , n with an average water depth of 1.34 m and a maximum depth Dxm()2 (1) of 2.3 m. The study area had a subtropical monsoon climate. iijij j1 The monthly variations of mean temperature and mean rainfall of the study area during 2001-2010 was obtained from where China Meteorological Data Sharing Service System. The av- Di, the Euclidian distance, x , the jth element of the cur- erage air temperature and annual rainfall during 2001-2010 ij rent input vector x; m , jth element of the weight vector x. were 17.8 ºC and 1367.6 mm. The highest monthly averaged ij temperature and rainfall, appeared in July, were 29.9 ºC and Then, a node (Equation 2) is chosen as the best match- 215.3 mm, respectively. The lowest monthly averaged tem- ing unit (BMU) and the weight vectors are updated conse- perature was 4.6 ºC in January; while the lowest monthly quently. The BMU and its topological neighbors are moved rainfall (39.1 mm) appeared in December. closer to the input vector on the basis of the update rule shown in Equation (3) [27]. 2.2 Sampling and analysis (2) Water quality data were collected by monitoring the x mxmcimin  surface water quality parameters for three times during where 2010-2011 (Jan 2010: 32 samples; Oct 2010: 33 samples; July 2011: 41 samples). The sampling sites were spread all x, the input vector; m, the weight vector; , the dis- over of Honghu Lake wetland and its main canals. Tem- perature, pH, conductivity, and dissolved oxygen (DO) tance measure. were analyzed by using a calibrated Hydrolab DS5 Multi- mt(1)()()()[()()]  mt th t xt  mt (3) probe Water Quality Sonde made by Hach Co., USA. ii cit Transparency was determined using Secchi disk. Total where phosphorus (TP) was measured by ammonium molybdate spectrophotometric method by a spectrophotometer.  ()t , the learning rate at t; mt(), weight vector at t; ht(), neighborhood function centered in the winner unit 2.3 Statistical analyses ci at t; x()t , input vector drawn from the input data set at t. All statistical analyses were performed using the standard software packages Origin 8, Matlab 7.0 and SPSS. After a competitive learning process, the clusters cor- Missing values are very common in field water quality data responding to characteristic features can be shown on a U- sets. Outliers are data that appear to be not consistent with matrix map. The quality of the mapping is measured by the the entire data set. Outliers can be identified by many meth- final quantization error and the final topographical error ods, such as visual inspection methods, z-score method, [27]. For prediction purposes, the model is initially trained modified z-score method, box plot method [14, 19, 21]. with the training data set. The vector is subsequently presented to the SOM to identify its BMU. The values for the 2.4 Self-organizing map missing variables are then obtained by their corresponding The SOM model, an unsupervised competitive learn- values in the BMU [19, 29].The SOM toolbox (version 2) ing neural network and algorithm, implements a character- for Matlab 7.0 used in this study was developed by the La- istic non-linear projection method from the high-dimen- boratory of Computer and Information Science at Helsinki sional space of sensory or other input signals onto a low- University of Technology. The toolbox is available online dimensional regular lattice of neurons [22, 23]. The SOM at http://www.cis.hut.fi/projects/somtoolbox [26].

3. RESULTS AND DISCUSSION of the 1980s, and increased sharply from the year 2000 [30, 31]. 3.1 Water quality analysis Table 1 summarizes the water quality parameters col- 3.2 Multivariable linear regression simulation lected in Honghu Lake during the monitoring period (2010- Table 2 summarizes the results from a correlation anal- 2011). Altogether 106 measurements per variable were ysis of six variables. It can be seen that the linear relation- taken, but three of them were identified as outliers accord- ships between five input variables transparency, tempera- ing to the visual inspection method supported by summary ture, pH, conductivity and DO, and the target variable TP statistics and excluded from the present analysis. The aver- were statistically significant. Multivariable linear regres- age transparency, water temperature, pH, conductivity, DO sion model was applied to determine the TP concentrations and TP concentration during the monitoring period were by using other more cost-effective, rapid and easier to 0.771 m, 20.3 ºC, 8.19, 0.407 mS/cm, 10.7mg/L, 0.211 mg/L, measure water quality variables such as transparency, tem- respectively. Several missing values were observed in the perature, pH, conductivity and DO. The results of the re- data set for transparency, temperature, pH, conductivity, gression analysis are shown in Table 3. TP concentrations DO and TP; these missing values would potentially mis- in Honghu Lake wetland during the monitoring period lead the data analysis. Honghu Lake wetland has an im- were determined well by the selected variables with r2 of portant role in the agricultural economy development in 0.4025, 0.2715, 0.3954, 0.4022, 0.3901 and 0.3341, re- Jianghan Plain, its water quality is required to meet the spectively (Table 3). However, the missing values of each Class III of the Chinese environmental quality standards pair were disregarded in this analysis, the simulation re- for surface water (GB3838-2002) with TP no more 0.05 sults were influenced by missing values in different de- mg/L, however, it was not satisfactory during the monitor- grees. ing period (Table 1). As a shallow freshwater wetland, DO in most surface water samples were supersaturated, sug- 3.3 Self-organizing map simulation gesting the areas of higher aquatic primary productivity. The SOM model was also applied to identify the rela- Algae blooms and the excessive loading of nutrients have tionships between the incomplete input data set of water also been observed in some bays of Honghu Lake wetland. quality parameters of Honghu Lake wetland. It is reported One of the most important reason causing this problem is that the SOM learning result can be affected considerably the enclosure culture for fishery, which started at the end when the input data are pre-processed appropriately [32].

TABLE 1 - Water quality variables collected during 2010 to 2011

Variable Mean ± standard deviation Samples Missing values Transparency (m) 0.771±0.3433 94 12 Temperature (ºC) 20.3±8.804 103 3 pH 8.19±0.5545 102 4 Conductivity (mS/cm) 0.407±0.10383 102 4 DO (mg/L) 10.7±5.395 103 3 TP (mg/L) 0.211±0.231 105 1

TABLE 2 - Correlation matrix (variable pairs in bracket) for transparency, temperature, pH, conductivity, DO and TP

Variable Transparency Temperature pH Conductivity DO TP Transparency 1.000 (94) 0.119 (91) 0.261 (90) -0.287 (90) -0.165 (91) 0.409 (94) Temperature 1.000 (103) -0.594 (102) -0.729 (102) -0.770 (103) 0.398 (102) pH 1.000 (102) 0.246 (102) 0.638 (102) -0.206 (101) Conductivity 1.000 (102) 0.445 (102) -0.245 (101) DO 1.000 (103) -0.493 (102) TP 1.000 (105)

TABLE 3 - Multivariate linear regression equations for TP as functions of transparency, temperature, pH, conductivity and DO.

Linear regression equation for determining phosphorus values Variable pairs r2 TP = 0.266  Transparency + 0.005  Temperature – 0.011  pH + 0.412  Conductivity – 0.020  DO + 0.053 90 0.4025 TP = 0.005  Temperature + 0.095  pH + 0.195  Conductivity – 0.024  DO – 0.487 101 0.2715 TP = 0.259  Transparency – 0.030  pH + 0.179  Conductivity – 0.024  DO + 0.457 90 0.3954 TP = 0.261  Transparency + 0.006  Temperature + 0.423  Conductivity – 0.021  DO – 0.044 90 0.4022 TP = 0.244  Transparency + 0.0004  Temperature – 0.024  pH – 0.022  DO + 0.462 90 0.3910 TP = 0.314  Transparency + 0.014  Temperature – 0.044  pH + 0.618  Conductivity – 0.204 90 0.3341

The numerical accuracy of statistical computations in of phosphorus previously retained in the sediment [12]. DO connection with the SOM algorithm is therefore improved by is an important variable influencing phosphorus transfor- utilizing a component scaling transformation (Equation 2) of mation in natural wetlands. The interrelationship between each vector component. DO and TP in an untreated effluent impacted urban stream xx was revealed by Girija et al. [12], concentration of TP was i min maximum while dissolved oxygen was the least. In terms of xi (new)  (4) xmax x min the effect of pH, phosphorus might precipitate as calcium phosphate within sediment pore water or in the water col- where umn near active phytoplankton growth when pH values were xi(new), pre-processed data of component i in the input relatively high [35]. The TP concentration in the Honghu vector x; xi, the original data of component i in the input Lake wetland tended to be low if the conductivities were la- vector x; xmin, the minimum component in the input vector ger than 0.44 mS/cm, it might be because the adsorption ca- x; xmax, the maximum component in the input vector x. pacity of phosphorus onto sediments increased with the en- Prior to simulation implementation, the raw data sets hancement of ionic strength, that in agreement with the for- were pre-processed by applying this component scaling mation of inner-sphere complexes [36]. transformation method (Equation 2) to ensure that no component would have excessive influence on the training result by virtue of its higher absolute value [32]. A two-dimensional lattice with a map size of 126 hexagonal units was used. The final quantization error and final topographic error (0.2062 and 0.0094, respectively) were relatively low if compared to the error values with other parameters, indicating that the mapping quality was relatively good. After training the original un-scaled data can be also obtained by reversing the Equation 2. The component planes for each scaled variable of the SOM model are shown in Fig. 1. The unified distance matrix (U-matrix) representation of the SOM visualizes the distances between the map neurons [16, 26]. The distances between the neighboring map neurons were calculated and visualized with grey shades between them in Fig. 1. The component plane shows the value of the variable in each map unit [16]. For example, the lighter grey shades are linked to the high relative component value of the corresponding weight vec- FIGURE 1 - Abstract visualization of the relationships between the tor. The relationships between TP and other input variables pre-processed input data set of water quality parameters using a self- could be identified and illustratively showed. The finding organizing map model. was in agreement with the result of correlation analysis (Table 2). High scaled TP(new) values (>0.22) were linked 3.4 TP prediction with the self-organizing map to high transparency (new) values (>0.36), high tempera- TP, an important parameter for water quality assess- ture(new) values (>0.50), low pH(new) values (<0.53), low ment, is the sum of reactive, condensed and organic phos- conductivity(new) values (<0.58) and low DO(new) phorous. In order to monitor TP in real time, the SOM (<0.39). model, which has been successfully used for nutrients mod- Phosphorus concentration in Honghu Lake wetland eling and prediction in pilot-scale constructed wetlands [9, tended to be low under the suitable conditions linked to 13], was tested in this study to predict the TP concentrations specific transparency, temperature, pH, conductivity, and in a large natural freshwater wetland. The input variables, DO concentrations. Phosphorus transformation and reduc- used for predicting, were easier to be measured online wa- ing mechanisms in wetlands include adsorption, desorp- ter quality parameters. tion, precipitation, dissolution, plant and microbial uptake, For modeling purposes, data set was randomly ar- fragmentation, leaching, mineralization, sedimentation ranged and then subdivided into three sets. The first set was (peat accretion) and burial [33, 34]. Those processes are used as training data, the second set was used as validation significantly affected by the crucial factors of water qual- data and the last set was used as test data. The model was ity, such as temperature, DO, pH and conductivity. High first trained using the training data set. Then, a prediction temperatures were linked to low DO concentrations in the could be made by seeking the BMU for a vector with un- Honghu Lake wetland (Fig. 2). Abundant growth of fila- known components. Finally, the missing values in the val- mentous algae, during high temperatures periods, would idation and test data sets were obtained from their corre- increase the sediment oxygen demand of the organics in sponding values in the BMU [19, 29]. The numbers of natural wetlands, the anoxic condition resulted in a break- training, validation and test data sets are shown in Table 4. down of sediment water boundary layer, leading to release The TP data were omitted from the test data set, implying

that TP values were in fact missing values, and then the ACKNOWLEDGEMENTS predicted TP values were compared with the actual values. The results of the SOM modeling performance in predict- The authors thank the support by the CRSRI Open ing TP concentrations are shown in Fig. 2. Table 4 shows Research Program (CKWV2013213/KY), National the summary statistics of the SOM model during test. SOM Natural Science Foundation of China (41001333), the model showed comparatively lower mean absolute scaled Youth Chenguang Project of Science and Technology of error (MASE < 1), a scale-free error metric proposed by Wuhan City (201150431072), and Hubei Province Natural Hyndman and Koehler, indicating that, on average, the fore- Science Foundation of China (2011CDB404). cast was satisfactory [16, 37]. In general, the SOM model performed better in predicting the TP concentrations in The authors have declared no conflict of interest. Honghu Lake wetland if compared with the multivariable linear regression model.

TABLE 4 - Summary statistics of the self-organizing map model applied for prediction purposes REFERENCES

Statistics Validation data set Test data set [1] Ryther, J.H. and Dunstan, W.M. (1971) Nitrogen, phosphorus, and eutrophication in coastal marine environment. Science, Number of training data sets 70 70 171(3975), 1008-&. Number of validation/test data 16 20 sets [2] Zhang, L., Du, Y., Wu, S.J., Yu, C., Feng, Q., Ban, X., Ren, 2 X.Y., Xue, H.P. (2012) Characteristics of nutrients in natural r 0.8897 0.8689 wetland in winter: a case study. Environmental Monitoring Mean absolute scaled error 0.3907 0.3111 and Assessment, 184(9), 5487-5495. [3] Jørgensen, B.B. and Richardson, K. [eds.] (1996) Eutrophica- tion in coastal marine ecosystems. American Geophysical Un- ion, Coastal and Estuarine Studies, Washington, D.C. 1.0 [4] Smith, V.H. (2003) Eutrophication of freshwater and coastal Actual marine ecosystems - A global problem. Environmental Sci- 0.8 Predicted ence and Pollution Research, 10(2), 126-139. [5] Schindler, D.W., Hecky, R.E., Findlay, D.L., Stainton, M.P., 0.6 Parker, B.R., Paterson, M.J., Beaty, K.G., Lyng, M. Kasian, S.E.M. (2008) Eutrophication of lakes cannot be controlled by reducing nitrogen input: Results of a 37-year whole-ecosystem 0.4 experiment. PNAS, 105(32), 11254-11258.

TP (mg/L) TP [6] Garcia-Garcia, V., Gomez, R., Vidal-Abarca, M.R., Suarez, M.L. (2009) Nitrogen retention in natural Mediterranean wet- 0.2 land-streams affected by agricultural runoff. Hydrology and Earth System Sciences, 13(12), 2359-2371. 0.0 [7] Garnier, M., Recanatesi, F., Ripa, M.N., Leone, A. (2010) Ag- ricultural nitrate monitoring in a lake basin in Central Italy: a further step ahead towards an integrated nutrient management Validation Test aimed at controlling water pollution. Environmental Monitor- ing and Assessment, 170(1-4), 273-286. FIGURE 2 - Comparing the actual and predicted total phosphorus (TP). [8] Xu, Y.Y., Cai, Q.H., Han, X.Q., Shao, M.L., Liu, R.Q. (2010) Factors regulating trophic status in a large subtropical reservoir, China. Environmental Monitoring and Assessment, 169(1-4), 237-248. 4. CONCLUSIONS [9] Zhang, L., Scholz, M., Mustafa, A., Harrington, R. (2008) As- sessment of the nutrient removal performance in integrated In this study, the feasibility of using SOM model to an- constructed wetlands with the self-organizing map. Water Re- alyze the water quality data in a large natural freshwater wet- search, 42(13), 3519-3527. land was assessed. The relationships between TP and other [10] Giusti, E., Marsili-Libelli, S., Mattioli, S. (2011) A fuzzy water quality parameters could be well identified and illus- quality index for the environmental assessment of a restored wetland. Water Science and Technology, 63(9), 2061-2070. tratively showed by using the SOM model. The SOM model allowed the utilization of more rapid and easily measurable [11] Qian, S.S., Reckhow, K.H. (1998) Modeling phosphorus trapping in wetlands using nonparametric Bayesian regression. online variables to predict TP concentrations. Compared Water Resources Research, 34(7), 1745-1754. with the multivariable linear regression model, the relative high r2 and low MASE values indicated that this method [12] Girija, T.R., Mahanta, C., Chandramouli, V. (2007) Water quality assessment of an untreated effluent impacted urban could be used as a potential approach, which was free from stream: the Bharalu Tributary of the Brahmaputra River, India. missing values, when monitoring and managing the water Environmental Monitoring and Assessment, 130(1-3), 221- quality in large natural freshwater wetlands. 236.

[13] Zhang, L., Scholz, M., Mustafa, A., Harrington, R. (2009) Ap- [31] Cheng, X.Y. and Li, S.J. (2006) Evolution and character of the plication of the self-organizing map as a prediction tool for an representative lakes in the middle and lower reaches regions integrated constructed wetland agroecosystem treating agri- of Yangtze River. Chinese Science Bulletin, 51(7), 848-855. cultural runoff. Bioresource Technology, 100(2), 559-565. [32] Obu-Cann, K., Fujimura, K., Tokutaka, H., Ohkita, M., Inui, [14] Burke, S. (1999) Missing values, outliers, robust statistics and M., Ikeda, Y. (2001) Data mining of power transformer data- non-parametric methods. LC*GC Europe Online Supplement base using self-organizing maps. In: Zhong, Y.X., Qin, S. Y. 19-24. (Eds.), Proceedings of International Conference on Info-tech & Info-net (29/10-01/11/2001), Beijing, China. 4, 44-49. [15] Kohonen, T., Oja, E., Simula, O., Visa, A., Kangas, J. (1996) Available online at http://ieeex- Engineering applications of the self organizing map. Preceed- plore.ieee.org/iel5/7719/21162/00983717.pdf. ings IEEE, 84(10), 1358-1384. [33] Hammer, D.A. and Bastian, R.K. (1989) Wetlands ecosys- [16] Lee, B.H., Scholz, M. (2006) Application of the self-organiz- tems: natural water purifiers. In: Hammer, D.A. (Ed.). Con- ing map (SOM) to assess the heavy metal removal perfor- structed wetlands for wastewater treatment. Lewis Publishers, mance in experimental constructed wetlands. Water Research, Chelsea, Michigan. 40(18), 3367-3374. [34] Vymazal, J. (2007) Removal of nutrients in various types of [17] Chon, T. (2011) Self-organizing maps applied to ecological constructed wetlands. Science of the Total Environment, sciences. Ecological Informatics, 6(1), 50-61. 380(1-3), 48-65. [18] Ki, S.J., Kang, J.H., Lee, S.W., Lee, Y.S., Cho, K.H., An, [35] USEPA (2000) Constructed wetlands treatment of municipal K.G., Kim, J.H. (2011) Advancing assessment and design of wastewaters. United States (US) Environmental Protection stormwater monitoring programs using a self-organizing map: Agency (EPA), Office of Research and Development, Cincin- Characterization of trace metal concentration profiles in nati, Ohio, USA. stormwater runoff. Water Research, 45(14), 4183-4197. [36] Karimaian, K.A., Amrane, A., Kazemian, H., Panahi, R., Zar- [19] Rustum, R. and Adeloye, A. (2007) Replacing outliers and rabi, M. (2013) Retention of phosphorous ions on natural and missing values from activated sludge data using Kohonen self- engineered waste pumice: Characterization, equilibrium, com- organizing map. Journal of Environmental Engineering- peting ions, regeneration, kinetic, equilibrium and thermody- ASCE, 133(9), 909-916. namic study. Applied Surface Science, 284, 419-431. [20] Kalteh, A.M., Hiorth, P., Bemdtsson, R. (2008) Review of the [37] Hyndman, R.J. and Koehler, A.B. (2006) Another look at self-organizing map (SOM) approach in water resources: measures of forecast accuracy. International journal of fore- Analysis, modelling and application. Environmental Model- casting, 22(4), 679-688. ling & Software, 23(7), 835-845. [21] Iglewicz, B. and Hoaglin, D.C. (1993) How to detect and handle outliers, American Society for Quality Control, Milwau- kee, WI.

[22] Kohonen, T. (1982) Self-organized formation of topologically correct feature maps. Biological Cybernetics. 43(1), 59-69. [23] Kohonen, T. (1995) Self-organizing Maps. Springer Series in Information Sciences, vol. 30. Springer, Berlin, Germany. [24] Kohonen, T. (2013) Essentials of the self-organizing map. Neural Networks, 37, 52-65.

[25] Alhoniemi, E., Hollmen, J., Simula, O., Vesanto, J. (1999) Process monitoring and modeling using the self-organizing map. Integrated Computer-Aided Engineering, 6(1), 3-14. Received: May 19, 2015 [26] Vesanto, J., Himberg, J., Alhoniemi, E., Parhankangas, J. Revised: June 22, 2015 (1999) Self-organizing map in matlab: the SOM toolbox. In: Accepted: July 23, 2015 Proceedings of the Matlab DSP Conference, November 1999, Espoo, Finland, pp. 34-40. Software available online at http://www.cis.hut.fi/projects/somtoolbox/. [27] Rustum, R., 2009. Modelling Activated Sludge Wastewater CORRESPONDING AUTHOR Treatment Plants Using Artificial Intelligence Techniques (Fuzzy Logic and Neural Networks). PhD Thesis, Heriot Watt Liang Zhang University, Edinburgh. Key Laboratory of Environment and Disaster Moni- [28] Mwale, F.D., Adeloye, A.J., Rustum, R. (2014) Application of toring and Evaluation of Hubei self-organising maps and multi-layer perceptron-artificial neu- Institute of Geodesy and Geophysics ral networks for streamflow and water level forecasting in Chinese Academy of Sciences data-poor catchments: the case of the Lower Shire floodplain, Malawi. Hydrology Research, 45(6), 838-854. Wuhan, Hubei Province P.R. CHINA [29] Rustum, R., Adeloye, J.A., Scholz, M. (2008) Applying Ko- honen self-organizing map as a software sensor to predict biochemical oxygen demand. Water Environment Research, Phone: +86 27 68881362 80(1), 32-40. Fax: +86 27 68881362 [30] Chen, Y.Y. and Xu, Y.X. (1995) Hydrobiological Resources E-mail: [email protected] and Environment in Lake Honghu. Beijing: Science Press, p.15. FEB/ Vol 24/ No 11a/ 2015 – pages 3865 - 3870

EFFECTS OF ZINC AND CADMIUM ON CONDITION FACTOR, HEPATOSOMATIC AND GONADOSOMATIC INDEX OF Oreochromis niloticus

Nuray Çiftçi1,*, Özcan Ay1, Fahri Karayakar1, Bedii Cicik1 and Cahit Erdem2

1University of Mersin, Faculty of Aquaculture, Yenişehir Kampüsü, C Blok, Kat 2, 33169 Mersin, Turkey 2University of Çukurova, Faculty of Art and Sciences, Biology Department, 01330 Balcalı, Adana, Turkey

ABSTRACT as Cd, Pb and Hg, are toxic even at very low concentrations with no known biological function [6]. Effects of Zn and Cd on condition factor (CF), hepato- Zinc, being a structural component of a number of en- somatic (HIS) and gonadosomatic (GSI) of Oreochromis zymes, plays an important role in protein, carbohydrate, li- niloticus, after exposing the animals to 10 % of their LC50 pid and nucleic acid metabolisms and have function in values, namely 6.0 ppm Zn and 1.6 ppm Cd, over 7, 15, growth, development and reproduction in animals. The and 30 days. Standard mathematical formulations were main source of zinc is the earth crust and it is widely used used in determining the mentioned parameters. in construction, automotive, dye and food industries and in Cadmium, which is a toxic metal, increased GSI and de- medicine [7]. creased HSI and CF at the beginning of the experiments. The Cadmium is xenobiotic which is widely used in elec- decrease in HSI and CF continued with increasing exposure tric, electronic, automotive, metal plating, battery, dye, periods. Zinc, however, decreased all three parameters com- plastic and synthetic fiber industries and in nuclear reactor pared with control at the end of experiments (P<0.05). control systems. It is known to have toxic effects on ani- Changes in CF, GSI and HIS might reflect metabolic mals even at very low concentrations [8]. and physiologic disturbances under the effect of metals. It was reported that Zn and Cd entering aquatic eco-

systems from these resources cause enzymatic and hormo-

KEYWORDS: Zinc, cadmium, Oreochromis niloticus, condition nal disorders, DNA damages, defectiveness in blood oxy- factor, gonadosomatic index, hepatosomatic index. gen carrying capacity and electrolyte losses in fish which

in overall result in alterations in condition factor and so-

1. INTRODUCTION matic index [9]. Together with hematologic and biochemical parame- The water cycle among atmosphere, lithosphere and ters, changes in condition factor, hepatosomatic and gona- hydrosphere is important in sustainability of life on earth. dosomatic index were also frequently used to determine Heavy metal containing wastes that enter lithosphere and at- and monitor heavy metal toxicity in aquatic animals. In mosphere by natural and anthropogenic activities, are washed general, condition factor, hepatosomatic and gonadosto- to the hydrosphere where they increase the levels of these met- matic index reflect the developmental, metabolic and re- als [1, 2]. Aquatic organisms uptake heavy metals directly productive status of the organisms respectively [10, 11]. from their environment and by through food chain. Excess Studying changes in biological functions of aquatic an- amounts of these metals are accumulated in various tissues imals under the effect of heavy metals allow to determine which result in loss of appetite, growth and reproduction the health condition of economic aquatic products which disorders and changes in metabolic and physiologic func- also reflect the state of pollution in the environment. Pre- tions as a result of weakening of immune system [3, 4]. sent study was undertaken to determine condition factor, Although these effects show differences among the spe- hepatosomatic and gonadosomatic index of O.niloticus ex- cies, they may alter the statue of a population and energy posed to 10% LC50 values of zinc and cadmium, namely flow within the ecosystem, thus resulting disturbances in 6.0 ppm Zn and 1.6 ppm Cd, over 7, 15, and 30 days. ecological balance [5]. The mode of action of heavy metals show differences among metals. Some of them, such as Cu, Zn and Fe, have 2. MATERIALS AND METHODS metabolic functions at low concentrations and some, such O.niloticus 18.34 ± 1.11cm in length and 107.89 ± 1.19 g * Corresponding author in weight were used in the experiments. Experiments were

run in the culture laboratory of the Aquaculture Faculty, determined. Weights of the liver and gonad tissues were also Mersin University. Fish were placed in three glass aquaria determined after dissecting from each fish. 40x100x40 cm in height and adapted to controlled labora- Total length, weight and organ wet weight values were tory conditions for one week. The laboratory had a constant used to determine the hepatosomatic index, gonadosomatic temperature of 24 ± 10C and a 12h light/12h dark illumina- index and condition factor parameters using the formulas tion period was applied. given below [12].

Three glass aquaria of the same size were used in the Liver Wet Weight (g) experiments. The first two aquaria were filled with 6.0 ppm % Hepatosomatic Index (HSI) = ------x 100 Zn and 1.6 ppm Cd, which are 96h LC50 concentrations of Total Body Wet Weight (g) these metals to this species, and the third aquarium was Gonad Wet Weight (g) filled with metal free tap water and used as control. Exper- % Gonadosomatic Index (GSI) = ------x 100 iments were run in triplicate being two fish in each repli- Total Body Wet Weight (g) cate. Hence 18 fish were placed in each aquarium. Aquaria were aerated using central aeration system. Total Body Wet Weight (g) Some physical and chemical parameters of experi- % Condition Factor (CF) = ------x 100 Total Length (cm) mental water are given in Table 1.

TABLE 1 - Physical and chemical parameters of experimental water Experimental data were statistically analyzed by a series of analysis of variance (ANOVA) and Student’s New- Temperature (0C) 24 ± 1 man Keuls’ Tests (SNK) using SPSS package program. pH 8.62 ± 0.16 Since data were expressed in percentages Arcsine transfor- Dissolved Oxygen (mgL-1) 5.29 ± 0.70 -1 mation was applied before analysis. Total Hardness (mgL CaCO3) 227 ± 0.48 -1 Alkalinity (mgL CaCO3) 332 ± 0.50

ZnSO4.7H2O and CdCl2 water soluble salts of the met- 3. RESULTS als were used in preparation of metal solutions. Experi- mental solutions were changed once in every two days to No mortality was observed in O.niloticus exposed to prevent concentration changes due to precipitation, evapo- 6.0 ppm Zn and 1.6 ppm Cd over 30 days. Exposure to both ration and adsorption. Fish were fed with a commercial fish metals decreased HSI significantly compared with control feed (Pınar; Bream feed, Pellet No:2) once a day in at a given period except 7 days exposure to Zn (P<0.05) amounts 2% of their total body weight. (Table 2). Cadmium was more effective in lowering HSI Six fish were removed from each aquarium at the end of compared to zinc (P<0.05) (Table 2). 7, 15 and 30 days of exposure periods and were anaesthe- tized using ethylene glycol monophenyl ether (=Phenoxy- Zinc decreased whereas Cd increased GSI compared to control at the end of exposure period (P<0.05). GSI also ethanol, C8H10O2; Merck). Fish were then washed with tap water, dried with Whatman paper to remove metal residues showed a decrease with increasing exposure periods to zinc from their skin and their total length and wet weight were and vice versa was true for Cd (P<0.05) (Table 2).

TABLE 2 - Effects opf Zinc and Cadmium on Hepatosomatic Index, Gonadosomatic Index and Condition Factor in O.niloticus.

Exposure Period (Days) Metal 0 (Control) 7 15 30 (Concentration) X  Sx * X  Sx * X  Sx * X  Sx * Zn 1,73±0,026 a 1,91±0,075 b 1,18±0,006 c 1,46±0,024 d HSI (6 ppm) (%) Cd 1,73±0,026 a 1,60±0,015 b 1,32±0,014 c 0,96±0,003 d (1.6 ppm)

Zn 0,56±0,012 a 0,67±0,006 a 0,60±0,050 a 0,44±0,012 b GSI (6 ppm) (%) Cd 0,56±0,012 a 0,46±0,019 b 0,56±0,042 a 0,66±0,018 c (1.6 ppm)

Zn 1,78±0,043 a 1,77±0,026a 1,75±0,107 a 1,52±0,014 b CF (6 ppm) (%) Cd 1,78±0,043 a 1,55±0,012 b 1,72±0,008 a 1,75±0,003 a (1.6 ppm) *SNK; Letters a, b, c and d show differences between exposure periods. Data shown with different letters are significant at the P<0.05 level. X  Sx : Mean± Standard Error.

The effects of Zn and Cd on CF differed depending on Gonadosomatic index is an important parameter that exposure period. These differences was statically im- reflects both the state of population for the continuity of portant on days 30 and 7 for Zn and Cd respectively. The generation and changes in organisms under the effect of decrease in CF was more pronounced on exposure to Cd heavy metals. GSI decreased in Mullus barbatus under the then to Zn (P<0.05) (Table 2). effect of Hg, Pb and Ar, possibly due to structural deformation of DNA and an increase in liver EROD (ethoxyresorufin-O deethylase) activity [21]. Heavy metals had 4. DISCUSSION negative effects on gonad size in Leuciscus cephalus possibly as a result of a decrease in the amounts of 11-keto- Toxic materials entering aquatic organisms through testesteron, EROD and vitelline [22]. The increase in GSI their food, skin and water are detoxified by various mech- under the effect of Zn, structural part of a number of en- anisms. The level of detoxification depends upon type of zymes and hormones, at the beginning of experiments with toxic material, its concentration, exposure period and vari- O. niloticus, might be due to stimulation of reproductive ous environmental factors. Sixty percent mortality was ob- enzymes and hormones. As in other trace elements, the in- served in C. carpio exposed to increasing concentrations of creasing levels of Zn with increasing exposure times might Fe, Zn, Mn, Ni, Cr, Cu, Pb and Cd under laboratory condi- have toxic effect in gonads, resulting a decrease in GSI. tions depended on concentration and exposure periods to a The decrease in GSI compared to control at the beginning given metal [13]. No mortality was observed in O. niloticus of exposure to toxic metal Cd was also probably due to in- exposed to 6.0 ppm Zn and 1.6 ppm Cd for 30 days which hibition of enzymes functioning in synthesis and release of was probably due to low concentrations of metals at the reproductive hormones whereas the increase in GSI with exposure periods tested and/or tested concentrations of prolonged exposure to this metal might activate synthesis metals stimulated the detoxification mechanisms. of metal binding proteins in gonads. Physical and chemical alterations occurring in the environment cause stress in aquatic organisms which in turn The authors have declared no conflict of interest. result in metabolic, physiologic, biochemical, behavioral changes that have negative effects on growth, development and reproduction [5]. REFERENCES Hepatosomatic index is the main indicator of metabolic activity in animal organisms. HSI of Leuciscus ceph- [1] Patrick, T.K., Yoke, M. and Ming, K.W. (1993) The Effects of alus sampled from a discharge area of a waste water plant Short-Term Acute Cadmium Exposure on Blue Tilapia, Oreo- was higher than those sampled from control station which chromis aureus. Environ. Biol. Fish, 37: 67-74. was attributed to hypertrophy or hyperplasia occurred in [2] Wepener, V., Van Vuren, J.H.J. and Du Preez, H.H. (2001) liver under the effect of pollutants [14]. Bekmezci [15] re- Uptake and Distribution of Copper, Iron and Zinc Mixture in ported that heavy metals decreased HSI in Clarias gariepi- Gill, Liver and Plasma of a Freshwater Teleost, Tilapia nus which was possibly due to depletion of energy reserves sparrmanii, Water SA, 27: 99-108. in liver. HSI decreased in O. niloticus exposed to Zn and [3] Wedemeyer, G.A., Barton, B.A. and McLeay, D. J. (1990) Cd singly compared to control. Stress condition developed Stress and Acclimation. In C. B. Schreck and P. B. Moyle (eds.), Methods for Fish Biology, 451–489. American Fisher- under the effect of metals and the excess usage of energy ies Society, Bethesda, Maryland. reserves in response to increase in requirement might cause the decrease in HSI. [4] Javed, M. and Usmani, J. (2011) Accumulation of Heavy Met- als in Fishes: A Human Health Concern, International Journal Condition factor is another parameter that reflects gen- of Environmental Sciences, 2(2):659-670. eral health state of fish which differs according to the envi- [5] Heath, A.G. (1995) Water Pollution and Fish Physiology, De- ronmental factors, age, sex and reproduction period [16]. CF partment of Biology Virginia Polytechnic Institute and State decreased in Gasterosteus aculeatus [17] and Gobio gobio University Blacksburg, Virginia, 4: 67-76. [18], increased in Astyanax fasciatus [19] and did not ef- [6] Hodson, P.V. (1988) The Effect of Metal Metabolism on Up- fected in C. gariepinus under the effect of heavy metals [20]. take, Disposition and Toxicity in Fish. Aquat. Toxicol., 11: 3- They concluded that the decrease in CF under the effect of 18. heavy metals might be a result of loss in appetite or exces- [7] Bryan, G.W. (1971) The Effects of Heavy Metals (Other than sive use of energy reserves to compensate requirements. The Mercury) on Marine and Estuarine Organisms. Proc. Roy. Soc. increase in CF under metal exposure can be explained by the London B., 177:389. stimulation of detoxification mechanisms which prevented [8] Sorensen, E.M. (1991) Cadmium. In: Metal Poisoning in Fish. metabolic reactions to be effected by heavy metals. The de- CRC Press, Boca Raton, Florida, pp 175–234 crease in CF in O. niloticus exposed to Cd and Zn was ob- [9] Filipovic, V. and Raspor, B. (2003) Metallothionein and Metal served at the beginning and at the end of experimental peri- Levels in Cytosol of Liver, Kidney and Brain in Relation to ods respectively. The decrease in CF in both groups can be Growth Parameters of Mullus surmuletus and Liza aurata explained by loss in appetite, whereas the difference between from the Eastern Adriatic Sea, Water Research., 37: 3253- the metals might be due to Cd being more toxic than Zn. 3262.

[10] Vives, I., Grimalt, J., Fernandez, P and Rosseland, B. (2004) Polycyclic Aromatic Hydrocarbons in Fish from Remote and High Mountain Lakes in Europa and Greenland. Science of the Total Environment, 324:67-77.

[11] Martin-Diaz, M., Tuberty, S., McKenney, C., Sales, D. and DelVals, T. (2005) Effects of Cadmium and Zinc on Procam- barus clarkii: Simulation of the Aznalcollar Mining Spill, Ciencias Marinas, 62 (1):113-124. [12] Biney, C., Amazu, A. T., Calamari, D., Kaba, N., Mbome, I. L., Naeve, H., Ochumba, P. B. O., Osibonjo, O., Radegonde, V. and Saad, M. A. H. (1994) Review of Heavy Metals in the African Aquatic Environment, Ecotoxicology and Environ- mental Safety, 31: 134-159.

[13] Saxena, M. and Saxena, H. (2007) Histopathological Changes in Lymphoid Organs of Fish After Exposure to Water Polluted With Heavy Metals, The Internet Journal of Veterinary Medi- cine, Volume 5, Number 1.

[14] Komsala, A., Migeon, B., Flammorion, P. and Garric, J. (1998) Impact Assessment of a Wastewater Treatment Plant Effluent Using the Fish Biomarker Ethoxyresorufin-O- Deethylase: Field and On-Site Experiments. Ecotoxicology and Environmental Safety, 41: 19-28. [15] Bekmezci, H.D. (2010) Aşağı Seyhan Ovası Drenaj Sistem- lerindeki Kirlilik Etmenlerinin Clarias gariepinus’da Toksik Etkileri. Çukurova Üniversitesi, Fen Bilimleri Enstitüsü, Biy- oloji ABD, Doktora Tezi, 145.

[16] Yılmaz, S., Yılmaz, M., Polat, N. and Bostancı, D. (2007) Altınkaya Baraj Gölü (Samsun, Türkiye)’nde Yaşayan Sudak Balığı Sander lucioperca (L., 1758)’nın Yaş ve Büyüme Özel- likleri. Fırat Üniversitesi Fen ve Mühendislik Bilimleri Dergisi, 19(3): 273-283. [17] Sanchez, W., Katsiadaki, I., Piccini, B., Ditche, J.M. and Porcher, J.M. (2008) Biomarker Responses in Wild Three- Spined Stickleback (Gasterosteus aculeatus L.) as a Useful Tool For Fresh Water Biomonitoring: a Multiparametric Ap- proach, Environment International, 34(4), 490-498.

[18] Bervoets, L. and Blust, R. (2003) Metal Concentrations in Wa- ter, Sediment and Gudgeon (Gobio gobio) from a Pollution Gradient: Relationship with Fish Condition Factor. Environ- mental Pollution, 126: 9-19.

[19] Alberto, A., Camargo, A. F. M., Veranı, J. R., Costa, O. F. T., Fernandes, M. N., 2005. Health Variables and Gill Morphol- ogy in the Tropical Fish Astyanax fasciatus from a Sewage- Contaminated River. Ecotoxicology and Environmental Safety, 61: 247-255. [20] Hmoud Fares Alkahemal-Balawi, Z.A., Al-Akel, A.S., Al- Misned, F., El-Amin, M.S. and Al-Ghanim, K.A. (2011) Tox- Received: January 27, 2015 icity Bioassay of Lead Acetate and Effects of its Sublethal Ex- Revised: June 23, 2015 posure on Growth, Haematological Parameters and Reproduc- Accepted: August 31, 2015 tion in Clarias gariepinus, African Journal of Biotechnology, 10(53):11039-11047. [21] Martínez-Gómez, C., Fernández, B., Benedicto, J., Valdés, J., CORRESPONDING AUTHOR Campillo, J.A., León, V.M. and Vethaak A.D. (2012) Health Status of Red Mullets from Polluted Areas of the Spanish Mediterranean Coast, with Special Reference to Portmán (SE Nuray Çiftçi Spain), Marine Environmental Research, 77:50-59. University of Mersin Faculty of Aquaculture [22] Randak, T., Zlabek, V., Pulkrabova, J., Kolarova, J., Kroupova, H., Siroka, Z., Velisek, J., Svobodova, Z. and Yenişehir Kampüsü, C Blok, Kat 2 Hajslova, J. (2008) Effects of Pollution on Chub in the River 33169 Mersin Elbe, Czech Republic, Ecotoxicology and Environmental TURKEY Safety. DOI:10.1016/j.ecoenv. 2008.09.020. E-mail: [email protected]

FEB/ Vol 24/ No 11a/ 2015 – pages 3871 - 3874

SUBJECT INDEX

A G acetylcholinesterase 3842 GHG emissions 3743 adsorption 3822 gonadosomatic index 3871 alluvial silt 3766 grass species 3759 artificial neural networks 3727 H B halloysite nanotube 3822 biological treatment 3695 health risk assessment 3853 biomass 3695 hepatosomatic index 3871 biomass 3703 heterogeneous Fenton 3717 biorefinery 3703 454 high-throughput pyrosequencing 3802 bottled drinking water 3836 highway 3759 Bushehr 3836 Honghu Lake wetland 3865 HPLC 3736 C hydraulic retention time 3695 cadmium 3871 cement 3766 I characterization 3822 infiltration-reducing effect 3766 chemical and microbial quality 3836 iron chloride 3794 chemical oxygen demand 3695 chlorpyrifos 3842 K chromium(III) ions 3774 K-mean cluster analysis 3853 condition factor 3871 Cylindrospermopsin 3736 L Cylindrospermopsis raciborskii 3736 land use types 3780 CYP1 3842 length-weight relation 3727 cytochrome P450 3842 life cycle assessment 3743 lignocellulosic 3703 D derivatization reagents 3813 M derivatization 3813 manganese oxide 3774 diatomite 3774 microbial community 3802 Dikilitas Pond 3727 morphological parts 3759 Direct Orange 26 3717 domestication 3802 N nitrogen and phosphorus losses 3780 E non-point source pollution 3780 element 3853 non-steroidal anti-inflammatory drugs 3813 emission reduction 3743 energy consumption 3743 O Enhanced UV-B 3831 operation models 3743 enthalpy 3754 Oreochromis niloticus 3871 estimated 3727 Oryzias javanicus 3842 estrogens 3794 oxidation 3717 experimental design 3717 oxidative degradation 3794 extracted complex 3754 extraction 3754 P papermaking wastewater 3802 F peanut 3831 Fe(III)-sepiolite catalyst 3717 phosphorus 3865 food waste 3703 photosynthesis 3831 physical 3836 G PM2.5 3853 gallic acid 3754 PMF 3853 gas chromatography 3813 Poland 3759

AUTHOR INDEX

P A postponed water feeding time 3766 Anagun, A. Sermet 3717 potassium Supply 3831 Aşçı, Yeliz 3717 prediction 3865 Ay, Özcan 3871 pretreatment 3703 printing and dyeing wastewater 3802 B Barragán-Huerta, Blanca E. 3703 R Benzer, Recep 3727 reactor systems 3695 Benzer, Semra 3727 refinery wastewater 3695 removal 3774 C Chen, Nan 3853 S Cicik, Bedii 3871 self-organizing map (SOM) 3865 Ciepiela, Grażyna Anna 3759 Serbia 3736 Çiftçi, Nuray 3871 silylation 3813 Ćirić, Andrija R. 3736 sodium nitrite 3794 source apportionment 3853 D surface runoff 3780 Demirtas, Ezgi A. 3717 Dobaradaran, Sina 3836 T Dong, Shuyu 3754 timber production 3743 Đorđević, Nevena B. 3736 transpiration 3831 Du, Yun 3865 tributyl phosphate 3754 Duan, Zhengchao 3794

V E victoria blue 3822 Erdem, Cahit 3871

W F wastewater 3813 Fan, Guisheng 3766 water quality 3865 Fard, Elham Shabankareh 3836 water soluble ion 3853 Fei Zhang 3802 Filik Iscen, Cansu 3717 Z Foglar, Lucija 3695 zeta potential 3822 Fu, Dongmei 3794 zinc 3759 Funahashi, Aki 3842 zinc 3871 G Gutiérrez-Macías, Paulina 3703

Han, Chenguang 3831

Han, Yan 3831

Hao, Fanghua 3780

Hayati, Roughayeh 3836

Hong-Li Zhao 3802

Hu, Chengda 3831

Hu, Hui 3853

Hu, Xi-sheng 3743

Itakura, Takao 3842

J Jankowska, Jolanta 3759

J W Jankowski, Kazimierz 3759 Wang, Rui-Fei 3802 Ji, Jinlan 3766 Wiśniewska-Kadżajan, Beata 3759 Wu, Yundong 3754 K Wu, Zhi-long 3743 Kai-Yue Zhuo 3802 Kaminishi, Yoshio 3842 Y Karayakar, Fahri 3871 Yang, Qing-Xiang 3802 Kolczarek, Roman 3759 Ying Xu 3802 Yu, Wei 3754 L Lacina, Petr 3813 Z Lai, Xuehui 3780 Zeng, Xiangying 3774 Li, Dan 3743 Zhang, Liang 3865 Li, Er 3774 Zheng, Li-feng 3743 Li, Meng 3831 Zhou, Cheng-jun 3743 Liang, Xinmiao 3794 Zhou, Kanggen 3754 Liu Yang 3802 Zhou, Xin-nian 3743 Liu, Hong 3853 Zhou, Yuan 3743 Liu, Li 3853 Zhuang, Yanhua 3865 Lou, Yunsheng 3831

M Malinowska, Elżbieta 3759 Margeta, Dunja 3695

Montañez-Barragán, Brenda 3703 Mravcova, Ludmila 3813

P Papić, Sanja 3695

Qu, Taoli 3794 Quan, Jihong 3853

R Ren, Xiaoli 3780

Sertić-Bionda, Katica 3695 Simić, Snežana B. 3736 Sosnowski, Jacek 3759 Sýkora, Richard 3813 Szulc, Wiesław 3759

T Tian, Yiping 3853 Tuan, Trieu 3842

Turan Beyli, Pınar 3822

V Vávrová, Milada 3813

W Wang, Lianzhi 3794

3877