sign in sign up
<<
Home , River ecosystem

water

Article Ecological Health Assessments Using Chemical Parameter Model and the Index of Biological Integrity Model

Jang HaRa, Md. Mamun and Kwang-Guk An *

Department of Bioscience and Biotechnology, Chungnam National University, Daejeon 34134, Korea * Correspondence: [email protected]; Tel.: +82-010-6404-9844; Fax: +82-42-882-9690

 Received: 26 June 2019; Accepted: 16 August 2019; Published: 20 August 2019 

Abstract: River health is one of the important issues today because of various threats by multiple anthropogenic stressors that have long-term impacts on the physical , , ecological functions, and their services. The main objectives of this study is to diagnose the chemical and biological river health in the watershed of Geum River with regard to the chemical regimes (N, P) and fish using multi-metric chemical pollution index (CPI), and the index of biotic integrity model (IBI), respectively. The empirical models of sestonic chlorophyll, nutrients (N, P), and nutrient ratios of N:P indicated that the watershed, including all sampling sites, was a phosphorus-limited system. Analysis of fish trophic and tolerance guilds showed that the fish species and tolerant fish species were dominant in the watershed, while the sensitive fish species decreased downstream because of nutrient enrichments (such as TN, TP) and organic matter pollutions (such as BOD, COD). The chemical model of CPI showed that 11 sampling sites were in the fair—good condition, and 8 sites were in poor—very poor condition. Species composition analysis indicated that Zacco platypus was most widely distributed in the watersheds and dominated the fish community. The biological health of the watershed, based on the multimetric IBI model, was in poor condition and was getting worse downstream. The degradation of the river health was matched with the chemical health and showed a decreased of and sensitive fish species. The outcomes of the river health were supported by principal component analysis (PCA) and cluster analysis (CA) of fish model metrics and the physicochemical parameters. Overall, our study suggests that river health was directly influenced by the chemical pollutions of nutrients and organic matter inputs.

Keywords: river health; nutrients; organic matters; chemical pollution index (CPI) model; index of biotic integrity (IBI) model; trophic and tolerance guilds

1. Introduction River and provide multiple eco-services including clean water in sufficient quantity and quality for agricultural, industrial, and residential uses, act as a hotspot of aquatic, riparian, and migratory biota, and support fisheries and recreation for a human beings in a sustainable manner [1,2]. With the rapid urbanization, industrialization, and intensive agricultural farming the structure and function of the river and are severely affected, which lead to the significant degradation of water quality and ecosystem integrity [3]. As a result, the river and streams ecosystem health assessment is getting more attention from scholars globally and has become a fundamental and increasingly important environmental management issue worldwide, especially in Australia [4], USA [5], Europe [6], Korea [7], and China [8]. The assessment of ecological health in river and streams cannot be diagnosed directly because it is affected by their surrounding environments such as hydrology, water quality, physical form, riparian

Water 2019, 11, 1729; doi:10.3390/w11081729 www.mdpi.com/journal/water Water 2019, 11, 1729 2 of 22 zone, and biological communities [9]. Earlier studies on the diagnosis of stream and river health assessment have focused on the water quality factors [10]. However, recent research pointed out that this method is not sufficient to determine the ecological health of the [11]. To address this problem, two important approaches have been used to evaluate the ecological health of the river and streams. One is chemical pollution index (CPI) based on water quality factors and another is an index of biotic integrity (IBI) based on fish assemblages. The CPI model was primarily suggested by Beach (1980) and after that, it was developed by many researchers in different regions of the world to evaluate the chemical health status of the ecosystem [12–16]. A variety of biological indicators and methods are being used to evaluate the ecological health of river around the world. The US Environmental Protection Agency (EPA) used the Rapid Bio-assessment Protocols (RBP) to diagnose the health of the [5]. In 1981, the index of biotic integrity based on fish assemblages was proposed by Karr to evaluate the biological health status of the freshwater system [17], since then, this index has been adopted throughout the world for assessing the effects of natural and human on the freshwater system [18–20]. The previous research indicated that fish assemblages respond significantly and predictably to almost all kinds of anthropogenic disturbances, including , acidification, chemical pollution, flow regulation, physical alteration and fragmentation, human exploitation, and [21–24]. That is why fish has been regarded as the best indicator to evaluate the ecological health of the system. In Korea, a model for health assessment of stream ecosystem using the fish community was developed by An et al. in 2006 [25]. It is used for the health assessment of the aquatic ecosystem of the four major . Korea established the biological water quality assessment guidelines in 2006 [26]. Since 2007, it has been surveyed and evaluated including the representative points of the nationwide watershed and the water quality measuring network [27]. Some previous research had been done to evaluate the ecological health of the Geum River mainstream using integrative approaches including water quality factors and biological communities [16,28]. The purpose of this study is (a) to diagnose the chemical health of Geum River using CPI multi-metric model; (b) to analyze the species composition, trophic and tolerance guilds and determine how are they correlated with water quality factors; (c) to evaluate the biological health using index of biotic integrity model based on fish community; (d) to predict the ecological health (chemical and biological) using artificial neural network model (ANN) along with estimating the validity of ANN; (e) to determine the most important water quality factors affecting the river health by principal component analysis (PCA).

2. Materials and Methods

2.1. Study Period and Sampling Sites Geum River watershed is composed of urban, mountain, and agricultural land-use pattern. The water quality parameters and fish composition are closely related to the land-use pattern. This study was conducted using 19 sampling sites in the mainstream of the Geum-River watershed (Figure1). Geum River originates from Jangsu-eup, North Jeolla Province. It lies between the coordinates of latitude 35◦34047”–37◦03003” and longitude 126◦40025”–128◦03053” and the watershed area is 17,537 km2 (Atique et al., 2019). The total reach length is 414 km and the average elevation is 85.31 m. composition and water quality parameters were investigated from May 2010 to June 2010 (The original dataset can be shared upon written request but is subject to approval from the funding agency). Water 2019, 11, 1729 3 of 22 Water 2019, 11, x FOR PEER REVIEW 3 of 22

FigureFigure 1.1. Sampling sitessites forfor GeumGeum RiverRiver watershed.watershed. 2.2. Water Quality Parameters 2.2. Water Quality Parameters During our study period, we measured eight parameters namely, total nitrogen (TN), total During our study period, we measured eight parameters namely, total nitrogen (TN), total phosphorus (TP), and chlorophyll-a (CHL-a), electrical conductivity (EC), total suspended solids phosphorus (TP), and chlorophyll-a (CHL-a), electrical conductivity (EC), total suspended solids (TSS), biological demand (BOD), chemical oxygen demand (COD), and dissolved oxygen (TSS), biological oxygen demand (BOD), chemical oxygen demand (COD), and dissolved oxygen (DO). A portable multi-parameter analyzer (YSI Sonde Model 6600) was used to measure the electrical (DO). A portable multi-parameter analyzer (YSI Sonde Model 6600) was used to measure the conductivity and dissolved oxygen. The chemical testing standard method was used to measure the electrical conductivity and dissolved oxygen. The chemical testing standard method was used to TN, BOD, and COD which was adopted by the Ministry of the Environment, Korea (MOE, 2006). The measure the TN, BOD, and COD which was adopted by the Ministry of the Environment, Korea ascorbic acid method was used to analyze the total phosphorus (TP) which was also standardized by (MOE, 2006). The ascorbic acid method was used to analyze the total phosphorus (TP) which was the Ministry of the Environment, Korea (MOE, 2006). Total suspended solids (TSS) were determined also standardized by the Ministry of the Environment, Korea (MOE, 2006). Total suspended solids by pre-weighted Whatman GF/C filters method. Chlorophyll-a (CHL-a) concentration was measured (TSS) were determined by pre-weighted Whatman GF/C filters method. Chlorophyll-a (CHL-a) by using a spectrophotometer (Bechman Model DU-65) after extraction in hot ethanol [29]. Nutrient concentration was measured by using a spectrophotometer (Bechman Model DU-65) after extraction analyses were performed thrice to ensure validity and CHL-a was measured twice. in hot ethanol [29]. Nutrient analyses were performed thrice to ensure validity and CHL-a was 2.3.measured Fish Sampling twice. Methods

2.3. FishFish Sampling sampling Methods was performed based on the Ohio EPA (1989) method [30], which was modified by An et al. 2001 for the regional purpose [31]. Fish assemblages were sampled overnight from the GeumFish River sampling using sets was of performed fyke nets (FN; based 20 mon longthe Oh andio 2.4 EPA m high,(1989) mesh method size: [30], 5 5which mm), was netsmodified (GN; × 50by mAn long et al. and 2001 2 m for high, the meshregional size purpose 45 45 mm),[31]. Fish trammel assemblages nets (TN; were 50 m sampled long and overnight 1.0 m high, from mesh the × sizeGeum 12 River12 mm). using The sets cast of fyke net (mesh nets (FN; size: 20 7 m7 lo mm)ng and and 2.4 kick m nethigh, (mesh mesh size: size: 4 5 4× mm)5 mm), were gill used nets × × × to(GN; catch 50 them long fish inand run, 2 m ri fflhigh,e, and mesh pool. size The 45 × fyke 45 mm), net, gill trammel net, and nets trammel (TN; 50 net m werelong and installed 1.0 m along high, themesh shoreline size 12 using× 12 mm). a small The boat cast while net (mesh the cast size: net 7 and× 7 mm) kick netand were kick usednet (mesh in nearshore size: 4 × waters 4 mm) of were the Geumused to River. catch The the samplingfish in run, distance , wasand 200pool. m The and fyke the time net, wasgill net, 60 min. and Aftertrammel collecting net were the samples,installed thealong fish the was shoreline identified using and a anysmall abnormalities boat while the in cast those net fish and were kick also net noted.were used Analysis in nearshore of trophic waters and toleranceof the Geum guilds River. was The done sampling by previous distance regional was studies200 m and [32]. the time was 60 min. After collecting the samples, the fish was identified and any abnormalities in those fish were also noted. Analysis of trophic and tolerance guilds was done by previous regional studies [32].

Water 2019, 11, 1729 4 of 22

2.4. Chemical Pollution Index (CPI) Model To diagnose the chemical health of the stream, the modified CPI model was performed [15]. The CPI 1 1 composed of eight metrics such as M1—total nitrogen (TN, mgL− ), M2—total phosphorus (TP, µgL− ), 1 M3—TN:TP ratio, M4—biological oxygen demand (BOD, mgL− ), M5—total suspended solids (TSS, 1 1 1 mgL− ), M6—electrical conductivity (µScm− ), M7—chlorophyll (CHL, µgL− ), and M8—dissolved -1 oxygen (DO, mgL ). The CPI was calculated using the following formula: CPI = 1/8 (CTN/CTN0 + C /C + C /C + C /C + C /C + C /C + C /C C /C ). TP TP0 TN:TP TN:TP0 BOD BOD0 TSS TSS0 EC EC0 CHL CHL0 − DO DO0 Where, CTN,TP,TN:TP,BOD,TSS,EC,CHL,DO is the measured value and CTN0,TP0,TN:TP0,BOD0,TSS0,EC0,CHL0,DO0 is the critical or standard value which was defined by the Korean Ministry of Environment. It is widely accepted that TP, TN, and TN:TP ratio are the indicators of nutrient pollution in the river system whereas BOD, TSS, and EC are the pointers of organic pollution, non-algal turbidity, and ionic pollution, respectively [7,8]. Trophic status of the waterbody can be determined by the concentration of CHL-a. DO can influence the ’s life within the waterbody and it is an important indicator of pollution and eutrophication in rivers [8]. The chemical health has been categorized as excellent (<0), good (0.0–1.0), fair (1.1–2.0), poor (2.1–3.0), and very poor (>3.1).

2.5. Index of Biotic Integrity (IBI) Model The biological health assessment of the Geum River was diagnosed by the multi-metric IBI model based on fish communities [32]. In the IBI model, we used eight metrics which were modified by An et al. (2006), such as, M1—total number of native fish species, M2—number of riffle benthic species, M3—number of sensitive species, M4—proportion of individuals as tolerant species, M5—proportion of individuals as omnivore species, M6—proportion of individuals as native species, M7—total number of native individuals, and M8—percent of individuals with anomalies [25]. Each metric had been assigned as scored 5, 3, and 1. The Geum River health condition was evaluated based on the obtained scores and the categories are excellent (36–40), good (28–34), fair (20–26), poor (14–18), and very poor (8–13).

2.6. Artificial Neural Network, Principal Component Analysis, and Cluster Analysis The nonlinear nature of water quality and fish data makes it difficult to interpret the spatio-temporal variations and ecological health of the river. For this reason, statistical learning approaches (ANN) and multivariate statistical approaches (PCA and CA) are used for providing the representative and reliable analysis and drawing out meaningful conclusions. An artificial neural network is a good tool for the simplification of complex non-linear physical, chemical, and biological data. This statistical tool has been regarded as one of the most effective tools for the prediction of ecological dynamics and variations [33]. We used various chemical and biological variables as the input layer to predict the ecological health of the Geum River watershed and to estimate the validity of ANN for forecasting of river health. To determine the accuracy of ANN, we used some accuracy metrics such as mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination (R2). The lower MAE and RMSE values indicate a better model, while the higher R2 value indicates a better model. The principal component analysis (PCA) was used to determine the most important water quality factors affecting the ecological health of the Geum River watershed. The cluster analysis (CA) was carried out to make a strong spatial association based on water quality parameters and fish composition.

2.7. Statistical Analysis Log-transformed regression analysis of water quality parameters was performed in Sigma Plot version 10 [34]. Principal component and cluster analysis were done using PAST software [35]. The artificial neural network analysis was performed by R 3.5.2 version. Water 2019, 11, 1729 5 of 22

Water 2019, 11, x FOR PEER REVIEW 5 of 22 3. Results and Discussion 3. Results and Discussions 3.1. Dynamics of Water Quality Parameters 3.1. Dynamics of Water Quality Parameters The physicochemical parameters of the Geum River watershed varied from upstream to downstreamThe physicochemical region and showed parameters a distinct of the heterogeneity Geum River featurewatershed (Figure varied2). from The concentrationupstream to of TPdownstream and TN wasregion lower and inshowed G01–G10 a distinct sites, heterogeneity while it was feature higher (Figure in G11–G19 2). The sites. concentration In site G10, of TP the TN:TPand TN ratio was was lower higher in G01–G10 compared sites, to while the other it was sites, high suggestinger in G11–G19 that sites. the siteIn site is highlyG10, the P-limited. TN:TP ratio The was higher compared to the other sites, suggesting that the site is highly P-limited. The value of BOD value of BOD increased from site G01 to G18. The other parameters (COD, TSS, EC, CHL) also increased from site G01 to G18. The other parameters (COD, TSS, EC, CHL) also followed the same followed the same pattern. Results of the present study revealed that the nutrient (TN, TP, TN:TP pattern. Results of the present study revealed that the nutrient (TN,TP, TN:TP ratios) regime, organic ratios) regime, organic content (BOD, COD), suspended solids (TSS), ionic contents (EC), and primary content (BOD, COD), suspended solids (TSS), ionic contents (EC), and primary (CHL) productivity (CHL) was highly affected by the land-use pattern of the Geum River watershed. As a was highly affected by the land-use pattern of the Geum River watershed. As a result, the upstream result,of Geum the upstream River watershed of Geum was River minimally watershed influenced was minimally by the water influenced quality by factors the water in comparison quality factors to in comparisonthe downstream to the region, downstream which is region, in line whichwith the is infindings line with of Atique the findings and An of (2018) Atique in and the AnKorean (2018) in thewatersheds Korean watersheds[16]. Kim et [16al.]. (2010) Kim etsuggested al. (2010) that suggested because that of the because higher of flushing the higher rate flushing from the rate fromupstream the upstream to downstream, to downstream, the concentrations the concentrations of physicochemical of physicochemical parameters parameters were were higher higher downstreamdownstream [36 [36].].

FigureFigure 2. 2.Variation Variation of of water water quality quality parameters parameters onon the basis of sites sites in in the the Geum Geum River River watershed. watershed.

Water 2019, 11, 1729 6 of 22 Water 2019, 11, x FOR PEER REVIEW 6 of 22

3.2.3.2. Spatial Spatial Distribution Distribution of of Fish Fish Assemblages Assemblages TheThe distribution distribution of of fish fish assemblages assemblages highly highly fluctuated fluctuated from from upstream upstream to to the the downstream downstream of of the the GeumGeum River River watershed watershed and and did did not not show show a a distinctivedistinctive patternpattern (Figure(Figure3 3).). TheThe highest number (18) of ofnative native fish fish species species was was found in G12G12 sitesite whereaswhereas itit waswaslower lower (3) (3) in in sites sites G16–G18. G16–G18. It It was was a a remarkableremarkable finding finding that that the the total total number number of native of native individuals individuals were were much much lower lower in the in downstream the downstream site G19.site TheG19. relative The relative abundance abundance of tolerant of speciestolerant fluctuated species fluctuated from site G01from to site G19. G01 Another to G19. significant Another findingsignificant was thatfinding no sensitive was that species no sensitive were foundspecies in we sitesre found G09 and in G10,sites especiallyG09 and G10, in the especially downstream in the region.downstream It means region. that the It downstream means that region the isdownstream highly polluted region by naturalis highly and polluted anthropogenic by natural activities, and makinganthropogenic it an unsuitable activities, habitat making for the it sensitivean unsuitable species. habitat Our resultsfor the concurred sensitive withspecies. some Our previous results findingsconcurred [16, 37with]. The some omnivore previous fish findings species [16,37]. was higher The omnivore in site G17 fish and species lower inwas G13. higher Insectivores in site G17 were and notlower observed in G13. in Insectivores sites G16 and were G17 not during observed the study in sites period. G16 and G17 during the study period.

FigureFigure 3. Variation 3. Variation of of fish fish abundanceabundance and and assemblages assemblages according according to the to sites the in sites the inGeum the River Geum River watershed. watershed.

3.3. Empirical Model of Chlorophyll and Nutrients An empirical model was used to define whether the aquatic ecosystem was phosphorus limited, nitrogen-limited, or phosphorus–nitrogen co-limited. The empirical model analysis among log- transformed CHL-TP, CHL-TN, and CHL-EC showed a strong positive linear relationship in the watershed (Figure 4). In CHL-TP (R2 = 0.45, p < 0.01, n = 19) model, the variation of CHL was explained

Water 2019, 11, 1729 7 of 22

3.3. Empirical Model of Chlorophyll and Nutrients An empirical model was used to define whether the aquatic ecosystem was phosphorus limited, nitrogen-limited, or phosphorus–nitrogen co-limited. The empirical model analysis among log-transformedWater 2019, 11, x FOR CHL-TP, PEER REVIEW CHL-TN, and CHL-EC showed a strong positive linear relationship in7 of the 22 watershed (Figure4). In CHL-TP (R 2 = 0.45, p < 0.01, n = 19) model, the variation of CHL was explained 2 45%45% forfor thethe concentrationconcentration of of TP TP wherein wherein CHL-TN CHL-TN (R (R2= =0.56, 0.56,p p< < 0.01,0.01, nn == 19) model, itit waswas 5656 percentpercent 2 forfor TN.TN. TheThe changechange ofof CHLCHL concentrationconcentration in in the the watershed watershed accounted accounted for for 18% 18% with with TN:TP TN:TP (R (R2= = 0.18,0.18, 2 pp< < 0.01,0.01, nn == 19) and 59%59% withwith ECEC(R (R2 == 0.59, p < 0.01, n = 19).19). The The TN:TP TN:TP ra ratiotio was more than 20,20, meansmeans thatthat thethe mainstreammainstream ofof GeumGeum RiverRiver watershedwatershed waswas P-limitedP-limited whichwhich isis similarsimilar withwith somesome previousprevious studiesstudies inin thethe KoreanKorean watershedwatershed [[16,37].16,37]. TheThe presentpresent studystudy indicatesindicates thatthat thethe electricalelectrical conductivityconductivity waswas highlyhighly alteredaltered inin thethe watershed,watershed, whichwhich meansmeans thatthat degradationdegradation ofof waterwater qualityquality happenshappens becausebecause ofof urbanurban andand agriculturalagricultural run-orun-offff [ 16[16].].

FigureFigure 4.4. EmpiricalEmpirical RelationshipRelationship among TP, TN,TN, TN:TP,TN:TP, andand CHL.CHL. (TN—total(TN—total nitrogen,nitrogen, TP—totalTP—total phosphorus,phosphorus, CHL—chlorophyll).CHL—chlorophyll).

3.4.3.4. FishFish TrophicTrophic andand ToleranceTolerance Guilds Guilds in in Relation Relation to to Water Water Quality Quality Parameters Parameters TheThe abundanceabundance ofof trophictrophic guildguild isis closelyclosely relatedrelated toto waterwater qualityquality factorsfactors inin thethe watershedwatershed µ 1 (Figure(Figure5 ).5). The The highest highest abundance abundance of omnivoresof was wa founds found in the in concentration the concentration between between 20 to 80 20 gLto −80 µ 1 ofμgL TP;−1 onof TP; the on other the hand,other hand, the lowest the lowest abundance abundance of omnivores of omnivores was observed was observed in 140 in gL140− μofgL TP.−1 of The TP. abundanceThe abundance of insectivores of insectivores with TPwith did TP not did display not display any specific any specific functional functional relationship. relationship. It is notable It is thatnotable the carnivoresthat the showed ashowed positive a functionalpositive func relationshiptional relationship with TP, suggestingwith TP, suggesting carnivore richness increasedrichness increased with the concentrationwith the concentration of TP; hence of TP; carnivores hence carnivores were abundant were inabundant the watershed. in the watershed. A similar patternA similar was pattern observed was for observed carnivores for carnivores in relation toin TN,relation BOD, to COD,TN, BOD, and CHL.COD, Theand omnivoresCHL. The omnivores suggested asuggested negative functionala negative relationshipfunctional relationship with the concentration with the concentration of TN, BOD, of COD, TN, BOD, and CHL. COD, The and abundance CHL. The ofabundance insectivores of pointsinsectivores out a positivepoints out linear a positive relationship linear with relationship nutrients with (TP, TN),nutrients organic (TP, matters TN), organic (BOD andmatters COD), (BOD and and primary COD), productivity and primary (CHL). productivity (CHL). Tolerance guilds of the watershed are highly influenced by the nutrients (TN and TP), organic contents (BOD and COD), and chlorophyll (Figure 6). The relative abundance of tolerant and intermediate species presented a matchless relationship with the concentration of TP. The abundance of sensitive species was lower in the Geum River watershed and they can live with a minimum concentration of TP, TN, BOD, COD, and CHL. The sensitive species showed quick response to increasing levels of nutrients, organic matters, and eutrophication in the watershed. The abundance of tolerant and intermediate species disclosed a positive linear functional relationship with TP, TN,

Water 2019, 11, x FOR PEER REVIEW 8 of 22

COD, BOD, and chlorophyll. In contrast, sensitive species exhibited a negative functional relationship with TP, TN, COD, and chlorophyll. It is widely accepted that nutrients, organic matters, and chlorophyll can alter the trophic and tolerance guilds in the watershed. Earlier research on trophic and tolerance guilds revealed that the tolerant species had a positive linear relationship and sensitive species had a negative functional relationship with water quality parameters, which supports our present research [16,30]. The higher

Waterabundance2019, 11, 1729of omnivores and carnivores indicates that the watershed is polluted because of excessive8 of 22 nutrient and organic matter loadings from upstream to downstream [37].

(a) Trophic guilds Vs. nutrients (TP, TN) 60 G17 100 G16 100 (r = 0.38 , p < 0.01) G18 (r = 0.38 , p < 0.01) G07 50 G19 G09 G08 G13 G10 80 G11 G15 80 G12 G03 G01 G02 G14 40 60 G04 30 60 G05 G06 G06 G15 40 G05 20 40 G04 G03 G01 G19 G10 G14 G13 G02 20 G08 % Carnivores G09 % Omnivores % G19 G11 G15 10 G06G05 G12 G14 G12 Insectivores % G07 G09 G02G03 20 G17 G16 G08G04G07G01 0 0 G10G17G16 G18 G11 (r = - 0.54 , p < 0.01) G13G18 0 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 -1 -1 -1 TP (μgL ) TP (μgL ) TP (μgL ) 60 G17G16 (r = - 0.48 , p < 0.01) 100 (r = 0.38 , p < 0.01) 100 G18 G19 G07 50 G09 G13 G10G08 80 G11 G15 80 G12 G03 G01 G02 G14 40 (r = 0.25 , p < 0.01) 60 G04 30 60 G05 G06 G06 G15 40 G05 G04 20 40 G03 G01G19 G13 G02 G10 G14 % Omnivores % 20 G08 % Carnivores G06G09 G14G19 G12 G11 G07 G15 10 G05 G12 % Insectiovores % G09 G02G03 20 G17G16 G04 G08G07G01 0 0 G10G17G16 G18 G11 G13 G18 (r = 0.38 , p < 0.01) 0 1234567 1234567 1234567 -1 -1 -1 TN (mgL ) TN (mgL ) TN (mgL ) (b) Trophic guilds Vs. organic matters (BOD, COD) 60 G17 G16 100 100 (r = 0.24 , p < 0.01) G18 G07 G19 G09 50 G10G08 G13 G15 80 (r = - 0.38 , p < 0.01) 80 G11 G03G01 G02 G12 40 G14 (r = 0.31 , p < 0.01) 60 G04 60 G05 30 G06 G06 G15 40 G05 G04 20 40 G03G01 G19 G13 G02 G10 G14 % Omnivores % 20 G09 G19G11 G14G12 G08 % Carnivores 10 G05 G06 G12 G15

% Insectivores % G07 20 G09 G03G02G01G08 G04 0 G17 G16 G10G07 G11 G17G16 G18 G13 G18 0 0 0123456 0123456 0123456 -1 -1 -1 BOD (mgL ) BOD (mgL ) BOD (mgL ) 60 G17 100 G16 100 (r = 0.17 , p < 0.01) G18 (r = 0.38 , p < 0.01) G07 50 G19 G09 G08 G13 G10 80 G11 G15 80 G12 40 G03 G01 G02 G14 60 G04 60 G05 30 G06 G06 G15 40 G05 G04 20 40 G03 G01 G19 G13 G10 G14 G02 20 G09 G08 Carnivores % % Omnivores % 10 G06 G11G19G14G12 Insectivores % G15 G05 G12

G07 G09 G02G03 20 G17 G16 G01G08G07G04 0 0 G10 G11G17 G18G16 (r = - 0.34 , p < 0.01) G18G13 0 23456789 23456789 23456789 -1 -1 -1 COD (mgL ) COD (mgL ) COD (mgL ) (c) Trophic guilds Vs. algal growth (CHL) 60 G17 G16 100 100 (r = 0.23 , p < 0.01) G18 G19 (r = 0.26 , p < 0.01) G07 50 G08 G09 G13 G10 80 G11 G15 80 G12 40 G03G01 (r = - 0.34 , p < 0.01) G02 G14 60 G04 60 G05 30 G06 G06 G15 40 G05 20 G04 G01 40 G03G19 G14G13 % Omnivores G02 G10 G19 G11 20 G08 Carnivores % 10 G05G06 G09 G12 G14 G12 % Insectivores G07 G15 G09 G02G03 20 G07G04G08G01 0 G17 G16 0 G10G17 G11 G16 G18 G13 G18 0 0 20406080100 0 20406080100 0 20406080100 -1 -1 -1 CHL (μgL ) CHL (μgL ) CHL (μgL ) FigureFigure 5. 5.Trophic Trophic guild analysis analysis with waterwith qualitywater factorsquality (TN—total factors (TN—total nitrogen, TP—total nitrogen, phosphorus, TP—total N:Pphosphorus, ratios, BOD—biological N:P ratios, BOD—biological oxygen demand, oxyge COD—chemicaln demand, COD—chemical oxygen demand, oxygen CHL—chlorophyll). demand, CHL— chlorophyll). Tolerance guilds of the watershed are highly influenced by the nutrients (TN and TP), organic contents (BOD and COD), and chlorophyll (Figure6). The relative abundance of tolerant and intermediate species presented a matchless relationship with the concentration of TP. The abundance of sensitive species was lower in the Geum River watershed and they can live with a minimum concentration of TP, TN, BOD, COD, and CHL. The sensitive species showed quick response to increasing levels of nutrients, organic matters, and eutrophication in the watershed. The abundance of tolerant and intermediate species disclosed a positive linear functional relationship with TP, TN, COD, BOD, and chlorophyll. In contrast, sensitive species exhibited a negative functional relationship with TP, TN, COD, and chlorophyll. Water 2019, 11, 1729 9 of 22 Water 2019, 11, x FOR PEER REVIEW 9 of 22

(a) Tolerance guilds Vs. Nutrients (TP, TN) 100 100 6 (r = 0.23 , p < 0.01) (r = - 0.33 , p < 0.01) G10 G07 G16 G15 5 G06 80 G14 80 G19 G09 G08 G13 G17 G02G03 G04 G01 4 G18 G12 60 G11 60 G05 G05G01 G06 3 G03 G11 G18 G12 40 40 G03 G08 G02 G17 2 G06G04 G13 G04 G09 G08 G14 G05G01 G19 20

20 Species Sensitive % 1 % Tolerant Species Tolerant % G10 G07 G15 G02 G16 (r = 0.13 , p < 0.01) Species Intermediate % 0 0 0 0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180 10 20 30 40 50 60 -1 -1 μ μ -1 TP (μgL ) TP ( gL ) TP ( gL ) 5.5 100 100 (r = - 0.42 , p < 0.01) G06 G10G07 G16 G15 (r = 0.15 , p < 0.01) 5.0 80 G14 80 G19 G09G08 G13 4.5 G17 G01 G04 G02 G03 G12 G18 4.0 60 G11 60 G05 G06 3.5 G03 G11 G12 G18 40 3.0 G05 G01 40 G03 G04G02 G17 G06 G13 G04 G09G08 G14 G05G01G19 2.5

20 20 Species Sensitive % G08 G15 % Tolerant Species Species Tolerant % G10G07 2.0 (r = 0.16 , p < 0.01) G02 G16 % Intermediate Species Species Intermediate % 1.5 0 0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 1234567 1234567 -1 -1 -1 TN (mgL ) TN (mgL ) TN (mgL ) (b) Tolerance guilds Vs. organic matters (BOD, COD) 100 100 5.5 (r = 0.16 , p < 0.01) (r = 0.37 , p < 0.01) G10G07 G16 G15 5.0 G06 80 G14 80 G19 G09G08 G13 4.5 G17 G01 G03G02 G04 G12 G18 4.0 60 G11 60 G05 G06 3.5 G03 G11 G12 G18 40 G05 G01 40 G03 3.0 G02 G17 G06G04 G13 G05 G04G19 G09G08 G14 G01 2.5 20

% Tolerant Species Tolerant % 20 G10G07 G15 Species Sensitive % G08

G02 G16 2.0

(r = 0.15 , p < 0.01) Species Intermediate % 0 0 1.5 0123456 0123456 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 -1 -1 -1 BOD (mgL ) BOD (mgL ) BOD (mgL ) 100 100 5.5 G10 G07 G16G15 (r = 0.13 , p < 0.01) (r = - 0.20 , p < 0.01) 5.0 G06 80 G19 80 G09 G14 G17 G08 G13 4.5 G01 G02 G03 G04 60 G11 G12G18 4.0 G05 60 G03 G06 3.5 G12G18 G11 40 40 3.0 G05G01 G02 G04 G03 G17 G09 G06 G13 G08 G14 G05G01 G04 G19 2.5 20 20

% Tolerant Species Tolerant % G08 G10 G07 G15 Species Sensitive % (r = 0.26 , p < 0.01) G02 G16 2.0 % Intermediate Species Intermediate % 0 0 1.5 23456789 23456789 2.5 3.0 3.5 4.0 4.5 5.0 5.5 -1 -1 -1 COD (mgL ) COD (mgL ) COD (mgL ) (c) Tolerance guilds Vs. algal growth (CHL) 100 100 5.5 G06 (r = - 0.21 , p < 0.01) G07G10 G16 G15 (r = 0.23 , p < 0.01) 5.0

80 G19 80 G08 G09 G14 4.5 G17 G13 G01 4.0 G02 G03 G04 60 G11 60 G12G18 G05 3.5 G03 G06 G11 G05 G01 40 G12G18 40 3.0 G02 G03 G04G06 G13 G17 G08 G09 G14 G05G04G19G01 2.5

% TolerantSpecies % 20 20 G08 % Sensitive Species Sensitive % G07G10 G16 G15 2.0 (r = 0.10 , p < 0.01) Species Intermedaite % G02 1.5 0 0 0 20406080100 0 20406080100 012345678 μ -1 -1 -1 CHL ( gL ) CHL (μgL ) CHL (μgL ) FigureFigure 6.6.Tolerance Tolerance guild guild analysis analysis with waterwith qualitywater factorsquality (TN—total factors (TN—total nitrogen, TP—total nitrogen, phosphorus, TP—total N:Pphosphorus, ratios, BOD—biological N:P ratios, BOD—biological oxygen demand, oxyge COD—chemicaln demand, COD—chemical oxygen demand, oxygen CHL—chlorophyll). demand, CHL— chlorophyll). It is widely accepted that nutrients, organic matters, and chlorophyll can alter the trophic and tolerance3.5. Chemical guilds Health in the Assessment watershed. Using Earlier Chemical research Pollution on trophic Index (CPI) and tolerance Model guilds revealed that the tolerant species had a positive linear relationship and sensitive species had a negative functional The chemical health assessment of the watershed was carried out by a modified chemical relationship with water quality parameters, which supports our present research [16,30]. The higher pollution index. In the CPI model, we used eight water quality factors (Table 1). The concentration of abundance of omnivores and carnivores indicates that the watershed is polluted because of excessive TN and TP exceeded the standard value according to the Ministry of Environment of Korea. The nutrient and organic matter loadings from upstream to downstream [37]. value of TN:TP ratio indicates that the Geum River watershed is a P-limited system and trophic status is mesotrophic to oligotrophic. The BOD concentration was >1 mgL−1 in all sites except site G03 and G10, suggesting that the watershed has been degraded. Suspended solids and ionic contents were higher in the downstream region because of high flushing rate, agricultural and urban run-off. The CHL values from the sites G09 to G18 were higher, suggesting that these sites are in eutrophic

Water 2019, 11, 1729 10 of 22

3.5. Chemical Health Assessment Using Chemical Pollution Index (CPI) Model The chemical health assessment of the watershed was carried out by a modified chemical pollution index. In the CPI model, we used eight water quality factors (Table1). The concentration of TN and TP exceeded the standard value according to the Ministry of Environment of Korea. The value of TN:TP ratio indicates that the Geum River watershed is a P-limited system and trophic status is 1 mesotrophic to oligotrophic. The BOD concentration was >1 mgL− in all sites except site G03 and G10, suggesting that the watershed has been degraded. Suspended solids and ionic contents were higher in the downstream region because of high flushing rate, agricultural and urban run-off. The CHL values from the sites G09 to G18 were higher, suggesting that these sites are in eutrophic condition and are highly deteriorated. The CPI model suggested that the upstream of the watershed was in good–fair condition whereas it was in a poor–very poor condition in the downstream region except for G19. Some previous studies advocated that TP, TN, and TN:TP are the key regulating factors for the algal growth in the waterbody [16,38]. The present research signposts that watershed health is in an anomalous state because of excessive nutrient and organic matter and ionic content enrichment. Organic pollution, non-algal turbidity, and ionic pollution can be determined by the concentration of BOD, TSS, and EC, correspondingly, in the watershed. Our results harmonized with some aforementioned findings that downstream is largely affected by water pollutants [16,39].

3.6. Fish Community Analysis Analysis of species composition in the watershed based on relative abundance (RA), the total number of individuals (TNI), and the total number of species (TNS) along with tolerance, trophic, and habitat guilds showed a substantial disproportion (Table2). A total of 44 species were observed in the mainstream of the Geum River watershed. The dominated species in the watershed were Zacco platypus (32.99%), Zacco koreanus (12.43%), and Squalidus chankaensis tsuchigae (11.19%), etc., which was similar with some previous studies on watershed [28]. One (Pseudopungtungia nigra) was found in the upstream of the watershed. Three exotic fish (Micropterus salmoides, Lepomis macrochirus, and Cyprinus carpio) were observed in the watershed. Analysis of tolerance guilds showed that 6 sensitive species, 25 intermediate species, and 13 tolerant species were detected in the watershed. Sensitive species with relatively high abundance was Zacco koreanus (12.43%), while tolerant species was Zacco platypus (32.99%), and intermediate species was Squalidus chankaensis tsuchigae (11.19%). In the Geum River watershed, the abundance of tolerant species was higher. According to the spatial tolerance guild, tolerant species accounted for 71.4% of the total species in G19 of downstream. The ratio of tolerant species increased as it moved from upstream to downstream. According to the category of trophic guilds, insectivore was 19 species, carnivore was 7 species, and the omnivore was 18 species. It has been reported that an increase in the inflow of organic and toxic substances decreases the relative abundance of sensitive species while increases the abundance of insectivore species [17]. Water 2019, 11, 1729 11 of 22

Table 1. Chemical health assessment using the chemical pollution index (CPI) model.

Model Metric G01 G02 G03 G04 G05 G06 G07 G08 G09 G10 G11 G12 G13 G14 G15 G16 G17 G18 G19 M1-TN 1 2.52 1.68 1.82 1.41 2.06 1.98 2.07 2.02 2.09 1.62 6.17 4.6 2.86 2.60 5.562 2.6425 2.406 5.53 2.92 (mgL− ) M2-TP 1 57.5 14 17 52 57 50.5 56 50 26 7.5 148 170 137.5 95.5 156.5 108 83 142 117 (µgL− ) M3-TN:TP 43.95 120.3 107.32 27.1 36.25 39.31 37.00 40.4 80.61 150 41.69 27.05 20.84 27.30 35.53 24.46 28.98 39 24.97 M4-BOD 1 1.2 1.05 0.95 1.8 0.7 1.5 1.3 1.2 1.15 0.85 2.45 3.6 3.5 3.15 4.85 3.85 3.05 4.35 2.15 (mgL− ) M5-TSS 1 12.6 3.55 2.4 4.8 11.55 10.45 4.65 4.6 4.3 2.8 22.3 28.6 21.45 22.1 28.55 16.35 13.05 22 8.3 (mgL− ) M6-EC 1 162.5 111.5 105.5 113 120.5 156 152 50 156 142 213 264.5 244 245 254.5 245 233.5 259.5 217 (µScm− ) M7-CHL 1 7.2 1 1.85 2.8 1.4 3.2 2 3 24.95 4.35 30.55 60.45 44.05 43.45 87.2 53.1 20.2 68.8 4 (µgL− ) M8-DO 1 10.95 9.15 10 8.4 11 10.25 9.5 9.3 10.5 11.1 10.65 10.15 9.9 6.25 10.8 8.9 8.1 8.85 8.2 (mgL− ) CPI Value 1.32 0.69 0.65 0.82 0.89 1.046 0.82 0.76 1.72 0.98 3.46 5.02 3.79 3.63 6.32 3.94 2.26 5.27 1.49 Chemical Very Very Very Very Very Very Very Fair Good Good Good Good Good Good Good Fair Good Poor Fair Criteria Poor Poor Poor Poor Poor Poor Poor Water 2019, 11, 1729 12 of 22

Table 2. Species composition analysis (To—tolerance guild, Tr—trophic guilds, Ha—habitat guilds, TS—tolerant species, SS—sensitive species, IS—intermediate species, O—omnivores, I—insectivores, C—carnivores, RB—riffle benthic, TNI—total number of individuals, TNS—total number of species, RA—relative abundance, ¥—exotic fish).

Species To Tr Ha G01 G02 G03 G04 G05 G06 G07 G08 G09 G10 G11 G12 G13 G14 G15 G16 G17 G18 G19 TNI RA Zacco platypus TS O 201 86 113 38 90 34 171 35 30 46 7 4 3 19 7 884 32.99 Zacco koreanus SS I 32 161 46 32 37 22 3 333 12.43 Squalidus chankaensis IS O 1 1 61 237 300 11.19 tsuchigae Microphysogobio jeoni IS I 1 159 17 177 6.60 Opsarichthys uncirostris TS C 1 1 1 4 25 9 31 4 30 1 107 3.99 amurensis Hamibarbus labeo TS I 2 1 1 4 9 23 31 12 1 84 3.13 Squalidus japonicus TS O 14 11 1 10 26 12 1 75 2.80 coreanus Squalidus gracilis majimae IS I 5 11 49 4 69 2.57 Pungtungia herzi IS I 3 11 11 5 1 25 3 8 1 68 2.54 Acheilognathus koreensis IS O 3 60 4 1 68 2.54 Acheilognathus lanceolatus IS O 8 6 37 4 5 2 1 63 2.35 Pseudogobio esocinus IS I 6 2 3 11 12 4 4 1 3 12 3 2 63 2.35 Microphysogobio yaluensis IS O RB 22 14 5 12 1 54 2.01 Rhinogobius brunneus IS I RB 23 6 1 6 2 2 3 1 44 1.64 Tridentiger brevispinis IS I RB 1 12 6 1 16 36 1.34 Sarcocheilichthys variegatus IS I 9 18 1 28 1.04 wakiyae Acheilognathus yamatsuatea IS O 4 9 2 6 6 27 1.01 Odontobutis platycephala SS C 4 8 7 1 5 25 0.93 Squaliobarbus curriculus IS O 3 3 2 4 6 2 1 21 0.78 Coreoleuciscus splendidus SS I RB 3 17 1 21 0.78 Water 2019, 11, 1729 13 of 22

Table 2. Cont.

Species To Tr Ha G01 G02 G03 G04 G05 G06 G07 G08 G09 G10 G11 G12 G13 G14 G15 G16 G17 G18 G19 TNI RA Pseudopungtungia nigra SS I 3 2 11 2 2 20 0.75 Hamibarbus longirostris IS I 2 4 1 1 1 3 2 1 2 17 0.63 Hemiculter eigenmanni TS O 2 7 5 14 0.52 Pseudobagrus koreanus IS I RB 1 5 1 4 11 0.41 Iksookimia koreensis IS I RB 6 2 1 9 0.34 Micropterus salmoides¥ TS C 4 4 1 9 0.34 Acanthorhodeus gracilis IS O 2 4 1 7 0.26 Odontobutis interrupta IS C 3 4 7 0.26 Coreoperca herzi SS C 2 1 3 1 7 0.26 Rhodeus uyekii IS O 5 5 0.19 Erythroculter erythropterus TS C 1 1 2 4 0.15 Acanthorhodeus IS O 1 2 3 0.11 macropterus Carassius auratus TS O 1 1 1 3 0.11 Cyprinus carpio¥ TS O 1 2 3 0.11 Pseudorasbora parva TS O 1 2 3 0.11 Acheilognathus rhombeus IS O 2 2 0.07 Rhynchocypris oxycephalus SS I 2 2 0.07 Siniperca scherzeri IS C 1 1 0.04 Cobitis lutheri IS I 1 1 0.04 Sarcocheilichthys IS I 1 1 0.04 nigripinnis morii Pseudobagrus fulvidraco TS I 1 1 0.04 Misgurnus mizolepis TS O 1 1 0.04 Misgurnus anguillicaudatus TS O 1 1 0.04 Lepomis macrochirus¥ TS I 1 1 0.04 TNS 16 9 15 14 14 14 6 15 12 4 5 18 13 9 9 4 3 3 7 TNI 314 285 257 136 181 126 196 185 339 57 28 134 241 34 71 38 20 30 8 2680 Water 2019, 11, 1729 14 of 22

3.7. Biological Health Assessment Using the Index of Biotic Integrity (IBI) Model The index of biotic integrity (IBI) model based on fish assemblages had been used to assess the ecological health of the mainstream in Geum River watershed (Table3). The average IBI value in the mainstream of the watershed was 17.68, which indicates that the watershed is in poor condition. Among the 19 sites, 8 sites were in fair condition, 6 sites were in poor condition, and 5 sites were in very poor condition. The upstream of the watershed was in fair condition but degraded from upstream to downstream because of the higher abundance of tolerant and omnivore fish species as well as a lower amount of sensitive species and insectivores. In addition, anomalies were collected at G09. The water quality of Lake Daecheong deteriorated because of the inflow of nutrients from and dominant tolerant species. Sites G11, G12, and G13 are downstream of Lake Daecheong and categorized to be in fair and poor state. Sites G15, G16, and G17, down of Baekje Bridge and Baekma River Bridge, revealed that the ecological health was in very poor condition. The abundance of native, riffle benthic, and sensitive fish species declined with an increase in the excessive nutrients and organic matter content and non-algal turbidity, which strongly supports our present study [16]. In 1999 and 2000, the USEPA put forward that the higher abundance of omnivores, tolerant, carnivores was found in the degraded system [5,40].

Table 3. Biological health assessment using index of biotic integrity model (TNS: total number of , RBS: total number of riffle benthic species, SS: total number of sensitive species, % TS: % individuals as tolerant species, % OS: % individual as omnivores, % IS: % individuals as native insectivores, TNI: total number of individual, % AI: % individuals with anomalies.

Sampling Model Metrics Sites Health TNS RBS SS % TS % OS % IS TNI % AI IBI Status G01 16(3) 4(3) 3(3) 64.97(1) 72.29(1) 26.11(3) 314(5) 0(5) 24 Fair G02 9(1) 4(3) 4(3) 30.18(1) 30.18(3) 67.02(5) 285(5) 0(5) 26 Fair G03 15(3) 4(3) 4(3) 44.75(1) 71.21(1) 25.29(3) 257(5) 0(5) 24 Fair G04 14(1) 1(1) 4(3) 29.41(1) 38.97(3) 59.56(5) 136(1) 0(5) 20 Fair G05 14(1) 0(1) 3(3) 52.49(1) 55.80(1) 37.57(3) 177(3) 0(5) 18 Poor G06 14(1) 4(3) 5(5) 26.98(1) 47.62(1) 44.44(3) 126(1) 0(5) 20 Fair G07 6(1) 2(3) 0(1) 87.76(1) 90.31(1) 9.18(1) 196(1) 0(5) 14 Poor G08 15(3) 3(3) 2(1) 22.70(1) 83.24(1) 14.59(1) 185(1) 0(5) 16 Poor G09 11(1) 0(1) 0(1) 24.19(1) 86.14(1) 5.31(1) 335(3) 2.7(1) 10 Very Poor G10 4(1) 1(1) 0(1) 87.72(1) 80.70(1) 19.30(1) 57(1) 0(5) 12 Very Poor G11 5(1) 0(1) 0(1) 57.14(1) 25.00(3) 75.00(5) 28(1) 0(5) 18 Poor G12 18(3) 3(3) 0(1) 39.55(1) 23.13(3) 69.40(5) 134(1) 0(5) 22 Fair G13 12(1) 0(1) 0(1) 26.97(1) 5.81(5) 80.91(5) 240(3) 0(5) 22 Fair G14 9(1) 0(1) 0(1) 23.53(1) 23.53(3) 64.71(5) 34(1) 0(5) 18 Poor G15 8(1) 1(1) 0(1) 87.32(1) 47.89(1) 9.86(1) 69(1) 0(5) 12 Very Poor G16 3(1) 0(1) 0(1) 89.47(1) 97.37(1) 0.0(1) 37(1) 0(5) 12 Very Poor G17 3(1) 0(1) 0(1) 70.00(1) 100.00(1) 0.0(1) 20(1) 0(5) 12 Very Poor G18 3(1) 1(1) 0(1) 40.00(1) 6.67(5) 93.3(5) 30(1) 0(5) 20 Fair G19 6(1) 1(1) 0(1) 75.00(1) 25.00(3) 25.00(3) 7(1) 0(5) 16 Poor

3.8. Ecological Health in Relation to Chemical Parameters and Fish Model Metrics The ecological health of the watershed is closely related to water quality parameters (Figure7). The ecological health of the watershed showed a linearly decreasing trend with increasing the concentration of TP, TN, BOD, COD, TSS, and CHL. The ecological health are related with TP (r = 0.37, p < 0.01), − Water 2019, 11, 1729 15 of 22

CHL (r = 0.37, p < 0.01), BOD (r = 0.42, p < 0.01), and COD (r = 0.47, p < 0.01) in the watershed. − − − It indicates that water quality parameters affect the ecological health of rivers. The ecological health declined with the increased relative abundance of omnivores (R2 = 0.31, r = 0.55, n = 19, p < 0.01), and − tolerant species ((R2 = 0.25, r = 0.50, n = 19, p < 0.01), Figure8). In the IBI model, regression analysis − of IBI-omnivores and IBI-tolerant species was explained as 31% and 25%, respectively. The ecological health of the watershed showed an upward trend with the increasing richness of insectivores, sensitive species, native fish and individuals. The empirical relationship between IBI-insectivores (R2 = 0.45, r = 0.67, n =19, p < 0.01) and IBI-total number of native fish species (R2 = 0.30, r = 0.55, n = 19, p < 0.01) showed a strong positive relationship in comparison to IBI-sensitive species (R2 = 0.19, r = 0.44, n = 19, p < 0.01) and IBI-total number of native individuals (R2 = 0.15, r = 0.39, n = 19, p < 0.01). The present resultWater concurred2019, 11, x FOR with PEER the REVIEW previous finding of Lee et al. (2018) and USEPA (1999 and 2000) [5,28,4015 ].of 22

28 28 26 ( r = - 0.37, p < 0.01 ) 26 ( r = - 0.29, p < 0.01 ) 24 24 22 22 20 20 18 18 16 16 14 14 12 12 Index of Biotic Integrity (IBI) Index of Biotic Integrity 10 (IBI) Index of Biotic Integrity 10 8 8 0 20 40 60 80 100 120 140 160 180 123456 -1 -1 TP (μgL ) TN (mgL ) 28 28 ( r = - 0.47, p < 0.01 ) 26 ( r = - 0.42, p < 0.01 ) 26 24 24 22 22 20 20 18 18 16 16 14 14 12 12 Index of Biotic Integrity (IBI) Index of Biotic Integrity 10 (IBI) Indexof Biotic Integrity 10 8 8 0123456 2468 -1 -1 BOD (mgL ) COD (mgL ) 28 28 ( r = - 0.37, p < 0.01 ) 26 ( r = - 0.16, p < 0.01 ) 26 24 24 22 22 20 20 18 18 16 16 14 14 12 12 Index of Biotic Integrity (IBI) Index of Biotic Integrity Index of Biotic Integrity (IBI) Indexof Biotic Integrity 10 10 8 8 0 5 10 15 20 25 30 0 20406080100 -1 CHL (μgL-1) TSS (mgL ) FigureFigure 7. 7. IndexIndex of of biotic biotic integrity integrity (IBI) (IBI) in inrelation relation to towater water quality quality parameters parameters in the in Geum the Geum River Riverwatershed watershed (TN—total (TN—total nitrogen, nitrogen, TP—total TP—total phosphorus, phosphorus, BOD—biological BOD—biological oxygen oxygen demand, demand, COD— COD—chemicalchemical oxygen oxygen demand, demand, TSS—total TSS—total suspended suspended solids, solids, CHL—chlorophyll). CHL—chlorophyll).

Water 2019, 11, 1729 16 of 22 Water 2019, 11, x FOR PEER REVIEW 16 of 22

28 30 IBI = 22.23 - 0.08 * Omnivores IBI = 13.60 + 0.10 * Insectivores 28 26 2 2 (R = 0.31, r = - 0.55, n = 19, p < 0.01) (R = 0.45, r = 0.67, n = 19, p < 0.01) 24 26 24 22 Upper 95% CI 22 20 Upper 95% CI 20 18 18 16 16 14 Lower 95% CI 14 Lower 95% CI

12 Integrity Biotic of Index 12

Index of Biotic Integrity (IBI) Integrity Biotic of Index 10 10 8 8 0 20 40 60 80 100 120 0 20406080100 % Omnivores % Insectivores 28 35 IBI = 22.61 - 0.10 * TS IBI = 15.25 + 1.65 * SS 26 2 2 (R = 0.25, r = - 0.50, n = 19, p < 0.01) (R = 0.19, r = 0.44, n = 19, p < 0.01) 24 30

22 Upper 95% CI 20 Upper 95% CI 25 18 16 20

14 Lower 95% CI Lower 95% CI 12 15 % Index of% Biotic Integrity (IBI) % Index of% Biotic Integrity (IBI)

10 8 10 20 40 60 80 100 2345 % Tolerant Species % Sensitive Species

28 28 IBI = 12.54 + 0.52 * TNS IBI = 15.15 + 0.01 * TNI 26 26 2 2 (R = 0.30, r = 0.55, n = 19, p < 0.01) (R = 0.15, r = 0.39, n = 19, p < 0.01) 24 24 Upper 95% CI 22 22 Upper 95% CI 20 20 18 18 16 16 Lower 95% CI Lower 95% CI 14 14 12 12 Index of Biotic (IBI) Integrity Index of Biotic Integrity (IBI) Integrity Biotic of Index 10 10 8 8 2 4 6 8 10 12 14 16 18 20 0 100 200 300 400 Total Number of Native Species Total Number of Native Individuals

FigureFigure 8. 8.IBI IBI with with trophic trophic and and tolerance tolerance guilds guilds and and TNS TNS and and TNI TNI in in the the watershed. watershed.

3.9.3.9. PredictionPrediction of of Chemical Chemical Pollution pollution IndexIndex andand IndexIndex ofof BioticBiotic IntegrityIntegrityUsing UsingArtificial Artificial Neural Neural Network Network (ANN)(ANN) Model Model TheThe performance performance and and validity validity of of ANN ANN is is shown shown in in Figures Figures9 –9–11.11. The The chemical chemical pollution pollution index index (CPI)(CPI) model model was was predicted predicted by by using using the the ANN ANN model model using using nine nine water water quality quality factors factors (Figure (Figure9). 9). The The 2 observedobserved CPI CPI are are highly highly correlated correlated with with predicted predicted CPI CPI (R (R=2 =0.99, 0.99, RMSE RMSE= =0.08, 0.08, MAE MAE= =0.06, 0.06,p p< < 0.01).0.01). InIn CPI CPI prediction prediction model, model, the the CHL CHL was was the the most most important important variable. variable. We usedWe used a diff aerent different biological biological and chemicaland chemical variable variable to predict to predict the IBI valuethe IBI in thevalue watershed in the watershed (Figures 10 (Figuresand 11). To10 seeand the 11). performance To see the 2 ofperformance the ANN model, of the weANN used model, a diff erentwe used input a di variable.fferent input The 17 variable. inputs ANNThe 17 (R inputs= 0.98, ANN RMSE (R2= =0.64, 0.98, 2 MAERMSE= 0.47,= 0.64,p

Water 2019, 11, 1729 17 of 22 Water 2019, 11, x FOR PEER REVIEW 17 of 22

FigureFigure 9. Prediction 9. Prediction of the of CPIthe modelCPI model by artificial by artificial neural neural network network model model (TN—total (TN—total nitrogen, nitrogen, TP—total TP— phosphorus,total phosphorus, N:P ratios, N:P BOD—biological ratios, BOD—biological oxygen demand, oxygen COD—chemical demand, COD—chemical oxygen demand, oxygen TSS—total demand, suspendedTSS—total solids, suspended EC—electrical solids, conductivity,EC—electrical CHL—chlorophyll, conductivity, CHL—chlorophyll, and DO—dissolved and oxygen). DO—dissolved oxygen).

Water 2019, 11, 1729 18 of 22 Water 2019, 11, x FOR PEER REVIEW 18 of 22 Water 2019, 11, x FOR PEER REVIEW 18 of 22

FigureFigureFigure 10. Prediction10. 10. Prediction Prediction of IBIof IBI using using artificial artificial neural neuralneural networknetwork network tool tooltool in in the the Geum Geum River River River watershed. watershed. watershed.

FigureFigure 11. Variable 11. Variable importance importance in ANN in toANN predict to IBIpredict model IBI (TN—total model (TN—total nitrogen, nitrogen, TP—total TP—total phosphorus, Figure 11. Variable importance in ANN to predict IBI model (TN—total nitrogen, TP—total N:P ratios,phosphorus, BOD—biological N:P ratios, BOD—biological oxygen demand, oxygen COD—chemical demand, COD—chemical oxygen demand, oxygen TSS—total demand, suspended TSS— phosphorus,total suspended N:P ratios,solids, BOD—biologicalEC—electrical conductivity, oxygen demand, CHL—chlorophyll, COD—chemical DO—dissolved oxygen demand, oxygen, TSS— solids, EC—electrical conductivity, CHL—chlorophyll, DO—dissolved oxygen, TNS—total native total suspended solids, EC—electrical conductivity, CHL—chlorophyll, DO—dissolved oxygen, fish species, RBS—riffle benthic species, SS—sensitive species, TS—tolerant species, IS—intermediate

species, TNI—total number of individuals, AI—anomalies). Water 2019, 11, x FOR PEER REVIEW 19 of 22

Water TNS—2019, 11 ,total 1729 native fish species, RBS—riffle benthic species, SS—sensitive species, TS—tolerant19 of 22 species, IS—intermediate species, TNI—total number of individuals, AI—anomalies).

3.10. Principal Principal Component Component Analysis Analysis (PCA) in Relation to Chemicals and Fish Metrics The principal component analysis was carried out to extract the most important factors and physicochemical parameters affecting affecting the ecological health of the watershed (Figure 1212).). Because of the complex relationshipsrelationships amongamong waterwater quality quality factors, factors, trophic trophic and and tolerance tolerance guilds, guilds, it wasit was di ffidifficultcult to todraw draw clear clear conclusions. conclusions. However, However, not only not can only principal can principal component component analysis extract analysis the informationextract the informationto some extent to andsome explain extent theand structure explain the of the structure data in of detail the data on spatial in detail characteristics on spatial characteristics by clustering theby clusteringwater quality the parameterswater quality and parame fish guilds,ters and but itfish can guilds, also describe but it theircan also diff erentdescribe characteristics their different and characteristicshelp to elucidate and the help relationship to elucidate between the waterrelation qualityship between parameters water and quality fish guilds parameters [41]. The and first fish and guildssecond [41]. axis The of PCA first accounts and second for approximatelyaxis of PCA accounts 60.97% andfor approximately 15.77% of the total60.97% variance, and 15.77% respectively. of the totalThe abundancevariance, respectively. of tolerant species, The abundance carnivores, of tolera omnivoresnt species, were carnivores, dominant inomnivores downstream, were indicatingdominant inthat downstream, the downstream indicating health that condition the downstream was deteriorated health condition because of was organic deteriorated and nutrient because pollutants of organic and andexcessive nutrient chlorophyll pollutants growth. and excessive The present chlorophyll results are growth. similar The to thatpresent of Atique results and are An similar (2018) to and that Lee of Atiqueet al. (2018) and An [16 ,(2018)28]. and Lee et al. (2018) [16,28].

Figure 12. Principal component analysis based on water quality factors and fishfish guilds.guilds. TN—total nitrogen, TP—totalTP—total phosphorus, phosphorus, N:P N:P ratios, ratios, BOD—biological BOD—biological oxygen oxygen demand, demand, COD—chemical COD—chemical oxygen oxygendemand, demand, TSS—total TSS—total suspended suspended solids, EC—electrical solids, EC—electrical conductivity, conductivity, CHL—chlorophyll, CHL—chlorophyll, SS—sensitive SS— sensitivespecies, TS—tolerantspecies, TS—tolerant species, species, IS—intermediate IS—intermediate species, species, % O—omnivores, % O—omnivores, % C—carnivores,% C—carnivores, % %I—insectivores). I—insectivores). 3.11. Site Grouping by Cluster Analysis (CA) 3.11. Site Grouping by Cluster Analysis (CA) The cluster analysis (CA) is a strong tool for solving classification problems (Figure 13). The The cluster analysis (CA) is a strong tool for solving classification problems (Figure 13). The objective of CA is to place factors or variables into groups such that the degree of association is strong objective of CA is to place factors or variables into groups such that the degree of association is strong between members of the same cluster and weak between members of different clusters [41,42]. In this between members of the same cluster and weak between members of different clusters [41,42]. In this study, CA showed a strong spatial association on the basis of variations of water quality factors and study, CA showed a strong spatial association on the basis of variations of water quality factors and fish assemblages. The dendrogram indicates pollution status as well as the effect of anthropogenic fish assemblages. The dendrogram indicates pollution status as well as the effect of anthropogenic activities at the sampling sites. It provides a visual summary of the clustering processes, presenting a activities at the sampling sites. It provides a visual summary of the clustering processes, presenting picture of the groups and their proximity. Cluster analysis (CA) was used to detect similarity among a picture of the groups and their proximity. Cluster analysis (CA) was used to detect similarity among 19 sampling sites. The Bay–Curtis similarity analysis confirmed that there is a similarity of 87% among 19 sampling sites. The Bay–Curtis similarity analysis confirmed that there is a similarity of 87% sites G11, G12, G18, G13, and G14. The generated dendrogram grouped into three clusters based on among sites G11, G12, G18, G13, and G14. The generated dendrogram grouped into three clusters 70% of their similarity of physicochemical parameters and fish trophic and tolerance guilds. Using the

Water 20192019,, 1111,, 1729x FOR PEER REVIEW 20 of 22

based on 70% of their similarity of physicochemical parameters and fish trophic and tolerance guilds. Bay–CurtisUsing the Bay–Curtis analysis (up analysis to 65% (up similarity) to 65% similari we categorizedty) we categorized the watershed the watershed into two regions into two including regions upstreamincluding (G01,upstream G02, (G01, G03, G04,G02, G05,G03, G06,G04, G05, G07, G06, G08, G07, G09, G08, G10) G09, and downstreamG10) and downstream (G11, G12, (G11, G13, G12, G14, G15,G13, G16,G14, G17,G15, G18,G16, andG17, G19). G18, and G19).

Figure 13. Cluster analysis of the Geum River watershed based on water quality factors and fish Figure 13. Cluster analysis of the Geum River watershed based on water quality factors and fish guilds. guilds. 4. Conclusions 4. Conclusions In the present study, the chemical parameter model of CPI and the biological model of the index of bioticIn the integrity present (IBI) study, were the used chemical to assess parameter the river model health of inCPI the and watershed. the biological Generally, model the of the chemical index healthof biotic of integrity the upstream (IBI) were was used in good-fair to assess condition the river health because in ofthe fewer watershed. human Generally, disturbances, the chemical while it washealth poor of the to veryupstream poor was condition in good-fair in the condition downstream because region of fewer because human of nutrient, disturbances, organic, while and it ionic was enrichments.poor to very Thepoor biological condition health in the based downstream on IBI model region also suggested because thatof nutrient, the watershed organic, health and was ionic in fairenrichments. to very poor The condition.biological health The ANN based model on IBI also model indicated also suggested that the chemical that the watershed and biological health health was ofin fair the Geumto veryRiver poor condition. was in fair The to veryANN poor model condition. also indicated The PCA that andthe chemical cluster analysis and biological revealed health that theof the abundance Geum River of tolerantwas in fair and to omnivore very poor species condition. was The dominant PCA and because cluster of analysis excessive revealed nutrients that and the organicabundance matter of intolerant the watershed. and omnivore The overall species ecological was dominant health of because the Geum of Riverexcessive was innutrients an abnormal and conditionorganic matter and needs in the immediatewatershed. actions The overall for proper ecological management. health of the Geum River was in an abnormal condition and needs immediate actions for proper management. Author Contributions: J.H. and M.M. conceived the idea, designed the experiments, analyzed the data, and wrote theAuthor manuscript Contributions: under the J. supervisionH. and M.M. of conceived K.-G.A. the idea, designed the experiments, analyzed the data, and Funding:wrote the manuscriptThis work was under conducted the supervision by Korea of EnvironmentK.-G.A. Industry & Technology Institute (KEITI) through “Exotic Invasive Fish Species Management Project”, funded by Korea Ministry of Environment (MOE) (Grant No.: 2018-1467-02)Funding: This and work Daejeon was Green conducted Environment by Korea Center Environmen under the Researcht Industry Development & Technology Program Institute (2016). (KEITI) Acknowledgments:through “Exotic InvasiveThe authors Fish are highlySpecies acknowledged Manageme tont theProject”, Korea Ministry funded of Environmentby Korea Ministry and Daejeon of GreenEnvironment Environment (MOE) Center (Grant for their No.: assistance. 2018-1467-02) and Daejeon Green Environment Center under the ConflictsResearch of Development Interest: The AuthorsProgram have (2016). no conflict Acknowledgments: of interest. The authors are highly acknowledged to the Korea Ministry of Environment and Daejeon Green Environment Center for their Referencesassistance.Conflicts of Interest: The Authors have no conflict of interest.

1.ReferencesCostanza, R.; Alperovitz, G.; Daly, H.E.; Farley, J.; Franco, C.; Jackson, T.; Kubiszewski, J.; Schor, J.; Victor, P. Building a Sustainable and Desirable Economy-in-Society-in-Nature; United Nations Division for Sustainable 1. Development:Costanza, R.; Alperovitz, New York, G.; NY, Daly, USA, H.E.; 2012. Farley, J.; Franco, C.; Jackson, T.; Kubiszewski, J.; Schor, J.; Victor, 2. Limburg,P. Building K.E. a Sustainable Aquatic Ecosystemand Desirable Services. Economy-in-Society-in-Nature In Encyclopedia of Inland; United Waters Nations; Likens, Division G.E., Ed.;for Sustainable Academic Press:Development: Oxford, New UK, 2009; York, pp. NY, 25–30. USA, 2012. 2. Limburg, K.E. Aquatic Ecosystem Services. In Encyclopedia of Inland Waters; Likens, G.E., Ed.; Academic Press: Oxford, UK, 2009; pp. 25–30.

Water 2019, 11, 1729 21 of 22

3. Dudgeon, D.; Arthington, A.H.; Gessner, M.O.; Kawabata, Z.; Knowler, D.J.; Leveque, C.; Naiman, R.J.; Prieur Richard, A.H.; Soto, D.; Stiassny, M.L. Freshwater biodiversity: Importance, threats, status and conservation challenges. Biol. Rev. 2005, 81, 163–182. [CrossRef][PubMed] 4. Davies, P.; Harris, J.; Hillman, T.J.; Walker, K.F. The sustainable rivers audit: Assessing river ecosystem health in the Murray–darling basin, Australia. Mar. Freshw. Res. 2010, 61, 764–777. [CrossRef] 5. Barbour, M.T.; Gerritsen, J.; Snyder, B.; Stribling, J. Rapid Bioassessment Protocols for Use in Streams and Wadeable Rivers: , Benthic Macroinvertebrates, and Fish, 2th ed.; USEPA: Washington, DC, USA, 1999. 6. Hering, D.; Borja, A.; Carstensen, J.; Carvalho, L.; Elliott, M.; Feld, C.K. The European water framework directive at the age of 10: A critical review of the achievements with recommendations for the future. Sci. Total Environ. 2010, 408, 4007–4019. [CrossRef][PubMed] 7. Atique, U.; Lim, B.; Yoon, J.; An, K.G. Biological Health Assessments of Lotic Waters by Biotic Integrity Indices and their Relations to Water Chemistry. Water 2019, 11, 436. [CrossRef] 8. Yu, X.; He, D.; Phousavanh, P. Balancing River Health and Hydropower Requirements in the Lancang River Basin; Springer: Singapore, 2019; ISBN 978-981-13-1564-0. Available online: https://www.springer.com/gp/book/ 9789811315640 (accessed on 20 August 2019). 9. Wu, W.; Xu, Z.; Zhan, C.; Yin, X.; Yu, S. A new framework to evaluate ecosystem health: A case study in the Wei River basin, China. Environ. Monit. Assess. 2015, 187, 460. [CrossRef] 10. Yoder, C.O. The Integrated Biosurvey as a Tool for 852 Evaluation of Aquatic Life Use Attainment and Impairment 853 in Ohio Surface Waters; EPA 440-5-91-005; USEPA: Washington, DC, USA, 1991; pp. 110–122. 11. Heatherly, T., II; Whiles, M.R.; Royer, T.V.; David, M.B. Relationships between water quality, habitat quality, and macroinvertebrate assemblages in Illinois streams. J. Environ. Qual. 2007, 36, 1653–1660. [CrossRef] 12. Bach, E. A chemical index for surveillance of river water quality. Dtsch. Gewasserkd. Mitt. 1980, 24, 102–106. 13. Trivedi, R.C.; De-Kruijf, H.A.M.; de-Zwart, D. The development and application of a yardstick for water quality evaluation. Sci. Total Environ. 1993, 134, 1191–1202. [CrossRef] 14. Dodds, W.K.; Jones, J.R.; Welch, E.B. Suggested classification of stream trophic state: Distributions of temperate stream types by chlorophyll, total nitrogen and phosphorus. Water Res. 1998, 32, 1455–1462. [CrossRef] 15. Yu, X.; Xia, J.; Yang, J.; Ma, W. Preliminary study on the index system and assessment method of green hydropower. J. Hydroelectr. Eng. 2011, 30, 71–77. (In Chinese) 16. Atique, U.; An, K.G. Stream Health Evaluation Using a Combined Approach of Multi-Metric Chemical Pollution and Biological Integrity Models. Water 2018, 10, 661. [CrossRef] 17. Karr, J.R. Assessment of biotic integrity using fish communities. Fisheries 1981, 6, 21–27. [CrossRef] 18. Simon, T.P.; Morris, C.C. Relationships among varying sampling distance and the IBI in warmwater, headwater streams of the Eastern Corn Belt Plain. J. Monit. Assess. 2014, 186, 6537–6551. [CrossRef] [PubMed] 19. Hallett, C.S.; Valesini, F.J.; Clarke, K.R.; Hesp, S.A.; Hoeksema, S.D. Development and validation of fish-based, multimetric indices for assessing the ecological health of Western Australian . Estuar. Coast. Shelf Sci. 2012, 104–105, 102–113. [CrossRef] 20. Republic of South Africa Department of Water and Sanitation (RSADWA). River Eco-Status Monitoring Programme; RSADWA: Pretoria, South Africa, 2016. 21. Hughes, R.M. Use of watershed characteristics to select control streams for estimating effects of metal mining wastes on extensively disturbed streams. Environ. Manag. 1985, 9, 253–262. [CrossRef] 22. Leonard, P.M.; Orth, D.J. Application and testing of an index ofbiotic integrity in small, coolwater streams. Trans. Am. Fish. Soc. 1986, 115, 401–414. [CrossRef] 23. Berkman, H.E.; Rabeni, C.F. Effect of on fish communities. Environ. Biol. 1987, 18, 285–294. [CrossRef] 24. Ormerod, S.J. issues with fish and fisheries: Editor’s overview and introduction. J. Appl. Ecol. 2003, 40, 204–213. [CrossRef] 25. An, K.G.; Lee, J.Y.; Bae, D.Y.; Kim, J.H.; Hwang, S.J.; Won, D.H. Ecological assessments of aquatic environments using a multi-metric model in major nationwide stream watersheds. J. Korean Water Qual. 2006, 22, 796–804. 26. MOE/NIER. Study on Development of Methods for Synthetic Assessment of Water Environment 2006; The Ministry of Environment/National Institute of Environmental Research: Sejong, Korea, 2006. Water 2019, 11, 1729 22 of 22

27. MOE/NIER. Survey and Evaluation of Aquatic Ecosystem Health in Korea 2013; The Ministry of Environment/National Institute of Environmental Research: Sejong, Korea, 2013. 28. Lee, S.J.; Lee, E.H.; An, K.G. Lotic Ecosystem Health Assessments Using an Integrated Analytical Approach of Physical Habitat, Chemical Water Quality, and Fish Multi-Metric Health Metrics. Pol. J. Environ. Stud. 2018, 27, 2113–2131. [CrossRef] 29. Marker, A.F.H.; Crowther, C.A.; Gunn, R.J.M. Methanol and acetone as solvents for estimating chlorophyll a and phaeopigments by spectrophotometry. Arch. Hydrobiol. Beih. Ergeb. Limnol. 1980, 14, 52–59. 30. Ohio Environmental Protection Agency. Biological Criteria for the Protection of Aquatic Life: Volume III. Standardized Biological Field Sampling and Laboratory Method for Assessing Fish and Macroinvertebrate Communities; Division of , Ecological Assessment Section: Columbus, OH, USA, 1989; p. 66. 31. An, K.G.; Jung, S.J.; Choi, S.S. An Evaluation on Health Conditions of Pyong-Chang River using the Index of Biological Integrity (IBI) and Qualitative Habitat Evaluation Index (QHEI). Korean J. Limnol. 2001, 34, 153–165. 32. An, K.G.; Kim, D.S.; Kong, D.S.; Kim, S.D. Integrative assessments of a temperate stream based on a multimetric determination of biological integrity, physical habitat evaluations, and toxicity tests. Bull. Environ. Contam. Toxicol. 2004, 73, 471–478. [CrossRef][PubMed] 33. Huang, J.; Gao, J. An ensemble simulation approach for artificial neural network: An example from chlorophyll a simulation in Lake Poyang, China. Ecol. Inform. 2017, 37, 52–58. [CrossRef] 34. Sigma Plot, Version 10; Systat Software, Inc.: San Jose, CA, USA, 2018; Available online: www.systatsoftware. com (accessed on 22 October 2018). 35. Hammer, Ø. The Past of the Future; PAST, Version 3.18 (Software); Natural History Museum, University of Oslo: Oslo, Norway, 2017. 36. Kim, J.H.; Yeom, D.H.; An, K.G. Diagnosis of Sapgyo stream watershed using the approach of integrative star-plot area. Korean J. Ecol. Environ. 2010, 43, 356–368. 37. Mamun, M.; An, K.G. Ecological Health Assessments of 72 Streams and Rivers in Relation to Water Chemistry and Land-Use Patterns in South Korea. Turk. J. Fish. Aquat. Sci. 2018, 18, 871–880. [CrossRef] 38. Xiaomei, S.; Alan, D.S.; Oudsema, M.; Hassett, M.; Xi, L. The influence of nutrients limitation on growth and microcystins production in Lake, USA. Chemosphere 2019, 234, 34–42. 39. Choi, J.W.; Han, J.H.; Park, C.S.; Ko, D.G.; Kang, H.I.; Kim, J.Y.; Yun, Y.J.; Kwon, H.H.; An, K.G. Nutrients and sestonic chlorophyll dynamics in Asian lotic ecosystems and ecological stream health in relation to land-use patterns and water chemistry. Ecol. Eng. 2015, 79, 15–31. [CrossRef] 40. U.S. Geological Survey. Biomonitoring of Environmental Status and Trends (Best) Program: Selected Methods for Monitoring Chemical Contaminants and Their Effects in Aquatic Ecosystems; UITR-2000–0005 Information and Technology Report; U.S. Geological Survey: Columbia, OR, USA, 2000; p. 81. 41. Bhat, S.A.; Meraj, G.; Yaseen, S.; Pandit, A.K. Statistical Assessment of Water Quality Parameters for Pollution Source Identification in Sukhnag Stream: An Inflow Stream of Lake Wular (), Kashmir Himalaya. J. Ecosyst. 2014, 2014, 898054. [CrossRef] 42. Brogueira, M.J.; Cabec, A.G. Identification of similar environmental areas in Tagus by using multivariate analysis. Ecol. Indic. 2006, 6, 508–515. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).