<<

Bioresource Technology 152 (2014) 107–115

Contents lists available at ScienceDirect

Bioresource Technology

journal homepage: www.elsevier.com/locate/biortech

Suitability of giant reed ( L.) for anaerobic digestion: Effect of harvest time and frequency on the biomethane yield potential ⇑ Giorgio Ragaglini a, ,1, Federico Dragoni a,1, Marco Simone b, Enrico Bonari a,c a Institute of Life Sciences, Scuola Superiore Sant’Anna, Piazza Martiri della Libertà 33, 56127 Pisa, Italy b Dipartimento di Ingegneria Chimica, Chimica Industriale e Scienza dei Materiali, Università di Pisa, Via Diotisalvi 2, 56122 Pisa, Italy c CRIBE – Centro di Ricerche Interuniversitario Biomasse da Energia, Via Vecchia Livornese 748, 56122 Pisa, Italy highlights

The Biochemical Methane Potential (BMP) of giant reed was assessed. Biochemical Methane Potential was significantly influenced by harvest time and frequency. The highest BMP and the best digestion kinetics was achieved at juvenile crop stages. Double harvesting increased by 20–35% the methane yield per hectare. article info abstract

Article history: This study aimed to investigate the potential of giant reed for biomethane production by examining the Received 27 August 2013 influence of harvest time and frequency on the Biochemical Methane Potential (BMP), the kinetics of Received in revised form 29 October 2013 biomethane accumulation in batch reactors and the expected methane yield per hectare. The crop was Accepted 2 November 2013 cut at five different times, regrowths from early cuts were harvested in autumn and BMP of each cut Available online 11 November 2013 was assessed. The highest BMP (392 NL kg VS1) and the best kinetics of methane production were asso- ciated to juvenile traits of the crop. By coupling the early cuts with the corresponding regrowths (double Keywords: harvest), the dry (from 35 to 40 Mg ha1) equaled that obtained by a single cut at end of the sea- son (38 Mg ha1), while the methane yield per hectare (11,585–12,981 Nm3 ha1) exceeded up to 35% the Double harvest 3 1 Kinetics methane produced with a single harvest at crop maturity (9452 Nm ha ). Biochemical Methane Potential Ó 2013 Elsevier Ltd. All rights reserved. Perennial energy crops

1. Introduction Perennial grasses, and particularly giant reed, have already been recognized as high-yielding crops that can minimize environmen- Anaerobic digestion is one of the most mature technologies for tal impacts, because of the reduced inputs requirements (Angelini biomass energy production, allowing energy conversion from a et al., 2009; Nassi o di Nasso et al., 2011). Giant reed is a perennial broad variety of substrates, such as wastes, sludges, manures, rhizomatous species that has been traditionally cultivated in and a wide range of crops and their residues. However, at present Southern Europe, North Africa, Asia and the Middle East and has biogas production relies greatly on co-digestion of animal slurries been recently introduced in the USA, where it is usually considered and annual crops, and particularly on (Bauer et al., 2010; an invasive species; it has already been studied for bioethanol pro- Herrmann and Rath, 2012). In Italy, maize cultivation is nowadays duction, direct combustion, and other thermal transformations expanding from traditional regions (mainly Northern Italy) to less (Pilu et al., 2012). Giant reed is considered a drought-tolerant spe- suitable areas because of the increase in biogas production, thus cies (Angelini et al., 2009; Pilu et al., 2012) and it can be grown in concerns may arise about water consumption and fertilizer marginal or sub-marginal lands (Dragoni et al., 2011; Nasso o di requirements. Moreover, maize use as is often criti- Nasso et al., 2013), thus reducing competition with food crops cized because of the changes it may cause on the land use and for soil use. on the price of food and feed commodities. Harvest time significantly influences biomass yield of giant reed and its characteristics (Nassi o di Nasso et al., 2011), and it is usu- ally seen as a major factor influencing digestibility and methane ⇑ Corresponding author. Tel.: +39 050 883521. yields of energy crops (Massé et al., 2010; Bélanger et al., 2012; E-mail address: [email protected] (G. Ragaglini). Kreuger et al., 2011; Gao et al., 2012; Kandel et al., 2012). The 1 These authors equally contributed to this study.

0960-8524/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.biortech.2013.11.004 108 G. Ragaglini et al. / Bioresource Technology 152 (2014) 107–115 proportion of the stems respect to the leaves is expected to change, Table 1 as well as chemical traits like nitrogen concentration, C/N ratio, date of harvest, total, and volatile solids content (TS and VS) on the fresh matter (FM), and ash concentration on the dry matter (DM) for first harvests (A1–A5) and second NSC (Non-Structural Carbohydrates), and cell wall components, harvests (RA1, RA2) of giant reed and for maize silage (M). thus influencing its biodegradability and methane output (Heaton et al., 2009; Smith and Slater, 2011; Slewinski, 2012; Nasso o di Cut Date TS (% of FM) Ash (% of DM) VS (% of FM) Nasso et al., 2013). A1 21-Jun 46.9 7.9 43.2 Previous studies assessed the capability of giant reed to regrow A2 15-Jul 49.1 6.6 45.9 A3 02-Aug 51.9 4.4 49.6 after an early cut (Dragoni et al., 2011), but regrowth of perennial A4 22-Aug 47.0 6.4 44.0 grasses is expected to vary according to the length of the growing A5 20-Sep 51.0 4.8 48.5 season and the environmental conditions (Kandel et al., 2012). RA1 18-Oct 42.3 7.7 39.1 Thus, biogas production could rely on multiple harvests per year, RA2 18-Oct 39.6 6.7 37.0 but several considerations are relevant to the question of whether M Sep 34.8 4.9 33.1 single harvest or multiple harvest of a perennial crop for biogas is preferable, and many of these questions are not directly related to biodegradability (e.g., crop duration, machinery requirements, was not negligible, that resulted those harvested in June and July nutrient uptake, environmental and economic sustainability). De- (RA1 and RA2). At each harvest time, biomass fresh weight was spite the fact that biogas production from giant reed has already determined by sampling a 2 m2 area within each plot (12 3 m). been hypothesized by some authors (Schievano et al., 2012; Di Nodes, green, and senescent leaves per were counted on a Girolamo et al., 2013), it is still a novel crop for this purpose. There- subsample (10 ), while lost leaves were determined by dif- fore, further questions about biomethane potential and digestibil- ference with the total node number; plant heights were also mea- ity of giant reed under different management conditions should be sured. Subsequently, leaves and stems were separated, weighed addressed. and their dry matter content (DM) was determined by oven drying Biochemical Methane Potential (BMP) has been widely used to at 65 °C until constant weight, in order to assess the aboveground assess methane yields of organic matter when degraded at anaer- dry biomass yield (AGB, Mg ha1) and its partitioning. Biomass obic conditions, including plant biomass (Chynoweth et al., 1993; yield from the summer harvest and regrowth from the same plots Angelidaki et al., 2009). Grasses and other lignocellulosic sub- were pooled to get the total biomass yield where double harvest strates have been extensively studied as interesting feedstocks was performed. for anaerobic digestion by several authors and remarkable meth- Samples for chemical analyses were prepared for each field rep- ane potentials are often reported (Nizami et al., 2009; Seppälä lication by milling in a Retsch SM1 rotor mill equipped with a et al., 2009; Massé et al., 2010; Kandel et al., 2012). Nevertheless, 1 mm grid. Subsamples for Biochemical Methane Potential (BMP) BMP is not the only parameter to be considered. Kinetics of anaer- determination were prepared by bulking biomass from the field obic digestion is also crucial, since rapid methane production is replications. All the samples were stored at 20 °C. fundamental to shorten the residence time and to achieve better methane yield in real-scale, continuous plants (Mähnert and Linke, 2009; Grieder et al., 2012). 2.2. Chemical analyses and nutrient evaluation Schievano et al. (2012) compared giant reed with other crops mainly in terms of Biochemical Biogas Potential (BBP) and biogas Total solids (TS), volatile solids (VS), and total Kjeldahl nitrogen production costs per hectare, while Di Girolamo et al. (2013) re- (TKN) were determined on fresh samples according to standard cently reported methane potentials for giant reed harvested in methods (APHA, 2005); C/N ratio was assessed by elemental anal- early autumn, as affected by several pre-treatments. ysis (Leco CHN 600). Lignin quantification was performed using the However, the BMP of giant reed as influenced by different har- acetyl bromide method (Fukushima and Hatfield, 2004), absor- vest times has not previously been studied and double harvest sys- bance was measured with a Beckman Coulter DU 800 UV/Vis spec- tems have not yet been evaluated respect to the overall trophotometer at 280 nm, then lignin content was calculated using biomethane yield per hectare. Thus, the aim of this study was to the Lambert–Beer equation. Non-Structural Carbohydrates (NSC) determine the BMP and the methanisation kinetics of giant reed were calculated as the sum of Water Soluble Carbohydrates at different harvest times as influenced by the crop characteristics (WSC) plus starch, determined according to Giovannelli et al. and the growth stage. Consequently, methane yields per hectare (2011). For each sample and parameter, three technical replicates were evaluated as affected by harvest time and frequency, in order were analyzed. to assess the suitability of giant reed as an alternative feedstock for biogas production systems. 2.3. Biochemical Methane Potential (BMP) assay

Biochemical Methane Potential (BMP) was determined in batch 2. Methods reactors (volume 2 L). The assays were conducted in triplicates on fresh biomass from the seven different cuts of giant reed (five first 2.1. Field experiment and samples preparation cuts, A1–A5, and two regrowths, RA1, RA2) and on maize silage (M), hybrid DKC6666, (wax ripeness, FAO 600, Dekalb) as a control A local ecotype of giant reed was cultivated since April 2007 in assay. Each reactor received 300 g of inoculum that was suspended San Piero a Grado, Pisa, Italy (43° 400 49.2100 North, 10° 200 47.1500 in a basal test medium up to a final filled volume of 1 L. Three blank East; 1 m above mean sea level and 0% slope). The establishment experiments were also performed with 300 g of inoculum, Milli-Q of the crop was performed using rhizomes, the canes were grown water, and minerals only (Angelidaki et al., 2009). at a population density of 20,000 plants ha1 in rows having 1 me- The medium was prepared using Milli-Q water, according to the ter of width. During the growing season of the year 2011, the crop ISO 11734 standard. The inoculum originated from the methano- was harvested at 5 different times from June to September (A1–A5) genic stage of a mesophilic anaerobic digester fed with energy (Table 1), each one replicated 3 times (15 plots). Resprouting after crops (maize and sorghum silages), agricultural residues, cattle first cut was expected, thus leading to perform a second cut in and poultry manure; the acidogenic stage was an horizontal early fall (18th October 2011) from plots where crop regrowth plug-flow, while the methanogenic stage was a CSTR. Inoculum G. Ragaglini et al. / Bioresource Technology 152 (2014) 107–115 109 was prepared according to the ISO 11734 standard, except for inor- used to calculate different kinetic parameters: the time, expressed ganic carbon removal procedure that has not been carried out, as in days, when 50% and 95% of methane production was reached proposed by Angelidaki et al. (2009). The anaerobic sludge was (respectively, T50 and T95), the maximum daily production rate sieved through a 1 mm mesh (Chynoweth et al., 1993; Rincón (Rmax) and the mean daily production rate from the beginning of et al., 2010); after removing large particles, it had a total solids the assay to T50 (R50). Curve fitting and model parameterization (TS) content of 84.5 g kg1, a volatile solids (VS) content of were performed using the R software, version 2.15.0, and then 56.3 g kg1, and a pH of 7,8. Before the beginning of the assay, mle and rootSolve packages (Pinheiro et al., 2013; Soetaert, 2013). the inoculum was left for 5 days at 37 °C to reduce the amount of readily available organic matter and to be degassed. 2.5. Methane yields per hectare The substrates were added to the reactors according to a ratio between the inoculum and the substrate (I:S) equal to 2:1 on the Biomass yield from first cuts and regrowths from the same plots basis of their volatile solids content (VS) (Table 1), as advised by were pooled to get the total biomass yield, in order to compare Chynoweth et al. (1993). Once the reactors were loaded with the AGB production of double harvest systems (A1 + RA1, A2 + RA2) different substrates, the batches were sealed and flushed with with that obtained from single harvest (A3–A5). Methane yields N , in order to obtain anaerobic conditions. Subsequently, the ves- 3 1 2 per hectare (Nm CH4 ha ) were determined by multiplying the sels were incubated at 37 ± 1 °C as long as the further production of mean BMP of each cut by the AGB production of each plot of the biogas became negligible (40 days). Biogas pressure in each reactor corresponding harvest time. Also in this case methane yields of was continuously measured by pressure piezo-resistive transduc- first cuts and regrowths were combined in order to compare meth- ers and continuously recorded by a dedicated Programmable Logic ane productivity of double harvest systems with single harvests. Controller (PLC) connected to a PC. The cumulative volume of biogas produced in each reactor at 2.6. Statistical analyses each time (Br,t) was calculated according to the Ideal Gas Law and to the Molar Volume of Ideal Gases at Standard Temperature Accumulated biomass and its partitioning, biometric values, and Pressure conditions (1 bar, 273.15 K). Methane concentration chemical traits, and anaerobic digestion parameters were com- (MC) was measured by gas chromatography (micro-GC Agilent pared for the different giant reed cuts by one-way ANOVA. Biomass 3000) (Angelidaki et al., 2009). Biogas was sampled and analyzed crop productivity and methane yield per hectare were also com- at five different times, considering intervals (i) of 2, 5, 10, 20, and pared by one-way ANOVA by coupling first and second harvest. 40 days after the start-up of the assay. When significant differences were evidenced, post hoc compari- Both the pressure reduction due to biogas removal at each sam- sons were made using the LSD test at the 0.05 p-level. Single linear pling interval and the biomethane content of the sampled gas were regressions were performed in order to point out the main factors considered in estimating the cumulative biogas production of each that influenced biogas and biomethane production and kinetics, batch, as reported in Appendix A (Eqs. (A.1), (A.2)). testing several predictors among the anaerobic digestion parame- Finally, in order to obtain the Biochemical Methane Potential ters and the crop traits. (BMP) of each substrate, expressed in NL of CH4 per kg of volatile solids (VS), the residual (or intrinsic) methane potential of the inoculum as obtained by blank experiments was removed accord- 3. Results and discussion ing to Eq. (A.3). Analogously, the Biochemical Biogas Potential (BBP) was obtained by subtracting the intrinsic biogas potential 3.1. Crop growth and biomass characteristics of the inoculum to the biogas produced in each reactor. In 2011, giant reed sprouted on the 20th of March and the growing season was characterized by an average daily mean tem- 2.4. Kinetics of the methane production perature of about 18 °C, while the maximum air temperature peaked above 35 °C in July and at the end of August (Fig. 1). Rainfall The kinetics of anaerobic digestion of giant reed and maize si- amounted to nearly 34% of the annual precipitations, mainly dis- lage were examined by regressing on time the daily cumulated tributed in two most intensive events (up to 45 mm/day) at the methane measured in each reactor. Six non-linear models were beginning and at the end of the season. Total reference evapotrans- evaluated (Table 2), namely Michaelis–Menten, Asymptotic piration (ET0) exceeded more than three times the total rainfall Regression, Weibull, Log–Logistic, Gompertz, and the five parame- (740 vs 212 mm) and the highest daily values were reached during ters Modified Gompertz (Beuvink and Kogut, 1993; Grieder et al., the third week of July, while after August ET0 rapidly decreased 2012). Goodness of fit (R2, RMSE) for the different models was as- (Fig. 1). sessed and their efficiency was tested by the Akaike Information The first considered harvest time, namely A1, occurred 93 days Criterion (AIC). Thus, the most efficient model was identified and after sprouting and the aboveground biomass (AGB) production

Table 2 Goodness of fit evaluation of investigated non-linear models. For each model are reported: the parameters description, the coefficient of determination (R2), the Root Mean Square Error (RMSE) and the Akaike Information Criterion (AIC) obtained from non-linear fitting of methane production curves (R2, RMSE, and AIC are means over all samples).

Models Parameters R2 RMSE AIC

Michaelis–Menten d d = maximum y, e = time at which y is half d 0.976 10.73 301.3 y ¼ eþx x Asymptotic regression y ¼ d ½1 expð eÞ d = upper limit, e = steepness of the increase 0.988 7.95 280.3 Weibull y ¼ d exp½ expðb ðlog x eÞ d = upper limit, b = slope, e=inflexion point 0.993 6.64 258.4 Log–Logistic d d = upper limit, b = slope, e = inflexion point 0.995 5.52 252.5 y ¼ 1 þ exp½b ðlog x log eÞ m Gompertz y ¼ d exp½ c expðc xÞ d = plateau, m = the initial relative growth rate, c = relative 0.983 9.48 301.2 higrowth rate at inflection Modified Gompertz mR mS d = plateau, m = the initial relative growth rate, c = relative 0.999 2.71 198.5 y ¼ d exp c expðcR xÞ c expðcS xÞ R S growth rate at inflection, R = rapid, S = slow 110 G. Ragaglini et al. / Bioresource Technology 152 (2014) 107–115

Fig. 1. Meteorological data of the investigated period on daily scale: rainfalls are represented by vertical bars, maximum and minimum air temperature are described by dashed and dotted lines respectively, while the solid line represents the daily reference evapotranspiration (ET0) averaged on a weekly basis. amounted to nearly 23 Mg DM ha1 (Fig. 2). AGB standard devia- The increase in C/N ratio over time is in line with previous findings tions increased sensibly in A2 and A3, probably depending on the on nutrient cycling in rhizomatous grasses and it can be explained variability of climatic conditions, thus not allowing to observe a by carbon accumulation (Kandel et al., 2012), by nitrogen reloca- significant yield increase, in spite of higher stem heights and leaves tion from the aerial parts to the rhizome and by leaf loss (Heaton number respect to A1. In general, crop growth was limited by et al., 2009; Nasso o di Nasso et al., 2013)(Fig. 2). water deficit, as indicated by the green leaves number remaining High acetyl bromide lignin contents were observed (between roughly unvaried from A1 to A4, while senescent leaves number in- 21% and 24% of DM), in line with those obtained using the same creased. The slow growth conditions are also evidenced by the fact method by Lygin et al. (2011), but no significant variations were that AGB in A4 was significantly higher than in A1 (P < 0.001) and showed between the considered harvest times (Fig. 2). Analo- appreciably but not significantly higher than in A2 and A3. Then, a gously, in a previous study giant reed showed no significant varia- substantial increase in stem height and green leaves number was tions in acid-detergent lignin (ADL) along the growing season achieved after A4, thus resulting in a significantly higher biomass (Nassi o di Nasso et al., 2011), although the values were substan- production in A5 (P < 0.001), 38 Mg DM ha1. As expected, senes- tially lower as typical for this method (Fukushima and Hatfield, cent leaves number increased from A3 to A5, and leaves loss was 2004). At contrary, a high level of variability was observed in the also observed from A4, thus determining a change in biomass par- Non-Structural Carbohydrates (NSC) content, with higher concen- titioning from early to late crop stages (Fig. 2). In particular, the trations from midseason to mature stages (A3–A5) and in second proportion of green leaves biomass on total AGB was higher in harvests (RA1–RA2) (Fig. 2). Sucrose resulted to be the main WSC A1 (33%) and decreased constantly in the following cuts except contained in giant reed biomass varying from 4.9 to 28.6 mg g1. A3, in which the increase in green leaves respect to A2 was proba- Starch content ranged from 1.9 to 33.5 mg g1. Although further bly due to leaf expansion, as consequence of the increased water investigations are advisable to clarify NSC dynamics in giant reed, availability occurred in the second half of July (Fig. 1). these observations are somehow consistent with previous studies Regrowths from A1 and A2, namely RA1 and RA2, were about 17 on other species. Rapid stem elongation occurring at early stages and 13 Mg DM ha1 respectively. RA1 and RA2 were harvested 119 of the growing season usually leads to relatively low NSC concen- and 95 days after resprouting from the first cuts, corresponding to trations (Slewinski, 2012), then biomass NSC content typically in- 1305 and 983 Growing Degree Day (GDD) respectively (Fig. 2). De- creases over time. In switchgrass, once accumulation is spite the cumulative heat units were comparable with those of A3 interrupted by harvesting, a new accumulation cycle is performed and A4, RA1 and RA2 showed much more juvenile traits, thus being and a steep increase in starch concentration can be achieved (Bél- quite similar to A1 in height, green leaves and senescent leaves anger et al., 2012). This fact, as well as the slower growth rate and number and green leaves biomass, (Fig. 2). the increased leafiness, might explain the higher NSC content in Biomass chemical traits are consistent with crop maturity and RA1 and RA2. Moreover, the NSC content in the stem is often in- biomass partitioning. Total Kjeldahl Nitrogen concentration (TKN) creased after the formation of new leaves and decreased when was much higher in early cuts and in second cuts (about 0.75%) the photosynthetic rate is reduced by drought conditions, in order than in the mature crop (0.41% in A5) and a decreasing trend from to mitigate energy losses (Slewinski, 2012), thus possibly explain- A1 to A5 was showed. A3 had a TKN content slightly higher than A2 ing deviations in NSC respect to the expected trend (A2 < A1; and comparable with RA1, probably as a consequence of the high A4 < A3). proportion of green leaves (Fig. 2). Conversely, C/N ratio peaked (>120) in A4–A5 and exhibited its lowest values in A1 and in sec- 3.2. Digestion kinetics and Biochemical Methane Potential ond cuts. Whereas RA2 had a higher TKN content and a lower C/ N ratio than RA1, it was evidenced that nitrogen concentration de- Among the non linear models used for fitting the daily methane creased at the increase of plant age both at first and second cuts. production data, the Modified Gompertz resulted the most efficient G. Ragaglini et al. / Bioresource Technology 152 (2014) 107–115 111

Fig. 2. Crop biometrics and yield, biomass partitioning and chemical traits according to the different cuts. On the bottom are reported the days after sprouting (DAS) and the Growing Degree Days (GDD) accumulated from the sprouting date to harvest. For RA1 and RA2, DAS and GDD were calculated from the date of the first cut. GDD were estimated according to NOAA method.

(Table 2). In fact, the average RMSE resulted the lowest and the evidenced by a higher methane production rate, with a maximum 1 1 cross-correlation test evidenced an homogeneous distribution of (Rmax)upto51NLkgVS day , that was comparable to results the prediction error, with an average R2 very close to 1. Moreover, obtained by Grieder et al. (2012) (61 NL kg VS1 day1). About the AIC showed the lowest level of redundancy, despite the fact giant reed, it was observed that T50 occurred at the end of the that the Modified Gompertz model has the highest number of 4th day in A1, while it was recorded around the 5th day in the parameters, thus making clear its efficacy in describing the dy- two regrowths. Harvest times from A2 to A4 reached the 50% of namic of methane production of each replication, as already re- the methane production during the 6th day, while in A5 the early ported by other authors (Beuvink and Kogut, 1993; Grieder et al., stage of digestion was much slower and T50 was reached at the 8th 2012). Hence, this model only was adopted to estimate the kinetics day. A similar pattern was found for T95, since RA1, RA2, A1, and A3 parameters T50, T95, Rmax, and R50. Details of the model parameter- reached the 95% of the cumulative production within 29 days, ization are reported in Appendix B. while A2, A4, and A5 were significantly delayed. In particular, T95 Significant differences were evidenced among the investigated occurred significantly earlier in RA1 respect to the other substrates substrates (P < 0.001) (Fig. 3). During the first days of digestion, (23rd day), while it was reached about 7 days later in A2 and A4, the methane production was quite faster in maize silage (M) than and 10 days later in A5. The 95% of total methane production in in giant reed, reaching the 50% of the accumulated production (T50) M was reached in 28 days, thus showing a noteworthy decrease before the 4th day from the beginning of the assay. This fact was in methane production after the early fast stage. 112 G. Ragaglini et al. / Bioresource Technology 152 (2014) 107–115

experimental conditions and maize silage characteristics, accord- ing to the variety, the maturity class, the harvest time and the cli- matic conditions. On average, the BMP of giant reed was 319.5 NL kg VS1 at the first cut, while at the second cut it was 381 NL kg VS1. Thus, it showed quite high BMPs that are comparable with other grasses (Seppälä et al., 2009), while much lower BMPs are reported for (Kreuger et al., 2011). According to literature, switchgrass showed lower methane potentials (Massé et al., 2010; Frigon et al., 2012), but also in this case the BMPs and the digestibility (Bélanger et al., 2012) of the regrowths were higher than the first cuts. The Biochemical Biogas Potential (BBP) of maize silage was 576.8 NL kg VS1, significantly lower than the BBP measured in A3 and RA2 (P < 0.001), that were 641.6 and 632.7 NL kg VS1 respectively (Fig. 4). BBP of RA1 was similar to that observed in maize, while the lowest BBP was measured in A5 (less than 450 NL kg VS1). Obtained results evidenced that the low methane production of A5 depended both on a low biogas production and on a markedly reduced methane content (56%). At contrary, A1 showed a quite low BBP (499 NL kg VS1) but a significantly (P < 0.001) higher MC than the other substrates (68%) (Fig. 4). In particular, MC showed a clear decreasing trend from early harvests to late harvests and second harvests seemed to behave similarly to A1–A2. In general, no significant correlation between MC and BMP was found, while BMP seemed to be determined mainly by BBP

(P < 0.001) and R50 (P < 0.01), as showed in Table 3. Interestingly, some crop characteristics showed a strong corre- lation with the digestion kinetics parameters (Table 4). Plant

height was positively correlated with T95 (P<0.001) and T50 Fig. 3. Kinetics of fermentation of giant reed harvested at different times; A1–A5 (P < 0.05), while Rmax and R50 were correlated to the green leaves refer to first cuts, while RA1–RA2 refer to regrowths from A1 and A2. M is maize percentage (P < 0.01 and P<0.001, respectively). silage. (a) cumulative methane production along 40 days, T (d), and T (j). (b) 50 95 The methane production rate decreased as the crop was more Rmax (N) and daily methane production rate estimated as the first derivative of cumulate production curves. The significance level of ANOVA is showed using ⁄⁄⁄ mature. Crop maturity is known to negatively affect specific meth- symbol for p-values < 0.001 and standard error bars are reported. ane yields and digestibility of grasses (Nizami et al., 2009; Seppälä et al., 2009). Double cutting (A1 + RA1, A2 + RA2) prevented the crop from senescing and the harvest of juvenile plants allowed to

Rmax clearly allowed to distinguish giant reed harvest times in obtain biomass with a higher proportion of leaves, particularly of accordance with the stage of development of the crop (Fig. 3). In photosynthetically active and nitrogen-rich leaves, as commonly the juvenile stages (A1, RA1, RA2) Rmax ranged between 37 and observed in giant reed (Nasso o di Nasso et al., 2013). In addition, 44.4 NL kg VS1 day1 and a large variability among the replica- the proportion of stems respect to the more digestible leaves (Niz- tions was showed, while in the mid season stages (A2, A3, A4) Rmax ami et al., 2009) increased with giant reed development. Given that was between 28.5 and 37.8 NL kg VS1 day1. At crop maturity TKN content and C/N ratio vary according to the harvest time and

(A5) Rmax was markedly the lowest, being less than to the green leaves proportion, these parameters also contribute to 1 1 20 NL CH4 kg VS day . explain how crop characteristics and biomass partitioning are re- Therefore, RA2, RA1, and A1 exhibited the highest methane dai- lated to kinetics, BMP, and BBP at different harvest times (Table 4). ly production in the early stage of the batch anaerobic digestion Nitrogen content possibly affected to some extent also the varia- 1 process, as evidenced also by R50 (36.9, 31.8, and 34.5 NL kg VS tions of the methane concentration between harvest times, since 1 day respectively) (Fig. 4). However, R50 in M was even higher a significant correlation between TKN and MC was found 1 1 (45.9 NL kg VS day ), consistently with Rmax. R50 was about (P < 0.05). Giant reed, and particularly giant reed leaves, are usually 30 NL kg VS1 day1 in A3 and it was slightly lower in A2 and A4, found to be richer in nitrogen than other substrates from perennial while in A5 it drastically decreased (16.0 NL kg VS1 day1), being energy crops (Smith and Slater, 2011). Anaerobic digestion of pro- less than half that of A1. teins is known to result in higher methane concentrations in the After 40 days of fermentation, the highest Biochemical Methane biogas produced respect to simple carbohydrates (Amon et al., Potential (BMP) was measured in RA2, A3, and RA1 (391.7, 374.3, 2007; Rincón et al., 2010). In fact, biogas from A1, RA1, and RA2 and 370.3 NL kg VS1, respectively), also showing a significantly was characterized by a higher MC, while NSC may have played a higher potential than M (P < 0.001), that yielded 345 NL kg VS1 role lowering the MC in late harvests respect to early harvests. (Fig. 4). The BMP observed in A1, A2, and A4 was 332.9, 325.2, However, despite the significant variations between harvest times, and 306.7 NL kg VS1 respectively, while the lowest value was the NSC content in giant reed showed no significant correlation measured in A5 (258.3 NL kg VS1), about 34% less than in RA2. with the anaerobic digestion parameters, as the overall perfor- By literature, BMP values obtained from maize silages at com- mances were probably more influenced by other factors. parable experimental conditions were found to range from 310 Cellulose is recognized as a high-potential substrate, consider- to 371 NL kg VS1 (Bauer et al., 2010; Herrmann and Rath, 2012), ing that its experimental BMP typically approaches the theoretical thus being in line with the results of the present study. However, maximum of 415 NL kg VS1 (Kreuger et al., 2011; Triolo et al., lower and higher values have also been reported (Amon et al., 2011). Its reduced bioavailability is commonly acknowledged as 2007; Gao et al., 2012), consequently to the large variability of the most important limiting factor in anaerobic conversion of G. Ragaglini et al. / Bioresource Technology 152 (2014) 107–115 113

Fig. 4. BBP, BMP, average MC, and R50 of the investigated substrates. Significance level of ANOVA and standard error bars are reported.

Table 3 Pearson’ r correlation matrix of anaerobic digestion parameters.

BMP BBP MC T50 T95 Rmax BBP 0.92 *** MC 0.35 ns 0.03 ns

T50 0.74 ns 0.5 ns 0.75 ns * T95 0.82 0.58 ns 0.59 ns 0.75 ns * ** * Rmax 0.87 0.65 ns 0.62 ns 0.9 0.85 ** ** * *** R50 0.89 0.68 ns 0.67 ns 0.95 0.84 0.97 p levels are represented: ns (p > 0.05). * (p < 0.05). ** (p < 0.01). *** (p < 0.001).

Table 4 Pearson’s r correlation between anaerobic digestion parameters, crop characteristics, and biomass chemical traits.

Height Leaves (% of DM) Green leaves (% of DM) TKN (% of DM) Carbon (% of DM) C/N Lignin (% of DM) NSC (mg g1)

BMP 0.84 * 0.82 * 0.84 * 0.78 * 0.53 ns 0.83 * 0.49 ns 0.44 ns BBP 0.58 ns 0.64 ns 0.58 ns 0.51 ns 0.16 ns 0.56 ns 0.48 ns 0.43 ns MC 0.69 ns 0.57 ns 0.72 ns 0.76 * 0.9 ** 0.77 * 0.06 ns 0.16 ns * * * * T50 0.87 0.65 ns 0.78 0.85 0.75 ns 0.86 0.16 ns 0.09 ns *** * * * T95 0.92 0.75 ns 0.87 0.75 0.71 ns 0.78 0.53 ns 0.47 ns *** ** *** * *** Rmax 0.98 0.74 ns 0.88 0.93 0.79 0.92 0.18 ns 0.42 ns *** * *** *** *** R50 0.94 0.82 0.91 0.93 0.75 ns 0.94 0.23 ns 0.33 ns p levels are represented: ns (p > 0.05). * (p < 0.05). ** (p < 0.01). *** (p < 0.001).

lignocellulosic materials, mainly due to lignin accumulation over since acetyl bromide lignin content in giant reed showed no signif- the vegetative season (Nizami et al., 2009; Triolo et al., 2011; Mon- icant change in the considered period according to harvest time lau et al., 2013). In fact, negative correlations between ADL content and frequency. Therefore, the reduced BMP and digestion rate in and BMP have often been reported (Triolo et al., 2011; Kandel et al., mature plants (A5) cannot be explained in terms of an increased 2012). However, no correlation was found in the present study, lignin content but the reduced bioavailability of structural carbo- 114 G. Ragaglini et al. / Bioresource Technology 152 (2014) 107–115

Fig. 5. Biomass yield (a) and methane yield per hectare (b) of first cuts (A1–A5) and second cuts (RA1–RA2) of giant reed. Significance level of ANOVA and LSD test results are reported.

hydrates can even so be inferred. In fact, an even higher lignin con- kinetics most favorable to the retention times of real scale contin- tent (about 25%) was reported by Di Girolamo et al. (2013) in giant uously fed digesters (Mähnert and Linke, 2009; Grieder et al., reed harvested at a maturity stage comparable to A5. Despite this, 2012). Also other perennial grasses, such as switchgrass, tall fes- the methane potential found by the authors was higher than in cue, cocksfoot, and reed canary grass, showed an increased meth- other perennials cropped at full maturity and more similar to ane potential when managed under multiple harvest systems grasses harvested at earlier stages (Frigon et al., 2012; Di Girolamo (Massé et al., 2010; Seppälä et al., 2009; Kandel et al., 2012). Nev- et al., 2013). Hence, it is perceived that the chemical traits of giant ertheless, the age of the plantation, the water availability during reed should be further investigated to better understand the rela- the season, the length of the growing season as well as the nutri- tionship between maturity and methane potential. Indeed, modifi- tional status of the soil are crucial to support the regrowth when cations of cellulose crystallinity, physicochemical properties of perennial grasses are harvested twice during the vegetative devel- hemicelluloses, lignin polymerization, and composition over the opment (Kandel et al., 2012). Increased nitrogen requirements can crop cycle have been proposed by several authors (Nizami et al., be expected, due to higher TKN contents at juvenile stages, thus 2009; Monlau et al., 2013) as key factors influencing the availabil- potentially leading to an intensification of the cropping systems ity of both structural and non-structural carbohydrates for anaero- that should be also taken into account. On the other hand, a single bic digestion. late harvest could be interesting, since reduced costs per hectare can be expected and AGB yield is maximized. Biomass pre-treat- 3.3. Methane yields per hectare ments could cope with the reduced bioavailability of carbohy- drates that was shown in A4–A5, thus increasing giant reed Despite the highest BMP was reached in A3 and it decreased be- biomethane potential (Di Girolamo et al., 2013). If compared with yond early August, methane yields per hectare was higher in late maize and other agricultural feedstocks (Amon et al., 2007; Rincón harvests as consequence of the highest biomass production et al., 2010), giant reed showed a high C/N ratio at all the consid- (Fig. 5). Considering first cuts only, A5 was in fact the most produc- ered harvest times that could be a disadvantage in commercial 3 1 plants and it can be overcome by co-digestion. tive harvest time, exceeding 9580 Nm CH4 ha , despite the rela- tively low BMP. Methane yields per hectare reached by giant reed resulted higher than those reported in other studies for hemp 4. Conclusion (Kreuger et al., 2011), reed canary grass (Kandel et al., 2012), and switchgrass (Massé et al., 2010), and equaled the levels reported This study showed that giant reed has a high potential for biom- by Amon et al. (2007) for maize. Schievano et al. (2012) reported ethane production, especially when managed under double har- 3 a higher methane production per hectare (7,170–11,280 Nm CH4 - vesting systems. Double harvest increased the methane yield per ha1, under single harvest management) respect to maize that hectare by 20–35% respect to the most productive single harvest 1 3 1 yielded about 20 Mg DM ha (6750 Nm CH4 ha ). Giant reed time, as consequence of the highest BMP achievable by juvenile substantially exceeded those production levels under the double stages, and the digestion kinetics was also favored. This cropping 3 1 harvest system, achieving 11,585 and 12,981 Nm CH4 ha in system could allow a considerable land use saving for biomethane A2 + RA2 and A1 + RA1 respectively, about 20% and 35% higher production respect to maize, but further researches are needed in than A5 (Fig. 5). Considering that the dry biomass produced by order to evaluate the long term effect of double harvesting on A1 + RA1 and A2 + RA2 substantially matched that obtained by the plantation and to assess the methane potential of the ensiled A5 (Fig. 5), the increase of methane yield achieved by double har- biomass. vests essentially depended on the higher BMP showed by the juve- nile stages. These results pointed out a high potential of giant reed Acknowledgements for biomethane production, and they may lead to consider this crop as a suitable alternative to maize especially in Mediterranean The research was carried out under the BIOSEA Project, funded environments, where its high productivity under rainfed condi- by MIPAAF (Italy). tions was already proven (Angelini et al., 2009; Nasso o di Nasso The authors wish to thank Sergio and Carlo Cattani for their et al., 2013). A substantial reduction of soil use for biogas produc- help in building, setting-up and managing the batch anaerobic tion could be allowed, especially in double harvest systems. Fur- digestion system. Thanks are due to Cristiano Tozzini, Fabio Taccin- thermore, the juvenile stages of giant reed showed a digestion i, Nicoletta Nassi o Di Nasso, and the CIRAA (Pisa, Italy) for their G. Ragaglini et al. / Bioresource Technology 152 (2014) 107–115 115 contribution in managing the field trials. Thanks also to Elisa Pel- Grieder, C., Mittweg, G., Dhillon, B.S., Montes, J.M., Orsini, E., Melchinger, A.E., 2012. legrino for her precious suggestions, to Federico Triana, Neri Ron- Kinetics of methane fermentation yield in biogas reactors: genetic variation and association with chemical composition in maize. Biomass 37, 132– cucci, Maria Valentina Lasorella, Valentina Giulietti, and Nico 141. Viligiardi for their support in field sampling, to Matteo Gnocato Herrmann, A., Rath, J., 2012. Biogas production from maize: current state, and Cooperativa Valle Bruna (Grosseto, Italy) for providing the challenges, and prospects. 1. Methane yield potential. Bioenergy Res. 5, 1027– 1042. inoculum, to Alessio Giovannelli and CNR IVALSA (Firenze, Italy) Heaton, E.A., Dohleman, F.G., Long, S.P., 2009. Seasonal nitrogen dynamics of for collaboration. x giganteus and . Global Change Biol. Bioenergy 1, 297–307. Kandel, T.P., Sutaryo, S., Møller, H.B., Jørgensen, U., Lærke, P.E., 2012. Chemical Appendix A. Supplementary data composition and methane yield of reed canary grass as influenced by harvesting time and harvest frequency. Bioresour. Technol. 130, 659–666. Supplementary data associated with this article can be found, in Kreuger, E., Prade, T., Escobar, F., Svensson, S.E., Englund, J.E., Björnsson, L., 2011. Anaerobic digestion of industrial hemp – effect of harvest time on methane the online version, at http://dx.doi.org/10.1016/j.biortech.2013.11. energy yield per hectare. Biomass Bioenergy 35, 893–900. 004. Lygin, A.V., Upton, J., Dohleman, F.G., Juvik, J., Zabotina, O.A., Widholm, J.M., Lozovaya, V.V., 2011. Composition of cell wall phenolics and polysaccharides of the potential bioenergy crop–Miscanthus. Global Change Biol. Bioenergy 3, References 333–345. Mähnert, P., Linke, B., 2009. Kinetic study of biogas production from energy crops Amon, T., Amon, B., Kryvoruchko, V., Zollitsch, W., Mayer, K., Gruber, L., 2007. and animal waste slurry: effect of organic loading rate and reactor size. Environ. Biogas production from maize and dairy cattle manure – influence of Technol. 30, 93–99. biomass composition on the methane yield. Agric. Ecosyst. Environ. 118, Massé, D., Gilbert, Y., Savoie, P., Bélanger, G., Parent, G., Babineau, D., 2010. Methane 173–182. yield from switchgrass harvested at different stages of development in Eastern Angelidaki, I., Alves, M.M., Bolzonella, D., Borzacconi, L., Campos, J.L., Guwy, A.J., Lier, Canada. Bioresour. Technol. 101, 9536–9541. J.V., 2009. Defining the biomethane potential (BMP) of solid organic wastes and Monlau, F., Barakat, A., Trably, E., Dumas, C., Steyer, J.-P., Carrere, H., 2013. energy crops: a proposed protocol for batch assays. Water Sci. Technol. 59, 927– Lignocellulosic materials into biohydrogen and biomethane: impact of 934. structural features and pretreatment. Crit. Rev. Environ. Sci. Technol. 43, 260– Angelini, L.G., Ceccarini, L., Nassi o Di Nasso, N., Bonari, E., 2009. Comparison of 322. Arundo donax L. and in a long-term field experiment in Nassi o Di Nasso, N., Roncucci, N., Triana, F., Tozzini, C., Bonari, E., 2011. Seasonal Central Italy: analysis of productive characteristics and energy balance. Biomass nutrient dynamics and biomass quality of giant reed (Arundo donax L.) and Bioenergy 33, 635–643. miscanthus (Miscanthus x giganteus Greef et Deuter) as energy crops. Ital. J. APHA, Awwa, WEF, 2005. Standard Methods for the Examination of Water and Agron. 6, 152–158. Wastewater, 21st ed. USA, Washington DC. Nassi o Di Nasso, N.N., Roncucci, N., Bonari, E., 2013. Seasonal dynamics of Bauer, A., Leonhartsberger, C., Bösch, P., Amon, B., Friedl, A., Amon, T., 2010. Analysis aboveground and belowground biomass and nutrient accumulation and of methane yields from energy crops and agricultural by-products and remobilization in Giant reed (Arundo donax L.): a three-year study on estimation of energy potential from sustainable crop rotation systems in EU- marginal land. Bioenergy Res. 3, 1–12. 27. Clean Technol. Environ. Policy 12, 153–161. Nizami, A.S., Korres, N.E., Murphy, J.D., 2009. Review of the integrated process for Bélanger, G., Savoie, P., Parent, G., Claessens, A., Bertrand, A., Tremblay, G.F., the production of grass biomethane. Environ. Sci. Technol. 43, 8496–8508. Babineau, D., 2012. Switchgrass silage for methane production as affected by Pilu, R., Bucci, A., Badone, F.C., Landoni, M., 2012. Giant reed (Arundo donax L.): a date of harvest. Can. J. Plant. Sci. 92, 1187–1197. weed plant or a promising energy crop? Afr. J. Biotechnol. 11, 9163–9174. Beuvink, J.M., Kogut, J., 1993. Modeling gas production kinetics of grass silages Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., 2013. nlme: linear and nonlinear mixed incubated with buffered ruminal fluid. J. Anim. Sci. 71, 1041–1046. effects models. R package version 3.1-110, . Biochemical methane potential of biomass and waste feedstocks. Biomass Rincón, B., Banks, C.J., Heaven, S., 2010. Biochemical methane potential of winter Bioenergy 5, 95–111. (Triticum aestivum L.): influence of growth stage and storage practice. Di Girolamo, G., Grigatti, M., Barbanti, L., Angelidaki, I., 2013. Effects of Bioresour. Technol. 101, 8179–8184. hydrothermal pre-treatments on Giant reed (Arundo donax) methane yield. Schievano, A., D’Imporzano, G., Corno, L., Adani, F., Badone, F.C., Pilu, S.R., 2012. Più Bioresour. Technol. 147, 152–159. biogas a costi inferiori con arundo o doppia coltura. Inf. Agr. 25, 21–25. Dragoni, F., Ragaglini, G., Nassi O Di Nasso, N., Tozzini, C., Bonari, E., 2011. Suitability Seppälä, M., Paavola, T., Lehtomäki, A., Rintala, J., 2009. Biogas production from of giant reed and miscanthus for biogas: preliminary results on harvest time boreal herbaceous grasses – specific methane yield and methane yield per and ensiling. Asp. Appl. Biol. 112, 291–296, Biomass and Energy Crops hectare. Bioresour. Technol. 100, 2952–2958. Conference IV. Slewinski, T.L., 2012. Non-structural carbohydrate partitioning in grass stems: a Frigon, J.C., Mehta, P., Guiot, S.R., 2012. Impact of mechanical, chemical and target to increase yield stability, stress tolerance, and production. J. Exp. enzymatic pre-treatments on the methane yield from the anaerobic digestion of Bot. 63, 4647–4670. switchgrass. Biomass Bioenergy 36, 1–11. Soetaert, K., 2013. rootSolve: non-linear root finding, equilibrium and steady-state Fukushima, R.S., Hatfield, R.D., 2004. Comparison of the acetyl bromide analysis of ordinary differential equations. R package version 1.6.4, . determining lignin concentration in forage samples. J. Agric. Food Chem. 52, Smith, R., Slater, F.M., 2011. Mobilization of minerals and moisture loss during 3713–3720. senescence of the energy crops Miscanthus x giganteus, Arundo donax and Gao, R., Xufeng, Y., Wanbin, Z., Xiaofen, W., Shaojiang, C., Xu, C., Cui, Z., 2012. arundinacea in Wales, UK. Global Change Biol. Bioenergy 3, 148–157. Methane yield through anaerobic digestion for various maize varieties in China. Triolo, J.M., Sommer, S.G., Moller, H.B., Weisbjerg, M.R., Jiang, X.Y., 2011. A new Bioresour. Technol. 118, 611–614. algorithm to characterize biodegradability of biomass during anaerobic Giovannelli, A., Emiliani, G., Traversi, M.L., Deslauriers, A., Rossi, S., 2011. Sampling digestion: influence of lignin concentration on methane production potential. cambial region and mature xylem for non structural carbohydrates and starch Bioresour. Technol. 102, 9395–9402. analyses. Dendrochronologia 29, 177–182.