methanogenic activity optimization using the response surface methodology, during the anaerobic...
TRANSCRIPT
ww.sciencedirect.com
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 7
Available online at w
ScienceDirect
ht tp: / /www.elsevier .com/locate/biombioe
Methanogenic activity optimization using theresponse surface methodology, during theanaerobic co-digestion of agriculture and industrialwastes. Microbial community diversity
J. Jim�enez a,*, Y. Guardia-Puebla b, O. Romero-Romero a,M.E. Cisneros-Ortiz c, G. Guerra d, J.M. Morgan-Sagastume c, A. Noyola c
a Universidad de Sancti Spıritus, Ave. de los M�artires, No 360, CP 60100 Sancti Spıritus, Cubab Universidad de Granma, Carretera de Manzanillo, Km 17 ½, CP 85100 Bayamo, Granma, Cubac Instituto de Ingenierıa, Universidad Nacional Aut�onoma de M�exico, Circuito Escolar S/N Ciudad Universitaria,
Delegaci�on Coyoac�an, CP 04510 D.F. M�exico, Mexicod Facultad de Biologıa, Universidad de La Habana, Calle 25 e/ I y J, Vedado, CP 10400 La Habana, Cuba
a r t i c l e i n f o
Article history:
Received 12 August 2012
Received in revised form
4 April 2014
Accepted 24 October 2014
Available online 14 November 2014
Keywords:
Clay
Pig manure
Optimization
Rice straw
Specific methanogenic activity
Microbial community
* Corresponding author. Tel.: þ53 41664137.E-mail addresses: [email protected] (J.
Cisneros-Ortiz), [email protected] (G. Guerra),(A. Noyola).http://dx.doi.org/10.1016/j.biombioe.2014.10.00961-9534/© 2014 Elsevier Ltd. All rights rese
a b s t r a c t
The anaerobic co-digestion of manure, agriculture and industrial wastes for methane
production depends on the nutritional condition to develop the microbial community. The
effect of each substrate concentrations, as well as their interactive effects on specific
methanogenic activity and microbial community diversity were investigated in this work.
A central composite design and the response surface methodology were applied for
designing the anaerobic co-digestion batch test at 35 and 55 �C. It was analyzed the
anaerobic sludge by specific methanogenic activity (SMA) and using molecular techniques
(terminal restriction fragment length polymorphism, TRFLP). The results showed a sig-
nificant interaction among the substrates and an enhancement of the methane production
and SMA response caused by the three components. Rice straw had lower influence on
SMA than clay residues, due to the mineral content and the beneficial ammonia nitrogen
adsorbent properties of the latter. The optimum condition for mesophilic and thermophilic
anaerobic co-digestion of pig manure, rice straw and clay mixture allowed SMA values of
1.31 and 1.38 gCH4-COD/gVSSd�1, respectively. The TRFLP analysis showed the effect of rice
straw and clay addition on microbial community diversity at both temperatures. The
acetotrophic methanogens belonging to the order Methanosarcinales (genera Methanosarcina
and Methanosaeta) dominated in mesophilic condition, whereas at thermophilic conditions
dominated Methanomicrobiales and Methanobacteriales order. The optimization allowed
identifying the substrate interaction effects in a concentration range with a reduced
number of experiments. Besides, the model validation proved to be useful for defining
optimal combination of wastes in anaerobic system.
© 2014 Elsevier Ltd. All rights reserved.
Jim�enez), [email protected] (Y. Guardia-Puebla), [email protected] ([email protected] (J.M. Morgan-Sagastume), [email protected]
23rved.
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 7 85
1. Introduction
The co-digestion of different wastes may improve nutrient
balance and cause synergy effects, overcoming substrate
deficits [1]. Moreover, this type of waste management may
improve methane yield and increase the efficient use of
equipment by processing different waste streams in a single
facility. The co-digestion of manure and industrial organic
wastes has been widespread in Europe [2] and reports on in-
dustrial applications of this concept have been published
[3e9].
Manures are an abundant source of organic material that
can be used as feedstock in anaerobic digesters [10,11]. How-
ever, manures often contain concentrations of ammonia
greater than necessary formicrobial growth, whatmay inhibit
the anaerobic digestion [12,13]. On such cases, the anaerobic
digestion of pig manure could be enhanced using agriculture
wastes as co-substrates, due to their high content of carbon
and subsequent improvement of carbon nitrogen (C/N) ratio
[1]. This way, rice straw could be a promising feedstock
biomass as co-substrate for pig manure anaerobic digestion,
mainly due to the low costs of this waste biomass [14].
In addition, some clays and zeolite have been described as
a means to reduce ammonia inhibition in the anaerobic
digestion of manure [12,15e19]. Angelidaki and Ahring [16]
used bentonite clay for the anaerobic thermophilic digestion
of cattle manure. Milan et al. [17], used a natural zeolite, a
modified zeolite [15] and a mixture of clinoptilolite, morden-
ite, montmorillonite and others [18], for the anaerobic diges-
tion of pig waste. Tada et al. [19], also found that natural
mordenite had a synergistic effect on the Ca2þ supply as well
as on NH4þ removal during the anaerobic digestion of a sludge
with high ammonia content (NH4Cl 4500mg L�1). On the other
hand, the metal contained on clays could be used by the
anaerobic microorganisms as part of their enzyme structure
and has a significant effect on the anaerobic degradation of
VFA, being this an ongoing research subject [20e24].
Despite the advantages of the co-digestion process, the
addition of co-substrate from a different typology can provoke
cell toxicity, that is why the optimization of substrate con-
centration, temperature and others factors that affect the co-
digestion process is necessary [1,25e27]. Response surface
methodology (RSM) is feasible to solve this kind of problem,
since it is a statistical technique for designing experiments,
building models, evaluating the effects of several factors and
searching optimum conditions for desirable responses,
maintaining a reduced number of experiments.With RSM, the
interaction of possible influencing parameters on methane
production can be effectively evaluated [28]. Furthermore,
central composite design is a fractional factorial design
effective for sequential experimentation to obtain a reason-
able amount of information for testing lack of fit while a large
number of design points are not involved [28].
In this sense, different molecular methods have allowed
[29,30] a better approach to the physiology of the microor-
ganisms involved in this process. The terminal restriction
fragments length polymorphism analysis (TRFLP) is an effec-
tive tool to determine the molecular composition of the mi-
crobial community and the relative abundance of certain
species [31]. This tool has been successfully applied in
different ecosystems such as methanogenic reactors [32,33].
The interactive effects of substrate concentration onmethane
production from co-digestion of pig manure, rice straw and
clay residues as inorganic additives have not been reported
yet. Consequently, the main objective of this work was to
investigate the effect of pig manure, rice straw and clay res-
idue concentration, as well as their interactive effect, on the
specific methanogenic activity and microbial community di-
versity at mesophilic and thermophilic conditions.
2. Methods
2.1. Inoculum and wastes sources
Two anaerobic inoculums were used depending on the tem-
perature tested: adapted to mesophilic (35 ± 2 �C) and to
thermophilic (55 ± 2 �C) conditions, both fed with pig manure
collected at the Veterinary School at Autonomy National
University of Mexico (UNAM). Rice straw was collected from
Rice Cuban Enterprise “Sur del Jibaro” and the clay residuewas
taken from the oil clarification process at Refinery and Petro-
chemical Industry “Sergio Soto”, both located in the province
of Sancti Spıritus. The pigmanurewas kept at 4 �C and the rice
straw and clay residual were kept at environmental temper-
ature until use. The characteristics of the substrates are
shown in Table 1.
Experiments were carried out in batch tests, containing
mineralmedium (1%), vitamins solutions (1%), micronutrients
(1%), resazurin (0.1%) and cystein (1 g L�1). Rice straw was
pretreated by size reduction. The oily residue was removed
from clay using absorbent paper. All bottles were inoculated
in an anaerobic chamber and incubated at mesophilic condi-
tion (35 ± 2 �C) or thermophilic condition (55 ± 2 �C), during30 days. Bottles with inoculum without substrate, and inoc-
ulum with pig manure were used as blanks.
2.2. Analytical techniques
The anaerobic process was monitored by means of total sus-
pended solids (TSS), volatile suspended solids (VSS), pH (pH-
conductivity meter, OAKTON, EUTECH Instrument,
Singapore), and alkalinity, determined according to the Stan-
dard Methods for the Examination of Water and Wastewater
[34]. The alkalinity ratio (a) was calculated as the quotient of
partial alkalinity (at pH 5.75) and total alkalinity (at pH 4.3).
2.3. Methane production and specific methanogenicactivity
The methane production was determined every day by gas
chromatography (Fisher Gas Partitioner model 1200 with a
thermal conductivity detector and a Porapak Q column).
Subsequently, the total methane in the bottle gas space was
determined. SMA was calculated with the slope of the accu-
mulatedmethane production curve (mL d�1) in the first 5 days,
divided by the amount of VSS introduced in the bottle
Table 1 e Characteristics of the inoculum and wastes used.
Parameters Inoculum Pig manure Rice straw Clay residual
Mesophilic Thermophilic
TS (%) 1.22 1.04 6.3 94.6 97.7
VS (%) 0.79 0.64 5.3 70.4 41.3
Total-N (g kg�1) e e 4.12 8.96 0.0
NeNH4þ1 (g kg�1) 0.68 0.53 2.24 0.28 0.0
C:N ratio e e 19 48 0.0
Alkalinity (gCaCO3 L�1) 2.1 2.2 2.7 e e
Alkalinity ratio 0.8 0.7 0.45 e e
pH 8.0 7.9 7.9 e e
Naþ (mg kg�1) e e 245.8 1972 393.5
Kþ (mg kg�1) e e 2100 12.186 527.5
Mg2þ (mg kg�1) e e 1824 18.76 42.48
Ca2þ (mg kg�1) e e 1761 1309 3708
PO43� (mg kg�1) e e 7019 2.95 <0.5
Al (mg kg�1) e e 271.9 5936 13.127
Fe (mg kg�1) e e 442.4 4754 1510
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 786
(inoculum) using the proper conversion factor to report it as:
gCH4-COD gVSS�1 d�1.
2.4. Central composite design experiments
In order to optimize the concentration of each waste and to
analyze the effect of their interaction, a 23 factorial central
composite design was used based on the STATISTICA software
(version 8.0, StatSoft Inc., USA). The factorial design was
amplified by six axial points, and two replications of center
points. The center runs provided a means for estimating the
experimental errors and a measure of lack of fit. The axial
points were added to the factorial design for estimating the
model curvature.
The experiments were tested in three replicates at both
temperature conditions (35 �C and 55 �C) for the selected
variables: manure (A), straw (B) and clay (C), in concentration
ranges of 9.1e36.6 gVSS L�1, 7.0e17.6 gVSS L�1 and
0.8e8.3 gVSS L�1, respectively. The effect of each factor was
evaluated under both low and high range of the remaining
factors. The concentration of each waste was chosen as three
independent variables in this experiment design; SMAwas the
dependent variable and was calculated as the average of the
replicates.
The dependent variable was fitted using a predictive
polynomial quadratic equation in order to correlate the
response variable to the independent variables. The general
shape of the predictive polynomial quadratic equation is:
Y ¼ N0 þXk
i¼1
Nixi þXk
i¼1
Niix2i þ
Xk
i¼1
Xk
j¼1
Nijxixj (1)
where Y is the response (dependent variable SMA); xi is the
input variable (concentration of manure, straw and clay),
which influence the response variable Y; N0 the i linear coef-
ficient; Nii the quadratic coefficient and Nij is ij interaction
coefficient.
Each independent variable and their interaction were
compared to verify the feasible hypothesis regarding the
treatment effect and their estimation with 0.05 of confidence
level. The data normality was confirmed through the
construction of the residue normal probability graphic. The
standardized effects of the independent variables and their
interactions on the dependent variable were also calculated
by preparing a Pareto chart.
The quality of fit of the model equation was expressed by
the determination coefficient (R2) and the adjusted determi-
nation coefficient (R2adj). The model statistical significance was
determined by a Fisher test (F-test) based on the p-value with
95% of confidence level. Also, the correlation coefficient (R), the
DurbineWatson (DW) statistic, the sum of squares (SS), the
middle sumof squares (MSS), and chi-square (c2) testwere used
to analyze the statistical significance of the quadratic model.
The total percentage of contribution (PC) achieved from the
lineal models, quadratic terms and interactions were calcu-
lated by the equations two, three and four as follows (Eqs
(2)e(4)):
TPCi ¼Pn
i¼1 SSiPni¼1
Pnj¼1 SSi þ SSj þ SSij
*100 (2)
TPCii ¼Pn
i¼1 SSiiPni¼1
Pnj¼1 SSi þ SSj þ SSij
*100 (3)
TPCij ¼Pn
i¼1 SSijPni¼1
Pnj¼1 SSi þ SSj þ SSij
*100 (4)
where: TPCi, TPCii, and TPCij are the total percentage contri-
butions (TPC) of first-order, quadratic and interaction terms,
respectively. Similarly, SSi, SSii and SSij are the computed sum
of squares for first-order, quadratic and interaction terms,
respectively [35].
Finally, three-dimensional (3D) response surfaces and two-
dimensional (2D) contour plots were constructed to give an
insight about the effects of the different testedwastes and their
interaction on the methane production measured as SMA.
2.5. Cultivation independent analysis of the microbialcommunity structure
Sampling and DNA extraction: For microbial community
analysis samples from the CSTR were taken (i) before the co-
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 7 87
digestion process, i.e. inoculum only fed with manure kept in
the control reactor, (ii) from the co-digestion of manure and
straw, and (iii) during and after clay supplementation.
Genomic DNA was isolated using the PowerSoil® DNA Isola-
tion Kit (Mo Bio Laboratories Inc., USA) according to the
manufacturer's instructions. For each crude sample, DNA
from two subsamples was extracted. The extracted DNA was
then used as template for TRFLP andQ-PCR analyses aswell as
for the construction of 16S ribosomal RNA gene clone libraries
to characterize the diversity and dynamic of the Bacterial and
Archaeal consortia during the experimental approach
described above.
TRFLP analysis: The TRFLP analyseswere in general carried
out following the protocol proposed above [33]. For the
amplification of the 16S rRNA gene the primer pairs 27f (50
AGAGTTTGATCMTGGCTCAG 30, [36,37]) and 926MRr (50
CCGTCAATTCMTTTRAGTTT 30, [38,39]) (for bacteria) and the
primer pairs Ar109f (50 ACKGCTCAGTAACACGT 30, [40]) andAr912r (50CTCCCCCGCCAATTCCTTTA 30; [41]) (for Archaea)
were used, where both forward primers were labeled with
Indodicarbocyanine (Cy5) at the 50-end. The TRFLP-PCR was
carried out in three replicates per DNA sample. After purifi-
cation of the PCR products, a total amount of 200 ng PCR
products were digested with MspI and Hin6I in case of the
Bacterial assay or with AluI for the Archaeal assay. All en-
zymes were provided by Thermo Scientific Fermentas, Ger-
many. As described in detail [33] the digestion fragments were
electrophoretically separated and detected by fluorescence
using a GenomeLab™ GeXP Genetic Analysis System (Beck-
man Coulter, Krefeld, Germany). The obtained data were pre-
analyzed using the GeXP analysis software (version 10.2)
considering the size calculation of the detected terminal re-
striction fragments (TRFs) based on the migration time of the
applied size standard. All fragments with a sequence length
between 60 and 650 bpwere used for further analyses with the
online available software package T-Rex [42]. The identifica-
tion of “true” peaks by distinguishing baseline “noise” from
Table 2 e Specific methane activity (SMA) observed (Obs.) andstraw and clay residue.
Run Pig manure(gVSS L�1)
Rice straw(gVSS L�1)
Clay residu(gVSS L�1)
1 9.10 7.00 0.80
2 9.10 7.00 8.30
3 9.10 17.60 0.80
4 9.10 17.60 8.30
5 36.60 7.00 0.80
6 36.60 7.00 8.30
7 36.60 17.60 0.80
8 36.60 17.60 8.30
9 9.10 12.30 4.55
10 36.60 12.30 4.55
11 22.85 7.00 4.55
12 22.85 17.60 4.55
13 22.85 12.30 0.80
14 22.85 12.30 8.30
15 22.85 12.30 4.55
16 22.85 12.30 4.55
a Specific methanogenic activity (g CH4-COD gVSS�1 d�1).
signals of fluorescently labeled fragments [43] as well as the
alignment (binning) of TRFs with a threshold of 0.5 [44] were
based on the evaluation of the peak height. In a last evaluation
step, TRFs were visualized by their relative distribution
considering that TRFs with a relative abundance lower than
2% were removed from the analyses.
3. Results and discussion
The results showed an increase on methane production dur-
ing the anaerobic co-digestion of pig manure, rice straw and
clay residues at all level mixture in respect to the blank used
(data not shown), similarly to results previously obtained
[11,14]. But the reactionmagnitudes were different in all cases
showing different metabolic behavior. With respect to the
anaerobic process, the values obtained at the end of each
experimental condition -pH (7.4e8.1), alkalinity
(2.98e3.80 gCaCO3 L�1), a (0.67e0.84) and the VSS removal
(72e85 %)- were acceptable, indicating the process stability.
3.1. Determination of the regression model and thewaste effect on specific methanogenic activity
The design matrix in actual terms and the experimental re-
sults of SMA are presented in Table 2. The experimental data
were fitted to equation one, resulting in a mathematical
regression second-ordermodel to explain the behavior of SMA
in mesophilic and thermophilic conditions respectively in
terms of uncoded as follows:
SMA ¼� 1:12217þ 0:156522$Aþ 0:0986102$B� 0:0533851$C
� 0:00239521$A2 � 0:00251286$A$B� 0:00267879$A$C
� 0:00366126$B2 � 0:0124214$B$C� 0:00233563$C2
(5)
predicted (Pred.) at various waste levels: pig manure, rice
e SMAa (35 �C) SMA (55 �C)
Obs. Pred. Obs. Pred.
0.27 0.36 1.11 1.15
0.42 0.36 0.61 0.54
0.42 0.40 1.02 0.89
1.26 1.32 0.29 0.29
1.26 1.20 0.25 0.27
0.56 0.58 0.69 0.84
0.38 0.45 0.15 0.25
0.84 0.75 0.85 0.83
0.84 0.76 0.74 0.90
0.84 0.89 0.96 0.72
1.12 1.12 1.45 1.32
1.26 1.23 1.14 1.19
1.26 1.17 1.31 1.28
1.26 1.32 1.32 1.27
1.26 1.29 1.27 1.35
1.26 1.29 1.27 1.35
Table 3 e Analysis of the variation of SMA for the
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 788
SMA ¼ 0:358717þ 0:0899109$Aþ 0:01027$B� 0:0793558$C2
experiments carried out in mesophilic (35 �C) andthermophilic (55 �C) conditions.
Source SSa DFb MSc F e ratio P e value
Mesophilic (35 �C)Model 2.1533 9 0.2393 25.61 0.00041d
A: Manure 0.0449 1 0.0449 4.81 0.07086
B: Straw 0.0281 1 0.0281 3.01 0.13359
C: Clay 0.0563 1 0.0563 6.02 0.04952d
A2 0.5549 1 0.5549 59.41 0.00025d
AB 0.3160 1 0.3160 33.83 0.00114d
AC 0.1891 1 0.1891 20.25 0.00411d
B2 0.0312 1 0.0312 3.34 0.11735
BC 0.4278 1 0.4278 45.80 0.00051d
C2 0.0040 1 0.0040 0.42 0.53873
Error 0.0560 6 0.0093
Total 2.2093 15
Thermophilic (55 �C)Model 2.3820 9 0.2647 9.24 0.00682d
A: Manure 0.0757 1 0.0757 2.64 0.15519
B: Straw 0.0436 1 0.0436 1.52 0.26365
C: Clay 0.0006 1 0.0006 0.02 0.88608
A2 0.7791 1 0.7791 27.20 0.00199d
AB 0.0276 1 0.0276 0.96 0.36412
AC 0.7021 1 0.7021 24.51 0.00258d
B2 0.0256 1 0.0256 0.90 0.38063
BC 0.0001 1 0.0001 0.004 0.95207
C2 0.0163 1 0.0163 0.57 0.47926
Error 0.1719 6 0.0286
Total 2.5538 15
a Sum of squares.b Degrees of freedom.c Mean of squares.d p-value <0.05 were considered to be significant.
� 0:00290271$A þ 0:00100343$A$Bþ 0:0056667$A$C
� 0:00369502$B2 þ 0:00091195$B$C� 0:0061364$C2
(6)
where SMA is the predicted value from regression equation
and A, B, and C are the substrate concentration (pig manure,
rice straw and clay residual, respectively). The optimum
values from the selected variables were obtained by Eqs. (5)
and (6) and also analyzing the response surface contour plots.
On the other hand, the analysis of variance (ANOVA) was
conducted to study the significance of fit from second-order
polynomial equation for the experimental data. The SMA re-
sults obtained at various wastes combinations using ANOVA
are summarized in Table 3 for mesophilic and thermophilic
conditions.
The p-values lower than 0.05 indicated that model terms
were significant. The ANOVA showed that quadratics models
were the most significant, illustrated by the F-test value
(Fmodel 35 �C¼ 25.61 and Fmodel 55 �C¼ 9.24) because the p-values
were lower than 0.05 (p-value ¼ 0.0041 and p-value ¼ 0.00682,
for mesophilic and thermophilic conditions, respectively).
Besides the F-value calculated (Fcal ¼ MSmodel/MSerror) were
higher than F-value tabulated (Fa;df,(n�dfþ1) F0.05,9,6 ¼ 4.1) at the
5% level of confidence. This indicated that the F-value ach-
ieved was enough to justify the quadratic model adequacy,
also showed that all waste combinations were significant.
These results are similar to reported by some authors
[35,45,46].
The higher F-test values and the p-value lower than 0.05
reject the null hypothesis and showed a significant effect from
both models. In mesophilic conditions the ANOVA (Table 3)
shows the significant effect of the quadratic effect of manure
(A2), manure-straw interaction (AB), manure-clay interaction
(AC) and straweclay interaction (BC), also the lineal effect of
Clay (C) was significant. Nevertheless, the p-value (p-
value¼ 0.04952) showed a little effect on the regressionmodel.
This result accorded with the quadratic effect of clay (C2)
which the least effect on the response showed. Besides, the
addition of straw did not show a significant effect on SMA. In
thermophilic condition, the quadratic effect of manure (A2),
and manure-clay interaction (AC) had a significant effect on
SMA, with p-values of 0.00199 and 0.00258, respectively.
However, the clay and the straweclay interaction showed the
slightest effect on SMA. The significant effect of temperature
condition on anaerobic processes is evident, essentially the
different substrate typology assimilation by the anaerobic
consortia previously reported [25,47].
The model adequacy was better-quality in mesophilic
conditions than in thermophilic conditions, where the deter-
mination coefficient (R2 ¼ 0.97463) showed that only 2.54% of
total variation could not be elucidated from regression model.
For thermophilic condition (R2 ¼ 0.9327) this value was higher
(6.73%) (Fig. 1). Nevertheless, both models confirm the
quadratic model adequacy to estimate the experimental data
with high-degree of fit. That was also corroborated by means
of the adjusted coefficient of determination (R2adj ¼ 0.93658 and
R2adj ¼ 0.83174 for mesophilic and thermophilic condition,
respectively). These values were close to the determination
coefficient (R2) values achieved for both condition described
above. It indicated the fine adjustment considering this size
sample and model component number. These results were in
agreement with the higher values for both coefficient ach-
ieved by Yetilmezsoy et al., and Liu et al. [35,46], using an
experimental design for response surfaces. On the other hand,
the correlation coefficient (R) was important to analyze the
relationship between the SMA values observed and the SMA
values predicted by the model. In mesophilic conditions the
elevated values of correlation coefficient (R35 �C ¼ 0.987235)
showed a positive and strong link due to the exponential slope
growth (Fig. 1a). In contrast, in thermophilic conditions
(Fig. 1b) the correlation coefficient (R55 �C¼ 0.804312) was lower
than in mesophilic condition, even though this values was
high too. In addition, the dispersion between the data
observed and the values predicted by models showed that the
points were close to the diagonal line, it evidences the high-
quality of fit in both temperature conditions.
The DurbineWatson (DW) statistic was used to determine
if autocorrelation of first-order existed and to test the linear
association between adjacent residuals, that is, whether
autocorrelation, or correlation between errors, is present in
the model [35]. In both cases, the DW statistic ranges were
nearly two, showing the high-quality of fit in both tempera-
ture conditions. In mesophilic conditions, DW was 2.38229 (p-
value35 �C ¼ 0.1623) while in thermophilic condition, DW was
2.47364 (p-value55 �C¼ 0.1241). In both cases, the p-values were
Fig. 1 e Graphics showing the correlation between the observed and predicted values for both temperatures.
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 7 89
over 0.05, what means that there is not a serial correlation
between the residues.
In addition, the chi-square (c2) test were carried out to
confirm whether there was a significant difference between
the data observed and the values predicted by the models. In
both temperature conditions the chi-square calculated (c2cal
35 �C ¼ 0.085; c2cal 55 �C ¼ 1.458) were lower than the chi-square
tabulated (c2tab (a; df�1) ¼ c2tab (0.05; 15) ¼ 7.261). Therefore there
is no significant difference between the response observed
and the response predicted by the models. Chi-square test
corroborates that the quadraticmodel proposedwas adequate
to optimize SMA as the response variable from this waste
anaerobic co-digestion process.
The SMA values had a wide range (0.27e1.26 gCH4-
COD gVSS�1 d�1 and 0.15e1.45 gCH4-COD gVSS�1 d�1 at mes-
ophilic and thermophilic conditions, respectively), and it
clearly showed that the SMA were strongly affected by the
selected variables (concentration andwastes typology). This is
corroborated by the wide range of term coefficient values.
The technique of arranging data according to priority or
importance in a problem-solving framework is called Pareto
analysis. Pareto charts are used extensively by improve-
ment teams all over the world; indeed the technique has
become fundamental to their operation for identifying the
really important problems and establishing priorities for
action [48]. Thus, the standardized effects of the indepen-
dent variables and their interactions on the dependent
variable, SMA, were investigated with a Pareto chart (Fig. 2).
In mesophilic conditions, the standardized effect of (C2), (B),
(B2) and A were below the reference line, considering 0.05 of
probability, even if the C component was at limit position.
This showed a lower effect on the SMA. However, the
negative coefficients for the model components (bars with a
minus sign): the quadratic component (A2) and the inter-
action coefficient of (AB) and (AC) indicated an antagonistic
effect on the SMA; while the positive coefficients for the
model component (bars with a plus sign), BC showed a
favorable or synergistic effect on the SMA (Fig. 2a). In
contrast, in thermophilic conditions, the standardized effect
of (BC), (C), (C2), (B2), (AB), (B) and (A) did not contribute to
the prediction model of SMA. The negative coefficient for
the model component A2 was the only one that showed an
unfavorable or antagonistic effect on the SMA, while the
positive coefficient for the model components (AC) showed
a favorable or synergistic effect on the SMA (Fig. 2b). Similar
results were obtained by Campos [1] during the co-digestion
of manure with agriculture waste and other kinds of clays
from olive oil clarification. Given its low effect in thermo-
philic conditions, straw apparently does not represent an
available substrate to methanogenic consortia, due to its
low rate of hydrolysis [49,50].
The significance of each coefficient determined by Stu-
dent's t-test and p-values were illustrated in Table 4. The t-
value represents the ratio of the estimated parameter effect at
an estimated parameter standard deviation. In mesophilic
conditions the principal effect (C) the quadratic effect (A2) and
the first-order interaction effect (AB), (AC) and (BC) were the
most significant on SMA during these waste co-digestion
processes. These results agreed with ANOVA (Table 3). How-
ever, in thermophilic conditions, only the interaction co-
efficients (AB) and (AC) were significant; what shows the effect
of temperature on the SMA response, taking into consider-
ation the substrate concentration and its typology. The per-
centages of contribution (PC) for each individual factor from
the model were calculated and tabulated (Table 4). The best
contributions were in accordance with the best t-values and
its lowest probability values, similar to the result achieved by
Yetilmezsoy et al. [35].
3.2. Three-dimensional (3D) response surface andcontour plots
The three-dimensional response surface plots are a function
of two factors, maintaining all other factors at fixed levels,
presented as a solid surface framed in a three-dimensional
space. This graphic presentation helps to understand the
main and the interaction effects of those two factors [35,51].
At the same time, 3D response surfaces and their corre-
sponding contour plots can facilitate the straightforward ex-
amination of the effects of the experimental variables on the
responses [52]. In this work, since the regression model had
three independent variables, one variable was held constant
at the center level. Figs. 3 and 4 showed the 3D response
surfaces plot for SMA and the corresponding contour plots (in
Fig. 2 e Pareto chart showing the standardized effect of independent variables and their interaction on the SMA at both
temperature conditions. The length of each bar in the chart indicates the standardized effect of each factor.
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 790
a two-dimensional plane) as the functions of two variables at
mesophilic and thermophilic condition, respectively. The
nonlinear nature of all 3D response surfaces and corre-
sponding contour plots demonstrated that there were
considerable interactions between each of the independent
variables. In other words, there was no direct linear relation-
ship among the selected independent variables, supporting
the adequacy of the second-order model to predict SMA
values.
In both temperature conditions, the manure had a note-
worthy effect on SMA (Figs. 3a, 3b, 4a and 4b). At extreme
values of manure, SMA decreased, but increased at central
levels (17.5 gVSS L�1). Themanure levels close to 36.6 gVSS L�1
decreased drastically the SMA values because of the ammonia
nitrogen content in manure, as it was described above.
According to the results, there was no evidence that straw
affect the SMA values. In mesophilic conditions a synergic
effect of straw and clay was achieved (Figs. 3a and 4a). In
thermophilic conditions the straw had not a significant effect
on the SMA. The synergic effect of manure and straw at low
concentration showed very low impact on SMA, however with
the straw concentration increase, the SMA values enhanced.
However, in thermophilic conditions a decrease was observed
on the SMAwith a little augmentation of straw concentration.
Actually, even at thermophilic conditions, straw apparently
does not represent an available substrate to methanogenic
Table 4 e Coefficient and significance of the components for the quadratic model.
Factor Coefficient Effect SEa t-ratiob p-value SSd PCe
35 �C Intercept 1.2859
A 0.067 0.134 0.0611 2.1922 0.0709 0.0449 2.72
B 0.053 0.106 0.0611 1.7341 0.1336 0.0281 1.70
C 0.075 0.150 0.0611 2.4540 0.0495c 0.0563 3.40
A2 �0.4588 �0.9176 0.1190 �7.7078 0.0002c 0.5549 33.59
AB �0.1988 �0.3975 0.0683 �5.8165 0.0011c 0.3160 19.13
AC �0.1538 �0.3075 0.0683 �4.4995 0.0041c 0.1891 11.45
B2 �0.1088 �0.2176 0.1190 �1.8277 0.1174 0.0312 1.89
BC 0.2313 0.4625 0.0683 6.7676 0.0005c 0.4278 25.89
C2 �0.0388 �0.0776 0.1190 �0.6517 0.5387 0.0040 0.24
55 �C Intercept 1.3524
A �0.087 �0.174 0.107000 �1.6255 0.1552 0.0757 4.53
B �0.066 �0.132 0.107046 �1.2331 0.2636 0.0436 2.61
C �0.008 �0.016 0.107046 �0.1495 0.8861 0.0006 0.04
A2 �0.5436 �1.0872 0.208481 �5.2151 0.0020c 0.7791 46.63
AB 0.0588 0.1175 0.119681 0.9818 0.3641 0.0276 1.65
AC 0.2963 0.5925 0.119681 4.9507 0.0026c 0.7021 42.02
B2 �0.0986 �0.1972 0.208481 �0.9461 0.3806 0.0256 1.53
BC 0.0038 0.0075 0.119681 0.0627 0.9521 0.0001 0.01
C2 �0.0786 �0.1572 0.208481 �0.7542 0.4793 0.0163 0.98
a Standard error.b t-ratio ¼ Effect
SE .c p-value <0.05 were considered to be significant.d Sum of squares.e Percentage contribution n ¼ SSP
SS*100.
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 7 91
consortia, as described above. However the great availability
of rice straw in the area studied could justify this study, which
showed again the necessity of pretreating rice straw to
enhance its methane potential.
A synergic effect between manure and clay was observed
(Figs. 3b and 4b). At mesophilic conditions, an increase of SMA
values were achieved with an increase of clay concentration,
maintaining low levels of manure, but at high levels of
manure the clay concentration decreased the inhibitory effect
due to the manure nitrogen content. This result could be
explained because, as it has been reported [12,15,19], the
ammonia could be absorbed by clay using similar compounds,
such as bentonite, zeolite, mordenite and others. In contrast,
at thermophilic condition the SMA values decreased with the
increase of the clay concentration. This suggests that tem-
perature affects the assimilation of thismineral (clay) and also
the ammonia nitrogen adsorbent properties of clays.
A similar response was achieved for SMA values when
straweclay interaction was analyzed maintaining fixed
manure level (Figs. 3c and 4c). At mesophilic condition SMA
values over 1.5 gCH4-COD gVSS�1 d�1 were obtained at high
levels of straw and low levels of clay, or low levels of straw and
high levels of clay. At thermophilic condition there was
almost not variation on SMA values, but those were lower
than the SMA values achieved at mesophilic condition.
In all cases, evidences of significant interaction among the
dependent variables and nonlinear relationship in the
response were obtained.
3.3. Optimal waste combination for maximizing SMA
The second-order polynomial model applied in this study was
used for determining the optimum condition to maximize
SMA and it was adjusted by minimal quadratic methods. The
effect on the dependent variable SMA at different mixture of
independents variables levels (manure, straw and clay) were
carried out to identify the waste concentration that produced
the maximum SMA value.
Based on the calculation steps defined for the optimization
algorithm, in mesophilic condition the optimal response was
close to the SMA value predicted. Fig. 5a shows the non-linear
behavior of the predicted response at the manure levels
evaluated, whose maximumwas at 18.73 gVSS L�1 of manure,
but the graphics of straw and clay showed the maximum
response at 17.6 gVSS L�1 and 8.3 gVSS L�1 respectively. Thus,
the optimal combination was determined for SMA of
1.31 gCH4-COD gVSS�1 d�1 using 28.35 gVSS L�1 of manure,
17.6 gVSS L�1 of straw and 8.3 gVSS L�1 of clay, with 0.05 of
confidence level in the range from 1.12 to 1.52 gCH4-
COD gVSS�1 d�1. In thermophilic conditions, (Fig 5b) the
optimal combination was obtained with 20.1 gVSS L�1 of
manure, 10.18 gVSS L�1 of straw and 3.05 gVSS L�1 of clay for a
maximized SMA value of 1.38 gCH4-COD gVSS�1 d�1 in a range
from 1.19 to 1.57 gCH4-COD gVSS�1 d�1 with 0.05 of confidence
level. These SMA results obtained at both temperature con-
ditions for the experimental set up assayedmay be considered
as representative for this kind of solid wastes [53e55].
3.4. Model validation
Finally, in order to validate the adequacy of the models pro-
posed to predict SMA for both temperature conditions during
the co-digestion of the wastes evaluated, new experiments
were carried out using the optimal waste concentrations
achieved before. The SMA values achieved were
1.23 ± 0.41 gCH4-COD gVSS�1 d�1 (middle value ± standard
Fig. 3 e Three-dimensional (3D) response surface and the corresponding contour plots showing the effects of interactions on
SMA under mesophilic conditions (35 �C) for two factors, maintaining the third variable fixed at their respective center level.
(a) Manure (A) and straw (B); (b) manure (A) and clay (C); (c) straw (B) and clay (C).
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 792
deviation) at mesophilic condition and 1.35 ± 0.26 gCH4-
COD gVSS�1 d�1 at thermophilic condition.
The Student t-test was carried out to evaluate the relation-
ship between the SMA observed at this optimal waste concen-
tration condition and the predicted SMA values by optimization
studiesdescribedabove.The tabulatedt-valuewereobtainedfor
some freedomdegree values (df¼ sample size (1)þ sample size
(2) e 2) and using 0.05 of confidence level. At mesophilic condi-
tions the t-value calculated (tcal ¼ 0.3038) was lower than the t-
value tabulated (ttab(a; df) ¼ ttab(0.05; 4) ¼ 2.1318) and the p-
Fig. 4 e Three-dimensional (3D) response surface and the corresponding contour plots showing the effects of interactions on
SMA under thermophilic conditions (55 �C) for two factors, maintaining the third variable fixed at their respective center
level. (a) Manure (A) and straw (B); (b) manure (A) and clay (C); (c) straw (B) and clay (C).
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 7 93
valuecal¼ 0.7765 (p-value >0.05). Therefore, this could not reject
the null hypothesis hence there was no significant difference
between the SMA observed and the optimal SMA values pre-
dicted by themodel. Similarly, at thermophilic conditions the t-
value calculated (tcal ¼ 0.1655) was lower than the t-value tabu-
lated, and thep-valuecal¼ 0.8766, hence therewasnosignificant
difference between the SMA observed and the optimal SMA
values predicted by themodel.
Fig. 5 e Level combination effect of independent variables (A: manure; B: Straw; C: clay) on the SMA and optimization. (a)
Mesophilic; (b) Thermophilic.
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 794
3.5. Cultivation independent analysis of the microbialcommunity structure
The diversity obtained in the initial and final samples (those
after 21 days of manure, straw and clay co-digestion) shows
that there is a tendency to the dominance of certain species
with co-digestion. According to Jaccard index (J0), at meso-
philic condition, 48.4% species were similar in both initial and
final samples, in contrast, at thermophilic condition, only
18.8% of similarity between the initial and final sample was
achieved. These results demonstrated the temperature effect
on the uptake of these co-substrates. It is noteworthy that
there is a 35.5% similarity between the initial samples of
mesophilic and thermophilic microbiota, which can be ex-
pected, taking into account that both inoculums were taken
from the same ecosystem, and were adapted to each tem-
perature condition. However, the community structure
developed at mesophilic condition, with the addition of co-
substrates (final sample), was significantly different from
those developed in thermophilic condition (J0 ¼ 13.5%).
The greatest variety of species of bacteria was obtained in
the final sample reflecting the anaerobic co-digestion of the
three substrates at mesophilic condition, although the Shan-
non index indicated that diversity was higher in the initial
samples, at both temperatures (Table 5). The community
organization expressed as the Evenness Index was similar, in
both initial and final samples; however, significant differences
were noted between each temperature.
The relative abundance of the terminal restriction frag-
ment of the initial samples in both temperature conditions
showed the predominance of the 66, 98 and 231 bp fragments
(Fig. 6). In the final samples that reflect the co-digestion of the
three substrates, the dominance of certain groups was noted
(more pronounced in mesophilic condition). Regarding this,
greater relative abundance of fragments of 89, 97 and 162 bp
in mesophilic condition was found, while in thermophilic
condition an increased in abundance of the 144, 151 and
231 bp fragments was observed. These fragments of 151 bp
belonged to Clostridium species [33].
Bacteroidetes phylum, held a 40% relative abundance (89
and 97 bp fragments) showing that this phylum favors the co-
digestion. It was also found in abundance in reactors that
degrade cow manure and in co-digestion of grass silage in
mesophilic conditions [56]. However, as far as the author
knows, prevalence of this bacteria phylum has not been re-
ported at swine manure anaerobic digestion.
As for the archaeal community, in initial sample of the
mesophilic temperature, predominant fragments of 108 bp,
with a 59.21% relative abundance, fragments of 343 bp (12.5%
relative abundance) and other fragments of 57 and 203 bp
Table 5 e Diversity and community organization in bacteria growing with the co-digestion of pig manure substrates, ricestraw and residual clays. Initial sample: time zero in the co-digestion at each temperature condition; Final sample:anaerobic co-digestion of co-substrates after 21 days of incubation at each temperature conditions.
Samples Temperature (�C) Richness Shannon Index Evenness Variance
Initial 35 22 2.46 0.80 0.10
Final 35 24 2.29 0.72 0.09
Initial 55 20 2.70 0.90 0.11
Final 55 18 2.51 0.87 0.10
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 7 95
were also highlighted (Fig. 6). In the final sample, reflecting
anaerobic co-digestion of the three substrates in mesophilic
condition, the dominance of the 108 bp fragmentwas reduced,
and other species (fragments of 431 and 470 bp) were favored,
where fragments of 470 bp was predominant.
In the initial sample of the thermophilic temperature,
predominantly fragments 343 and 629 bp were detected,
although fragments of 101, 203 and 257 bp were also detected.
In the final sample reflecting anaerobic co-digestion of the
three substrates, the community structure was similar to that
obtained in the initial stage. However, in this case 629 bp
fragments predominated, while the prevalence of 343 bp
fragments decreased. In this case, fragments of 82, 108, 203,
257, 339 bpwere also detected. It is noteworthy that fragments
of 108, 203 and 343 bp were detected, both in the initial sam-
ples and the final samples of both temperature conditions.
About archaea community, Rademacher et al. [33], also
obtained the predominant fragments of 108, 339 and 629 bp
within a two-phase leach bed biogas reactor supplied with rye
silage and straw in thermophilic conditions. Sequencing as-
says revealed that species belonged to the genera
Methanosaetas, Methanotermobacter and Methanosarcinas,
respectively. With the co-digestion at mesophilic temperature
predominated fragments 108 and 470 bp, corresponding to the
families Methanosarcinaceae and Methanosaetaceae respec-
tively, while at thermophilic temperatures proliferate Meth-
anothermobacter species and Methanobacterium, the order
Methanobacteriales (339 bp) and species Methanosarcinas
(629 bp), similar to that reported previously in the degradation
Fig. 6 e Relative abundance of terminal restriction fragments (FT
initial and final samples of the co-digestion of pig manure, rice
thermophilic conditions (55 �C).
of pig manure [57], because they are part of the pig gut flora
and thus could appear in their feces.
Fragments of 343 bp (Methanobacteriales order) were detec-
ted in all cases, indicating its stability against environmental
changes [32]. However, when the co-substrates (final sample
in both temperature conditions) were added, the abundance
of these fragments of 343 bp decreased, in contrast to the re-
sults of other authorswho demonstrated the predominance of
Methanobacteriales orders and Methanomicrobiales in batch re-
actors trying pig residual [30], even in co-digestion with
chickenmanure [58]. The 108 bp fragmentswere predominant
especially at mesophilic condition. These species belong to
the family Methanosaetaceae according Padmasiri et al. [32],
and Rademacher et al. [33],. Fragments of 470 bp also pre-
dominated in mesophilic conditions, species belonging to the
order Methanosarcinales but Methanosarcinaceae family as ob-
tained by Padmasiri et al. [32].
Certain abundance of fragments 431 bp in mesophilic
conditions, with the addition of the three substrates was also
obtained. Kim et al. [30], and Padmasiri et al. [32], obtained
fragments of this size and identified them within the Meth-
anomicrobiales order. Furthermore, the presence of fragments
of 203 bp, the phylogenetic affiliation was detected, what has
not been reported up to date.
Finally, in mesophilic methanogenic conditions aceto-
trophics methanogens of Methanosarcinales order prevailed
and at thermophilic conditions hydrogenotrophic metha-
nogens of Methanomicrobiales and Methanobacteriales orders
were predominant. During the co-digestion of the three
) and their sizes in bp of bacteria (a) and archaea (b) in the
straw and residual clays, at mesophilic (35 �C) and
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 796
substrates, the effect of temperature on the diversity and
community structure was demonstrated [59].
4. Conclusions
The results showed that there was a significant interaction
among the substrates evaluated and that the co-digestion of
the three substrates enhanced SMA. At mesophilic conditions
the optimal combinationwas determined for SMAof 1.31 gCH4-
COD gVSS�1 d�1 using 28.35 gVSS L�1 of manure, 17.6 gVSS L�1
of strawand 8.3 gVSS L�1 of clay. In thermophilic conditions the
optimal combination was 20.1 gVSS L�1 of manure,
10.18 gVSS L�1 of straw and 3.05 gVSS L�1 of clay for a maxi-
mizedSMAof 1.38 gCH4-CODgVSS�1 d�1. At central levels of pig
manure (22 gVSS L�1), SMA was enhanced by the co-digestion
with rice straw and clay residues. At higher levels
(36 gVSS L�1), clay residues reduced the inhibitory effect of pig
manure on SMA. In general, the presence of rice straw had a
lower influence on SMA than clay residues, due to the beneficial
adsorption properties of the latter. Thus, it was corroborated
the positive effect of clay as inorganic additive for stimulating
the anaerobic process as a result of its mineral content and its
adsorbent N-rich sludge properties. The T-RFLP analysis results
showed the effect of the anaerobic co-digestion of manure and
straw with clay addition on microbial community diversity at
different temperature conditions. The assays suggest the
prevalence of acetotrophic methanogens in mesophilic condi-
tion, belonging to the order Methanosarcinales (genera Meth-
anosarcina and Methanosaeta) whereas at thermophilic condition
it was showed the prevalence of Methanomicrobiales and Meth-
anobacteriales order. The statistical methods applied in this
study proved to be useful for defining optimal combination of
wastes considering their anaerobic co-digestion. Specific
methanogenic activity was also a good response variable for
that purpose. These results open the door to a new investiga-
tion about the use of these regression models to adjust a
phenomenological model in anaerobic digestion systems.
Acknowledgments
The first author would like to thank the Institute of Engi-
neering, UNAM, and the Leibniz Institute for Agricultural En-
gineering Potsdam-Bornim (ATB), Germany, for research
grants to acomplisch this investigation.
r e f e r e n c e s
[1] Campos PAE. Optimizaci�on de la digesti�on anaerobia depurines de cerdo mediante co-digesti�on con residuosorg�anicos de la industria agroalimentaria [PhD thesis].L�erida, Espa~na: Universitat de Lleida; 2001.
[2] Danish-Energy-Agency. Overview report on biogas plants inDenmark. Denmark: Danish Energy Agency Copenhagen;1995.
[3] Ward AJ, Hobbs PJ, Holliman PJ, Jones DL. Optimization of theanaerobic digestion of agricultural resources. BioresourTechnol 2008;99:7928e40.
[4] Angelidaki I, Ellegaard L. Co-digestion of manure and organicwastes in centralized biogas plants - status and futuretrends. Appl Biochem Biotechnol 2003;109:95e105.
[5] Bolzonella D, Battistoni P, Susini C, Cecchi F. Anaerobiccodigestion of waste activated sludge and OFMSW: theexperiences of Viareggio and Treviso plants (Italy). Water SciTechnol 2006;53:203e11.
[6] Gomez X, Cuetos MJ, Cara J, Moran A, Garcıa AI. Anaerobicco-digestion of primary sludge and the fruit and vegetablefraction of the municipal solid wastese conditions formixing and evaluation of the organic loading rate. RenewEnergy 2006;31:2017e24.
[7] Romano RT, Zhang RH. Co-digestion of onion juice andwastewater sludge using an anaerobic mixed biofilm reactor.Bioresour Technol 2008;99:631e7.
[8] Mata-Alvarez J, Mace S, Llabres P. Anaerobic digestion oforganic solid wastes. An overview of research achievementsand perspectives. Bioresour Technol 2000;74:3e16.
[9] Alatriste-Mondragon F, Samar P, Cox HHJ, Ahring BK,Iranpour R. Anaerobic co-digestion of municipal, farm, andindustrial organic wastes: a survey of recent literature. WaterEnviron Res 2006;78:607e36.
[10] Chambers BJ, Smith KA, Pain BF. Strategies to encouragebetter use of nitrogen in animal manures. Soil Use Manag2000;16:157e61.
[11] Moller HB, Sommer SG, Ahring BK. Methane productivity ofmanure, straw and solid fractions of manure. BiomassBioenergy 2004;26:485e95.
[12] Hansen KH, Angelidaki I, Ahring BK. Anaerobic digestion ofswinemanure: inhibitionbyammonia.WaterRes1998;32:5e12.
[13] Sung SW, Liu T. Ammonia inhibition on thermophilicanaerobic digestion. Chemosphere 2003;53:43e52.
[14] Wang G, Gavala HN, Skiadas IV, Ahring BK. Wet explosion ofwheat straw and co-digestion with swine manure: effect onthe methane productivity. Waste Manag 2009;29:2830e5.
[15] Mil�an Z, Villa P, S�anchez-Hern�andez EP, Montalvo SJ, Borja R,Ilangovan K, et al. Effect of natural and modified zeolite onanaerobic digestion of piggery waste. Water Sci Technol2003;48:263e9.
[16] Angelidaki I, Ahring BK. Thermophilic anaerobic digestion oflivestock waste: the effect of ammonia. Appl MicrobiolBiotechnol 1993;38:560e4.
[17] Mil�an Z, S�anchez E, Weiland P, Borja R, Martın A,Ilangovan K. Influence of different natural zeoliteconcentration on anaerobic digestion of piggery waste.Bioresour Technol 2001;80:37e43.
[18] Mil�an Z, Montalvo S, Ilangovan K, Monroy O, Chamy R.Influence of natural zeolite addition on methanogenesis ofnitrogen wastes. In: IWA, editor. VII Latin AmericanWorkshop and Symposium on anaerobic digestion. Punta delEste, Uruguay 2005.
[19] Tada C, Yingnan Y, Toshiaki H, Akinari S, Kenta O, Shigeki S.Effect of natural zeolite on methane production foranaerobic digestion of ammonium rich organic sludge.Bioresour Technol 2005;96:459e64.
[20] Feng XM, Karlsson A, Svensson BH, Bertilsson S. Impact oftrace metal addition on biogas production from foodindustrial waste-linking process to microbial communities.FEMS Microbiol Ecol 2010;74:226e40.
[21] Kumar A, Miglani P, Gupta RK, Bhattacharya TK. Impact of Ni(II), Zn (II) and Co (II) on biogasification of potato waste. JEnviron Biol 2006;27:61e6.
[22] Espinosa A, Rosas L, Ilangovan K, Noyola A. Effect of tracemetals on the anaerobic degradation of volatile fatty acids inmolasses stillage. Water Sci Technol 1995;32:121e9.
[23] Mil�an Z, Montalvo SJ, Ruiz-Tagle N, Urrutia H, Chamy R.Identification of anaerobic microbial groups influenced byheavy metal supplementation to batch digesters. In: IWA,
b i om a s s a n d b i o e n e r g y 7 1 ( 2 0 1 4 ) 8 4e9 7 97
editor. VII Latin American Workshop and Symposium onanaerobic digestion. Punta del Este, Uruguay 2005.
[24] LoHM, LinKC, LiuMH, Pai TZ, LinCY, LiuWF, et al. Solubility ofheavy metals added to MSW. J Hazard Mater 2009;161:294e9.
[25] S�anchez E, Borja R, Weiland P, Travieso L, Martın A. Effect ofsubstrate concentration and temperature on the anaerobicdigestion of piggery waste in a tropical climate. ProcessBiochem 2001;37:483e9.
[26] S�anchez E, Borja R, Travieso L, Martın A, Colmenarejo MF.Effect of influent substrate concentration and hydraulicretention time on the performance of down- flow anaerobicfixed bed reactors treating piggery wastewater in a tropicalclimate. Process Biochem 2005;40:817e29.
[27] Wang X, Dong-jie N, Xiao-shuang Y, You-cai Z. Optimizationof methane fermentation from effluent of bio-hydrogenfermentation process using response surface methodology.Bioresour Technol 2008;99:4292e9.
[28] Montgomery DC. Design and analysis of experiments. NewYork: John Wiley & Sons (Eds); 2005.
[29] Nettmann E, Bergmann I, Pramschufer S, Mundt K,Plogsties V, Herrmann C, et al. Polyphasic analyses ofmethanogenic archaeal communities in agricultural biogasplants. Appl Environ Microb 2010;76(8):2540e8.
[30] Kim W, Lee S, Shin SG, Lee C, Hwang K, Hwang S.Methanogenic community shift in anaerobic batch digesterstreating swine wastewater. Water Res 2010;44(17):4900e7.
[31] Schutte UME, Abdo Z, Bent SJ, Shyu C, Williams CJ,Pierson JD, et al. Advances in the use of terminal restrictionfragment length polymorphism (T-RFLP) analysis of 16SrRNA genes to characterize microbial communities. ApplEnviron Microb 2008;80:365e80.
[32] Padmasiri SI, Zhang JZ, Fitch M, Norddahl B, Morgenroth E,Raskin L. Methanogenic population dynamics andperformance of an anaerobic membrane bioreactor (AnMBR)treating swine manure under high shear conditions. WaterRes 2007;41(1):134e44.
[33] Rademacher A, Nolte C, Sch€onberg M, Klocke M.Temperature increases from 55 to 75 �C in a two-phasebiogas reactor result in fundamental alterations within thebacterial and archaeal community structure. Appl MicrobiolBiotech 2012;96:565e76.
[34] APHA-AWWA-WPCF. Standard methods for examination ofwater and wastewater American Public Health Association.19 ed. 1995. Washington DC, USA.
[35] Yetilmezsoy K, Demirel S, Vanderbei RJ. Response surfacemodeling of Pb(II) removal from aqueous solution by Pistaciavera L.: BoxeBehnken experimental design. J Hazard Mater2009;171:551e62.
[36] Lane DJ. 16S/23S rRNA sequencing. In: Stackebrandt EGM,editor. Nucleic acid techniques in bacterial systematics. NewYork: Wiley; 1991. p. 115e47.
[37] Sipos R, Szekely AJ, Palatinszky M, Revesz S, Marialigeti K,Nikolausz M. Effect of primer mismatch, annealingtemperature and PCR cycle number on 16S rRNA gene-targeting bacterial community analysis. FEMS Microbiol Ecol2007;60:341e50.
[38] Weisburg WG, Barns SM, Pelletier DA, Lane DJ. 16S ribosomalDNA amplification for phylogenetic study. J Bacteriol1991;173:697e703.
[39] Despr�es VR, Nowoisky JF, Klose M, Conrad R, Andreae MO,P€oschl U. Characterization of primary biogenic aerosolparticles in urban, rural, and high-alpine air by DNAsequence and restriction fragment analysis of ribosomalRNA genes. Biogeosciences 2007;4:1127e41.
[40] Großkopf R, Janssen PH, LiesackW. Diversity and structure ofthe methanogenic community in anoxic rice paddy soilmicrocosms as examined by cultivation and direct 16S rRNAgenesequenceretrieval.ApplEnvironMicrobiol1998;64:960e9.
[41] Lueders T, Friedrich M. Archaeal population dynamicsduring sequential reduction processes in rice field soil. ApplEnviron Microbiol 2000;66:2732e42.
[42] Culman SW, Bukowski R, Gauch HG, Cadillo-Quiroz H,Buckley DH. T-REX: software for the processing and analysisof T-RFLP data. BMC Bioinforma 2009;10:171.
[43] Abdo Z, Schuette UME, Bent SJ, Williams CJ, Forney LJ,Joyce P. Statistical methods for characterizing diversity ofmicrobial communities by analysis of terminal restrictionfragment length polymorphisms of 16S rRNA genes. EnvironMicrobiol 2006;8:929e38.
[44] Smith CJ, Danilowicz BS, Clear AKF, Costello J, Wilson B,Meijer WG. T-Align, a web-based tool for comparison ofmultiple terminal restriction fragment length polymorphismprofiles. FEMS Microbiol Ecol 2005;54:375e80.
[45] Sen R, Swaminathan T. Response surface modeling andoptimization to elucidate and analyze the effects ofinoculum age and size on surfactin production. Biochem EngJ 2004;21:141e8.
[46] Liu HL, Lan YW, Cheng YC. Optimal production of sulphuricacid by Thiobacillus thiooxidans using response surfacemethodology. Process Biochem 2004;39:1953e61.
[47] Kim M, Ahn YH, Speece RE. Comparative process stabilityand efficiency of anaerobic digestion; mesophilic vs.thermophilic. Water Res 2002;36:4369e85.
[48] Oakland J. Statistical process control. 5th ed. ElsevierScience; 2003.
[49] Mussatto SI, Fernandez M, Milagres AMF, Roberto IC. Effect ofhemicellulose and lignin on enzymatic hydrolysis ofcellulose from brewer's spent grain. Enzyme MicrobiolTechnol 2008;43:124e9.
[50] Zhang R, Zhang Z. Biogasification of rice straw with ananaerobic-phased solids digester system. Bioresour Technol1999;68:240e5.
[51] Adinarayana K, Ellaiah P. Response surface optimization ofthe critical medium components for the production ofalkaline protease by a newly isolated Bacillus sp. J PharmPharm Sci 2002;5:272e8.
[52] Wu D, Zhou J, Li Y. Effect of the sulfidation process on themechanical properties of a CoMoP/Al2O3 hydrotreatingcatalyst. Chem Eng Sci 2009;64:198e206.
[53] Soto M, M�endez R, Lema JM. Methanogenic and non-methanogenic activity tests. Theoretical basis andexperimental set up. Water Res 1993;27:1361e76.
[54] Riffat R, Krongthamchat K. Specific methanogenic activity ofhalophilic and mixed cultures in saline wastewater. Int JEnviron Sci Technol 2006;2:291e9.
[55] Laubie B, Buffi�ere P, Benbelkacem H, Bayard R. Anaerobicdigestion in dry conditions: Influence of the moisturecontent on the specific methanogenic activity. In: IWA,editor. 12th World Congress on anaerobic digestion.Guadalajara, Jalisco, Mexico 2010.
[56] Wang H, Tolvanen K, Lehtomaki A, Puhakka J, Rintala J.Microbial community structure in anaerobic co-digestionof grass silage and cow manure in a laboratorycontinuously stirred tank reactor. Biodegradation2010;21(1):135e46.
[57] Boopathy R. Isolation and characterization of amethanogenic bacterium from swine manure. BioresourTechnol 1996;55(3):231e5.
[58] Xia Y, Masse DI, McAllister TA, Kong Y, Seviour R, Beaulieu C.Identity and diversity of archaeal communities duringanaerobic co-digestion of chicken feathers and other animalwastes. Bioresour Technol 2012;110:111e9.
[59] Van Lier JB, Hulsbeek J, Stams AJM, Lettinga G. Temperaturesusceptibility of thermophilic methanogenic sludge e
implications for reactor start-up and operation. BioresourTechnol 1993;43(3):227e35.