Targeting Bacteria and Methanogens To Understand the Role of Residual Slurry as an Inoculant in Stored Liquid Dairy Manure

ABSTRACT

Microbial communities in residual slurry left after removal of stored liquid dairy manure have been presumed to increase methane emission during new storage, but these microbes have not been studied. While actual manure storage tanks are filled gradually, pilot- and farm-scale studies on methane emissions from such systems often use a batch approach. In this study, six pilot-scale outdoor storage tanks with (10% and 20%) and without residual slurry were filled (gradually or in batch) with fresh dairy manure, and methane and methanogenic and bacterial communities were studied during 120 days of storage. Regardless of filling type, increased residual slurry levels resulted in higher abundance of methanogens and bacteria after 65 days of storage. However, stronger correlation between methanogen abundance and methane flux was observed in gradually filled tanks. Despite some variations in the diversity of methanogens or bacteria with the presence of residual slurry, core phylotypes were not impacted. In all samples, the phylum Firmicutes predominated (∼57 to 70%) bacteria: >90% were members of Clostridia. Methanocorpusculum dominated (∼57 to 88%) archaeal phylotypes, while Methanosarcina gradually increased with storage time. During peak flux of methane, Methanosarcina was the major player in methane production. The results suggest that increased levels of residual slurry have little impact on the dominant methanogenic or bacterial phylotypes, but large population sizes of these organisms may result in increased methane flux during the initial phases of storage.

IMPORTANCE Methane is the major greenhouse gas emitted from stored liquid dairy manure. Residual slurry left after removal of stored manure from tanks has been implicated in increasing methane emissions in new storages, and well-adapted microbial communities in it are the drivers of the increase. Linking methane flux to the abundance, diversity, and activity of microbial communities in stored slurries with different levels of residual slurry can help to improve the mitigation strategy. Mesoscale and lab-scale studies conducted so far on methane flux from manure storage systems used batch-filled tanks, while the actual condition in many farms involves gradual filling. Hence, this study provides important information toward determining levels of residual slurry that result in significant reduction of well-adapted microbial communities prior to storage, thereby reducing methane emissions from manure storage tanks filled under farm conditions.

INTRODUCTION

Dairy farming is responsible for large amounts of non-CO₂ greenhouse gas emissions, mainly from enteric fermentation and manure management. In large dairy farms, manure is usually stored as slurry for extended periods (i.e., months to 1 year) (1), creating a medium that supports the growth of microbial communities involved in anaerobic degradation and CH₄ production (2). During pumping out of slurry tanks for land application, complete removal is difficult in practice, and residual slurries (RS) of various amounts are usually left in storage. In view of the potential CH₄ emission mitigation strategies from these systems, the inoculum effects of RS in new storages have become the focus of several studies (3–8). Different levels of RS left in storage tanks, especially under actual farm conditions where manure is transferred to tanks gradually throughout the storage period, result in varied ratios of RS to fresh manure throughout the storage period. Thus, understanding the effects of these ratios on the dynamics of microbial communities that derive CH₄ emission and linking them with CH₄ flux during storage are critical to developing effective mitigation strategies. For instance, accelerated starts of CH₄ production from liquid dairy manure have been observed after supplementation with 7.6% inoculum material (manure that has been stored for 45 to 60 days at 20°C) (7). About 56 and 97% reductions of CH₄ emission have also been demonstrated by complete removal (4) and reduction of RS levels from ∼26% to 13% (5), respectively. Using pilot-scale outdoor storage tanks, ∼56% reduction of cumulative CH₄ emission by complete removal of old manure compared to partial emptying was demonstrated (8). Using the same tanks, ∼26% lower cumulative CH₄ emission was also reported after reduction of old residual manure from 15% to 5% (6). Interestingly, these decreases in CH₄ emissions were presumed to be explained by reductions of well-adapted consortia of microorganisms in the RS that might readily start functioning in new storages, although no study so far has investigated these microbial communities.

With long-term storage of dairy manure, bacterial and methanogenic communities of different sources (e.g., rumen, soil, water, and bedding material) may be enriched and acclimatized to storage conditions (9, 10). All microbial communities functioning in rumen environments may not be able to establish in manure storage systems, as these environments can differ with a number of parameters, such as temperature and substrate composition. Thus, after long-term storage of manure, a consortium of microbial communities that are well adapted to existing storage conditions may readily break down manure organic polymers (polysaccharides, proteins, and lipids), an important step toward CH₄ production. To convert complex polymers into CH₄, four physiological groups of microorganisms are important: hydrolytic, acidogenic, acetogenic, and methanogenic microbial communities (11, 12). First, hydrolytic bacteria degrade residues of organic polymers into simple compounds such as glucose, fatty acids, amino acids, alcohols, H₂, and CO₂. As the efficiency of the hydrolysis step determines the rate of anaerobic degradation (13) particularly when the ratio of hydrolytic bacteria to methanogenic archaea is high (14), the relative composition of these bacteria in RS may be critical. By-products of hydrolysis are substrates for acidogenic bacteria that further convert them into different organic acids (e.g., acetic, lactic, propionic, formic, and butyric acids), ethanol, H₂, and CO₂. Acetogenic bacteria then convert higher organic acids, such as butyric and propionic acids, into acetic acid, H₂, and CO₂, which are important substrates for methanogenesis (12). As long-term storage of dairy slurry may enrich these hydrolytic, acidogenic, acetogenic, and methanogenic microorganisms, investigating these communities in RS and their effects on CH₄ emission in stored manure is an important step toward the refinement and/or development of efficient mitigation strategies.

It is critical to assess the effects of RS on CH₄ emission and structure of slurry microbial communities using dairy manure that is stored and managed under conditions similar to farm conditions. Slurry storage systems of dairy farms are usually filled gradually (i.e., daily to weekly loadings) throughout the storage period by transferring barn collections or slurries from temporary storages (1). Compared to batch-filled tanks that have often been used to study CH₄ flux from manure storage systems (6, 8), repeated loading in gradually filled tanks may affect the stability of slurry environments and microbial communities. For instance, with repeated loadings, crusts that have been presumed to be the potential locations where aerobic oxidation of CH₄ may occur (15–18) could be impacted. Although available substrates for microbial communities in batch-filled tanks can be high in the initial phases of storage, gradual filling continually supplies fresh substrates. Thus, the impact of frequency of filling on CH₄ flux and microbial communities needs to be assessed.

Advanced molecular methods such as quantitative real-time PCR (qPCR) and massively parallel sequencing of fragments of universal marker genes, such as 16S rRNA, are revealing the structures of microbial communities from complex environments (19, 20). These techniques have also been successfully used to study functional groups of microorganisms, such as methanogens, by targeting a gene or transcript related to their function (21–23). Three different physiological groups of methanogens are known (hydrogenotrophic, aceticlastic, and methylotrophic) and have the gene encoding the alpha subunit of methyl coenzyme A reductase (mcrA), which catalyzes the last step of methanogenesis (24). Since 16S rRNA gene- and mcrA gene-based phylogenies of methanogens are congruent (25), targeting the mcrA genes or transcripts may provide both phylogenetic and functional information about the methanogen community.

The objective of this study was to assess the effects of different RS levels (0%, 10%, and 20%) on CH₄ emissions and structures of methanogenic and bacterial communities in stored dairy slurries filled gradually and in batch.

RESULTS

Methane flux and manure characteristics.

The absence of RS (0%) had an impact only in gradually filled tanks, where peak flux occurred about 3 weeks later than in the tanks with 10% or 20% RS (Fig. 1). In the absence of RS, the maximum mean daily flux reached was ∼43 to 47% lower than in the tanks with 10% and 20% RS. This resulted in up to 85% reductions of the cumulative emissions compared to those of the tanks with RS. There was little variation in cumulative fluxes (∼11.25 ± 0.53 kg CH₄ m⁻²) from tanks with 10% and 20% RS. In batch-filled tanks, no significant differences in CH₄ flux or cumulative emissions (∼9.09 ± 0.23 kg CH₄ m⁻² of slurry) due to the presence or absence of RS were detected. On the other hand, manure characteristics (e.g., dry matter [DM] and volatile solid [VS] contents, volatile fatty acids [VFAs], and pH) varied with storage time. Dry matter and VS contents of fresh dairy slurries were higher (∼8.7-and 8.4-fold, respectively) than in RS; thus, different levels of RS combined with heterogeneity of the farm manure resulted in varied contents of these parameters in stored slurries on day 1 of storage (Table 1). Regardless of RS levels and filling type, these initial contents were reduced with storage time. On day 65, relatively higher levels of VFAs and hence lower pHs were observed in slurries without RS than in slurries that contained 10% or 20% RS (Tables 1 and 2). Nevertheless, the pHs of slurries from most treatments decreased initially (until day 18) and then increased gradually after, reaching 7.8 ± 0.2 on day 120. Most of the VFAs detected during peak flux of CH₄ (day 65) were not detected in day 120 samples, which corresponded to up to a 5.3-fold reduction of CH₄ flux.

FIG 1.

Mean daily methane (CH₄) flux from liquid dairy manure storage tanks containing 0%, 10%, and 20% residual slurry (RS), for gradually filled or batch-filled tanks.

TABLE 1.

Dry matter contents, volatile solid contents, and pHs of stored dairy slurries with 0%, 10%, and 20% residual slurry and filled gradually or in batch^a

Sample	Dry matter (%)					Volatile solids (%)					pH
	Day 1	Day 18	Day 65	Day 87	Day 120	Day 1	Day 18	Day 65	Day 87	Day 120	Day 1	Day 18	Day 65	Day 87	Day 120
FM	20.2					10.4					7.0
RS	2.3					1.2					7.6
0%, GF	19.6	13.8	11.4	7.8	6.3	8.2	6.3	4.9	3.2	2.7	6.9	6.7	7.6	7.5	8.2
10%, GF	9.5	13.2	10.6	7.5	5.7	5.3	6.2	4.3	2.8	2.3	6.9	6.8	8	7.8	7.8
20%, GF	5.8	8.92	10.2	8.0	5.4	3.6	4.8	4.2	3.0	2.0	7.0	7.1	7.8	7.5	7.6
0%, BF	14.1	13	7.1	6.9	6.2	6.7	5.6	3.5	3.0	2.6	6.9	6.7	7.6	7.7	7.7
10%, BF	10.7	17.3	7.0	6.2	6.1	5.7	7.0	3.6	2.6	2.5	7.1	6.7	8	7.7	7.7
20%, BF	11.8	12.6	6.0	7.8	7.9	6.4	6.2	3.0	3.4	3.6	7.1	6.9	7.9	7.7	7.8

^aFM, fresh manure; RS, residual slurry; GF, gradually filled; BF, batch filled.

TABLE 2.

Volatile fatty acid contents of stored dairy slurries with 0%, 10%, and 20% residual slurry and filled gradually or in batch

Sample	Volatile fatty acids (g liter−1 of dairy slurry), mean ± SDa
	Formic	Acetic	Propionic	Isobutyric	Butyric	Isovaleric	Valeric
Day 65
0%, GF	0.02 ± 0.01	2.66 ± 1.07	5.32 ± 1.59	4.21 ± 1.08	6.77 ± 2.50	5.36 ± 1.39	1.93 ± 0.62
10%, GF	0.02 ± 0.00	0.26 ± 0.00	3.01 ± 1.26	0.86 ± 0.12	0.01 ± 0.00	3.20 ± 1.57	0.13 ± 0.03
20%, GF	0.03 ± 0.01	0.47 ± 0.12	3.14 ± 0.86	0.44 ± 0.18	0.02 ± 0.00	0.74 ± 0.50	0.14 ± 0.02
0%, BF	0.03 ± 0.01	2.41 ± 1.09	5.81 ± 0.79	4.15 ± 0.20	5.56 ± 2.0	5.42 ± 0.26	1.80 ± 0.22
10%, BF	0.02 ± 0.01	0.66 ± 0.20	4.01 ± 1.14	2.76 ± 0.76	0.37 ± 0.39	4.26 ± 0.97	0.87 ± 0.55
20%, BF	0.02 ± 0.00	0.33 ± 0.15	1.83 ± 0.55	0.47 ± 0.27	0.01 (B)	0.12 (T)	ND
Day 120
0%, GF	0.02 (T)	0.32 ± 0.14	4.11 ± 2.28	0.27 ± 0.11	ND	0.72 ± 0.79	0.16 ± 0.07
10%, GF	0.02 ± 0.00	0.10 ± 0.02	ND	ND	ND	ND	ND
20%, GF	0.01 ± 0.00	0.09 ± 0.00	ND	ND	ND	ND	ND
0%, BF	ND	0.17 ± 0.02	2.38 ± 0.04	0.09 (T)	ND	ND	ND
10%, BF	0.02 (T)	0.13 ± 0.10	ND	ND	ND	ND	ND
20%, BF	0.05 ± 0.04	0.23 ± 0.16	ND	ND	ND	ND	ND

^aValues are averages of concentrations obtained from the bottom (∼10 cm from tank floor) and top (5 cm from surface) samples of each tank and time point. A “T” or “B” in parentheses indicates that a value from either the bottom or top of the tank was not detectable (ND).

Abundance and activity of methanogens and bacteria.

There were more (between ∼11 and 458%) copies per gram of dry manure (log₁₀ transformed) of mcrA and bacterial 16S rRNA genes and transcripts in the 1-year-old RS than in fresh dairy slurry used in this study (Fig. 2). During the initial 65 days of storage, these abundance differences resulted in significantly varied (one-way analysis of variance [ANOVA], P < 0.05) abundances of methanogens and bacteria in stored slurries that received different levels of RS. In samples collected on day 65, the presence of 10% and 20% RS in gradually filled tanks increased the mcrA gene copies between ∼7 and 12% (as calculated from log₁₀-transformed values) above the value for the control (Fig. 2a). Consistently, the activities of methanogens in these slurries, as observed from log₁₀ copies of mcrA transcripts, increased up to ∼12% (Fig. 2a). On day 65, in slurry samples from batch-filled tanks, 10% and 20% RS also resulted in ∼6.5 to 8% increases of mcrA gene copies (log₁₀ transformed) (Fig. 2a). Mean copies (log₁₀ transformed) of mcrA transcripts were also increased between ∼2.5 and 6%. In both gradually filled and batch-filled tanks, most of RS-related increases in the copies of mcrA genes and transcripts were statistically significant (Tukey's post hoc test using log₁₀-transformed values, P < 0.05). However, positive correlations between daily flux of CH₄ during the week of day 65 and mcrA gene (r² = 0.6; P < 0.05) or transcript (r² = 0.53; P < 0.05) copies were observed only in gradually filled tanks.

FIG 2.

Abundance and activity of bacteria (a) and methanogens (b) in fresh manure, old slurry (residual slurry), and stored dairy slurry (day 65 and day 120), as indicated by mcrA/bacterial 16S rRNA gene and transcript copies (log₁₀ transformed). Results shown are means and SEs of duplicate biological samples. Letters shown on the bars for gradually filled tanks indicate significance levels obtained after Tukey's post hoc test using log₁₀-transformed copies; for abundance or activity, different letters indicate statistically significant differences at a P value of <0.05. FM, fresh manure; RS, residual slurry; GF, gradually filled; BF, batch filled.

The day 65 slurry samples collected from batch-filled tanks had more copies (per gram of dry manure) of mcrA genes (∼5 to 10%) and transcripts (∼4 to 9%) than did slurries from gradually filled tanks (Fig. 2a). While these increases were statistically significant (Tukey's test, P < 0.05), they were not reflected in CH₄ flux. This is because during the week of day 65, daily mean fluxes of CH₄ from gradually filled tanks with RS were ∼23.2% higher than from the corresponding batch-filled tanks. However, in the absence of RS, the gradually filled tank showed the lowest CH₄ flux during the week of day 65 (Fig. 1).

Except the tank with 10% RS and filled in batch, RS-related increases in the abundance of bacteria coincided with the changes in methanogen abundance (Fig. 2b). In day 65 samples, slurries with 20% RS of both filling types had more copies (up to 5%), as calculated using log₁₀-transformed copy numbers, of 16S rRNA genes and transcripts (Fig. 2b). On day 120, the abundance and activities of methanogens and bacteria did not show clear trends between RS levels or filling types, indicating little effect of RS in later stages of storage. Thus, qPCR results suggested that increased levels of RS could increase the number methanogen and bacterial populations during the initial phases of storage, which may have increased CH₄ flux.

Diversity of archaeal and bacterial communities.

After a series of quality inspections and denoising of raw sequences, on average, 35,074 (1,255 unique) and 8,549 (2,056 unique) quality reads per sample were obtained for archaeal and bacterial 16S rRNA amplicons, respectively. From the RNA samples that were obtained from slurries during peak flux of CH₄, 9,841 (688 unique) quality reads per sample of archaeal 16S rRNA transcripts were also retrieved. Prealigned sequence lengths of archaeal and bacterial 16S rRNA amplicons were 348 to 360 bp and 379 to 405 bp, respectively. The final quality reads of archaeal and bacterial 16S rRNA amplicons were used to define operational taxonomic units (OTUs) at a cutoff of 0.03 and to perform further OTU-based analyses. Two samples with low numbers of amplicon sequences were excluded from OTU-based analysis in archaea. Rarefaction curves (Fig. S2a and S2b), particularly for the archaeal 16S rRNA gene amplicon library, indicated reasonable sampling efforts to capture most of the species in slurry samples. This was consistent with Good's coverage estimates, which indicated sufficient capture of archaeal (>99%) and bacterial (80 to 92%) species from these samples (see Table S2 in the supplemental material).

Diversity indices (sobs, Chao1, and inverse Simpson index) calculated for archaeal 16S rRNA amplicon sequences indicated little difference in the diversity of archaea between RS and fresh dairy manure (Table S2). With the presence of 20% RS, the diversities of archaea in slurries from both gradually filled and batch-filled tanks were slightly increased. These diversity differences gradually intensified with storage time, resulting in groupings of slurry samples with and without RS (Fig. 3a). Despite some overlap, the observed separation of slurries with and without RS in the nonmetric multidimensional scaling (NMDS) plots (stress = 0.18; r² = 0.95) were significant (analysis of molecular variance [AMOVA], P < 0.001). A consistent result (parsimony, P = 0.038) was also observed in a tree that was generated using Yue and Clayton theta distance matrices of archaeal 16S rRNA amplicon sequences (see Fig. S4a). Analysis of shared OTUs (Fig. S3a to d) indicated that RS and fresh manure share more than 50% of their total OTUs. Most (25–33) of these shared OTUs were also present in stored slurries from all tanks. These shared OTUs include the most dominant ones in all tanks, which might govern microbial processes in slurries. Thus, RS-related differences in the diversity of archaea might be less important, as the dominant OTUs were consistently present among all slurries.

FIG 3.

Nonmetric multidimensional scaling (NMDS) plots of the archaeal (a) and bacterial (b) communities in fresh slurry, residual slurry, and stored slurries. Yue and Clayton theta distance matrices of archaeal and bacterial 16S rRNA gene amplicons were used to calculate the respective NMDS axis values, which were plotted in R. D65 and D120 represent day 65 and day 120, respectively. The dark green, blue, red, and black colors represented samples from fresh manure, RS, gradually filled tanks, and batch-filled tanks, respectively.

The diversity of bacteria in RS and fresh slurries showed little variation. Unlike archaea, stored slurries with and without RS of both filling types did not show noticeable variation in day 65 or day 120 samples (Table S2). However, when day 65 and day 120 samples are compared, the number of OTUs (sobs and Chao1 estimates) and inverse Simpson diversity indices are shown to be reduced, indicating a gradual reduction of diversity. In NMDS analysis (stress = 0.06; r² = 0.99), there was significant separation (AMOVA, P < 0.001) of day 65 and day 120 samples (Fig. 3b). The Yue and Clayton theta distance matrix-based tree of slurry samples further confirmed the differences (parsimony, P = 0.002) between day 65 and day 120 samples (Fig. S4b). Even though the diversity of bacteria was reduced with storage time, analysis of shared OTUs among slurry samples indicated little change in proportion of the predominant OTUs (data not shown). Hence, like for archaea, changes in the diversity of bacteria due to the presence of RS or storage time might not be important, as the predominant OTUs, which might govern the microbial processes in slurries, were consistently present in all samples.

Phylogenetic analysis of archaea and bacteria.

Almost all (>99.9%) of the archaeal 16S rRNA gene and transcript reads were phylogenetically assigned to methanogenic archaea, indicating the suitability of stored dairy slurry for these organisms. Twenty-eight archaeal genera were identified from all slurry samples, which were dominated by the genus Methanocorpusculum, accounting for ∼56.6 to 88% of the reads. The genus Methanosarcina also represented a significant proportion of the reads (up to ∼36.7%). Other major methanogenic genera identified in this study were Methanobrevibacter, Methanomassiliicoccus, Methanoculleus, and several uncultured members, all representing up to 17.5% of the reads in each sample (Fig. 4a). On the other hand, RNA samples obtained during peak flux of CH₄ were dominated by the genus Methanosarcina, accounting for ∼57% of the reads (Fig. 4b).

FIG 4.

Relative proportions of archaeal genera in fresh, residual (old), and stored dairy slurry samples (day 65 and day 120) (a) and active archaeal genera during peak methane (CH₄) flux from each tank (b). Each bar indicates the proportion of archaeal genera in each sample, and the paired bars are for duplicate samples collected from storage tanks. Data outputs from shared OTUs and taxonomic analysis of OTUs were used to generate the graph.

Regardless of RS levels, no significant variations in the relative abundances of major methanogenic genera were observed (Fig. 4a). The genus Methanocorpusculum, which was predominant in all samples, represented ∼8.5% more reads in RS than fresh slurry. However, this difference did not result in noticeable variation in stored slurries of both filling types. On the other hand, Methanosarcina and Methanobrevibacter were relatively abundant in fresh dairy slurry, representing about ∼1.2% and ∼10% more reads, respectively (Fig. 4a). The relative proportion of Methanosarcina, the second most abundant methanogens in stored slurries (∼7 to 28%), varied more with storage time than the presence or absence of RS. Its mean relative abundance increased from ∼14.3% (on day 65) to 36.4% (on day 120) (Fig. 4a). While Methanocorpusculum dominated 16S rRNA gene sequences of all samples, Methanosarcina predominated in the RNA samples obtained during peak flux of CH₄ in each tank (Fig. 4b). It represented 42.4 to 57.6% of the reads in day 65 samples of most tanks. In gradually filled tanks without RS, which had peak flux of CH₄ on day 87, Methanosarcina represented ∼75.3% of the reads. This indicated a gradual increase of this genus with storage time. The lower relative proportion of Methanosarcina in one of the day 65 samples from a batch-filled tank with 20% RS might be due to the spatial heterogeneity of VFA levels in the tank (Table 2). Thus, in addition to Methanocorpusculum, both DNA and RNA data indicated the importance of methanogens related to the genus Methanosarcina as key players in CH₄ production.

A total of 27 phyla were identified from bacterial 16S rRNA amplicon sequences. The phylum Firmicutes predominated in all slurry samples, representing 57 to 70% of the reads (Fig. 5a). While 16 of the 27 phyla represented <0.5% of the reads in each sample, the phyla Bacteroidetes, Synergistetes, Spirochaetes, Actinobacteria, Proteobacteria, and Chloroflexi represented between ∼1 and 15% of the total reads in each sample (Fig. 5a). Romboutsia and Clostridium_XI in day 65 samples and Sedimentibacter and Syntrophomonas in day 120 samples were the predominant genera of the phylum Firmicutes (Fig. 5b). As the relative abundances of different bacterial phyla in fresh and RS were not significantly different (Fig. 5a), the presence of 20% RS did not alter the relative proportions of these phyla in stored slurries of both filling types. During storage, however, the relative abundance of Bacteroidetes in all slurries gradually decreased, while Synergistetes showed the reverse.

FIG 5.

Relative proportions of bacterial phyla (a) and genera (b) in fresh, residual, and stored dairy slurries. Bars indicate the proportion in each sample, and the paired bars are for duplicate samples collected from each tank. Data outputs from shared OTUs and taxonomic analysis of OTUs were used to make the graph. uncl., unclassified.

Despite the difference in the relative proportions of Sedimentibacter between fresh manure and 1-year-old RS (1.4% and 3.3%, respectively), no significant difference was observed with the presence of RS in either day 65 or day 120 slurry samples (Fig. 5b). However, its relative proportion increased with storage time, reaching ∼20.3% on day 120. Similar trends were observed for Cloacibacillus and Syntrophomonas, where gradual increases (up to ∼11% and 6%, respectively) in relative proportions were observed in day 120 slurries. The relative proportions of the genus Romboutsia in fresh manure and RS were ∼4% and 1%, respectively, which did not result in significant variation in stored slurries of either day 65 or day 120 samples. However, its relative proportion in day 65 slurries (12 to 15.5%) was higher than in day 120 slurries (∼8%) (Fig. 5b). Similarly, the relative proportion of Clostridium cluster XI was higher in fresh manure (∼2.4%) than in RS (∼0.8%) and showed a gradual reduction, from ∼9.5 to 11.6% in day 65 slurries to ∼4.5% in day 120 slurries (Fig. 5b).

DISCUSSION

In dairy farming, which is a source of large amount of non-CO₂ greenhouse gases (i.e., CH₄ and N₂O), liquid manure storage systems have been identified as hot spots for CH₄ emission (2). In these storage systems, the inoculum effect of RS in stimulating anaerobic degradation processes and CH₄ production has been evident from early studies (26, 27). The authors demonstrated shorter lag phases for CH₄ productions from stored animal manure when about 15% old manure is left after emptying storage tanks. In the current study, despite the RS-related increases in the abundances of bacterial and methanogenic communities in day 65 slurries, impacts on CH₄ flux were observed only in gradually filled tanks. With gradual filling, a practice that represents the actual manure storage conditions in many dairy farms, the longer lag phase (∼3 weeks) and lower peak flux of CH₄ from the 0% RS tank resulted in up to 85% reduction of cumulative emission. With the presence of 10% or 20% RS, increases in CH₄ flux and cumulative emission did not vary, indicating that 10% RS might be sufficient to stimulate the start-up of anaerobic degradation and CH₄ production. As complete removal of residual manure is often difficult, future studies need to examine the range between 0% RS and 10% RS for levels that have minimal inoculum effect during storage. In other studies, linear increases in CH₄ production with the presence of up to 20 to 30% inoculum (digested old manure slurry) in stored liquid dairy manure (6) or food wastewater (28) have been reported; however, the source of inoculum, storage condition, and characteristics of fresh manure may determine the activities of microbial inoculants. In this study, low CH₄ production in the tank without RS coincided with low abundance of methanogens and bacteria in day 65 samples.

Gradual filling of tanks means increasing volumes of manure throughout the storage period; thus, considering the surface area-to-volume ratio in the CH₄ flux calculation was necessary to assess if the available volume of slurry impacted the observed flux levels. However, CH₄ flux per cubic meter of slurry had trends similar to those for the flux calculated using surface area (Fig. S1). The influence of RS-derived microbial abundance on CH₄ production can also be realized from the positive correlation that occurred between daily CH₄ fluxes during the week of day 65 and RS level-related increases in the abundance (r² = 0.6; P < 0.05) and activity (r² = 0.53; P < 0.05) of methanogens.

Unlike in previous batch-based studies (4, 6, 8), in this study, no significant difference in CH₄ flux and cumulative emission was observed among batch-filled tanks with and without RS. After complete removal of old manure sludge from batch-filled tanks, ∼97% reduction of fugitive CH₄ emissions (4) and ∼56% reduction of cumulative CH₄ emissions (8) have been reported. In the current study, what obscured the expected differences in CH₄ flux among batch-filled tanks (with and without RS) is not yet clear. Unlike the small difference in CH₄ flux from these tanks, there was a positive correlation between RS levels and bacterial or methanogen abundance. However, abundance of these organisms and CH₄ flux was poorly correlated during the week of day 65. Thus, with high abundance and activity of bacteria and methanogens at high RS levels, some unknown factors rather than RS might have played a role in reducing CH₄ emission from these tanks. Undisturbed crusts that could physically block gaseous emissions and potential methanotrophy (17, 18, 29) are among the factors that might impact CH₄ emission from these tanks. In different storage systems, characteristics of crust (e.g., thickness) and methanotrophy may vary with the nature of manure (e.g., total solid level) (30).

Regardless of filling type, the influences of RS on the abundance and activities of bacterial and methanogenic communities were not important after day 120, thus playing a significant role only during the initial phases of storage, when population sizes rather than available substrates may be limiting. Using different sources of inoculum material (wetland sediment, landfill leachate, and mesophilic digestate), a recent study (31) indicated poor correlations between methanogen population size and biochemical methane potential. However, in different storage systems, these correlations may vary with factors that may affect flux, such as the presence of crust and hence potential methanotrophy (17, 18, 29, 32–34) and activity of methanogens (e.g., substrate level, pH, and moisture level) (35). Thus, in dairy farms where liquid manure is gradually loaded in open storage tanks or lagoons, reduction or complete removal of RS can lower the number of well-adapted bacteria and methanogens that may serve as inoculants. It is important to note, however, that RS removal should occur as close as possible to the time when high emissions are expected; therefore, late-spring RS removal is recommended and fall RS removal is unlikely to be effective (3).

In all slurry samples, a lower diversity of archaea than of bacteria was observed, a finding in line with previous reports for manure-based mesophilic anaerobic digesters (20, 36–38). Differences in the diversity of archaea among all samples were small. As archaeal communities in rumen and manure environments are generally dominated by methanogens (39) and many of them can establish in both environments (12), storage of dairy slurry might not result in significant shifts in the diversity of archaea or methanogens (40). Indeed, an earlier study (41) identified similar methanogenic archaeal communities from fresh diary manure and a mixed inoculum obtained from different biogas reactors, although with a higher abundance in the latter, which is consistent with the current study. The relatively higher diversity of bacteria in RS than in fresh manure was not expected, but the source of RS might have influenced the diversity. Nevertheless, the stability of core communities throughout the storage, as observed from the stability of dominant bacterial OTUs in all samples, could be an indication that diversity differences might not be important. A study (40) also showed little difference in the diversity of bacteria between several-month-adapted inoculum and swine slurry treated with anaerobic digesters and supplemented with the inoculum. Hence, RS might be more important in contributing to the starting number of methanogen and bacterial populations than adding new phylotypes to dairy slurry storage systems.

The dominant bacterial phyla identified in this study (i.e., Firmicutes, Bacteroidetes, and Synergistetes) were typical of anaerobic digester and rumen environments (20, 36, 42–45). The dominance of Firmicutes throughout the storage might suggest members of this phylum as key players in the anaerobic degradation of organic substrates in stored dairy manure. Dairy manure contains large amount of lignocellulose biomass (e.g., celluloses, hemicellulose, lignin, and pectin) and some amounts of protein, which are substrates for hydrolytic microorganisms. Thus, these substrates might have supported the hydrolytic members of Firmicutes, such as Clostridium cluster XI and Romboutsia (46), that were dominant in day 65 slurries. While the reduction in the relative proportion of these bacteria in day 120 samples might indicate gradual limitation of available substrates, their by-products (i.e., organic acids) that might have supported acidogenic bacteria, such as Sedimentibacter (47), were predominant in day 65 and day 120 samples.

The predominance of hydrolytic and acidogenic bacteria especially in day 65 and day 120 slurries might suggest accumulation of methanogenic substrates (CO₂/H₂ and acetate), coinciding with the exclusive dominance (>99.9%) of methanogen-related archaeal OTUs. Although methanogens that are closely related to the genus Methanocorpusculum have rarely been reported from rumen environments (48–50), we identified these methanogens as predominant in all samples. The predominance particularly in fresh dairy slurry that was expected to be the main source of these organisms might indicate these methanogens as predominant members of rumen archaea. Moreover, the predominance of these methanogens in RS and stored dairy slurries with 10% and 20% RS might suggest stored liquid dairy manure as suitable habitats for these methanogens. Members of Methanocorpusculum are hydrogenotrophic and have been isolated from anaerobic systems where fermentative by-products such as H₂/CO₂ or formate may be accumulated (12, 49, 51). Although these substrates are also used by the genus Methanobrevibacter, its relative proportion showed gradual reduction with storage. This might be due to its poor adaptability to storage conditions; this methanogen was commonly reported as dominant in rumen environments (48, 52, 53).

Despite the low abundance of Methanosarcina in fresh slurries and RS, its relative abundance gradually increased with storage time. Moreover, during peak flux of CH₄ from each tank, the most active methanogens were members of this genus. This might be explained by its versatile metabolic capability, as it uses a range of substrates, including CO₂/H₂, acetate, methylated compounds, and methanol (12, 49). Moreover, this methanogen has relatively higher tolerance to different environmental stresses (e.g., high ammonium concentration, high VFA contents, and wider pH range) (54, 55). Particularly, its gradual increase with storage time coincided with the gradual increase in the proportion of Synergistetes that was dominated by the genus Cloacibacillus. This bacterium uses different amino acids (e.g., arginine, lysine, histidine, and serine) and releases direct (acetate, H₂, and CO₂) and indirect (propionate, butyrate, and valerate) methanogenic substrates (56, 57). Moreover, accumulation of these methanogenic substrates could be contributed by members of Firmicutes, such as Sedimentibacter, an amino acid-utilizing bacterium (47) that increased with storage time. Moreover, higher organic acids, such as propionic acid, could be converted into acetate, formate, H₂, and CO₂ by acetogenic bacterial communities, such as Syntrophomonas (58), that were relatively abundant in day 120 samples. While acetate was detected in both day 65 and day 120 samples, OTUs related to the genus Methanosaeta (methanogens that can grow solely on acetate) (12, 49) were detected in very low proportions (<0.5%). Methanosaeta has high affinity for acetate (55) but has often been described to be less competitive at high acetate concentrations (31), although a recent study found strong competitiveness of these methanogens in animal wastewater with an elevated acetate level (59).

Conclusion.

On many farms, complete removal of dairy slurry from storage tanks is difficult. Residual slurry may contain large numbers of well-adapted bacterial and methanogenic communities. As a result, the inoculum effect is significant. This study demonstrated the effects of different levels of RS on CH₄ emissions and abundance and diversity of methanogenic and bacterial communities. Regardless of filling type (gradual or batch), increased RS levels resulted in higher abundances of methanogens and bacteria in day 65 slurry samples. However, the increased abundances of these populations were significantly correlated with CH₄ emissions only in gradually filled tanks, which represent the typical slurry storage practices in many dairy farms. Regardless of RS and slurry filling method, little variation in diversity of bacteria and methanogens was observed among slurry samples. In all samples, Firmicutes and Methanocorpusculum were the predominant bacteria and methanogen, respectively. In addition to its gradual increase with storage, Methanosarcina represented the most active methanogen during peak flux of CH₄ from each tank, indicating its significant role in the overall CH₄ emitted. Overall, this study revealed the importance of RS in providing newly stored manures with large number of active and well-adapted bacteria and methanogens that may accelerate the start of anaerobic degradation and methanogenesis, and complete removal of RS through alternative pumping systems may provide a low-cost option in reducing CH₄ emissions from dairy slurry storage systems.

MATERIALS AND METHODS

Experimental setup and measurement of CH₄ flux emissions.

The study was conducted from 2 June to 24 September 2016 (120 days) at Dalhousie University's Bio-Environmental Engineering Center (BEEC) in Truro, NS, Canada (45°45′N, 62°50′W). Six pilot-scale (3.9 m by 1.75 m by 1.8 m) manure storage tanks enclosed by flowthrough steady-state chambers were used. This site has been previously described by Wood et al. (30). The RS was obtained from a previous study conducted at the site and consisted of dairy slurry stored for about 1 year. Dairy slurry (freshly excreted to 2 weeks old) was obtained from a nearby farm that had about 185 dairy animals (lactating and dry cows, calves, and heifers), used washed sand as bedding material, and stored dairy waste (wash water, dung, urine, and bedding material) in a concrete lagoon. Tanks with two levels of RS (10% or 20%) and without any RS were filled with fresh dairy slurry either in batch (on day 1 to 100% volume, 10.6 m³) or gradually with incremental manure additions on day 1 (∼33.3% volume), day 20 (∼33.3% volume), and day 40 (∼33.3% volume). Gas samples were drawn continuously from each tank's outlet, and ambient air and CH₄ concentrations were measured at the site using a model TGA 100A tunable diode laser trace gas analyzer (Campbell Scientific Inc., Logan, UT). The CH₄ flux (F, g m⁻² s⁻¹) was calculated as described previously (30):

$$F=\frac{C_{\text{out}}-C_{\text{in}}}{A_s}Q$$

where C_out and C_in are the chamber's outlet and inlet CH₄ concentrations (g m⁻³), respectively, A_s is the surface area of the storage tank (m²), and Q is the airflow rate (m³ s⁻¹) as determined from the wind velocity in each tank's venturi using cup anemometers (Davis Instruments Corp., Hayward, CA). Fluxes were then calculated into daily averages.

Slurry sample collection and analyses.

Prior to storage, 2-g slurry samples were collected in duplicate from fresh and 1-year-old slurries using 15-ml Falcon tubes containing 5 ml of LifeGuard soil preservation solution (MoBio Laboratories Inc., Carlsbad, CA). During storage, slurry from the top (∼5 cm from surface) and bottom (∼10 cm from tank's floor) sections of each tank was sampled every 15 days for 120 days. From a preliminary test on a few samples, significant shifts in the abundance of methanogenic communities were observed on day 65 (when peak CH₄ emissions for most of the tanks were detected); hence, samples from fresh manure, old slurry, and day 65 and day 120 samples were selected for further microbial analysis. On each date, two samplings were made from each tank (top and bottom). During each sampling, nine samples were collected from distinct locations of a tank and homogenized in a clean bucket using a sterile metal rod. Subsamples (2 g each) were then collected in duplicate sterile 15-ml Falcon tubes each containing 5 ml of LifeGuard soil preservation solution. Samples were then transported in coolers to the lab and stored in a −20°C freezer until nucleic acid extraction. Further subsamples of appropriate volume were also collected to analyze pH, dry matter (DM) and volatile solid (VS) contents, and volatile fatty acids (VFAs). Dry matter and VS contents and pH were determined at the Nova Scotia Department of Agriculture's Laboratory Services (Harlow Institute, Bible Hill, NS, Canada) using standard methods. The VFA analysis was performed by InnoTech Alberta (Edmonton, AB, Canada) via headspace gas chromatography.

Nucleic acid extractions and quantitative real-time PCR.

Slurry samples in LifeGuard soil preservation solution were centrifuged, and pellets were used to coextract total RNA and DNA using RNA PowerSoil total RNA isolation with DNA elution accessory kits (MoBio Laboratories, Inc., Carlsbad, CA) by following the manufacturer's protocol. As the massive research projects (such as Earth Microbiome Project, which aimed to characterize microbial life on this planet) also use nucleic acid extraction kits from MoBio, efficient extractions of nucleic acids were expected when following the optimized and recommended protocol. Duplicate RNA and DNA samples per tank and sampling date were prepared by pooling extractions made from the top and bottom sections of individual samplings. Using triplicate reaction tubes, 8-μl RNA samples were treated with RQ RNase-free DNase (Promega) to remove contaminant DNA and then reverse transcribed into cDNA using a high-capacity cDNA reverse transcription kit with RNase inhibitor (Applied Biosystems) according to the recommended protocols. Before further downstream analyses, both cDNA and DNA samples were diluted and assessed for potential inhibitory effects as described earlier (60). Appropriate dilutions were then selected to conduct quantitative real-time qPCR and amplicon sequencing.

Clear multiplate 96-well PCR plates (Bio-Rad Laboratories, Inc., Hercules, CA) were used to prepare duplicate reaction mixes from each sample, and qPCR was performed using the CFX96 real-time system on a C1000 Touch thermal cycler (Bio-Rad Laboratories, Inc.). Methyl coenzyme A reductase (mcrA) genes and transcripts were quantified using primers mlas-mod F and mcrA-rev-mod R (24, 61). Bacterial population size and activities were also estimated by targeting 16S rRNA genes and transcripts using primers Bac 338F and Bac 518R (62). The 20-μl qPCR mix contained 10 μl of Ssofast EvaGreen supermix (Bio-Rad Laboratories, Inc.), 1 μl (10 pM) of each primer, 2 μl of template DNA or cDNA (1 to 10 ng μl⁻¹), and 6 μl of PCR-grade water. Thermal cycling for mcrA gene and transcript quantifications involved initial denaturation at 98°C for 2 min followed by 40 cycles of dissociation (95°C for 5 s), annealing (57°C for 10 s), and extension (72°C for 15 s) and a final step at 72°C for 30 s before the temperature was increased by 0.5°C from 65°C to 95°C for 5 s to analyze melting curves. For bacterial 16S rRNA gene and transcript quantification, the optimized cycling conditions used were initial denaturation at 98°C for 2 min followed by 34 cycles of dissociation (98°C for 5 s), annealing (55°C for 5 s), and extension (65°C for 5 s), with a final extension 72°C for 3 min before the temperature was increased by 0.5°C from 65°C to 95°C for 5 s. Known copies of plasmid standard curves for mcrA (10⁸ to 10¹) and bacterial 16S rRNA (10⁹ to 10¹ copies) gene and transcript quantifications were prepared from Methanosarcina mazei (ATCC 43340) and pure culture of Clostridium thermocellum, respectively. Efficiency, r², and slope of the plasmid standard curve for the mcrA gene were 93.6% ± 2.8%, 0.99, and −3.4 ± 0.1, whereas for the 16S rRNA gene, these values were 98.0% ± 0.7%, 0.99, and −3.30 ± 0.01, respectively. Cycles of quantification (CQs) for the highest diluted standard point of mcrA (10e1) and no-template controls (NTCs) were 36.1 ± 0.4 and 39.1 ± 1.1, respectively, whereas for the bacterial 16S rRNA gene, these values were 28.1 ± 0.2 and 32 ± 0.0, respectively. CFX Manager software version 3.1 (Bio-Rad Laboratories, Inc., Hercules, CA) was used to analyze the qPCR data. To determine the effects of RS on the abundance of methanogens and bacteria, one-way analysis of variance (ANOVA) was carried out using log-transformed copy numbers of the mcrA and 16S rRNA genes and transcripts in day 65 and day 120 slurries. When statistically significant variation (P < 0.05) was observed among means of gene copies in slurries that received 0%, 10%, and 20% RS, Tukey's post hoc test was conducted using GraphPad Prism version 7 (GraphPad Software, Inc.) to determine the sample that resulted in variation.

Amplicon library preparation and sequencing.

Amplicon libraries of archaeal and bacterial 16S rRNA genes were prepared from DNA samples obtained from RS, fresh slurry, and stored slurries (0% and 20%). To identify the key players during peak flux of CH₄ from each tank, a library of archaeal 16S rRNA transcripts was also constructed from RNA samples. As CH₄ flux from tanks with 10% and 20% RS of both filling types did not show significant variation, only tanks with 0% and 20% RS were used to determine the identities and relative proportions of bacterial and archaeal phylotypes. For bacteria, amplicons were prepared using primers 341F and 805R, which target the V3-V4 region of the 16S rRNA gene (63), whereas primers Arch349F and Arch806R were used to prepare archaeal 16S rRNA gene and transcript libraries (64). On both archaeal and bacterial primers, Illumina adapter sequences A and B (Table S1) were added to the 5′ ends of the forward and reverse primers, respectively.

For both genes, amplicons for MiSeq sequencing were prepared in two PCR steps with a total of 33 (bacteria) or 37 (archaea) cycles. First, archaeal and bacterial 16S rRNA genes were amplified for 25 PCR cycles using the primer sets modified as described above. For each sample, the 25-μl PCR mix contained 5 μl of 5× Phusion HF buffer, 0.25 μl of Thermo Scientific Phusion Hot Start II high-fidelity DNA polymerase (Thermo Scientific), 0.5 μl of 10 mM deoxynucleoside triphosphates (dNTPs; Thermo Scientific), 0.5 μl of each primer (10 μM), 2 μl of diluted DNA (10 to 50 ng/μl) or cDNA, and 16.25 μl of nuclease-free water. Thermal cycling for both gene targets was as follows: initial denaturation at 98°C for 3 min followed by 25 cycles of dissociation at 98°C for 10 s, primer annealing at 55°C for 30 s, extension at 72°C for 30 s, and a final extension for 5 min. Duplicate PCR tubes were pooled and products were cleaned using silica spin columns (Wizard SV Gel and PCR Clean-Up system; Promega) by following the recommended protocol. The second-step PCR was performed for 8 (bacteria) or 12 (archaea) cycles to attach Illumina index tags to the ends of the amplicons that were obtained from the first-step PCR. For each sample, a different combination of index primers 1 (N7xx) and index primers 2 (S5xx) of Illumina's Nextera XT DNA library preparation kit (Illumina Inc., San Diego, CA) were used to perform PCR. This was performed in a single 50-μl reaction mixture per sample; the same proportion of reagents and thermal cycling conditions similar to those of the first-step PCR were used, except that 4 μl of purified amplicons was used as the template. PCR products were then purified by magnetic beads (Agencourt AmPure XP; Beckman Coulter, Brea, CA) and resuspended in 25 μl. Purified PCR products were tested for correct amplicon length using gel electrophoresis and submitted to the University of Guelph Advanced Analysis Centre Genomic Facility (Guelph, ON, Canada) for sequencing. Prior to sequencing, libraries were normalized by Sequalprep (Thermo Fisher Scientific, Hampton, NH) and library quality was assessed from a random sample of 12 samples using a DNA1000 chip bioanalyzer (Agilent, Santa Clara, CA). Multiplexed sample sequencing was conducted using MiSeq reagent kit version 2 (500 cycles) (Illumina Inc., San Diego, CA), producing paired-end reads 250 bp in length. Unprocessed FASTQ files were received for subsequent analysis.

Sequence data analysis.

Raw sequence data for archaeal and bacterial 16S rRNA genes and transcripts were processed and analyzed in mothur version 1.39.5 (65) following the recommended pipeline for MiSeq 16S rRNA gene sequences (66), with some modifications. Briefly, after the forward and reverse reads of each sample were merged, target-specific primer sequences were removed and sequences were screened for ambiguity and length. Then sequences were aligned against the Silva gold reference file v128 and further screened for length and homopolymer; overhangs and common gaps were filtered, and sequences were preclustered to further denoise sequencing errors. After removal of potential chimeric sequences, a mothur-formatted version of the RDP's 16S rRNA reference (version 16) was used to classify sequences into phylotypes at a cutoff of 80%, at which undesirable targets that might have been picked by primers were filtered. Finally, purified sequences were clustered into operational taxonomic units (OTUs) at a cutoff of 0.03 (97% similarity), phylotypes of OTUs were identified using the RDP's 16S rRNA reference database, and rarefaction curves were calculated. For OTU-based alpha diversity (e.g., number of OTUs, coverage, Chao1, and inverse Simpson diversity estimate) and beta diversity (e.g., nonmetric multidimensional scaling and shared OTUs) analyses, sequence reads were subsampled and rarefied. Phylogenetic trees generated using the Yue and Clayton theta distance matrices (calculated by relaxed neighbor joining algorithm of Clearcut in mothur) (66–68) were used to construct dendrograms showing community structure-based variations between samples, and significance of variations was assessed using AMOVA and parsimony analyses.

Accession number(s).

The unprocessed sequence reads of bacterial 16S rRNA gene, archaeal 16S rRNA gene, and archaeal 16S rRNA transcripts obtained in this study have been deposited in NCBI's Sequence Read Archive as FASTQ files with accession numbers SRR6132422 to SRR6132441, SRR6132442 to SRR6132461, and SRR6132462 to SRR6132469, respectively.