Comparative Metagenomics Reveals Enhanced Nutrient Cycling Potential after 2 Years of Biochar Amendment in a Tropical Oxisol

The incorporation of biochar into soil is a promising management strategy for sustainable agriculture owing to its potential to sequester carbon and improve soil fertility. Expanding the addition of biochar to large-scale agriculture hinges on its lasting beneficial effects on the microbial community. However, there exists a significant knowledge gap regarding the specific role that biochar plays in altering the key biological soil processes that influence plant growth and carbon storage in soil. Previous studies that examined the soil microbiome under biochar amendment principally characterized only how the composition alters in response to biochar amendment. In the present study, we shed light on the functional alterations of the microbial community response 2 years after biochar amendment. Our results show that biochar increased the abundance of genes involved in denitrification and carbon turnover and that biochar-amended soil microcosms had a reduction in cumulative CO₂ production.

ABSTRACT

The complex structural and functional responses of agricultural soil microbial communities to the addition of carbonaceous compounds such as biochar remain poorly understood. This severely limits the predictive ability for both the potential enhancement of soil fertility and greenhouse gas mitigation. In this study, we utilized shotgun metagenomics in order to decipher changes in the microbial community in soil microcosms after 14 days of incubation at 23°C, which contained soils from biochar-amended and control plots cultivated with Napier grass. Our analyses revealed that biochar-amended soil microbiomes exhibited significant shifts in both community composition and predicted metabolism. Key metabolic pathways related to carbon turnover, such as the utilization of plant-derived carbohydrates as well as denitrification, were enriched under biochar amendment. These community shifts were in part associated with increased soil carbon, such as labile and aromatic carbon compounds, which was likely stimulated by the increased available nutrients associated with biochar amendment. These findings indicate that the soil microbiome response to the combination of biochar addition and to incubation conditions confers enhanced nutrient cycling and a small decrease in CO₂ emissions and potentially mitigates nitrous oxide emissions.

IMPORTANCE The incorporation of biochar into soil is a promising management strategy for sustainable agriculture owing to its potential to sequester carbon and improve soil fertility. Expanding the addition of biochar to large-scale agriculture hinges on its lasting beneficial effects on the microbial community. However, there exists a significant knowledge gap regarding the specific role that biochar plays in altering the key biological soil processes that influence plant growth and carbon storage in soil. Previous studies that examined the soil microbiome under biochar amendment principally characterized only how the composition alters in response to biochar amendment. In the present study, we shed light on the functional alterations of the microbial community response 2 years after biochar amendment. Our results show that biochar increased the abundance of genes involved in denitrification and carbon turnover and that biochar-amended soil microcosms had a reduction in cumulative CO₂ production.

INTRODUCTION

Burgeoning global population and accelerated urbanization of developing countries is increasing the competition for land, water, and energy resources and amplifying the imperative for agricultural intensification without compromising the environment for future generations (1). To feed this growing and urbanized population, global food production must increase by ∼70% (2), putting additional pressure on existing natural resources already under unsustainable management practices. Past increases in global food production were accomplished by an intensification facilitated by massive inputs of synthetic nitrogen (N) fertilizers that ultimately conferred high environmental costs. A significant amount of applied N is lost from agricultural fields, which can cause eutrophication of aquatic ecosystems, loss of diversity, and increased nitrate leaching as well as increased nitrogen oxide (NO_x) production (3–5), leading to increased greenhouse gas (GHG) flux to the atmosphere.

Integrative solutions are required to restructure productive systems into “climate smart agriculture” and models of “sustainable intensification” in order to increase food production from existing farmland in ways that reduce environmental impacts, such as a reduction in GHG emissions and through enhancing carbon (C) sequestration (6–8). In this context, the incorporation of biochar into soil is a promising management strategy for sustainable agriculture owing to its potential to sequester C and improve soil fertility (9, 10). Biochar is a C-rich product of biomass pyrolysis and contains large portions of aromatic compounds that influence its stability and C sequestration potential in soil (11, 12). The documented beneficial effects of adding biochar to soil include increases in moisture retention, pH and cation exchange capacity (CEC) (10, 13), decreases in N₂O and CH₄ emissions (14–18), and decreases in N leaching from soil (19, 20).

As part of a sustainable management practice, biochar addition induces changes in the soil physical and chemical properties and shifts in the soil microbiome. However, the microbial response to biochar addition depends strongly on the soil type and cropping system as well as the properties of the biochar being added (21–27). The observed effects of biochar on microbial processes are varied. Some studies have observed an increase in soil respiration (17, 28, 29), although decreases or no changes have also been observed (25, 30, 31). The effects of biochar on microbial community composition have also been reported with some contradictory findings. Several studies have observed increases in Actinobacteria, Proteobacteria, Bacteroidetes, and Gemmatimonadetes (32–35) and decreases in Acidobacteria in biochar-amended soils (22, 24), while others have reported decreases in Proteobacteria and Bacteroidetes (33, 35, 36). Shifts in the microbiome composition and function influenced by biochar addition have the potential to impact GHG emissions, alter nutrient mineralization, and influence plant growth promotion. However, the dynamics and mechanisms of the impacts of biochar on microbial community composition as well as function remain poorly understood. A number of studies have attempted to assess the influence of biochar pyrolysis temperature (37, 38), feedstock (39), soil type, cropping system (18, 22), and addition with N fertilizer (40, 41) on the soil microbiome.

Many studies on the effect of biochar amendment on the soil microbiome have been based on analysis of the 16S rRNA gene sequence, which revealed important shifts in community composition. However, there remains a lack of information concerning functional gene content and diversity, which limits our understanding of the impacts of biochar on potential microbial function. The potential of the soil microbiome to control the fate of C and N in soils may be investigated using a shotgun metagenomic approach to better elucidate the functional significance of a shift in community composition in response to biochar amendment than amplicon sequencing alone. To date, only one study has used a shotgun metagenomic approach to investigate the microbial community of aged biochar and the adjacent soils collected from a northern forest (42). Our work applied shotgun metagenomics to agricultural biochar-amended soils.

In this study, we report on the shotgun metagenomic analysis of the soil microbial community of tropical oxisol soils that experienced 2 years of biochar amendment under Napier grass cultivation. Our previous analyses of samples collected from the same soils in the first year of biochar amendment using targeted amplicon sequencing coupled with molecular ecological networks revealed that biochar amendment induced significant shifts in the microbial community and increased diversity and network complexity. However, whether the observed changes in the community composition reflected a shift in the functional gene diversity, whether the responses to biochar amendment were attributed to a few taxa as opposed to being global, and what the functional adaptations underlining the response were remained unknown in our previous study (22). The objective of the present study was to provide a high-resolution description of the community complexity, the genes and taxa responding to biochar amendment, and their potential effects on the soil C and N cycling. We expect the functional gene content of the biochar-amended metagenomes to reflect the shift in the community composition observed in the previous study and the functional potential to reflect the results of the network analysis from our previous study, particularly with regard to enhanced potential for N cycling and competition for resources associated with biochar.

RESULTS

Soil chemical characteristics and respiration.

We focused our study on soils collected 2 years after the initial addition of biochar to an oxisol under Napier grass cultivation. Eight samples from the control and biochar-amended soils were used to determine the soil chemical characteristics. These soils, collected preharvest, had few measurable differences in elemental concentration, nutrient status, and associated crop yield (Table 1). As expected, the mean C concentration (C%) of the biochar-amended soils was significantly higher than soil control soils. Otherwise, biochar-amended soils contained slightly higher soil N%, but the differences were not statistically significant. Soil pH was close to neutral in both biochar-amended and control soils. No statistical differences were observed in soil moisture or base cations (calcium [Ca²⁺], sodium [Na²⁻], magnesium [Mg²⁺], and potassium [K⁺]), although soil base cations were generally higher in biochar-amended samples, with the exception of Ca (Table 1). The Napier grass crop yield, harvested approximately 1 month after the soil core collection, was higher in biochar-amended plots than in control plots (see Fig. S1 in the supplemental material), but the difference also was not statistically significant (P = 0.233, paired t test).

TABLE 1.

Soil characteristics of the oxisol used in the microcosms

Soil characteristic	Value (mean ± standard error)	P value
Ca (mg kg−1 soil)	1,554.14 ± 106.00	0.828
Na (mg kg−1 soil)	35.85 ± 1.83	0.608
Mg (mg kg−1 soil)	235.10 ± 12.40	0.254
K (mg kg−1 soil)	861.98 ± 56.05	0.199
pH	6.70 ± 0.123	0.709
Moisture (%)	35.51 ± 0.88	0.152
Carbon (%)		0.001a
Total	1.73 ± 0.11
Control	1.37 ± 0.02
Biochar	2.08 ± 0.13
Nitrogen (%)	0.170 ± 0.003	0.162

^aLevels are significantly different between biochar-amended and control soils.

Soil respiration, or cumulative CO₂ production from the soil, was measured as a proxy for microbial activity for each of the microcosms over a 14-day period. Overall, the concentration of CO₂ in the headspace from the control soil was significantly higher than in the biochar-amended soil at all time points measured (Fig. 1). The mean CO₂ production rate in the biochar-amended microcosm was slightly lower (0.011 μg C-CO₂ g⁻¹ soil day⁻¹) than the control (0.014 μg C-CO₂ g⁻¹ soil day⁻¹), although the difference in respiration rate was not significant (P = 0.14, two-sample t test) (Fig. 1). However, when substrate quality was taken into account, the biochar-amended microcosm respiration rate was significantly lower (0.005 μg C-CO₂ g⁻¹ soil-C day⁻¹) than the control (0.010 μg C-CO₂ g⁻¹ soil-C day⁻¹) (P < 0.001, two-sample t test) (Fig. 1).

FIG 1.

Cumulative CO₂ production over a 14-day incubation period. (Top) CO₂ production per gram of soil. (Bottom) CO₂ production per gram of soil carbon. Insets show equations for best-fit lines. Points represent the average microcosm CO₂ concentrations and error bars represent the standard errors of the means (n = 6). Green circles, control soil microcosms; black circles, biochar-amended soil microcosms; ***, P < 0.001; **, P < 0.01; *, P < 0.05.

Statistics of metagenomes and community complexity.

Sequencing from three replicate samples representing the biochar-amended soils (BC1 to BC3) and three samples representing the nonbiochar control soil (NBC1 to NBC3) yielded approximately 10 to 40 gigabytes of paired-end sequence data per sample. Four metagenomes (NBC1, NBC2, NBC3, and BC3) each had between 11 and 15 Mbp of sequences, and two metagenomes (BC1 and BC2) had approximately 45 to 50 Mbp of sequences (see Table S1). The estimated coverage based on the read redundancy value calculated using Nonpareil revealed an average coverage of approximately 0.15 and 0.28 for the metagenomes obtained from control and biochar-amended oxisol samples, respectively. The application of Nonpareil estimates revealed that large sequencing efforts were required for these soil samples, where up to 1 terabyte (Tb) of sequence data was expected to be necessary to achieve nearly complete (99%) abundance-weighted average coverage (Fig. 2). Sequence diversity values, a measure of alpha diversity derived from Nonpareil curves, showed no differences between biochar-amended (average, 28.29) and control (average, 28.22) soils. The assembly of the metagenomes recovered ∼280,000 and ∼1.5 million contigs of at least 1 kbp in length from the control and biochar-amended samples, respectively. The N₅₀ values averaged from biochar-amended soil metagenomes were slightly higher than the controls (1,133 versus 915 bp), reflecting the lower sequence coverage determined by Nonpareil for the control metagenomes (Fig. 2; Table S1). For MG-RAST sequence statistics, see the supplemental material.

FIG 2.

Average coverage, estimated from the portion of nonunique reads as a function of the size of subsamples randomly drawn from the metagenomes of biochar-amended and control soils. The solid lines indicates the fitted models based on subsampling, the empty circles mark the actual sizes and estimated coverage of the metagenome data set, the red and pink horizontal dashed lines indicate the 95% and 100% average coverage levels, respectively. BC, biochar-amended samples; NBC, nonbiochar control samples. Numbers following abbreviations indicate the sequencing technical replicate.

Microbial community structure and diversity.

Prokaryote sequences represented the majority of each microcosm community sampled, with >99% of the total number of genes recovered with best matches against bacterial and archaeal genomes in MG-RAST (Table S1). Bacterial sequences predominated the prokaryotic sequences, averaging ∼99% of sequences in all samples. Archaeal sequences comprised approximately 0.43% and 0.63%, and eukaryotes represented approximately 0.41% and 0.45% of the total sequences in the control and biochar-amended samples, respectively. Domain-level differences in abundance were significant for archaea and viruses (P < 0.05, two-tailed paired t test); archaeal abundance was higher in controls and viruses were greater in biochar-amended samples (Table S1). Significant differences in archaeal relative abundance were observed for Euryarchaeota and Thaumarchaeota (P < 0.05, paired t test). Within eukaryotes, no significant differences were observed for fungal abundance. Within bacteria, several significant differences were observed at the phylum level (Fig. 3A). In biochar-amended samples, the relative abundances of Proteobacteria and Bacteroidetes were significantly higher than in the controls. In the controls, the relative abundances of Actinobacteria, Firmicutes, Chloroflexi, and Cyanobacteria were significantly higher than in the biochar-amended samples.

FIG 3.

Shifts in taxon abundance as effects of biochar amendment. (A) Rings represent the average abundances of phyla that make up at least 1% of the whole community. Phyla that are significantly different in abundance between biochar-amended and control samples are marked by an asterisk (P < 0.05, two-tailed paired t test). (B) Heatmap of the normalized abundance at the class level for bacteria in the six microcosm metagenomes. Color code based on higher relative abundance in control (red) or in biochar-amended (blue) metagenomes (see scale on the top right).

A heatmap of the relative abundances at the class level confirmed that the samples from the two different treatments clustered separately (Fig. 3B), although the differences between biochar-amended and control samples were not significant (P = 0.1, permutational multivariate analysis of variance [PERMANOVA]). Biochar-amended metagenomes had higher relative abundances of Proteobacteria belonging to Alphaproteobacteria, Betaproteobacteria, Gammaproteobacteria, and Deltaproteobacteria, and Bacteroidetes belonging to the classes, Bacteroidia, Cytophagia, Flavobacteriia, and Sphingobacteriia. Phyla that comprised less than 1% of the community were generally higher in the controls than in biochar-amended samples with the exception of the Spirochaetes, Fibrobacteres, and Tenericutes.

Relative abundances of metabolic pathways in biochar-amended versus control metagenomes.

Of the total pathways at the SEED subsystem level 3, 380 of the 1,035 pathways were significantly differentially abundant (P < 0.05). The majority of the significant differences in pathway abundances between biochar-amended and control samples were small, typically a log₂-fold change of <1 (see Table S4). Nonetheless, several significant changes were noted, and these changes were consistent among the replicates. For example, clustering of samples and replicates based on level 1 subsystems showed that the biochar-amended versus the control samples clustered together (see Fig. S2). A large portion of pathways involved in carbohydrate metabolism showed significant changes in relative abundance (Fig. 4). In particular, several pathways that are involved in central carbohydrate metabolism were enriched in controls, including some pathways involved in methane metabolism (Fig. 4). Pathways involved in CO₂ fixation, such as CO₂ uptake and genes that encode carboxysomes (e.g., carbonic anhydrase) and labile carbon source metabolism, were enriched in biochar-amended samples. In the respiration subsystem, control samples were enriched in pathways involved in carbon monoxide dehydrogenases, succinate dehydrogenase and respiratory complex 1 (e.g., NADH ubiquinone oxidoreductase). Biochar-amended samples were enriched in pathways involved in NiFe-hydrogenase maturation, respiratory dehydrogenase 1, which included several dehydrogenases involved in amino acid, sugar and alcohol dehydrogenation, and several cytochrome oxidases (Fig. 5A). The majority of pathways in the metabolism of aromatic compounds subsystem were enriched in biochar-amended samples (Fig. 5B). Additionally, the benzoate degradation and carbazole degradation clusters were also enriched in biochar-amended samples. Conversely, pathways in the secondary metabolism subsystem were enriched in the controls (Fig. 5B), including cinnamic acid degradation and genes for 4-coumarate-coenzyme A (CoA) ligase 1.

FIG 4.

Significant changes in abundance of carbohydrate pathways as an effect of biochar addition. Row labels on the left indicate the level 2 subsystem classification and labels on the right indicate the level 3 subsystem classification for each microcosm metagenome (columns). Color code is based on the magnitude of change; scale values indicate the log₂-fold change (see scale on the top).

FIG 5.

Significant changes in abundance of different pathways in respiration, metabolism of aromatic compounds, and secondary metabolism as an effect of biochar addition. Row labels on the left indicate the level 2 subsystem classification and labels on the right indicate the level 3 subsystem classification for each microcosm metagenome (columns). Color code is based on the magnitude of change; scale values indicate the log₂-fold change (see scale above each heatmap). (A) Heatmap repesenting pathways in the respiration subsystem. (B) Heatmap representing pathways in secondary metabolism and metabolism of aromatic compounds. Labels in boldface indicate the level 1 subsystem classification.

With respect to the nitrogen metabolism subsystem, pathways involved in allantoin utilization and in denitrification, which included genes for nitrous and nitric oxide reductases, their maturation, and activation proteins as well as genes for the copper-containing nitrite reductase accessory protein, were enriched in biochar-amended samples (Fig. 6A; see Fig. S3). Control samples were enriched in pathways involved in ammonia assimilation, specifically, genes for ammonia transporters and glutamine synthetases (Table S4). Pathways involved in N fixation, specifically, nitrogenase transcriptional regulators, were higher in controls (Fig. 6A; Fig. S3). Control samples were significantly enriched for dissimilatory nitrite reductase; however, it is important to note that the genes belonging to this subsystem were involved in c-type cytochrome and heme d₁ biosynthesis rather than genes encoding the nitrite reductase enzymes (Fig. 6A; Fig. S3). Additionally, the majority of significantly differentially present pathways involved in amino acid degradation/utilization and their derivatives were enriched in the controls (Fig. 6A). The majority of genes related to the cycling of other nutrients, such as phosphorus (P), potassium (K), and iron (Fe) metabolism were enriched in biochar-amended samples (Fig. 6B). Control samples were enriched in pathways involved in P uptake in cyanobacteria and hemin uptake in Gram-positive bacteria. Finally, the pathways in the sulfur metabolism subsystem were enriched in biochar-amended samples (Fig. 6B). Organic sulfur assimilation was enriched in controls: this included pathways such as alkanesulfonate assimilation and utilization of glutathione, which included ABC-type nitrate/sulfonate/bicarbonate transport systems and putative glutathione transporters, respectively.

FIG 6.

Significant changes in abundance of different pathways for nutrient acquistition and metabolism. Row labels on the left indicate the level 2 subsystem classification and labels on the right indicate the level 3 subsystem classification for each microcosm metagenome (columns). Color code is based on the magnitude of change and scale values indicate the log₂-fold change (see scales above each heatmap); labels in boldface indicate the level 1 subsystem classification. (A) Heatmap repesenting pathways in the amino acid degradation and biosynthesis and N metabolism. (B) Heatmap repesenting pathways in phosphorus, potassium, sulfur, and iron metabolism.

DISCUSSION

Composition and stability of microbial community responding to biochar.

The use of metagenomics to probe putative functional changes in the microbiome within microcosms containing oxisol soils under Napier grass cultivation 2 years after the initial addition of biochar revealed a deep level of insight into microbiome responses to this amendment. Shifts in the microbial community composition were much less pronounced than in our previous study (22) where the overall changes in the composition of the microbiome were small though significant. This may reflect the effects of homogenization of the soils by sieving, since there is evidence that different soil fractions support distinct microbial communities (43). Communities may differ between the <2-mm fraction and the large soil aggregates (>2 mm). Conversely, the majority of functional genes/pathways did not exhibit significant differences in abundance between control and biochar-amended metagenomes. It is important to note that the metagenomic snapshots reported here might have missed short-term changes to microbial community composition or functional gene abundances due to biochar amendment. Additionally, the number of rRNA gene sequences identified in the metagenomic data was low compared to the number of gene-encoding sequences. Nonetheless, the taxon shifts within our metagenomic data set reflected the shifts observed in our previous analysis (22).

Consistent with previous studies based on 16S rRNA gene amplicons, we showed that Proteobacteria and Bacteroidetes relative abundances significantly increased in biochar-amended samples and that the relative abundance of Acidobacteria decreased (22, 24, 35, 44). Overall, all classes of Proteobacteria were more abundant in the biochar-amended samples, consistent with our previous study (22) in which we observed that biochar increased Proteobacteria abundance in the oxisol as early as 1 month and up to 1 year after the initial amendment. Notably, the orders Rhizobiales and Burkholderiales were most abundant and higher in biochar-amended samples and have been characterized as N fixers and more broadly as N-cycling generalists, carrying genes for the majority of the assimilatory and dissimilatory N pathways (45). Furthermore, some members of both bacterial orders have the ability to degrade recalcitrant compounds, including some naturally occurring aromatic compounds and organic contaminants such as those used as pesticides, herbicides, or fungicides. For example, members of Burkholderiales have been shown to degrade pentachlorophenols (46) and members of Rhizobiales can degrade compounds such as 4-fluorocatechol and catechols (47). In this study, biochar-amended samples nearly doubled the relative abundance of Bacteroidetes. Bacteroidetes in agricultural soils are less well characterized than their characterization in aquatic ecosystems, with higher abundances suggested as an indicator of good soil quality (48). Flavobacteriia seem to be an important class of Bacteroidetes in soils; their abundance has been observed to be influenced by electric conductivity, pH, and soil Na, Zn, Mg, and Ca (48). The slight increases in soil base cations and pH in biochar-amended samples appear to support the increase of Flavobacteriia in biochar-amended soils. Additionally, members of this class participate in organic matter turnover and the degradation of various aromatic compounds (48, 49).

As a number of families belonging to Actinobacteria have been characterized as a group that plays an important role in the decomposition of plant cell wall polymers and recalcitrant organic matter (50), we expected the increased soil C% associated with the biochar and Napier grass root exudates would favor the Actinobacteria. Unexpectedly, we observed a decrease in the Actinomycetales relative abundance in biochar-amended samples, which was contrary to the findings of previous studies (32, 34, 35, 38, 51). However, previous reports on biochar functionality related to shifts in microbial community composition were often carried out in short-term studies (e.g., less than 1 year) (32, 34, 35, 38, 51). In contrast, studies have reported decreases in their relative abundance at least 2 years after the initial addition of biochar (44, 52). Thus, the observed changes in Actinobacteria abundance are likely related to temporal shifts or driven by changes in organic carbon quality originating from plant litter and/or biochar (53). The decreased abundance of Actinobacteria may suggest a lower C degradation rate that may, in part, explain the lower CO₂ production in the biochar-amended microcosms 2 years after amendment.

Biochar decreases the abundance of assimilatory N pathways but increases denitrification.

Although the observed difference in soil N concentration was not statistically significant, pathways involved in N assimilation were significantly higher in the controls. Conversely, biochar-amended samples contained significantly higher abundances within the denitrification pathway. These findings indicate that the addition of biochar to an oxisol potentially resulted in better retention of N, as soil N% increased over the 2 years since the initial addition of the soil amendment. Compared to our previous study, soil N% measured 2 years after the initial addition of biochar was higher in both controls and biochar-amended soils than the soil N% after 1 year of biochar addition. The increase in soil N% may reflect the higher abundance of N fixation genes and assimilatory N pathways in our controls soils. Furthermore, the relative abundances of N-cycling generalist and denitrifying Proteobacteria, such as the Rhizobiales, Pseudomonadales, and Burkholderiales, were higher in the biochar-amended samples, consistent with our previous findings (22). In addition, the archaeal phylum Thaumarchaeota was lower in biochar-amended samples. This phylum belongs to a population of ammonia oxidizers that is likely a major driver of nitrification and is influenced by soil organic carbon and pH (54, 55). Although, ammonia-oxidizing archaea are not capable of nitrification-denitrification and thus do not contribute to N₂O; their contribution to global N₂O may occur indirectly through the oxidation of nitrogenous compounds that are converted into electron acceptors for denitrifying organisms (56). The decrease in abundance of ammonia oxidation and nitrification pathways and increase in denitrifying Proteobacteria and denitrification-related genes in biochar-amended samples may increase the potential to mitigate N₂O emissions and reduce N losses from soil (56–60).

These observations suggest that the addition of biochar to an oxisol resulted in better retention of inorganic N, as reflected in increased soil N% over the 2 years since biochar addition, as well as increased N availability and the potential to mitigate N₂O emissions, in agreement with studies that have shown increased N bioavailability with biochar amendment (61) and studies which showed that biochar amendment decreased N₂O emissions (17, 62–64). The increased potential for denitrification is also consistent with the observations of the higher abundances of genes encoding soluble cytochromes, specifically, cytochrome c, and respiratory dehydrogenases in biochar-amended samples (65).

Enhanced potential for acquisition for nutrients associated with biochar and other compounds.

The significantly enriched pathways in biochar-amended soils within the carbohydrate subsystem indicated conspicuous responses to plant growth activity, such as the utilization of a number of labile carbon sources, mainly, plant-derived sugars (66). This may be linked with the slightly increased, though nonsignificant, Napier grass yield from the biochar-amended plots. Though the increase in yield was nonsignificant, there may be increases in root exudates that increase labile carbon input into the soil that were not measured in this study (66, 67). Additionally, pathways involved in the metabolism of aromatic compounds were significantly higher in biochar-amended samples, including pathways for the degradation of some naturally occurring aromatics from plants and polycyclic aromatic hydrocarbons (PAHs), likely associated with the biochar (68). This finding contrasts with previous studies that observed decreased degradation of PAHs in biochar-amended soils due to limited bioavailability, such as PAH adsorption onto the biochar surface (69, 70). Interestingly, control samples were enriched in the pathways involved in the degradation of lignin precursors, (i.e., cinnamic acid) (71) and several ring-cleaving enzymes within the 4-hydroxyphenylacetic acid catabolism pathway (72).

In the respiration subsystem, biochar-amended samples were generally enriched in genes for cytochromes, while the pathways or genes enriched in controls were primarily dehydrogenases. The increased abundance of genes encoding cytochromes may be related to the increased abundance of denitrification and metabolism of other nutrients in the biochar-amended samples. Abundant respiration genes in the controls included some associated with methylotrophy, carbon monoxide dehydrogenases, NADH dehydrogenase (quinone), and ATP synthase. With respect to the cycling of P, K, S, and Fe, overall pathways involved in the metabolism of these nutrients were significantly higher in biochar-amended samples, except for organic S assimilation. This may link to the slightly increased concentration of soil base cations in the biochar-amended soil, suggesting that they are bioavailable. The function-level descriptions of these pathways could be broadly categorized as uptake and transport systems and generally included ATP transporter and permeases. Additionally, regarding Fe metabolism, biochar-amended samples showed an increased abundance of genes involved in siderophore synthesis and uptake, which may be a result of increased soil pH, since iron is insoluble at higher pH values (73). Although there are few studies focused on the effects of biochar on Fe and S metabolism, there is evidence to suggest that the redox state of Fe can enhance P, N, and S availability in biochar-amended soils (74, 75). The increased abundance of these genes might also coincide with the synthesis of the cytochromes, hemes, and other electron transport proteins as well as Fe-S cluster enzymes, including some of the dehydrogenases and hydrogenases, as Fe-S clusters proteins are involved in many fundamental processes, including respiration and denitrification (76).

While previous studies have used a single phylogenetic marker to investigate the effects of biochar on the soil microbial community (18, 24, 32, 33, 35), to date, only one has utilized shotgun metagenomics to examine the effects of biochar on the microbial community (42). Our results from the biochar-amended soil metagenomes showed some similarities to those of this previous study of biochar metagenomes (42). The average abundance of genes in our metagenomes at the level 1 subsystem group of carbohydrates, clustering-based subsystems, and amino acids and derivatives accounted for the majority of functional genes across all our metagenomes. Additionally, genes related to iron acquisition and metabolism were more abundant in metagenomes associated with biochar than in controls. However, our main findings contrast with the results reported by Noyce et al. (42). Here, we observed significant differences in functional gene abundances for genes related to N, P, and S cycling. In addition, the biochar-amended soil metagenomes in our study showed increases in abundance for genes related to the metabolism of aromatic intermediates, and genes related to amino acids and derivatives were less abundant than in the control soil metagenomes, in contrast to the previous study (42). It is important to note that the previous study analyzed the microbial community of aged biochar particles and the adjacent soil, whereas our analyses examined the bulk soil with and without biochar in soil microcosms. Additionally, they examined forest soils, which typically have higher microbial diversity than agricultural soils (77).

Conclusions.

These data reveal that the soil microbial community response to biochar is not transient over 2 years and that the shifts observed previously underlie functional adaptations to changes in nutrient availability induced by biochar. Our data showed that biochar increased soil C% and the abundance of genes involved in substrate acquisition and utilization. In agreement with our previous study, biochar enhanced the potential for denitrification, specifically, genes for nitric oxide and nitrous oxide reductases, and decreased the abundance of ammonia oxidizers, which may have large implications for decreasing the emissions of a potent GHG such as N₂O. Biochar increased the abundance of genes involved in the utilization of labile C and aromatic compounds, and the fitted respiration rate between the control and biochar-amended microcosms exhibited a significant difference. Additionally, the cumulative CO₂ emissions from the control soil microcosms were significantly elevated at all time points compared to those from the biochar-amended soil microcosms. This may suggest that the microbial community utilizes the plant- or biochar-associated carbon more efficiently for the production of microbial products and incorporation into microbial biomass (78). However, the effects of biochar on the microbial carbon use efficiency remains experimentally unresolved and outside the scope of this study. Disentangling the direct and indirect effects of biochar on the soil microbial community remains a challenge. Additional samples across time should be examined and coupled with flux data and measures of C use efficiency before robust conclusions can emerge with respect to biochar effects on C and N cycling by the soil microbiome.

MATERIALS AND METHODS

Sample collection.

A field experiment was established in November 2013 on the island of Oahu, HI, USA, at the Poamoho agricultural experimental research station managed by the College of Tropical Agriculture and Human Resources, University of Hawaii Manoa (21°32′30″N, 158°05′15″W). Detailed information on the experimental setup of the field experiment, biochar, and application rates were described in a previous study (22). Briefly, the soil at Poamoho is an acidic oxisol with 44% clay rich in kaolinite and iron oxides with low CEC (NRCS Web Soil Survey). Napier grass yield was determined as the total dry weight normalized by the number of plants before calculating the total dry weight per hectare of land, because the number of plants in each plot varied slightly.

Our previous analysis of samples collected from the same field study during the first year and using 16S rRNA gene amplicon sequencing revealed that biochar had a significant effect on the soil community of the oxisol compared to a mollisol, and the effect of biochar was consistently more pronounced under Napier grass (Pennisetum perpereum var. green bana, a C₄ tropical perennial grass cultivated as a potential biofuel feedstock) cultivation than with the annual cropping system (22). Therefore, for this study, we selected a critical subset and analyzed the oxisol under Napier grass, which is managed as a zero-tillage (i.e., ratoon harvested) system that retains the belowground environment and live root mass during harvest, approximately 2 years after the initial biochar amendment. Soils were collected in November 2015 from four replicate plots for biochar-amended and control soils (collecting and compositing eight samples from randomly selected locations within each plot) prior to harvest and transported on dry ice to the laboratory. Soil chemical properties were determined as previously described (22) and are summarized in Table 1. Soils samples were frozen at −80°C without the addition of any protective agent until ready for further processing.

Experimental setup.

Biochar-amended and control samples collected in November 2015, previously frozen field-moist, were thawed, composited, and then sieved using a 2-mm sieve. Six soil microcosms were set up in by adding 10 g of soil to 150-ml serum bottles, three bottles containing biochar-amended soil and three with control soil. Microcosms were preincubated at 4°C open to the ambient atmosphere for 7 days to allow the soils to equilibrate and then closed with bromobutyl rubber septa before incubation at 23°C. For the determination of cumulative CO₂, 200 μl of headspace from each microcosm was sampled in triplicates using a gas-tight syringe (VICI Precision Sampling, Baton Rouge, LA). Soil microcosms were not continuously aerated. The CO₂ concentrations in the bottle were measured using a gas chromatograph equipped with a flame ionization detector (SRI instruments, Torrance, CA). The headspace was measured after microcosm setup (day 0) and after 2, 4, 6, 8, 10, 12, and 14 days of incubation. Details of the standard curve can be found in the supplemental material. The rate of CO₂ production was calculated from the best-fit line. After 14 days of incubation, soils were stored at −80°C until DNA extraction.

DNA extraction, sequencing, and preprocessing.

Genomic DNA was extracted from 4 g of soil using the DNeasy PowerMax Soil kit (Qiagen Company, Hilden, Germany) as described previously (22), and final DNA concentrations were quantified by a Qubit dsDNA high-sensitivity kit (Thermo Fisher Scientific, Waltham, MA, USA) using the Qubit 3.0 (Thermo Fisher Scientific, Waltham, MA, USA). From each sample, 1 μg of genomic DNA was used for library preparation. Briefly, DNA was first fragmented using the Covaris system (Covaris, Woburn, MA, USA) and ligated with Illumina TruSeq paired-end adaptors. Sequencing was carried out using the 2 × 150 high-output platform on the Illumina NextSeq 500 instrument. The resulting sequencing reads were quality filtered and trimmed to remove the Illumina adaptors using Trimmomatic version 0.36 (79), and paired end reads were interleaved using the interleave-reads.py script from khmer version 2.1.1 (80) before assembly.

Metagenome assembly and gene annotation.

The assembly of metagenomes was carried out using MEGAHIT v1.1.2 (81, 82) and the quality of the assemblies was assessed with QUAST version 3.0 (83). Assembled reads were used as input into the MG-RAST pipeline (84) for downstream processing and annotation. Details can be found in the supplemental material. The best match for each read using a cutoff E value of <1E−7, an alignment length of 25 amino acids, and an amino acid identity of >60% against the SEED (85) genes was recorded, and the number of best-hit reads was taken as a proxy for the abundance of SEED genes and subsystems in each sample. The relative abundances of domains in the metagenomes were estimated based on the best match of amino acid sequences against the RefSeq database (86) using MG-RAST.

Estimating community complexity and determination of differentially present pathways.

The relative abundances of different phyla/classes in each sample were quantified by the number of reads assigned to a taxon using the same cutoffs as described above and normalized by sample size. To examine differences in the abundances of bacteria as a result of biochar amendment, the average abundances from three replicates (biochar-amended and control soils) of phyla that made up less than 1% of the whole community were analyzed separately from those which comprised at least 1% of the whole community. Data sets were subject to Hellinger transformation for Bray-Curtis dissimilarity matrices, and significant differences in the microbial community between treatments were tested with permutational multivariate analysis of variance (PERMANOVA) (87) using the ADONIS function in R’s vegan package (88). Phyla that were significantly differentially present were identified using the paired t test. Differences were considered significant at a P value of <0.05. Estimates of average coverage and sequence diversity for each metagenomic data set were carried out with Nonpareil 3 using default settings (89, 90). Additional details of the metagenomic diversity and coverage are given in the supplemental material.

To identify pathways that were significantly differentially present between the control and biochar-amended samples, the DESeq2 package (91) was employed in RStudio (version 1.0.136). A count table of the functional annotation was generated with the SEED level 3 subsystem, which is similar to a Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway. Each column represented a sample and each element was the number of reads from the sampled assigned to the SEED subsystem. DESeq2 was then used, with the default settings, to estimate the effective library size and variance. The size factor of the metagenomes was then used to normalize the counts prior to the detection of differences between biochar-amended and control samples for each SEED subsystem.

Accession number(s).

Raw sequences were deposited in GenBank under PRJNA497915 and the assembled reads were deposited in MG-RAST under project identifier (ID) mgp83293.