Soil biochar amendment affects the diversity of nosZ transcripts: Implications for N2O formation

Microbial nitrogen transformation processes such as denitrification represent major sources of the potent greenhouse gas nitrous oxide (N2O). Soil biochar amendment has been shown to significantly decrease N2O emissions in various soils. However, the effect of biochar on the structure and function of microbial communities that actively perform nitrogen redox transformations has not been studied in detail yet. To analyse the community composition of actively denitrifying and N2O-reducing microbial communities, we collected RNA samples at different time points from a soil microcosm experiment conducted under denitrifying conditions and performed Illumina amplicon sequencing targeting nirK, typical nosZ and atypical nosZ mRNA transcripts. Within 10 days, biochar significantly increased the diversity of nirK and typical nosZ transcripts and resulted in taxonomic shifts among the typical nosZ-expressing microbial community. Furthermore, biochar addition led to a significant increase in transcript production among microbial species that are specialized on direct N2O reduction from the environment. Our results point towards a potential coupling of biochar-induced N2O emission reduction and an increase in microbial N2O reduction activity among specific groups of typical and atypical N2O reducers. However, experiments with other soils and biochars will be required to verify the transferability of these findings to other soil-biochar systems.


Material and Methods
The RNA extracts analysed in this study were obtained from soil samples collected in a microcosm experiment previously described by Harter, et al. 34 . In the following we briefly describe the soil and biochar used for this soil microcosm experiment and its experimental design. For a detailed description of the experiment please refer to Harter, et al. 34 .
Soil sampling and biochar production. The soil for the soil microcosm experiment was collected from the top 10 cm of an Anthrosol (World Reference Base for Soil Resources, 2014) located at an urban gardening site of the University of Tuebingen, Germany (48° 32′ N, 9° 4′ E). The soil is classified as sandy clay loam (USDA) with a particle size distribution of 49.8% sand, 25.8% silt and 24.4% clay. Biochar was produced by Swiss Biochar Sàrl (Belmont-sur-Lausanne, Switzerland) from green waste by slow pyrolysis (620 °C) according to the standards of the European Biochar Certificate (EBC) (http://www.european-biochar.org). Important physical and chemical properties of soil and biochar are summarized in Table 1. Detailed descriptions of the methods used to determine the physical and chemical properties of soil and biochar can be found in Harter, et al. 34 .
Soil microcosm experiment. Prior to microcosm setup the field moist (water content: 29% w/w) soil and the biochar were passed through a 2 mm sieve and thoroughly homogenized. Soil microcosms were set up in 250 ml glass beakers. Microcosms contained either the field moist equivalent of 100 g dry soil (control microcosms) or the field moist equivalent of 95 g dry soil and 5 g dry biochar resulting in a final biochar content of 5% (biochar microcosms). Control and biochar microcosms were set up in triplicates. After preparation, soil microcosms were homogenized and compacted (10 repetitions of dropping a 125 g hammer from a height of 20 cm onto the soil surface). In order to simulate conventional fertilization, all soil microcosms were fertilized with a NH 4 NO 3 solution at a rate of 332 mg N kg −1 (equivalent to 100 kg N ha −1 estimated based on the soil surface area in the microcosms). The amount of water added with the NH 4 NO 3 solution was calculated to adjust the water filled pore space (WFPS) to 90%. While the experiment was running the soil microcosms were covered with a perforated aluminium foil to allow gas exchange with the ambient atmosphere and to decrease evaporation. Soil microcosms were incubated at a constant temperature of 20 °C and the WFPS was held constant by periodically replenishing the evaporated water. Soil microcosms were incubated for 10 days with sampling taking place after 0, 1, 2, 4, 7 and 10 days of incubation. To allow destructive soil sampling of a set of 6 microcosms (3 control and 3 Scientific RepoRts | 7: 3338 | DOI: 10.1038/s41598-017-03282-y biochar microcosms) at each time point of sampling, we set up 36 soil microcosms in total. During soil sampling the entire soil of the soil microcosm was transferred into a sterile container, homogenized using a spatula and aliquoted for parallel RNA extraction. All samples for RNA extraction contained the equivalent of 2 g dry soil and were directly frozen at −80 °C. For the present sequencing study only RNA extracts obtained from soil samples collected at day 0, 4, and 10 were analysed. RNA extraction and reverse transcription. Total RNA was extracted using the RNA PowerSoil Total RNA Isolation Kit according to the manufacturer protocol (MO BIO Laboratories, Carlsbad, CA, USA). RNA concentration and quality of the resulting extracts were determined spectrophotometrically (NanoDrop 1000, Thermo Scientific, Waltham, MA, USA), fluorometrically (Qubit 2.0 Fluorometer, Life Technologies, Carlsbad, CA, USA), and by gel electrophoresis (Experion Automated Electrophoresis Station, Bio-Rad Laboratories, Hercules, CA, USA). Residual DNA in RNA extracts was digested using the Ambion TURBO DNA-free Kit (Life Technologies, Carlsbad, CA, USA) according to the instructions given by the manufacturer. Successful removal of DNA was confirmed by PCR using primers 27F (5′-AGAGTTTGATCMTGGCTCAG-3′) 39 and PC5 (5′-TACCTTGTTACGACTT-3′) 40 with the following conditions: hot start at 70 °C, 5 min at 95 °C, 35 cycles with 1 min at 95 °C, 1 min at 44 °C and 3 min at 72 °C followed by a final elongation step of 10 min at 72 °C. If the PCR revealed no PCR products, RNA extracts were used for cDNA synthesis. Reverse transcription of pure RNA extracts was performed with the SuperScript III Reverse Transcriptase using random primers according to the manufacturer's protocol (Life Technologies, Carlsbad, CA, USA). cDNA samples were quality checked and quantified using agarose gel electrophoresis and Nanodrop (NanoDrop 1000, Thermo Scientific, Waltham, MA, USA).
Illumina amplicon sequencing. In order to sequence nirK, typical nosZ and atypical nosZ mRNA transcripts we amplified the corresponding fragments from the cDNA samples from the control and biochar microcosms at day 0, 4 and 10 (n = 18) using PCR. PCRs were performed using target-specific primers fused with overhang adapter sequences at the 5′-end to allow the addition of index sequences and sequencing adapters in a second index PCR. Transcripts of nirK genes were amplified using primer F1aCu (5′-ATCATGGTSCTGCCGCG-3′) and R3Cu (5′-GCCTCGATCAGRTTGTGGTT-3′) 41 . For typical nosZ gene transcripts primer nosZ2F (5′-CGCRACGGCAASAAGGTSMSSGT-3′) and nosZ2R (5′-CAKRTGCAKSGCRTGGCAGAA-3′) 42 were used. Transcripts of atypical nosZ genes were amplified with the primers nosZ-II-F (5′-CTNGGNCCNYTKCAYAC-3′) and nosZ-II-R (5′-GCNGARCARAANTCBGTRC-3′) developed by Jones, et al. 26 . PCRs were performed with the FastStart High Fidelity PCR system (Roche Diagnostics, Rotkreuz, Switzerland) using the thermal protocols and reaction mixtures described in Table S1 in the supplementary information. The produced amplicons were purified using AMPure XP beads (Beckman Coulter, Brea, CA, USA) at a ratio of 1: Sequence analysis. Quality control of raw nirK, typical nosZ and atypical nosZ read pairs was performed using Cutadapt v1.9 43 and USEARCH v8.1.1812 44,45 . At first target-specific primers were detected and trimmed using Cutadapt. Afterwards, paired-end nirK and typical nosZ reads were merged using the fastq_mergepairs algorithm implemented in USEARCH. Due to the large size of atypical nosZ amplicons (>600 bp) paired-end reads could not be merged. Instead, the forward read was trimmed to 200 bp to ensure the removal of bases with low phred scores and to obtain sequences of equal length. Trimming was performed with fastx_truncate using USEARCH. Merged (nirK and typical nosZ) and trimmed (atypical nosZ) sequences were quality filtered using the USEARCH fastq_filter function with a maximum error rate threshold of 1%. Chimeric sequences were de novo detected and removed using the uchime_denovo algorithm implemented in USEARCH 46 . All quality filtered non-chimeric nirK, typical nosZ and atypical nosZ sequences were mapped against the NCBI Reference Sequence (RefSeq) protein database (release 66). Mapping was performed using DIAMOND (v0.7.9.58) 47 in blastx mode with a minimum protein sequence identity cut-off of 70% and an e-value cut-off of 10 −10 . The top 50 database hits of each sequence were used for further analysis with MEGAN6 Ultimate Edition 48,49 . For taxonomic placement the Lowest Common Ancestor (LCA) algorithm of MEGAN was applied 50 . The LCA analysis parameters 'Top Percent' and 'Min Support' were set to 0.5% (all hits within the top 0.5% of the best bit score are considered) and 10 (taxa need to obtain a minimum of 10 reads to be considered), respectively. Shannon-Weaver and Simpson-Reciprocal diversity indices were computed on the species level in MEGAN.
Absolute sequence counts and relative sequence abundances of all assigned taxa were exported from MEGAN for further analyses. Relative sequence abundances were calculated based on all sequences that matched database entries in the RefSeq protein database.
Typical and atypical nosZ transcript copy numbers of individual species at day 4 were calculated by multiplying absolute transcript copy numbers determined by qPCR in a previous study conducted by Harter,et al. 34 with the corresponding relative sequence abundances of individual nosZ-expressing species obtained in this study. This was possible because sequencing and qPCRs were performed on the same RNA samples with the same sets of primers in the present and the previous study by Harter, et al. 34 . Individual species with an average relative sequence abundance below 1%, were not considered.
Figures 1, 2 and 3 were generated with GraPhlAn 51 using the LCA-based taxonomy and the corresponding relative sequence abundance information from MEGAN. According to the basic principle of the LCA algorithm, sequences that are conserved among different species (e.g. as consequence of horizontal gene transfer) will only be assigned to taxa of higher rank 50 . Nonetheless, it is very difficult to directly prove that a given nirK or nosZ gene (typical and atypical) appears in a specific microbial taxon. Thus, whenever we mention specific taxa in the results and discussion we refer to a taxonomic group of microorganisms that contains a nirK or nosZ gene closely related to the nirK or nosZ gene of the respective taxa. (Tables 2 and 3) and the relative sequence abundance of each assigned microbial species (Figs 1,2,3), were statistically evaluated at each time point of sampling (0, 4, and 10 days) using t-tests. In addition, we used t-tests to determine significant biochar effects on the absolute nosZ transcript copy numbers of individual species at day 4 ( Fig. 4). Differences in diversity indices among the average (all control and biochar microcosms at all time points of sampling, n = 18) nirK, typical nosZ and atypical nosZ-expressing microbial communities (Tables 2 and 3) were determined using one-way ANOVA with Tukey's HSD post-hoc test. T-tests and one-way ANOVAs were carried out in SAS (SAS 9.2, SAS Institute, Cary, NC, USA) using PROC TTEST and PROC GLM, respectively. Dissimilarities among nirK, typical nosZ and atypical nosZ-expressing microbial communities in control and biochar microcosms and over time were statistically evaluated using permutational multivariate analysis of variance (PERMANOVA) with biochar addition and time as main factors and 10 5 permutations (Table 4). PERMANOVAs were performed based on Bray-Curtis dissimilarity matrices using the adonis function implemented in the R package vegan 52 .

Statistical analysis. The effects of biochar addition on diversity indices
Data availability. Illumina
Quality processing resulted in 375,665 merged high quality nirK sequences of which 374,783 (20,821 ± 8,672 per sample) matched entries in the RefSeq database (protein sequence identity cut-off: 70%, e-value cut-off: 10 −10 ). Merged high quality nirK sequences had an average sequence length of 436 bp.
Quality processing of atypical nosZ sequences led to 288,167 high quality sequences with a sequence length of 200 bp. 166,786 sequences (9266 ± 2243 per sample) matched database entries (protein sequence identity cut-off: 70%, e-value cut-off: 10 −10 ).
Diversity of the microbial communities expressing nirK, typical nosZ, and atypical nosZ. In order to investigate the impact of soil biochar amendment on the diversity of expressed nirK and nosZ genes, we computed Shannon-Weaver and Simpson-Reciprocal diversity indices for all samples.
Shannon-Weaver indices ranged from 0.99 to 3.49 and were generally highest for atypical nosZ and lowest for nirK (Table 2). On average (all control and biochar microcosms at all sampling time points, n = 18), the nirK-expressing community had a value of 1.56. In comparison with nirK, typical and atypical nosZ-expressing communities showed significantly higher values of 2.74 (p < 0.001) and 3.20 (p < 0.001), respectively. Atypical Large, filled, circles depict microbial taxa with average relative sequence abundances above 1%. Light colour shaded areas indicate orders with a relative sequence abundance above 1%. Dark colour shaded boxes with numbers (1-5) represent species with a relative sequence abundance above 1%. External rings show the relative sequence abundance of the corresponding species. Ring 1 shows the average relative sequence abundance in percent (all microcosms at all time points of sampling, n = 18). Rings 2-7 show the relative sequence abundances in control and biochar microcosms at day 0, 4, and 10 as circular heatmaps (n = 3).
nosZ had a significantly higher average Shannon-Weaver index than typical nosZ (p < 0.001). Biochar addition altered the Shannon-Weaver index at all time points of sampling. For nirK and typical nosZ a biochar-induced increase in diversity was observed ( Table 2). Atypical nosZ diversity slightly decreased in the presence of biochar. While this slight decrease was not statistically significant at any time point of sampling, the values for nirK at day 10 (p = 0.002) and typical nosZ at day 4 (p = 0.014) were significantly higher in biochar compared to control microcosms.
The Simpson-Reciprocal diversity index showed a very similar pattern and ranged from 1.53 to 6.97 (Table 3). With an average value of 5.21, the highest diversity was determined for the atypical nosZ-expressing microbial community. The average value for typical nosZ was slightly, but not significantly, lower (4.72). However, with a value of 2.27 the nirK-expressing community had a significantly lower average diversity compared to the microbial communities producing typical (p < 0.001) and atypical (p < 0.001) nosZ transcript. Biochar addition significantly increased the Simpson-Reciprocal diversity index of typical nosZ transcripts at day 4 (p = 0.016) and 10 (p = 0.049) ( Table 3). In addition, significantly higher values in biochar compared to control microcosms were also determined for nirK at day 10 (p = 0.003). For atypical nosZ soil biochar amendment resulted in significantly lower values at day 4 (p = 0.012).
In order to determine if biochar addition or time caused significant dissimilarities among nirK, typical nosZ and atypical nosZ-expressing microbial communities, we performed two-way PERMANOVAs. As shown in Table 4, PERMANOVAs revealed a significant time effect for nirK (p < 0.001), typical nosZ (p < 0.001) and atypical nosZ-expressing microbial communities (p < 0.001) and a significant biochar effect for the typical nosZ-expressing microbial community (p = 0.017). Different classes are shown with specific branch colours. Large, filled, circles depict taxa with average relative sequence abundances above 1%. Light colour shaded areas indicate orders with a relative sequence abundance above 1%. Dark colour shaded boxes with numbers (1-8) represent species with a relative sequence abundance above 1%. External rings show the relative sequence abundance of the corresponding species. Ring 1 shows the average relative sequence abundance in percent (all microcosms at all time points of sampling, n = 18). Rings 2-7 show the relative sequence abundances in control and biochar microcosms at day 0, 4, and 10 as circular heatmaps (n = 3). Different phyla are shown with specific branch colours. Large, filled circles depict taxa with average relative sequence abundances above 1%. Light colour shaded areas indicate orders with a relative sequence abundance above 1%. Dark colour shaded boxes with numbers (1-15) represent species with a relative sequence abundance above 1%. External rings show the relative sequence abundance of the corresponding species. Ring 1 shows the average relative sequence abundance in percent (all microcosms at all time points of sampling, n = 18). Rings 2-7 show the relative sequence abundances in control and biochar microcosms at day 0, 4, and 10 as circular heatmaps (n = 3).  Table 3. Simpson-Reciprocal diversity index for nirK, typical nosZ, and atypical nosZ, in control and biochar microcosms after 0, 4, and 10 days. "average" shows average values of all microcosms at all time points of sampling. Values represent means ± standard deviation (individual samples: n = 3, average: n = 18). Significant differences between individual control and biochar samples are shown in the column "sig. " (ns = not significant, *p < 0.05, **p < 0.01). Significant differences (p < 0.05) between the average values of nirK, typical nosZ and atypical nosZ are indicated by superscripted letters next to the corresponding values.   Table 4. Results from Bray-Curtis dissimilarity based two-way PERMANOVAs. The table shows pseudo F-statistics (F), explained variances (R 2 ), and the level of significance (p) for the two main effects "biochar" and "time" and their interaction "biochar*time". Significant effects indicated by p-values below 0.05 are shown in bold font.
Scientific RepoRts | 7: 3338 | DOI:10.1038/s41598-017-03282-y sp. BNC1 (2.05%) showing a relative sequence abundance above 1% (Fig. 1). A complete list with the relative sequence abundances of all nirK-expressing microbial species identified in this study can be found in Table S2 in the supplementary information.
The taxonomic composition and distribution of nirK transcripts slightly differed in control and biochar microcosms at all time points of sampling (0, 4, and 10 days) (Fig. 1). However, biochar related changes in relative sequence abundance of specific species did not follow a clear trend. Accordingly, significantly higher and lower relative sequence abundances in biochar compared to control microcosms were only detected for Mesorhizobium australicum at day 0 (control: 3.0%, biochar: 6.3%, p = 0.048) and Mesorhizobium ciceri at day 10 (control: 44.0%, biochar: 33.6%, p = 0.011), respectively (Fig. 1).
The impact of soil biochar amendment on the taxonomic composition and distribution of atypical nosZ transcripts was minor. Although differences in relative sequence abundance between control and biochar microcosms were observed for several species at all time points of sampling (0, 4 and 10 days), most of them were not statistically significant. Significantly higher relative sequence abundances in biochar compared to control microcosms were determined for Ignavibacterium album at day 0 (control: 0%, biochar: 0.3%, p = 0.035) and 10 (control: 0.7%, biochar: 1.5%, p = 0.039) and for Melioribacter roseus at day 4 (control: 37.2%, biochar: 43.7%, p = 0.015). A biochar-induced statistically significant decrease was not observed.

Discussion
Highest N 2 O emission rates usually occur after heavy rain fall when high amounts of organic carbon and mineral nitrogen are mobilized and oxygen availability decreases due to high water filled pore spaces (WFPS) 17,18 . These conditions promote the growth and activity of microbial strains capable of performing anaerobic nitrogen transformation processes such as denitrification 19,20 . In order to simulate such conditions, we set up a soil microcosm experiment at a constant WFPS of 90% to which we added a NH 4 NO 3 solution at concentrations that reflect common agricultural application rates. Although the amount of N 2 O emitted under these conditions can account for large fractions of total annual emissions, a high WFPS is usually occurring only a few days after rain events in most soils 17,18 . Accordingly, our experiment was designed to mimic the frequently observed high N 2 O emissions occurring in response to soil water saturation over a short time period (10 days of incubation, 6 time points of sampling).
Harter, et al. 34 described and discussed the impact of biochar addition on nitrogen transformation rates, absolute gene and transcript copy numbers of denitrification marker genes, and the formation and release of N 2 O and N 2 . According to Harter, et al. 34 , biochar addition significantly decreased N 2 O emissions and increased transcript copy numbers of specific functional denitrification genes. While transcripts of the denitrification genes napA, narG and nirS did not show significant biochar effects or were subject to a significant interaction effect (biochar*time), nirK, typical nosZ and atypical nosZ transcript copy numbers were significantly increased by biochar addition. Thus, the current Illumina amplicon sequencing study described here focuses on biochar-induced changes in diversity and relative abundance among the microbial communities actively expressing nirK, typical nosZ and atypical nosZ transcripts.
According to both diversity indices (Shannon-Weaver and Simpson-Reciprocal) the nirK-expressing microbial community had a significantly lower diversity than the microbial communities producing typical nosZ and atypical nosZ transcripts. A lower diversity of the nirK-harbouring community compared to the microbial community carrying typical nosZ has also been reported by Palmer and Horn 53 . The microbial community that actively expressed atypical nosZ genes was significantly more diverse (Shannon-Weaver diversity index) than the community expressing typical nosZ genes. These findings confirm the results of Jones, et al. 29 who determined an up to 3-fold higher Faith's phylogenetic diversity (PD) for atypical nosZ compared to typical nosZ genes. In accordance with our results, Jones and colleagues also demonstrated a significant relationship between the relative abundance and the phylogenetic diversity of the nosZ community and the ability of the soil microbial community to reduce N 2 O. Using network and co-occurrence analyses, Jones, et al. 29 found that functional groups that were identified as significant indicators of reduced soil N 2 O emissions were dominated by atypical nosZ communities. Notably, Jones and colleagues found specific atypical nosZ groups in high abundance in soils for which the lowest N 2 O emissions were observed. These atypical nosZ lineages lacked either nir genes but seemed to be critical for the N 2 O sink capacity of the respective soil. Interestingly, the observation made by Jones, et al. 29 with respect to nirK and nosZ gene abundance and diversity are very similar to what we report here for a biochar-amended soil and nirK and nosZ transcript abundance and diversity under conditions of reduced N 2 O emissions.
The taxonomic compositions of the microbial communities expressing nirK, typical nosZ, and atypical nosZ described here, were in good agreement with other sequencing studies performed on soil. In accordance with the findings of Bremer, et al. 54 and Henry, et al. 55 , the nirK-expressing community was largely dominated by the order Rhizobiales and the genus Mesorhizobium. The typical nosZ-expressing soil community was dominated by Alphaand Betaproteobacteria and species affiliated with the genera Oligotropha, Bradyrhizobium, Hyphomicrobium, Ensifer, and Azoarcus. The importance of these genera within the typical nosZ gene-carrying community, as well as the dominance of microbial taxa that belong to the classes Alpha-and Betaproteobacteria has been reported in several previous soil studies 26,28,29,33,53,56 . While all nirK and typical nosZ transcripts were produced by microbial taxa affiliated with the phylum Proteobacteria, the atypical nosZ-expressing community comprised a board variety of different microbial species affiliated with several phyla including Bacteroidetes, Chlorobi, Chloroflexi, and Firmicutes. These results support the findings of Jones, et al. 26 and Sanford,et al. 27 who showed that N 2 O reduction is a widespread functional trait among soil microbes. Furthermore, the determined species reflected the taxonomic composition of atypical nosZ-carrying microbial species identified in similar sequencing studies and a recent soil metagenome analysis 28,29,33 .
Soil biochar amendment significantly increased the diversity indices computed for nirK (day 10) and typical nosZ (day 4 and 10). Higher diversity indices in biochar-amended compared to biochar-free soils have been reported frequently in studies targeting 16S rRNA genes 31,35,57 . The reason for the biochar-induced increase in microbial diversity is mostly unknown. However, as mentioned in other studies, it seems likely that organic compounds attached to the biochar particles 58,59 , the high pH of biochar-amended soils 35 , or other biochar properties (surface area, pore space) promoted the formation of specific niches, which supported the growth and activity of a diverse range of taxa 33,35,57,60 .
PERMANOVA analyses indicated that the composition of nirK, typical nosZ and atypical nosZ-expressing microbial communities were significantly altered over the whole duration of the experiment (p < 0.001) and that the composition of typical nosZ transcripts was significantly affected by biochar addition (p < 0.017). Among the typical nosZ-expressing community biochar addition resulted in a significant increase in relative abundance of Ensifer meliloti (day 4), Azoarcus sp. BH72 (day 4 and 10), Bradyrhizobium diazoefficiens (day 10), and Ensifer fredii (day 10). Interestingly all of these microbial species are primarily known for their ability to perform nitrogen fixation and usually form a symbiotic relationship with legumes or grasses [61][62][63][64] . Except for Azoarcus sp. BH72, which does not contain nirK or nirS genes, all species carry the full set of denitrification genes (napA/narG, nirK/nirS, norB, nosZ) 65 . Furthermore, it is known that many Rhizobia species are able to denitrify in their free-living states as well as in association with legume root nodules 66 . Bacterial strains affiliated with the genera Ensifer, Azoarcus, and Bradyrhizobium are ubiquitous in soils and have been identified in numerous 16S rRNA and typical nosZ gene-based sequencing studies in natural and biochar-amended soils 33,35,53,67 .
In order to further our understanding of the effects biochar has on microbial N 2 O reduction, we calculated the absolute numbers of typical and atypical transcript copies produced by individual species based on the qPCR data from Harter, et al. 34 and the relative transcript abundances determined in the present sequencing study. Species-specific transcript expression was specifically determined at day 4, because biochar most significantly affected total transcript copy numbers and N 2 O emission rates at this day 34 .
Biochar addition to soil significantly increased the calculated production of nosZ transcripts in 5 typical and 5 atypical nosZ-carrying species. Typical nosZ transcript production was enhanced in Oligotropha carboxidovorans, Bradyrhizobium diazoefficiens, Ensifer meliloti, Ensifer fredii, and Azoarcus sp. BH72. An increased transcription of atypical nosZ genes in biochar microcosms was determined for Melioribacter roseus, Pedobacter saltans, Belliella baltica, Flavobacteriaceae bacterium 3519-10, and Runella slithyformis. The calculated production of typical and atypical nosZ transcripts in all other relevant microbial species we identified based on the sequencing of typical and atypical nosZ transcripts was not significantly affected by biochar addition. Recent studies revealed that well-studied strains of some of these species (Ensifer meliloti 1021 and Bradyrhizobium diazoefficiens USDA 110) are able to grow with externally supplied N 2 O as sole electron acceptor 68,69 . Thus, these strains take up N 2 O from the environment and further reduce it to N 2 . This is in contrast to other denitrifying strains such as Pseudomonas aeruginosa PAO1 which cannot grow on exogenous N 2 O as sole electron acceptor 70 . In addition to these species that are presumably able to physiologically specialize on N 2 O reduction many other species that had a significantly higher nosZ transcript production in biochar microcosms such as Azoarcus sp. BH72, Flavobacteriaceae bacterium 3519-10, Belliella baltica, Melioribacter roseus, Pedobacter saltans, and Runella slithyformis lack at least one of the functional genes encoding nitrate, nitrite, and nitric oxide reductases 65 . Thus these species are forced to perform a truncated version of the denitrification pathway by taking up denitrification intermediates (e.g. N 2 O) from the environment that have been produced by other denitrifiers 71 . Taken together these findings suggest that soil biochar amendment significantly increased nosZ transcript expression in microbial species capable of reducing exogenous N 2 O from the environment.
Our results are in good agreement with the findings from a recent DNA-based sequencing study in which biochar addition significantly increased relative sequence abundances of microbial species capable to use exogenous N 2 O as electron acceptor directly from the environment 33 . The cause for the increase in nosZ expression in these specific species might be due to the higher quantities of entrapped N 2 O in biochar microcosms 34 that can serve as electron acceptor and thus lead to an increase in their transcriptional activity. Based on the high availability of N 2 O, species specialized on N 2 O reduction that are able to reduce exogenous N 2 O from the environment might gain a competitive advantage over species performing the full denitrification pathway considering the thermodynamic advantages associated with directly reducing N 2 O to N 2 72 . However, although N 2 O entrapment in biochar-amended soils might be an important stimulant for increased nosZ gene expression in exogenous N 2 O reducers, we cannot exclude that other factors that are directly related to the physicochemical properties of biochar might have also contributed to the observed increase in nosZ transcription among the identified taxa. As reported in other studies several properties of biochar have the potential to significantly alter microbial communities by directly affecting microbial growth and activity of microorganisms (e.g. pH, biochar's redox activity, nutrient sorption) 60,[73][74][75][76] . In our study biochar's high pH might have stimulated alkaliphilic Ensifer meliloti and Ensifer fredii strains 61,77 . Furthermore, it is possible that the high aromaticity of biochar or organic compounds that were attached to the biochar particles promoted Bradyrhizobium diazoefficiens strains that are able to degrade aromatic substances 78 .
Independent of the exact mechanism causing higher transcript production in these species, a higher N 2 O reduction activity of microbes that are potentially capable to reduce exogenous N 2 O from the environment will result in a decrease of the N 2 O/(N 2 O + N 2 ) ratio. Thus, together with the findings from our previous study 34 , the decreased N 2 O emissions from biochar-amended soils are most likely caused by N 2 O entrapment resulting in retention and increased availability of N 2 O for microbial reduction in the water-filled pore space. The lower N 2 O/ (N 2 O + N 2 ) ratio is then caused by a higher microbial N 2 O reduction activity of nosZ-containing microorganisms able to directly reduce exogenous N 2 O from the environment.
In conclusion, this soil microcosm study showed that soil biochar amendment can lead to dynamic changes in the diversity of active denitrifier populations over time. Sequencing of nirK, typical nosZ and atypical nosZ transcripts revealed a significant biochar-induced increase in nirK and typical nosZ diversity and a higher relative sequence abundance of typical nosZ-expressing Rhizobia species. Furthermore, biochar addition led to a significantly higher calculated typical and atypical nosZ transcript expression among microbial species that are capable of directly reducing exogenous N 2 O from the environment. Our results suggest that the increased activity of these specific microbial species might be responsible for the observed dynamic changes in N 2 O/(N 2 O + N 2 ) ratios in biochar vs control microcosms. We relate the observed increase in soil N 2 O sink capacity to the higher activity of specific nosZ gene taxa in response to N 2 O entrapment in the water-filled pore spaces of the biochar-amended soil. These findings further improve our understanding of the mechanisms responsible for biochar-induced N 2 O emission mitigation. However, it has to be taken into account that species-specific transcript expression was calculated based on qPCR and sequencing data collected from a short-term experiment performed in plant-free, fertilized soil microcosms with a high water content. Hence, generalizations and extrapolation to field scales and extended time periods should be done cautiously until the findings from the present study have been evaluated using different biochars and soils at spatial and temporal scales that more closely translate to field conditions.