Land use in the Prairie Pothole Region influences the soil bacterial community composition and relative abundance of nitrogen cycling genes

Abstract The undulating topography of Prairie Pothole Region of North America creates spatial and temporal variability in soil moisture and nutrient levels, affecting microbial community processes and greenhouse gas emissions. By identifying differences in soil bacterial and archaeal community composition and the abundance of nitrogen cycling genes in permanent cover versus annual crop land over two growing seasons (2017 and 2018), we were able to assess the effects of topography and land use on the functional capacity of the soil microbiome. Permanent grassland cover was associated with higher bacterial diversity in upland positions and lower diversity in low-lying depressions. Bacterial community composition was also significantly different between cultivated and permanent cover at all points along the topographic slope, with the largest effects seen in the footslope and backslope positions. Compared to permanent cover, soil from annual cropland had consistently more abundant nitrifiers, including Nitrospira in the toeslope and backslope, and Nitrososphaeraceae in the shoulder and knoll samples while soils from permanent cover had a greater abundance of several Alphaproteobacteria from Rhodospirillales and Hyphomicrobiaceae across multiple upland positions. Upland soils from annual cropland also had consistently higher abundance of both bacterial and archaeal ammonia oxidizing (amoA) genes and a higher ratio of nirK:nirS genes compared to those from permanent cover. These differences in microbial community composition were associated with higher N2O and CO2 emissions in upland soils in annual cropland; however, there were no differences in GHG emissions between the two systems in low-lying positions.


Introduction
The North American Prairie Pothole Region (PPR), which covers approximately 800 000 km 2 across western Canada and the northern United States, is a landscape characterized by an undulating topography populated by thousands of small wetland depressions (Tangen et al. 2015).Economic incentives have increased cultivation in this area, resulting in 40%-70% of the PPR in Canada being converted to cultivated land over the last century (Lawley 2019).This land use change is not without consequences as intensively managed land requires greater chemical inputs including fertilizer, pesticides, and herbicides that can affect surrounding wildlife (Main et al. 2014;McMurry et al. 2016) as well as increase greenhouse gas (GHG) emissions (Dunmola et al. 2010;Tangen et al. 2015).
The topography of the PPR includes both upland and lowland areas of varying soil moisture levels, with the topographical depressions also subject to fluctuating hydro-periods and O 2 concentrations (Gleason et al. 2009).When comparing natural wetland areas to annually cropped lands, Hayashi et al. (2003) showed that the dense nesting cover at the St. Denis National Wildlife Area (SDNWA) site reduced water input to wetlands by trapping more snow than the adjacent cropland, increasing water infiltration into frozen soil and preventing runoff that is typically observed at spring snowmelt on annually cropped lands at the site.These significant moisture events have been shown to have both short-term and longterm effects on soil microbial processes and the relative abundance of nitrification and denitrification activity (Ma et al. 2008;Banerjee et al. 2016).Both soil moisture and inorganic N have been shown to be higher in low-lying depressions compared to upland areas (Ashiq et al. 2021) and this combination of high soil moisture, C and N availability significantly impacts microbial community composition and GHG emissions (Dunmola et al. 2010).
GHG fluxes in the PPR, including N 2 O, CO 2 , and CH 4 , are influenced by both the variable topography and land management (Tangen et al. 2015).While the majority of soil N 2 O production has been attributed to denitrification (Liang et al. 2020), cultivation introduces large amounts of synthetic nitrogen (N) fertilizer, a primary driver of higher nitrification and nitrifier denitrification related N 2 O emissions (Pennock et al. 2010;Rochette et al. 2018).Compared to permanent cover, cultivated annual croplands have also been associated with lower soil organic carbon (SOC) as well as changes in SOC composition (Nelson et al. 2008;Gleason et al. 2009).These differences may influence GHG production directly by limiting electrons available for denitrifying enzymes (Giles et al. 2012) and decreasing soil respiration (Tangen et al. 2015) or indirectly via changes in microbial community composition (Giles et al. 2012).
The effects of land use change on the composition and function of the microbial community in the soil have been previously documented in perennial versus annual systems, with increased N 2 O production in annual systems associated with changes in the relative abundance of nitrifying and denitrifying bacteria (Ma et al. 2011).The abundance and diversity of various genes from the N cycling pathways have been previously shown to predict the biological potential for N 2 O emissions from soil microbial communities (Braker and Conrad 2011).During nitrification, ammonia monooxygenases in both bacteria (BamoA) and archaea (AamoA) oxidize ammonia to NH 2 OH, which is subsequently reduced to nitrite (NO 2 − ) by hydroxylamine oxidoreductase (hao).Nitrite reductases, including nirK and nirS, represent the second step in the denitrification pathway reducing NO 2 − to NO, which is then further reduced to N 2 O by nitric oxide reductases (NOR).Bacteria encoding nitrous oxide reductases (nosZI and nosZII) then convert N 2 O to N 2 , and the balance of NOR and nitrous oxide reductase (NOS) activities is influenced by multiple biotic and abiotic factors (Highton et al. 2020).
This study compares managed permanent cover fields to annual cropland to evaluate soil GHG fluxes, microbial composition, and functional attributes along a topographic gradient in the PPR of the Canadian Prairies.Soils were collected over two growing seasons and the bacterial community characterized by 16S rRNA gene amplicon sequencing along with the enumeration of genes involved in both nitrification and denitrification pathways.We hypothesized that land use intensification would result in changes to the structure of the soil microbiome as well as the relative abundance of N cycling genes.By characterizing these differences, we were able to examine the relationship between changes in soil bacterial community composition, N cycling functional attributes, and GHG fluxes.

General site description
The SDNWA is located approximately 40 km east of Saskatoon, Saskatchewan in the Parkland Ecozone and the Dark Brown soil climatic zone.Hummocky terrain is dominant and the soils are mostly medium textured, Dark Brown Chernozems and Orthic Regosols (Miller et al. 1985).The permanent cover location is vegetated by a dense nesting cover consisting smooth brome grass (Bromus inermis) and alfalfa (Medicago sativa).The cropland is leased to a local farmer and was previously cropped to canola (Brassica napus) in 2015 and wheat (Triticum aestivum) in 2016.In 2017, the annual cropland field was planted to flax (Linum usitatissimum) and fertilized with 90 kg ha −1 N as urea at seeding.In 2018, the cropland field was planted to barley (Hordeum vulgare) and with 95 kg ha −1 N as urea at the time of seeding.Both fields were dominated by volunteer vegetation in the area adjacent to the pond edge where soils from position 1 were collected (Fig. 1).In both years, the first sampling event occurred prior to seeding of the crop and the second sampling event occurred after crop emergence.A detailed site description, including hydrological data, is available in Bam et al. (2019).

Sample collection
Sampling transects were established in the permanent cover, adjacent to pond 120 on the southwest side (52.21072, −106.08295) with the lowest elevation at 553.98 m and the highest at 555.55 m for a total elevation gain of 1.57 m.In the annual cropland, sampling transects were established on the west side of Pond 60 (52.20923, −106.10280) with the lowest elevation at 553.70 m and the highest at 555.47 m for a total elevation gain of 1.77 m.For both permanent cover and cultivated fields, soil samples were collected from three parallel transects spaced 5 m apart at six positions spanning the topographic gradient of one such wetland area.Starting immediately adjacent to the edge of the accumulated water in the low-lying zone (position 1), samples were collected 5 m apart going up the slope with positions 2-6 located in the toeslope, footslope, backslope, shoulder, and knoll regions, respectively (Fig. 1).For each soil sample, six (0-10 cm) cores were collected using a Backsaver™ soil probe (3.175cm diameter; JMC Soil Samplers, IA, USA) and stored at 4 • C. Within 24 h, samples were mixed by hand and a aliquot of each composite sample was stored at −80

Soil gas flux measurements
Soil gas fluxes were measured using vented soil chambers (12 cm in diameter and 13.5 cm in height) embedded 3 cm into the soil as described previously (Lemke et al. 1998).Briefly, on each sampling date, gas samples were drawn from the headspace at the beginning of the collection period and after 1 h using a disposable 30 mL polypropylene syringe and injecting the full volume into a pre-evacuated 22 mL vacutainer.Samples were collected mid-day between 10:00 am and 2:30 pm, beginning with the permanent cover followed by the cropland.Atmosphere (check) gas samples were taken immediately prior to gas sampling at each site on every sampling date.The concentrations of CH 4 , CO 2 , and N 2 O in the gas samples was determined with a gas chromatograph equipped with a 63Ni electron capture detector.

Sample processing
Soil samples were thoroughly mixed and sieved using a 2 mm brass sieve.Aliquots of 0.25 g were used for total genomic DNA extraction using the DNeasy Power Soil kit (Qiagen, Germany) according to the manufacturer's instructions, and total genomic DNA was eluted in 50 μL of Tris-EDTA.Extracted gDNA concentrations ranged from 10.5  and 2018 (May 7, May 28, June 25, July 18, and August 16) growing seasons.to 138.6 ng/μL and are available in Table S1, and all samples were then diluted to 10 ng/μL in water.For the amplification of the 16S rRNA gene, 20 ng of genomic DNA was used as template for 30 cycles of amplification of the V3-V4 region of 16S rRNA using primers 515f (GTGYCAGCMGCCGCGGTAA) and 926R (CCGYCAATTYMTTTRAGTTT).Each 25 μL amplification reaction contained 1X PCR Buffer (Thermo Fisher Scientific, Waltham, MA, USA), 2.5 mmol/L MgCl 2 , 0.5 mmol/L dNTPs, 0.4 μmol/L of each primer, and 1 U of Platinum Taq polymerase (Thermo Fisher Scientific, Waltham, MA, USA).Reactions were purified using 0.8:1 ratio of Nucleomag SPRI beads (Macherey-Nagel, Düren, Germany) to PCR reaction volume.The amplicons were prepared and sequenced using NexteraXT (Illumina, San Diego, CA, USA) library preparation as per the manufacturer's instructions and 600 cycles of Illumina Miseq V3 chemistry (Illumina, San Diego, CA, USA).

Quantification of nitrogen cycling genes
Quantitative PCR assays were used to enumerate gene copies for nirS, nirK, nosZI, nosZII, BamoA, and AamoA genes in the soil samples.The primer sequences and reaction conditions used for each assay are outlined in Table S2.Each 25 μL reaction contained 1X PowerUp SYBR Green Master Mix (Applied Biosystems, Waltham, MA, USA), 0.3 μmol/L of each primer and either 20 or 30 ng of genomic DNA, depending on the assay (Table S2).Plasmids containing the cloned target sequence for each gene were used as standards (102-108 gene copies per reaction) to calculate the number of gene copies per gram of soil and standard curve descriptors are included in Fig. S1.Correlation of gene copy abundances with soil gas fluxes and gravimetric moisture were tested using Spearman's rank correlation (ρ) with the "stats" package in R.

Statistical methods
Soil gas emission data was analyzed by multivariate analysis of variance with false discovery rate correction using the "stats" package in R. Microbial community alpha and beta diversity statistics were calculated after rarefying to the smallest library size (8000 reads/sample).Analysis of the rarefaction curves indicated sufficient sequencing depth was achieved for alpha and beta diversity analyses (Fig. S2).A Bray-Curtis dissimilarity matrix was calculated using the "phyloseq" package (McMurdie and Holmes 2013) and significance was determined by permutational multivariate analysis of variance (PERMANOVA) using the "vegan" (Oksanen et al. 2007) package in R. The number of observed ASVs, Simpson evenness, and Shannon diversity statistics were also calculated using "phyloseq" with significance determined by the Kruskal-Wallis test (α = 0.05) using the "stats" package in R. Correlation of changes in microbial community diversity (Bray-Curtis dissimilarity) to soil gas fluxes and gravimetric moisture were calculated with Mantel tests using spearman correlation with 999 permutations using the "vegan" package in R (Legendre and Legendre 2012).Differentially abundant sequences were identified by analysis of composition of microbiomes (ANCOM) (Mandal et al. 2015) using the q2-ANCOM plugin for QIIME2 (Bolyen et al. 2019) with ASVs retained that were significantly differentially abundant (α = 0.05) at one or more topographic positions in both 2017 and 2018.For microbial co-occurrence network analysis, sequencing libraries for both years were pooled and filtered to include only ASV with ≥50 reads in 80% of the samples for each management type at every topographic position.Network associations were calculated at each topographic position with a semi-parametric rank based approach (SPRING) (Yoon et al. 2019) with λ = 10 and 10 replications using the "Netcomi" package for R (Peschel et al. 2020).Hub ASVs were identified as having values above the 95% quantile for the degree (number of adjacent nodes), betweenness (proportion of times a node lies on the shortest path between two other nodes), and closeness (average distance between all other nodes) centralities at the same time.Full sequence and taxonomic information for hub ASV is available in Table S3.

Soil gas fluxes
Soil gas flux data were collected for CH 4 , CO 2 , and N 2 O in triplicate samples across all six topographic positions in late May and June for both 2017 and 2018.There were no consistent differences in either soil CO 2 or CH 4 fluxes (Figs.2B  and 2C), although CO 2 was significantly higher in selected upland samples from the cultivated fields in 2018 but not 2017 (Fig. 2B).Fig. 3. Nitrogen cycling gene abundances at each topographical position over the course of five collection timepoints during the 2017 and 2018 growing seasons in fields under permanent cover or cultivated management.Significance was tested using ANOVA and significant differences between permanent cover and cultivated samples at each time point are indicated by * (Benjamini-Hochberg corrected p ≤ 0.05).Data from positions 2-6 for nirK, nirS, nosZI, and nosZII are in Fig. S3.
In soils from cultivated fields, N 2 O fluxes were significantly higher in most samples from positions 2 (toeslope) through 5 (shoulder) during late May sampling, but there were no significant differences at the June sampling.The increase was significant in both the 2017 and 2018 growing seasons, with average N 2 O fluxes being 8-10× higher than samples from permanent cover field (Fig. 2A).Soil moisture was consistently very high in all samples from near the pond edge transition (position 1), averaging 46%-71% and 42%-61% between early May and August in 2017 and 2018, respectively; however, moisture in cultivated soils was significantly lower (p ≤ 0.05) than in soil from permanent cover fields.This trend was reversed in the toeslope (position 2) and footslope (position 3) samples where soil moisture was higher in cultivated soils (p ≤ 0.05); however, there were no consistent significant differences between upland samples (positions 4-6) from the two systems (Fig. 4D).

Nitrogen cycling gene copy numbers
Quantitative analysis of several N cycling genes indicated significant differences in potential nitrification and denitrification functional capacity by the soil bacterial community in response to land use change.Cultivated fields had a significantly higher abundance of bacterial ammonia oxidizing (BamoA) gene copies in most upland position samples from May to August for both growing seasons (Fig. 3A) while increases in archaeal ammonia oxidizing (AamoA) gene copies were restricted mostly to the 2017 growing season (Fig. 3B).The ratio of BamoA:AamoA was also consistently higher in upland samples from cultivated soils in both 2017 and 2018 (Fig. 4C).
Denitrification genes were significantly more abundant (p ≤ 0.05) in samples from the pond edge in the permanent cover fields, with an average 2.37× higher nirS abundance between May and August and average 1.33× higher nirK abundance during the June-August sampling times (Fig. 3C; Fig. S3).The ratio of nirK:nirS gene copies was consistently higher (p ≤ 0.05) in low-lying positions from cultivated fields at most sampling times in 2017 and in May only for 2018 (Fig. 4A); however, it was lower in upland positions in both early and late May compared to soils from permanent cover fields (p ≤ 0.05).
NosZI and nosZII were significantly more abundant (p ≤ 0.05) in low-lying samples from the permanent cover fields; however, increases in nosZII abundance were consistent from May to August in both 2017 and 2018, while nosZI was more abundant only at selected time points between June and August (Fig. 3C; Fig. S3).Additionally, the ratio of nir:nosZ gene copies was also significantly higher (p ≤ 0.05) in soils from the permanent cover fields in footslope, backslope, shoulder, and knoll samples at multiple time points in both 2017 and 2018 (Fig. 4B).

Soil bacterial community structure and composition
There were significant differences in microbial community composition in response to land use change at all topographic positions during both growing seasons (PER-MANOVA, p ≤ 0.01) (Table 2).The difference in Bray-Curtis dissimilarity between permanent cover and cultivated soil at each position increased as sampling moved upland, with the largest differences found in samples from the footslope (PERMANOVA, F = 25.66 in 2017 and F = 26.34 in 2018, p.adj = 0.001) and backslope (PERMANOVA, F = 22.03 in 2017 and F = 19.65 in 2018, p.adj = 0.001).Generally, the effect of cultivation was smaller in samples from the pond edge compared to those from midslope and upland positions in both years (Table 2).These differences in microbial community composition also correlated with differences in both N 2 O soil moisture at multiple positions (Mantel test, p ≤ 0.05) (Table 3).
Samples from fields converted from permanent cover to cultivated land also showed changes in bacterial community richness and diversity across the topographical range.In cultivated soils, the number of observed ASV, Shannon diversity, and Simpson evenness metrics were all higher in samples from the pond edge (Kruskal-Wallis, p ≤ 0.01) but lower in the backslope, shoulder, and knoll samples (Kruskal-Wallis, p ≤ 0.05) (Fig. 5).
Differential abundance analysis identified several ASV sequences that were consistently differentially abundant between samples from cultivated and permanent cover fields across growing seasons.The majority of the differentially  abundant bacteria were Proteobacteriota, with multiple ASVs (number in parentheses) classified as Rhodomicrobium (2), Geminicoccaceae (2), Dongia, Xanthobacteraceae, and Desulfuromonales significantly higher in upland samples from the permanent cover fields.In samples from the cultivated fields, differentially abundant Proteobacteriota included Xanthobacteraceae (2) and Acidibacter (2) (Fig. 6).
The majority of the differentially abundant Acidobacteriota, Actinobacteriota, and Bacteroidota were significantly higher at multiple positions along the topographic gradient from cultivated fields as well as multiple sequences classified as Vicinamibacterales (4) as well as Nitrososphaeraceae (2) and Nitrospira (1) (Fig. 6).One particularly abundant ASV sequence classified as Chryseolinea, represented an average of 0.13%-1.5% of the sequencing reads and was 5.5-6.9×more abundant in samples from the toeslope and footslope in cultivated fields (Fig. 6).

Microbial co-occurrence network analysis
There was significant variation in the taxonomy of bacterial hub taxa identified in soils collected from permanent cover versus cultivated fields.In the upland samples (footslope, backslope, shoulder, and knoll), hub ASVs from permanent cover fields were primarily Actinobacteriota (4) and Vicinamibacterales (2), while those from cultivated fields included Nitrososphaeraceae (3) as well as multiple Alphaproteobacteria (4) classified as Bradyrhizobium, Skermanella, and Microvirga (Fig. 7).At the lowest position (pond edge), the two-hub ASVs in permanent cover samples were all classified as Anaeroliniae, while only one hub sequence, classified as Fig. 5. Alpha diversity metrics calculated from 16S sequencing libraries generated for triplicate samples collected at five points during the growing season along the topographical gradient in fields under permanent cover or cultivated management.Significance was tested using the Kruskal-Wallis test and significant differences between permanent cover and cultivated samples at each time point are indicated by * .

Discussion
In the PPR, the wide-spread system of shallow wetlands creates unique soil environments that can influence the fate of synthetic N fertilizer.When comparing soil microbial community composition and the abundance of N cycling genes between soils from permanent cover and cultivated fields, there were multiple indicators that changes in both nitrifier and denitrifier communities are contributing to observed differences in soil N cycling and GHG emissions.
The almost universally significant increase in BamoA gene copies in both growing seasons along with the increase in AamoA gene copies in 2017 suggests a role for increased ammonia oxidation as a potential driver of the higher N 2 O fluxes seen in the midslope and upland areas of cultivated fields.The role of ammonia oxidation activity in shaping microbial community composition was also indicated by the higher abundance of specific ammonia oxidizing bacteria (AOB) and ammonia oxidizing archaea (AOA) including Nitrospira and Nitrososphaeraceae in samples from cultivated fields.Additionally, network analysis identified hub ASV classified as Nitrospira in samples from the pond edge and Nitrososphaeraceae in samples from midslope and upland regions, suggesting bacteria and archaea involved in nitrification are playing a key role in shaping microbiome composition.In the permanent cover fields, a significantly higher abundance of nirK, nirS, and nosZI + nosZII gene copies in samples from the pond edge suggests an increase in denitrification activity, which may have contributed to reduced N 2 O emissions in these samples.While there were no differentially abundant ASV associated with these samples, both hub taxa from the pond edge were classified as Anaerolineaceae, of which some members are known to contain nirS (Wei et al. 2015).
The increased abundance and influence of AOB and AOA within the microbial community network is likely related to the application of ammonia-based fertilizer in the cultivated system (Tourna et al. 2011), and the identification of Nitrospira as a hub taxa in samples from the pond edge in cultivated fields suggests that N inputs applied to upland areas Fig. 6.Bacterial ASV sequences that were consistently differentially abundant (ANCOM, α = 0.05) between samples from fields under permanent cover or cultivated management in both 2017 and 2018 growing seasons for at least one position along the topographic gradient.Color intensity reflects the log2 fold change in abundance between permanent cover and cultivated and point size reflects the mean proportion in the sequencing libraries.ASV labels indicate the highest taxonomic level to which the sequences could be classified based on the Naïve−Bayesian classification to the SILVA reference database.is influencing microbial community composition and nitrification/denitrification activity.Gravimetric moisture at the pond edge was also consistently higher in samples from the permanent cover field at the early season sampling points, and the resulting reduction in available O 2 is likely driving the observed increases in denitrifier abundance (Bedard-Haughn et al. 2006b;Butterbach-Bahl et al. 2013;Banerjee et al. 2016).In our system, soil moisture also correlated positively with the abundance of denitrification genes (nirK, nirS, and nosZII); however, further analysis of the N 2 fluxes would be needed to confirm whether this is associated with an increase in complete denitrification in permanent cover soils.Lower moisture conditions at low-lying positions in cultivated fields, combined with the potential for N accumulation at the footslope and toeslope of the cultivated fields, are likely contributing to increased nitrification activity resulting in the observed higher N 2 O fluxes.
Additionally, the increase in the abundance of denitrification gene copies for nirS, nirK, and nosZII along with the lower nirK:nirS ratio in the pond edge samples from permanent cover fields suggests that cultivation is affecting nitrification/denitrification processes in adjacent low-lying areas.While the understanding the role of these changes soil microbial processes on the quantity of specific soil N pools is limited by the absence of soil nutrient data for this study, the trends observed here are consistent with previous research showing that upland cultivation results in higher amounts of soluble N and C accumulating in adjacent lowlying regions, affecting soil microbiome composition and activity (Bedard-Haughn et al. 2006b).Both the abundance of nosZII genes along with a lower ratio of nirK:nirS have been previously linked to the ability of soil to act as a sink for N 2 O (Jones et al. 2014), suggesting that complete denitrification to N 2 is more prevalent in low-lying soils from permanent cover fields.While the ratio of nir:nosZ gene was observed to be lower in cultivated soils in this system, the higher available O 2 in upland soil likely inhibited complete denitrification.Fig. 7. Hub taxa identified using SPRING microbial co-occurrence network analysis in permanent cover or cultivated soils at each position along the topographic gradient.Hub taxa were identified as having values above the 95% quantile for the degree (number of adjacent nodes), betweenness (fraction of times a node lies on the shortest path between two other nodes), and closeness (minimum shortest path to all other nodes) centralities at the same time.Point size reflects the mean proportion in the sequencing libraries and points are colored according to taxonomic class.Labels indicate the highest taxonomic level to which the hub sequences could be classified based on the Naïve-Bayesian classification with the SILVA reference database.Labels representing multiple hub ASVs include the number of ASV sequences in brackets.
In addition to N fertilization, it is likely that changes in the amount and composition of available plant biomass as a result of cultivation are also playing a critical role in soil microbiome assemblage and N cycle dynamics (Bedard-Haughn et al. 2006a).While higher synthetic N inputs have been directly associated with increased N 2 O in previous studies (Giles et al. 2012), in other situations, N 2 O emissions were not directly related to the amount of N introduced in the system, but rather other factors including soil moisture, temperature, and SOC stocks (Rochette et al. 2010).Cultivation has been previously shown to reduce SOC stocks over time (VandenBygaart et al. 2003;Bedard-Haughn et al. 2006a), and the combination of high N, high O 2 , and low C conditions has been previously associated with higher nitrifier-denitrification activity, which is more likely to result in N 2 O production (Braker and Conrad 2011).Soils from cultivated fields were also significantly more abundant in several taxa classified as Chryseolinea and Xanthobacteraceae and which all have been previously associated with soils with higher C inputs in managed soils (Kim et al. 2013;Milkereit et al. 2021) and their abundance may be indicative of changes in the composition of soil carbon pools in cultivated fields, further altering nitrification/denitrification dynamics (Giles et al. 2012).
The effects of cultivation on GHG emission in the PPR are mediated by multiple factors including soil moisture and C and N accumulation, which in turn exert strong influence on the composition and activity of the soil microbiome.Cultivated land use in this system resulted in higher N 2 O and CO 2 fluxes, with the greatest differences between permanent cover and annual cropland found in the upland areas.The higher N 2 O emissions were associated with increased nitrifier abundance, more abundant amoA gene copies, and lower soil moisture, suggesting N 2 O emissions are resulting from nitrifier-denitrification or from incomplete denitrification.Further understanding of how annual cultivation affects soil moisture, microbial N cycling dynamics, and the composition of soil C and N pools can help predict effects on GHG emissions and N accumulation in catchment areas.

Fig. 1 .
Fig. 1.In both permanent cover and cultivated fields, individual soil samples (0-10 cm) were collected at six positions 5 m apart in three parallel transects along the topographic gradient.(A) and (B) Photographs from the site of the cultivated field taken on June 28, 2017, with flags indicating sampling positions 1, 2, 3, 4, and 5 for two of the transects.The division between volunteer vegetation and the cultivated area is visible between positions 1 and 2. (C) Photograph of the permanent cover field taken on August 21, 2017 with flags indicating positions 1, 2, 3, and 4 for two of the transects.(D) A schematic diagram showing a transect of the sampling area.Samples were collected 5 m apart going up the slope with sampling positions corresponding to the following topographic features: P1 = pond edge, P2 = toeslope, P3 = footslope, P4 = backslope, P5 = shoulder, and P6 = knoll.Sampling was conducted at five collection times during the 2017 (May 11, May 31, June 28, July 24, and August 21) and 2018 (May 7, May 28, June 25, July 18, and August 16) growing seasons.

Fig. 2 .
Fig. 2. Soil gas fluxes at each topographical position during the 2017 and 2018 growing seasons in fields under permanent cover or cultivated management.Significance was tested using ANOVA and significant differences between permanent cover and cultivated samples at each time point are indicated by * (Benjamini-Hochberg corrected p ≤ 0.05).

Fig. 4 .
Fig. 4. Ratio of gene abundances and soil moisture over the course of five collection time points during the 2017 and 2018 growing seasons in fields under permanent cover or cultivated management.Significance was tested using ANOVA and significant differences between permanent cover and cultivated samples at each time point are indicated by * (Benjamini-Hochberg corrected p ≤ 0.05).
• C prior to DNA extraction.

Table 1 .
Spearman's rank correlation of N cycling gene abundances with soil N 2 O fluxes and gravimetric moisture at each sampling position.
Note: Significant correlations (p ≤ 0.05) are indicated in bold.

Table 2 .
PERMANOVA analysis of the Bray-Curtis dissimilarity between cultivated and permanent cover fields along the topographic gradient for both the 2017 and 2018 growing seasons.P-values were adjusted for multiple comparisons using the Benjamini-Hochberg correction.

Table 3 .
Correlation of differences in microbial community composition (Bray-Curtis dissimilarity) with differences in soil N 2 O fluxes and gravimetric moisture (Euclidean distances) at each sampling position.Mantel tests were conducted using Spearman's rank correlation and 999 permutations.
Note: Significant correlations (p ≤ 0.05) are indicated in bold.