Groundwater Depth Overrides Tree-Species Effects on the Structure of Soil Microbial Communities Involved in Nitrogen Cycling in Plantation Forests

: Microbial communities found in soil ecosystems play important roles in decomposing organic materials and recycling nutrients. A clear understanding on how biotic and abiotic factors influence the microbial community and its functional role in ecosystems is fundamental to terrestrial biogeochemistry and plant production. The purpose of this study was to investigate microbial communities and functional genes involved in nitrogen cycling as a function of groundwater depth (deep and shallow), tree species (pine and eucalypt), and season (spring and fall). Soil fungal, bacterial, and archaeal communities were determined by length heterogeneity polymerase chain reaction (LH-PCR). Soil ammonia oxidation archaeal (AOA) amo A gene, ammonia oxidation bacterial (AOB) amo A gene, nitrite oxidoreductase nrx A gene, and denitrifying bacterial nar G, nir K, nir S, and nos Z genes were further studied using PCR and denaturing gradient gel electrophoresis (DGGE). Soil fungal and bacterial communities remained similar between tree species and groundwater depths, respectively, regardless of season. Soil archaeal communities remained similar between tree species but differed between groundwater depths in the spring only. Archaeal amo A for nitrification and bacterial nir K and nos Z genes for denitrification were detected in DGGE, whereas bacterial amo A and nrx A for nitrification and bacterial nar G and nir S genes for denitrification were undetectable. The detected nitrification and denitrification communities varied significantly with groundwater depth. There was no significant difference of nitrifying archaeal amo A or denitrifying nir K communities between different tree species regardless of season. The seasonal difference in microbial communities and functional genes involved in nitrogen cycling suggests microorganisms exhibit seasonal dynamics that likely impact relative rates of nitrification and denitrification.


Introduction
Soil teems with diverse microorganisms, which regulate ecosystem functions such as decomposing organic materials and recycling nutrients [1]. Plant growth in terrestrial ecosystems is largely dependent on the activities of soil microorganisms [2][3][4]. Soil fungi have been recognized as contributing to degradation of recalcitrant lignocellulose complexes [5] and positively affecting soil groundwater depth and season), whereas the fungal community would be influenced more strongly by biotic factors (i.e., tree species).

Study Site Description and Sample Collection
The study site is located at the US Department of Energy's Savannah River Site in New Ellenton, SC, USA. The gradient of groundwater depth across the experimental watershed was planted with loblolly pine (Pinus taeda) in March 2013, an indigenous pine species to southeast USA, and Camden white gum (Eucalyptus benthamii) in October 2013, a non-native eucalypt. Each plot contains 168 individual trees with an area of 0.15 hectare. Three paired plots were located in deep groundwater (groundwater level > 10 m) and three were located in shallow groundwater (groundwater level < 2 m). Each of the six boxes ( Figure 1 in red color) indicates the location of paired loblolly pine and eucalypt plots-three paired plots located in deep groundwater (U6, U7, and U10) and three paired plots located in shallow groundwater (U3, U5A, andU5B), which makes six repetitions for tree species and six repetitions of groundwater level. Soil type belongs to Fuquay loamy for plot U7 (https://soilseries.sc.egov.usda.gov/OSD_Docs/F/FUQUAY.html) and Dothan fine-loamy (https://soilseries.sc.egov.usda.gov/OSD_Docs/D/DOTHAN.html) for rest of the plots. Twenty soil cores around the two tree species from each plot were collected from top 20 cm of the soil layer and pooled together to obtain the soil samples in the spring (March) and the fall (August) of 2015 for this study.

Soil Chemical Analysis
The chemical analysis was performed by A&L Plains Agricultural Laboratory, Inc (Lubbock, TX, USA) and Environmental Microbiology Laboratory, Georgia Southern University (Statesboro, GA, USA). The components measured were organic matter (%), nitrate, phosphorus, potassium, magnesium, calcium, soil pH, and cation exchange capacity (A&L Plains Agricultural Laboratory), and soil water was analyzed according to the methods in Methods of Soil Analysis, Agronomy Book, No. 9 (Page et al. 1982, Madison, Wisconsin, USA). Soil ammonium was extracted with 2 M KCl and followed with the salicylate method [39].

DNA Extraction
DNA of soil samples was extracted using the PowerSoil DNA Isolation kit (Mobio Laboratories Inc., Carlsbad, CA, USA) following the manufacturer's instruction manual. All samples were rapidly and thoroughly homogenized. The extracted DNA was stored at −20 °C to be used for PCR analysis and other downstream applications.

Quantitative PCR (Q-PCR)
Quantitative PCR was performed using the QuantStudio™ 6 Flex Real-Time PCR System (Life Technologies, Carlsbad, CA, USA). Quantification utilized the fluorescent dye SYBR-Green I which binds to double-stranded DNA during amplification. Primer details and sequences for the target genes are listed in Table 1. All PCR mixtures contained the recommended 25 μL protocol consisting of 12.5 μL of 2× GoTaq® Colorless Master Mix (Promega, USA), 0.5 μL of 10 μM forward and reverse primer, 1 μL BSA, 2 μL SYBR® of 1×, 6 μL of nuclease-free water, and 2 μL of DNA template. The Q-PCR of each sample was run in triplicate to control for mechanical and technical errors. The mean of all three DNA quantities was used for statistical analysis.

Statistical Analysis
Univariate statistics: Treatment effects (groundwater depth, tree species, and season) on soil chemical characteristics were analyzed using analysis of variance (ANOVA). The homogeneity of variances and normality of distribution were tested with the Levene and Kolmogorov-Smirnov tests, respectively. In addition, the non-normally distributed dependent variables were log10-transformed before analysis. All data were analyzed by three-way ANOVA on groundwater depth, tree species, and season with means and a post-hoc method (Duncan's method) for multiple comparisons at a 5% significance level. All the statistical analyses were performed with JMP® Pro 12.1.0 (SAS Institute Inc., USA).
Statistical analysis of fragments of microbial DNA: Nonparametric multivariate analyses procedures including pairwise Bray-Curtis similarity of LH-PCR fragment profiles and band matrices of DGGE obtained above were applied for microbial community analysis. A square-root transformation was applied to the data before construction of the similarity matrices. Cluster analysis was performed to compare the similarity microbial and functional gene communities; and analysis of similarity (ANOSIM) and permutational multivariate analysis of variance (PERMANOVA) including principal coordinates analysis (PCO) with Spearman's correlations of variables with the PCO axes were applied to test the significance of microbial community difference and its correlation with environmental factors. A significance level of 5% for ANOSIM (ANOSIM p < 0.05) was applied to detect a difference of microbial community between groups. The correlations between nitrogen functional genes and environmental factors were determined by ENVIRO-BEST function. All nonparametric multivariate analyses procedures, including calculation of Bray-Curtis similarity matrices, cluster analysis ANOSIM, PERMANOVA, and PCO and ENVIRO-BEST were conducted using the PRIMER-E & PERMANOVA plus statistical (PRIMER-E Ltd, Plymouth Marine Laboratory, UK) software.

Soil Chemical Characteristics
Groundwater depth, tree species, and season differently affected the soil chemical characteristics of soil organic matter (%), pH, phosphorus, potassium, magnesium, and nitrate content, soil CEC, as well as water content (%) ( Table 2). Soil at the site of deep groundwater had significantly higher soil pH and lower water content (%) than at shallow groundwater depth (p < 0.05). Eucalypt trees on the plots significantly increased soil organic matter (%), as well as soil water content (%), CEC, phosphorus and magnesium contents, compared with pine field plots (p < 0.01). The spring samples had significantly higher nitrate, potassium, magnesium, and CEC contents than in the fall samples (p < 0.05), and the nitrate concentration was almost four times higher in the spring than in the fall (Table  2). However, significantly lower soil phosphorus contents were observed in the fall than in spring. No interactions (p > 0.05) among groundwater depth, tree species, and season were observed for all soil organic matter, nitrate, potassium, and soil CEC, but interaction between tree species and season was observed for soil phosphorus and magnesium contents ( Table 2). Table 2. Soil chemical characteristics and the significance (p-value) of statistical analysis under different groundwater depth, tree species and season. Means ± SE (n = 3) are presented. Soils were collected from top 20 cm of soil layer. Means for main effects followed by different letters are significantly different at p < 0.05.

Soil fungal Communities
Soil fungal communities determined by LH-PCR in the spring significantly differed from soil fungal communities in the fall (ANOSIM p < 0.01, Figure 2A). For the spring samples, tree species significantly affected the soil fungal community structure (ANOSIM p = 0.01, Figure 2A,B), whereas no significant effect (ANOSIM p = 0.13) of groundwater depth on soil fungal community structure was detected. For the fall samples, the loblolly pine and eucalypt also significantly affected the soil fungal community structure (ANOSIM p < 0.01, Figure 2A,C), but no significant effect of groundwater depth on soil fungal community structure was detected (ANOSIM p = 0.69). The effect of environmental factors on soil fungal community structure can be seen in the PCO ordination with the length and direction of each vector through the relationship between the variable and the PCO axes, where soil organic matter (%) had a strong negative relationship with PCO1 to the separation of soil fungal communities in both the spring ( Figure 2B) and the fall ( Figure 2C), whereas soil ammonium ( Figure 2B) and soil pH ( Figure 2C) positively contributed to the different soil fungal communities in the spring and fall with PCO1, respectively.

Soil Bacterial and Archaeal Communities
Soil bacterial communities determined by LH-PCR in the spring were significantly different from soil bacterial communities in fall (ANOSIM p < 0.01, Figure 3A). Soil bacterial communities in the spring samples ( Figure 3A,B) differed between groundwater depths (ANOSIM p = 0.06) but remained similar between plant species (ANOSIM p = 0.89). Soil bacterial communities in the fall samples ( Figure 3A,C) also differed between groundwater depths (ANOSIM p = 0.02) but remained similar between plant species (ANOSIM p = 0.85). The PCO ordination indicated that water content (%) had a positive relationship with PCO1, whereas soil pH negatively contributed to the separation of soil bacterial communities in the spring but contributed positively in the fall with PCO2 ( Figure  3B,C).

Spring-deep
Spring-shallow Fall-deep Fall-shallow Soil archaeal communities determined by LH-PCR in the spring were significantly different from soil archaeal communities in fall (ANOSIM p < 0.01, Figure 4A). Soil archaeal communities in the spring samples ( Figure 4A,B) differed between groundwater depths (ANOSIM p = 0.02) but remained similar between plant species (ANOSIM p = 0.85). Soil archaeal communities in the fall samples ( Figure 4A,C) were not significantly affected by neither groundwater depth (ANOSIM p = 0.20) nor plant species (ANOSIM p = 0.42). Soil pH, water content (%), and ammonium contents all positively contributed to the different soil archaeal communities with PCO1 in both the spring and fall samples ( Figure 4B,C). Soil nitrate contents positively and negatively contributed to the different soil archaeal communities with PCO1 in the spring and fall samples, respectively ( Figure 4B,C).

Soil Nitrification Archaea and Bacteria
There was a significantly different soil nitrification community determined by DGGE with ammonia-oxidizing archaea amoA AOA gene (Table 3, Figure 5A,B) at shallow groundwater compared with deep groundwater in the spring (ANOSIM p = 0.02), but not in the fall (ANOSIM p = 0.58). The communities of amoA AOA gene were significantly different between the spring and the fall (ANOSIM p = 0.01). Tree species did not influence the communities of ammonia-oxidizing archaea amoA AOA gene in either the spring (ANOSIM p = 0.68) or the fall (ANOSIM p = 0.86). ENVIRO-BEST analysis indicated that nitrate, Mg, and Ca contributed to the observed communities of ammoniaoxidizing archaea amoA AOA gene with correlation coefficient of >10% (Table 4). Soil nitrification communities of ammonia-oxidizing bacteria amoA AOB were not detected by DGGE in the soil samples (Table 3).   The nitrite oxidoreductase gene nrxA of nitrite oxidizer was not detected by regular DGGE procedures using extracted soil DNA, but it was detected in the samples when a two-step nested PCR reaction for DGGE was applied. Q-PCR indicated that threshold cycles (CT) of nrxA gene were significantly shortened from 26 to 28 cycles (with low DNA amounts) of regular PCR to only around 10 cycles (with high DNA amounts) of two-step nested PCR (p < 0.05).

Soil Denitrification Bacteria
In the spring samples, obvious bands of nosZ gene products were detected by DGGE in ten of twelve samples. We did not detect bands in a sample from pine grown in shallow groundwater (U3P) and another in deep groundwater (U6P). Based on ANOSIM analyses of the nosZ gene obtained from the 10 samples ( Figure 6A,B), the communities of nosZ gene were significantly different between the spring and the fall (ANOSIM p < 0.05). There was no significantly different soil denitrification community under the shallow groundwater depth sites compared with deep groundwater depth sites (ANOSIM p = 0.83, Table 3); no difference between tree species (ANOSIM p = 0.61, Table 3) was observed either in the spring samples. In the fall samples, there were no significantly different soil denitrification communities of nosZ gene under the shallow groundwater sites compared with the deep groundwater sites (ANOSIM p = 0.32, Table 3). Likewise, no difference was observed between tree species (ANOSIM p = 0.88, Table 3) either. ENVIRO-BEST analysis indicated that correlation coefficients of nosZ gene with all of the environmental factors were <10% (Table 4). There was a significantly different soil denitrification community of nirK gene detected by DGGE under the shallow groundwater depth sites compared with deep groundwater depth sites in both the spring (ANOSIM p = 0.04) and the fall (ANOSIM p = 0.02) samples ( Figure 7A,B, Table 3). No difference of denitrifying bacterial community of nirK gene was observed between different plant species in either the spring (ANOSIM p = 0.63) or the fall (ANOSIM p = 0.46) samples ( Figure 7A,B, Table 3). Likewise, the communities of nirK gene were significantly different between the spring and the fall (ANOSIM p = 0.01). ENVIRO-BEST analysis indicated that nitrate, Mg, K, and water content (%) contributed to the observed communities of nirK gene with correlation coefficients of >10% (Table  4).

Discussion
Both biotic and abiotic factors influenced the diversity and functional genes of soil microbial communities. As we predicted, the soil fungal communities were affected more strongly by tree species, whereas soil bacterial and archaeal communities were influenced more strongly by groundwater depth. Different plant species and their root exudates have been reported to affect soil microbial community and diversity [25,51,52]. Plant root exudates and soil organic matter content play an underappreciated role in shaping the soil fungal communities [25,26]. It is demonstrated that beech and spruce trees species, with different litter quality, selected different soil fungal communities expressing different set of genes involved in organic matter degradation [52]. Another study indicated that 35%-37% of the dominant fungal "species" (operational taxonomic units, OTUs) were restricted to one or two tree species, whereas only about 15%-45% of fungal OTUs were common under six or seven tree species in the study [53], suggesting the significant effects of plant species on fungal community compositions. The research of our study indicated that tree species of eucalypt significantly increased soil organic matter (%), soil phosphorus, magnesium, soil water content (%) and CEC over tree species of pine, and the interaction between tree and season for phosphorus and magnesium was also detected ( Table 2), suggesting that the different root growth and activities with season as well as different characteristics of pine and eucalyptus roots could affect the nutrient uptake differently, therefore leading to the observed interaction on soil phosphorus and magnesium concentration. Furthermore, the difference of the root growth and activities with seasonal change could affect the nutrient uptakes differently and lead to different nutrient concentration and CEC with season, thus perhaps contributing to significant differences in soil fungal community. Table 4. The ENVIRO-BEST data for the correlation between amoA AOA, nirK, and nosZ genes and environmental factors under deep and shallow groundwater depths of pine and eucalypt plantation in the spring and fall. Soil water content and availability were related to decomposition and microbial growth as well as to changes in the microbial community structure [54]. Our study suggested that groundwater depths affected soil bacterial communities in both spring and fall and affected soil archaeal communities and functional genes of N cycling in the spring. Significantly higher soil pH but lower soil water content (%) were observed under deep than shallow water depths ( Table 2). The deep root systems of a Eucalyptus tree requesting a lot of water, with a huge capacity to get such water from groundwater, may greatly affect soil water content (%), total soil bacteria, and nitrifying/denitrifying bacterial communities observed in this study. The spring and fall water contents were not significantly different in the collected soil samples of two sampling times in the present study, however, the seasonal fluctuation of the water content and the effects of anoxic conditions of the soils for the effect of anaerobic soils may cause the seasonal difference between the functional genes. Study on the relationship of soil water contents and microbial functional genes indicated that increased soil moisture rapidly and distinctly changed soil AOA abundances in two temperate forest soils [22]. However, contrary to our present study, it was also observed that fungal and actinobacterial community changes due to short-and long-term water-level changes at three different sites in a boreal peatland complex, indicating that fungal community responds to persistent water-level drawdown, whereas actinobacterial community was less sensitive to hydrological change of shortterm groundwater depth drawdown [55]. The interaction between abiotic and biotic factors affects microbial community structure in the natural environment, with effects of plant species on fungal community structure being statistically significant; effects of moisture on bacterial community structure were also significant [35,52].

Genes
Functional genes regulating the N cycle, including nitrification and denitrification, have important implications for N-use efficiency in agricultural and forest ecosystems as well as for environmental quality. Nitrification in soil converts relatively immobile ammonium-N to nitrite first and then to highly mobile nitrate-N. Soil matrix, water status, aeration, temperature, and pH have strong influence on nitrification rates [56]. Genes involved in the nitrification include amoA AOA and amoA AOB from archaea and bacteria, respectively. Groundwater depth significantly changed the amoA AOA community in this study, indicating that the groundwater depths and consequent changes of soil pH are the major factors influencing nitrification. The amoA AOB was not detected in this study. The importance of archaea or bacteria in the process of nitrification has been observed in agricultural soils [19,20]. The contribution of bacteria and archaea to nitrification processes in soil are affected by soil pH. In acidic soil conditions, archaea may be more important than bacteria for the ammonium oxidation. Another study indicated that AOA might influence the nitrogen cycle in lownutrient, low-pH, and sulfide containing environments [21]. The soil pH in our study site was acidic, ranging from 3.4 to 6.0 in raw data of pH value, confirming that the archaeal amoA AOA, but not bacterial amoA AOB, dominated in the nitrification process in low-pH soil. The soil pH at the site with shallow groundwater depth was significantly lower than that at the site with deep groundwater depth in this study (Table 2), which may explain the observed dominance of amoA AOA in the study sites as well as different communities of nitrification functional genes under deep and shallow groundwater depths.
Denitrification is closely related to loss of nitrogen, when dinitrogen is the dominant product, or to an environmental concern as the intermediate product N2O of denitrification is not only harming the ozone layer but also a potent greenhouse gas [8,50]. Denitrification is known to be influenced by temperature, soil carbon, and oxygen conditions of soil controlling aerobic and/or anaerobic microbial processes [57,58]. Because many factors are involved in the processes, denitrification activities may not always correlate with the denitrifying microbial biomass and community structure found in each environment. Using the DGGE for studying the denitrification functional genes, the nirK and nosZ were determined as the most abundant denitrifying functional genes, and the nirK gene performed unique community structure under different groundwater depths. Similar to total bacterial 16S rRNA gene and nitrification amoA AOA gene, the nirK denitrifying gene was also affected by groundwater depth, but not tree species. This is consistent with a recent study that water content differences have stronger effects than plant functional groups on soil bacteria in a steppe ecosystem [59], and nirK gene abundance rapidly increased in response to wet conditions until substrate (NO3 − ) became a limiting factor [60]. In contrast, abundant denitrifying nosZ gene was detected, but it was not affected by either groundwater depth or plant species. Previous studies indicated that denitrifying functional genes interacted with complex soil environment, including soil C, moisture, N contents, etc., and less is known about how the genes are influenced by soil environment or affect the denitrification rate [8]. The abundance of denitrification genes had been reported as significantly different in forest and agricultural sites. A study indicated that the nosZ gene was significantly dominant in forest sites, while nirS gene was more abundant than nosZ gene in agricultural sites [61]. Our study also confirmed the dominant nosZ gene but undetected nirS gene in the forest site of present study. The intense bands of nosZ gene were detected across soil samples in this study but did not vary with respect to fluctuating groundwater depths or tree species. Differences in microbial communities with respect to groundwater depths on nosZ gene may have been affected by other factors, such as temperature, redox potential, and soil pH. Some studies have shown a positive correlation between denitrifying genes abundance and denitrification activity, while others have shown quite the opposite [62]. No correlation between groundwater depths and soil microbial community of nosZ gene suggested that the abundance of denitrifying nosZ gene was independent of denitrifying activity across groundwater depth. There have been several past studies supporting this claim. For example, nirK and nirS positively correlate in abundance with oxygenation, and nosZ positively correlates with soil pH [63]. However, significant difference in soil microbial community related to nirK genes was observed across the groundwater depth gradient in samples of both spring and fall, and ENVIRO-BEST analysis suggested that nirK genes were related to soil nitrate, Mg, K, and water content (%). Soil microbial communities with various functional genes related to denitrifying bacteria and their relationships with complicated environmental factors should also be considered to further understand the functioning of these denitrifying genes pertaining to soil.
Seasonal changes in the soil microbial community as well as nitrogen-cycle gene (nirK) abundance have been observed in agricultural, forest, and grassland ecosystems [64][65][66]. Statistical analysis of the DGGE banding patterns revealed significant differences of soil microbial community between samples taken in different seasons [65]. We observed unique soil bacterial community and functional genes under different groundwater depths in both the spring (March) and the fall (August). However, unique community of functional gene amoA AOA was only detected under different groundwater depths in the spring. Significantly higher nitrate contents in the spring than that in the fall ( Figure 2B) suggest that seasonal dynamics of nitrate contents may greatly affect both nitrifying and denitrifying communities. In a manipulating precipitation study in beech and conifer forest plots, decreased soil water content and their effects on the total bacterial community structure were negligible, but significant effects for the active bacteria were observed during growing season [32]. The observed different microbial community structure and nitrogen functional genes of the present study may be related to differences in plant growth, belowground allocation of carbon, and obviously higher nitrate contents in the spring compared to the fall. The functional genes related to nitrification/denitrification are critical in regulating soil N cycling but account for only a small portion of the total bacterial population, such as 0.5% to 5% of the total bacterial for denitrifying microorganisms [8,67]. A relationship between seasonal temperature changes and the number of psychrophylic and mesophylic isolates was observed from the sediment-water interface [68]. Significantly different soil microbial communities and functional genes involved in nitrogen cycling were observed in spring and fall in our study, however, other parameters, such as water content, humidity and nutrient supply in soils, and their seasonal fluctuations on microbial communities and especially the nitrification/denitrification functional genes under different groundwater depths could also play a crucial role for the biogeochemical cycling processes and woody biomass production in plantation forests.

Conclusions
Soil fungal communities remained similar between tree species, whereas, soil bacterial communities are determined by groundwater depth regardless of season. Groundwater depth only affected soil archaeal communities in the spring. Soil microbial functional genes involved in nitrogen cycling varied with respect to groundwater depth. The nitrifying bacteria determined by archaeal amoA (AOA) nitrifying gene, and denitrifying bacteria through nirK genes, varied significantly with respect to groundwater depths. The tree species did not have an effect on the bacterial/archaeal and nitrifying/denitrifying microbial communities of neither the spring nor the fall samples, thus groundwater depth overrides tree-species effects on the structure of soil microbial communities involved in nitrogen cycling in plantation forests. Understanding how different nitrifying/denitrifying genes affect the function of nitrogen cycling and which stages of the nitrification/denitrification process they are involved in, through targeting respective enzymes nitrite reductase (nirS and nirK) and nitrous oxide reductase (nosZ), could be a key step to improve plant production and sustain our environment, such as by reducing N2O emissions.
Author Contributions: All authors have read and agree to the published version of the manuscript and contributed to either field or laboratory experimental design, formal analysis, and writing. Acknowledgments: Thanks to Dr. Scott Harrison for assistance in using ABI 3500. We also thank Georgia Research for Academic Partnership in Engineering (GRAPE) Project supporting on the microbial nitrification and denitrification studies and the Department of Biology Charles Chandler Foundation of Georgia Southern University for financial support. We thank M. Bitew for preparing Figure 1 and Garret Strickland for help on optimizing DGGE conditions for nitrogen functional gene analyses.

Conflicts of Interest:
The authors declare that research was conducted in the absence of any commercial or financial relationship that could be constructed as a potential conflict of interest.