Across 1 661 km in temperature grassland of Inner Mongolia, China, copy numbers of N functional genes from soil microbial communities tended to decrease from northeast to southwest, similar to the changing trend of precipitation. Consistently, spatial patterns of biological communities including microbial functional genes have been observed to be shaped by climatic factors (e.g. precipitation or temperature) by numerous studies at regional scales [16, 18, 48–51]. However, most previous studies presented linear relationships between microbial diversity [52, 53] or functional gene abundances [16] and precipitation. In fact, the precise response curves of biological communities to abiotic factors could be more complex but seldom linear if ranges of abiotic factors are wide enough [54, 55]. In this regional scale study, we discovered that the saturation curve dominated the response patterns of most N genes and identified breaking points automatically along the natural precipitation gradient ranged from 215 to 501 mm. Copy numbers of most genes (i.e. AOB, nirS, narG, AOA, nxr and nirK) increased with the precipitation when it was lower than 321 or 403 mm, but did not change after these breaking points, indicating the precipitation effect became saturated after breaking points on microorganism containing these genes.
Differently, the copy number of nifH increased with precipitation almost along the full range of natural precipitation gradient of this study. Consistently, nifH was observed to increase continuously on Qinghai-Tibet Plateau along a natural precipitation gradient ranged from 62 to 614 mm [11]. The accelerated increasing trend of nosZ copy number after its breaking point of 367 mm likely implied a stimulating effect of intensified anaerobic condition [56]. As far as we know, this is the first study to recognize these diverse patterns of various N functional genes along the natural precipitation gradient.
The changing trend of N functional genes were contrary to that of temperature. Inconsistently, number studies found that microbial functional gene abundances increased with higher temperature [51, 57, 58]. Such discrepancy implied that intensified water loss or deficiency underlying the temperature increase in our studying sites may surpass the stimulating effect of warming itself and limit abundances of these functional genes (e.g. AOB, nxr, nosZ and nifH) in temperate grassland, which was generally deficient in water [59]. Consistently, Aridity Index (AI) representing the degree of drought was positively correlated with MAT significantly in this study, and abundances of all measured N genes decreased significantly with the increase of AI after their breaking points (see details in Additional file 1 and Additional file 2: Fig S2). Moreover, the copy numbers of nirK and AOA did not change before their breaking points of 1.70 and 3.79℃ for MAT, respectively, indicating microorganism containing these genes may be less responsive to the lower arid under lower temperature in our sampling sites. Higher resistance of nirK and AOA was also observed in the polar regions and dryland ecosystems [51, 60]. Furthermore, increased copy numbers of nirS and narG after a breaking point of 5.44℃ indicated stimulating effect of temperature itself may become dominant under such condition.
Significant geographic distance-decay relationships were discovered between microbial N gene community similarity and geographic distance, indicating dispersal limitation and history mattered [28, 33, 37]. This is the first study to explore such relationship for microbial N genes, as far as we know, though distance decay relationships have been found for a comprehensive set of microbial functional genes in GeoChip studies [10, 41–43]. Interestingly, SEM analysis showed that geographic distance only had direct effect on the N gene community similarity in the temperate desert steppe, likely implying the importance of dispersal limitation and relatively less environmental heterogeneity in such ecosystem. Similarly, a previous study found that, geographical distance was the only reason for explaining the similarity of bacterial community in the desert [61], even harsh than the temperate desert steppe. However, its indirect effects via plant community dissimilarity, short-term and long-term environmental distances were observed in temperate meadow and temperate desert steppe, implying that environmental heterogeneity may be more responsible for geographic distance-decay relationships in these ecosystems. Consistent with previous discoveries, geographic distance-decay relationship could be attributed to environmental heterogeneity depending on their niche preferences [36, 61, 62].
Geographic factor or distance used to represent history contingencies in previous studies [16, 30, 33], though the underlying logic or its representativeness on history contingencies was never clarified or proved. Our results presented a clear evidence for the first time that the representativeness of geographic distance on history contingencies was dependent on ecosystem types. In temperate meadow, geographic distance was dominantly explained by history contingencies characterized by long-term environmental distance, which were highly correlated with geographic distance with a R2 of 0.81 by Pearson or a coefficient of -0.90 by SEM. In typical grassland, the representativeness of geographic distance on history contingencies was also high, as their correlation with a R2 of 51.8% by Pearson or a coefficient of -0.72 by SEM, as double as the interpretation power of short-term environmental distance. However, in the temperate desert steppe where geographic distance had a direct effect on the N gene community similarity, the interpretation of long-term environmental distance for geographical distance is still > 50%, but comparable to that of short-term environmental distance, revealing less representativeness of geographic distance on history contingencies in the temperate desert steppe.
SEM and attribution analysis based on MLR showed that environmental distance based on short-term variables representing contemporary factors had the greater influence on N gene similarity in temperate meadow. Microbial community composition shifts were previously observed to result from gradual or sudden changes in abiotic and biological factors such as soil moisture, temperature, weather, season and plant species in a relatively short period of time in forest, farming-pastoral ecotone and wetland ecosystems [63–66]. In the typical steppe, the N gene community similarity was dominantly affected by plant community dissimilarity, while other factors only attributed < 15%. The plant coverage range was widest in the typical steppe with higher plant richness (Additional file 1 and Additional file 2: Fig S8), indicating that plant community might be more dynamics in typical steppe, likely responsible for its dominant role in N gene community similarity. Consistently, plant community shifts in typical steppe also exhibited a much stronger explanatory power for explaining the regional variation of ecosystem respiration than that in temperate meadow and temperate desert steppe [59]. In temperate desert steppe, N gene community similarity was attributed to long-term environmental distance and geographic distance. Temperate desert steppe had higher AI and lower soil resources like soil organic carbon (SOC)and total N (TN), creating a harsh environment for microbial communities [67]. Thus, soil microorganisms that can survive in the temperate desert steppe may have developed higher tolerance for short-term environmental disturbances in such harsh environment and adapted to the long-term environmental dynamics.
Other than deterministic processes (e.g. abiotic and biotic factors) that have been studied for a long period of time [68, 69], neutral mutation and random genetic drift theory asserts that stochastic processes (e.g. birth, death, immigration, and limited dispersal) are responsible for shaping the microbial community structure [70]. However, in this study, NST revealed that deterministic processes were predominant for N gene community in the temperate grassland. Significant decay relationships between the N gene community similarity and environmental distances, either long-term or short-term, observed in this study were consistent with many previous microbial studies [37, 71–73], though not focusing on microbial N genes. As revealed by Partial Mantel, such distance-decay relationships based on long-term environmental distance was driven by TOC, TN and climatic factors, e.g. precipitation or temperature while such relationship based on short-term environmental distance was driven by available nitrogen and phosphorus, AI-m, plant richness and so on. The importance of these environmental variables was consistent with many previous studies [26, 74–76].