Assessing the relative importance of soil properties, tree type and distance on soil microbial community composition at multiple spatial scales

Background/Question/Methods

Soil microbial communities influence a host of biogeochemical processes. An understanding of the factors regulating their composition may improve our ability to project changes in ecosystem function. Recent studies demonstrate that soil microbial communities are strongly influenced by soil pH. However, whether the influence of pH is direct (e.g. caused by the physiological constraints of pH on cell membrane integrity) or indirect (e.g. via the influence of plants on soil properties) is uncertain. We designed a study to assess the relative importance of tree type, soil properties, and geographic distance between samples on turnover in composition of soil microbial communities at spatial scales ranging from meters to hundred of kilometers. Samples were collected from twelve forest sites in the eastern United States in close proximity to either Sugar Maple (Acer saccharum), American Beech (Fagus grandifolia), White Oak (Quercus alba) or Yellow Birch (Betula alleghaniensis). Bacterial DNA was extracted from the samples and amplified using bar-coded primers specific to the 16S rDNA gene. The resulting amplicons (~270 bp in length) were sequenced using the Roche 454 GS-FLX system. Analysis of the amplicons was based on a principal coordinates analysis of the Unifrac metric (Lozupone and Knight 2007). Generalized Dissimilarity Modeling (GDM), a non-linear form of matrix regression, was used to relate compositional turnover of soil microbial communities to environmental gradients and geographic distance.  

Results/Conclusions

We obtained 1,024,059 high-quality amplicons from 700 samples (60 per site). The first principal coordinate (PCO1) revealed that six of the twelve sites could be distinguished from all other sites based on microbial community composition. However, the differences among samples were unrelated to site or proximity to a specific tree species. Instead, variation among samples along PCO1 was driven by soil pH (r = 0.88). Other soil properties had no correlation with PCO1. GDM demonstrated that turnover in soil microbial communities was explained almost entirely by differences in pH among sites. These patterns were robust regardless of microbial phyla. A Unifrac analysis of samples with similar pH (4.6 – 5.1) revealed that some sites harbored distinct soil microbial communities, however we were unable to relate these differences to any of the other variables that we measured. Our results confirm that soil microbial community composition is strongly influenced by the pH of soil and not tree type or distance between samples. Additional investigations are needed to understand what other factors might be driving variation in soil microbial community composition among sites.