Incorporating biotic factors into species distribution modeling: Are interactions with soil microbes important?

Background/Question/Methods

Species distributions are driven by both abiotic factors and biotic interactions. Biotic interactions between plants and soil microbial communities are an understudied yet potentially important driver of plant distributions. Some soil bacteria promote plant growth by cycling nutrients, while others are pathogenic. Thus, we predicted that bacteria would influence plant distributions through both positive and negative interactions, and that these interactions would improve prediction of plant species distributions. We investigated the influence of soil bacterial clades on the distributions of bryophytes and 12 vascular plant species in a high elevation tundra ecosystem in the Rocky Mountain Front Range, Colorado, USA. We used an information-theoretic criteria (AICc) modeling approach to compare species distribution models (SDMs) with the following different sets of predictors: abiotic variables, abiotic variables and other plant abundances, abiotic variables and soil bacteria clade relative abundances, and a full model with abiotic, plant, and bacteria factors.

Results/Conclusions

Inclusion of either plant or bacteria biotic predictors improved the fit, deviance explained, and predictive power of the SDMs, and for the majority of the species, adding information on both other plants and bacteria yielded the best model. Interactions between the modeled species and biotic predictors were both positive and negative, suggesting the presence of competition, parasitism, and facilitation. The improvement of the SDMs attributed to bacteria was not related to plant traits or functional group, suggesting this phenomenon could be widespread. While our results indicate that plant-plant interactions appear to be a stronger driver of plant distributions than plant-bacteria interactions, they also show that bacteria can explain parts of plant distributions that remain unexplained by abiotic and plant predictors. Our results provide further support for including biotic factors in SDMs, and suggest that belowground factors be considered as well.