The impact of microbial community diversity and composition on ecosystem functioning has long been recognized as significant. Despite evidence that incorporating microbial functional groups leads to more accurate predictions of ecosystem process rates, the inclusion of microorganisms in ecosystem models remains limited. Microbial ecologists have long endeavored to link microbial function to phylogeny, but methods such as metagenomics and amplicon sequencing fail to provide quantitative, insitu, taxon-specific measures of microbial function. Quantitative stable isotope probing has recently emerged as a reliable method of quantitatively measuring taxon-specific functions such as ammonium uptake and cell growth at fine phylogenetic resolution in natural communities. The combination of quantitative functional traits and phylogenic placement allows for the development of phylogenetic approaches to model the evolution of these functional traits and predict trait values for uncharacterized taxa. Owing to the great diversity of microorganisms, it is not feasible to measure the functional traits of all microbial species. Thus, the ability to use phylogeny to predict the functional traits of organisms that have been detected within ecosystems using sequencing approaches but remain functionally uncharacterized could help connect microbial biodiversity and ecosystem function.
Results/Conclusions
Using measurements of taxon-specific carbon substrate utilization, we developed an approach for phylogenetic prediction of functional traits and tested this technique using in silico simulations. By combining qSIP data with phylogenetic trees we used a Brownian motion model of evolution to reconstruct ancestral traits. These ancestral traits were then used as the basis to predict trait values and variance estimates for uncharacterized taxa. To test the efficacy of this approach trait values were randomly removed from a portion of the taxa to simulate uncharacterized organisms, traits were predicted and compared to the actual values. This simulation process was performed for traits with varying strength of phylogenetic conservatism and repeated with varying proportions of the taxa as unmeasured. Confidence intervals (95%) for predicted trait values consistently encompassed the observed trait value. Our in silico analysis suggests the strength of a phylogeny based prediction is determined by 1) the evolutionary conservatism of the trait being predicted, 2) the proportion of taxa in the phylogeny with observed trait measurements, and 3) the evolutionary distance from organisms with observed trait values. Our results suggest that phylogenetic information may be used to predict the functional traits of uncharacterized taxa. This represents a step toward connecting microbial phylogeny with the functioning of species and their effect on ecosystem process rates.