OOS 76-9
Trait-based ecological strategies explain microbial responses to environmental change

Thursday, August 13, 2015: 4:20 PM
328, Baltimore Convention Center
Kelly Gravuer, Graduate Group in Ecology and Department of Plant Sciences, University of California, Davis, CA, USA
Anu Eskelinen, Department of Environmental Science and Policy, University of California, Davis, CA, USA ; Department of Ecology, University of Oulu, , Finland
Susan Harrison, Department of Environmental Science and Policy, University of California, Davis, CA, USA
Background/Question/Methods

Although bacteria and archaea (hereafter, "microbes") drive many element cycles, the complexity and inscrutability of microbial communities have largely prevented their inclusion in ecological and biogeochemical models.  One promising approach is to distill microbial communities down to a manageable number of ecological strategies, each of which has characteristic environmental responses and ecosystem process effects.  Efforts to date have struggled to develop quantitative, broadly applicable methods to achieve this distillation.

Estimating ecologically important microbial traits using 16S rRNA sequence-based phylogenetic placements could provide a way forward.  We developed a reference tree of 1475 high-quality annotated microbial genomes and used it to test for phylogenetic conservatism of seven ecologically important traits.  Highly-conserved traits were then estimated for soil microbial communities from a grassland environmental change experiment on adjacent high- and low-productivity soils. For three years prior to sampling, replicate plots on each soil type received late spring precipitation addition and NPK nutrient addition in a factorial design.  We used cluster analysis to identify co-occurring suites of traits ("ecological strategies") and linear mixed models to determine how relative abundances of microbes with these strategies varied across soils and treatments.

Results/Conclusions

Four traits showed strong phylogenetic signal: rRNA gene copy number (maximum growth rate), genome size, sporulation capability, and anaerobic tolerance.  Cross-validation indicated strong correlations between observed and estimated values (0.88, 0.89) for continuous traits and high percentages of correct classification (92%, 96%) for binary traits.  Results were also robust to exclusion of the least certain 20% of estimates.

Cluster analysis defined three ecological strategies.  The first group had high rRNA copy numbers and large genomes and were more abundant on high-productivity soil.  They responded to both treatments, increasing with nutrients (F1,117=12.9, p<0.001) and decreasing with precipitation (F1,117=19.8, p<0.001). The second group had low rRNA copy numbers, medium-sized genomes, and low anaerobic tolerance.  They were more abundant on low-productivity soil and had opposite responses, decreasing with nutrients (F1,117=37.7, p<0.001) and increasing with precipitation (F1,117=29.5, p<0.001). The final group had low rRNA copy numbers, small genomes, and high anaerobic tolerance.  They were more abundant on high-productivity soil and their treatment responses were similar to those of the first group.

These relationships suggest that defining and quantifying microbial ecological strategies via trait estimation holds promise for ecologically-meaningful prediction of their environmental change responses and, ultimately, of the consequences of these responses for important ecosystem functions.