Over gradients of increasing abiotic stress, phylogenetic community structure may shift from even to clustered (i.e., co-occurring species less vs. more related than expected by chance). I tested the predictions of this “stress-relatedness” hypothesis by measuring plant community composition within 109 field sites (each 2000-m²) as well as 14 nested quadrats (1-m²) distributed on serpentine soils throughout California. For each site and quadrat, climate variables were derived and soil metrics were measured as indices of abiotic stress. A net relatedness index value for each site and quadrat was calculated using the Comstruct module of Phylocom based on a phylogeny made using Phylomatic.

Results/Conclusions

At the site-level scale, I found strong linear relationships of climate with relatedness (precipitation - r² = 0.54, P <0.0001, n = 109; temperature - r² = 0.30, P < 0.0001, n = 109): Serpentine plant communities in environmentally benign, cool and wet sites were predominantly even (chaparral and woodland) while communities in stressful, hot and dry sites (barren and grassland) were clustered. Phylogenetic evenness in relatively rich sites may imply competition prevents local community coexistence of closely related taxa. Alternatively, if rich sites support multistoried, functionally diverse communities, co-occurring species may be less related than expected by chance. Clustering may reflect environmental filtering for closely related stress tolerators in stressful sites. Edaphic variables did not significantly explain relatedness, likely because site selection was stratified to include only serpentine soils. At the smaller 1-m² scale, variability in relatedness among the 14 quadrats within each site markedly increased over a gradient of site-level precipitation (r² = 0.29, P <0.0001, n = 109). This increasing spatial turnover in community relatedness with decreasing abiotic stress may suggest that local interactions, environment heterogeneity, and contingency strongly influence community formation in lower stress sites. The results generally support the stress-relatedness hypothesis, and represent a first attempt to incorporate evolutionary relatedness into a predictive framework of community assembly.