Environmental filtering increases in intensity at both ends of climatic gradients, though driven by different factors, across woody vegetation types of the southwest USA

Species distributions are theorized to be more intensively constrained by abiotic factors in severe than in benign environments. A similar concept can be applied to assemblages of species: environmental filtering is expected to increase in intensity in colder and drier environments. To assess the filtering effects of climate on vegetation at a regional scale, climate niche values were estimated for 338 woody species across 93 vegetation types from arid sub-tropical to alpine ecosystems of the southwest USA. The standardized range and spacing of climatic niche values in each vegetation type - used as estimates of the intensity of climatic and micro-environmental filtering, respectively - were correlated with the mean niche values of those vegetation types - used as surrogates for climatic gradients - in order to assess how filtering of vegetation composition varies along broad climatic gradients. The range of climatic niche values was narrower than expected in most vegetation types, indicating significant climatic filtering, with frost having the strongest average effect. Niche spacing differed little from null expectations. Variation in the intensity of climatic filtering along gradients of the same climate variable was primarily asymmetrical, and provided support for the hypothesis that abiotic filtering is most intense in cold and growing season dry environments. However, filtering patterns of at least one climatic factor along gradients of other climatic factors ran counter to the trend of increasing filter intensity in cold or dry environments. In other words, climatic factors exhibited interactive effects on vegetation filtering, often in antagonistic ways. The majority of these interactions were compatible with interspecific niche relationships that correspond with anatomical and physiological tradeoffs among drought, frost and heat tolerances. Filtering patterns and interspecific tradeoffs are likely to vary across taxa and biomes, and application of the methods presented here could help to explain such variation.