Drought stress limits the geographic ranges of two tree species via different physiological mechanisms.

Range shifts are among the most ubiquitous ecological responses to anthropogenic climate change and have large consequences for ecosystems. Unfortunately, the ecophysiological forces that constrain range boundaries are poorly understood, making it difficult to mechanistically project range shifts. To explore the physiological mechanisms by which drought stress controls dry range boundaries in trees, we quantified elevational variation in drought tolerance and in drought avoidance-related functional traits of a widespread gymnosperm (ponderosa pine - Pinus ponderosa) and angiosperm (trembling aspen - Populus tremuloides) tree species in the southwestern USA. Specifically, we quantified tree-to-tree variation in growth, water stress (predawn and midday xylem tension), drought avoidance traits (branch conductivity, leaf/needle size, tree height, leaf area-to-sapwood area ratio), and drought tolerance traits (xylem resistance to embolism, hydraulic safety margin, wood density) at the range margins and range center of each species. Although water stress increased and growth declined strongly at lower range margins of both species, ponderosa pine and aspen showed contrasting patterns of clinal trait variation. Trembling aspen increased its drought tolerance at its dry range edge by growing stronger but more carbon dense branch and leaf tissues, implying an increased cost of growth at its range boundary. By contrast, ponderosa pine showed little elevational variation in drought-related traits but avoided drought stress at low elevations by limiting transpiration through stomatal closure, such that its dry range boundary is associated with limited carbon assimilation even in average climatic conditions. Thus, the same climatic factor (drought) may drive range boundaries through different physiological mechanisms - a result that has important implications for process-based modeling approaches to tree biogeography. Further, we show that comparing intraspecific patterns of trait variation across ranges, something rarely done in a range-limit context, helps elucidate a mechanistic understanding of range constraints.