LAUSR

Along elevational gradients, species richness often peaks at intermediate elevations and not the base. Here we refine and test eight hypotheses to evaluate causes of a richness peak in trees of the eastern Himalaya. In the field, we enumerated trees in 50 plots of size 0.1 ha each at eight zones along an elevational gradient and compared richness patterns with interpolation of elevational ranges of species from a thorough review of literature, including floras from the plains of India. The maximum number of species peaks at similar elevations in the two data sets (at 500 m in the field sampling and between 500 m and 1,000 m in range interpolation); concordance between the methods implies that statistical artefacts are unlikely to explain the peak in the data. We reject most hypotheses (e.g., area, speciation rate, mixing of distinct floras). We find support for a model in which climate (actual evapotranspiration [AET] or its correlates) sets both the number of species and each species optimum, coupled with a geometric constraint. We consider that AET declines with elevation, but an abrupt change in the association of AET with geographical distance into the plains means that the location of highest AET, at the base of the mountain, receives range overlaps from fewer species than the location just above the base. We formalize this explanation with a mathematical model to show how this can generate the observed low-elevation richness peak.