Influences of interpolation of species ranges on elevational species richness gradients

Interpolation of species ranges has been a common approach to compensate for the unevenness or incompleteness in sampling effort in studies of geographic species richness gradients. However, potential biases introduced by this estimation method remain unclear. Here, we presented an explicit examination of the influences of one‐dimensional interpolation on elevational species richness gradients, and discussed potential causes and processes of these influences. We conducted intensive surveys of birds along the elevational gradients of the Ailao Mountains, southwestern China, and compared richness patterns based on interpolation with raw data as well as estimated data from rarefaction and Chao1 non‐parametric estimator; we also compared results of multiple linear regressions and hierarchical partitioning analyses explaining these four measures of richness. Actual evapotranspiration (AET) and the mid‐domain effect (MDE) were highly correlated and separately provided a good potential explanation for the unimodal richness pattern in the Ailao Mountains, with modifying and suppressive effects of other variables such as area. Interpolation consistently and significantly increased the effects of AET/MDE, while it reduced contributions of area and human disturbance. Our results demonstrated that while compensating for biases in sampling effort, interpolation may also spuriously fill genuine distribution gaps, and tend to underestimate the effects of the non‐monotonic or discontinuous influencing factors that are responsible for these gaps, and overestimate the effects of other factors actually suppressed by these factors. These influences were most strong for species with relatively medium elevational ranges. We conclude that at the regional scale, interpolation method is a potential source of bias in identifying and explaining species richness gradients and should be used with careful consideration. It may be advantageous to adopt other robust estimation methods besides interpolation to gain a more accurate assessment of species richness and a more objective understanding of their underlying mechanisms.