Spatial biases
Biases in ERM data were inconsistent across space and taxa. Relative to their global area, provincial and country boundaries played disproportionate roles in delineating species ranges, especially in mammals and amphibians. Political boundaries near temperate-tropical transitions in particular have high levels of bias, including borders in South China and northern Southeast Asia (Figure 2, Figure S1), as well as the southwestern Brazil, making the use of these data for these areas exceptionally risky. In these cases, careful assessment for the possibility of strong administrative biases is needed, as using ERMs at these transitions may cause significant errors in analysis, such that alternate approaches such as models or trimmed MCPs should be used where such data exist.