Testing for potential survey bias: the effect of roads, urban areas and nature reserves on a southern African mammal data set

Mammalian record collection densities and bias attributable to observer survey preferences are assessed for the former Transvaal region of South Africa using a geographic information system (GIS). A method is presented which may be used to distinguish between highly biased and relatively unbiased sampling densities. Small mammal survey records within the region appear to have been collected relatively independently of the road and urban area infrastructure and are not biased towards nature reserves. Large mammal data on the other hand has mostly been collected within existing conservation areas. Although this latter data set would appear to be highly biased, it does accurately reflect the current distributions of larger mammalian taxa, which are presently almost entirely restricted to conservation areas.