Measures of geographic range size: the effects of sample size

Abstract

A number of methods have been used for quantifying the sizes of the geographic ranges of species. The consequences of different levels of sampling (the proportion of actual spatial occurrences) are explored for eight of these, using data on the occurrences of butterfly species on a 10 x 10 km grid across Britain. For all methods, the percentage error of estimation (PEE) decreases with the number of 10 x 10 km squares which a species occupies, most rapidly for extent measures, and more rapidly for area measures than for measures of numbers of units occupied. The rate of decline in PEE itself falls as sampling effort increases. At a given sampling level, rank correlations between range sizes measured by different methods are generally high. but there is no consistent change in the magnitude of these correlations as the level of sampling increases. The composition of the set of species with the smallest range sizes changes with the level of sampling.