Progressive bias in species status is symptomatic of fine-grained mapping units subject to repeated sampling

Atlas maps of butterflies invariably fail to distinguish the status of reco rds, that is, whether they are observations of breeding populations or vagr ant individuals. Yet, for conservation purposes, it is clearly important to know whether records relate to breeding populations in suitable habitats o r not. The high mobility of butterfly adults carries two expectations. Firs t, vagrants will frequently be recorded within grid squares, particularly w hen mapping is fine grained. Second, the frequency of vagrants is likely to be higher in mapping units comprising biotopes with few butterfly habitats , than in those rich in butterfly habitats, if only because there is a grea ter regional pool of potential vagrants for biotopes depleted in butterfly habitats. It follows that with repeated sampling of squares or biotopes dif ferences in species breeding status will be cumulative between those rich a nd poor in butterfly habitats. In this paper, these predictions have been t ested using data from 30, 1 ha grid squares on Alderley Edge Site of Specia l Scientific Interest (SSSI) (UK), half of which are in woodland (few butte rfly habitats) and half with open habitats (non-woodland; rich in butterfly habitats). Butterflies have been recorded during 38 visits over 4 years. W ithin squares, vagrants have been distinguished from breeding species on th e basis of the presence of larval hostplants and other suitable resources f or seasonal maintenance. Owing to the occurrence of vagrants, number of bre eding species was found to be significantly fewer than total species in map ping units. Vagrants were found to be significantly more abundant in woodla nd squares, which contain few species larval hostplants, than in squares do minated by open biotopes with many species larval hostplants. Repeated samp ling resulted in a cumulative increase in the frequency of vagrant species. Furthermore, the relative frequency of vagrants increased at a greater rat e in woodland squares compared to non-woodland squares. Suggestions are mad e as to how recording for atlases may be improved.