Abstract
This study explores the use of TWINSPAN (Hill, 1979; Hill & Šmilauer, 2005) to analyse microlepidoptera data gathered through opportunistic recording by amateurs in 1991-2015 in the vice-county of Shropshire (VC40). Biases in the database are explored and found to be consistent with those identified for opportunistic data by the literature. Following an initial test the database was divided into two files, a Day file containing data obtained by daytime recording methods (searching, sweep-netting, beating vegetation) and a Night file containing data obtained by light-trapping. A minimum number of species per sample was established experimentally. Each file was then analysed separately using TWINSPAN.
Most groups in the Day file output may be interpreted along ecological lines, and suggest the influences chiefly of seasonality, soil type and associated vegetation, altitude and anthropogenic factors including habitat management. Indicator species in the Day file output are mostly monophagous or near-monophagous moths, allowing the interpretation of groups through the ecologies of larval host plants. Ecological trends suggested by the Night file output are not strong, and light-trapping appears to offer little potential for extracting ecological information. The main influence found to affect the Night file output is sample size: samples with fewer species are separated by the program from those with more species. This finding may have implications for the use of light-trapping data in understanding and mapping species distributions. Some methodological issues arising from the use of TWINSPAN in this study are also discussed.