Pii: S0006-3207(02)00190-8

This paper introduces and discusses the consistency and effectiveness of an inexpensive modification of Pollard transects for assessing the diversity and abundance of tropical butterfly communities in two national parks in Rio de Janeiro, Brazil. To measure butterfly abundance, students walked simultaneous timed transects at the interface of forest and cleared areas. They either counted the number of individuals observed or tabulated the number of specimens collected with nets. After a short training period, the number of butterflies observed or collected on simultaneous transects was statistically indistinguishable among student groups, and there was a significant positive correlation between observation counts and collected number of individuals. As a measure of species richness, the number of butterfly species sampled on each simultaneous transect was tabulated and did not differ statistically. To measure diversity, alpha of the logseries model was calculated for each collected sample, and statistical fit to a logseries model was determined. Although virtually all daily samples and the year's accumulated sample at one park fit the logseries model, about 35% of the daily samples and the entire year's sample did not fit at the other park. Despite these differences between the two parks, values for alpha from daily samples at both parks varied similarly (from 15 to 50 in almost all cases), and values from the entire year's samples were statistically indistinguishable. The repeatability of results among novices, such as students, suggests that timed transects have great promise for furthering our understanding of butterfly community demographics. Published by Elsevier Science Ltd.