LAUSR

Abstract Surveys on tropical invertebrates must gather as much information as possible over the shortest period, mainly because of financial limitations and hyperdiversity. Fruit-feeding butterflies in the subfamilies Biblidinae, Charaxinae, Nymphalinae and Satyrinae (Nymphalidae) are attracted to decaying material and can be sampled with standardized methodologies, but (1) some groups can be difficult to collect, despite being quite common in Amazonian forest understorey; moreover, (2) the duration of the sampling period is not consistent among studies and (3) the sufficient effort for financially limited projects remains unknown. With this study, we aimed to fulfill points 1–3 in order to recommend a less costly protocol for monitoring purposes in the Amazon. In 25 km2 of rainforest in the state of Amazonas, Brazil, we evaluated the performance of sampling schemes for these butterflies (four, three, two and one visit in 250 m-long plots), using both nets and bait traps, while considering reductions in sampling effort and the removal of rare and infrequent taxa to optimize field and laboratory work. Reduced-effort schemes are only validated if they reflect both taxonomic and ecological information provided by the maximum effort. Procrustes superimposition was used to estimate the dissimilarity in the spatial distribution of species between schemes. Spatial turnover in herb, liana, palm tree and diameter-at-breast-height > 10 cm tree species composition was used as predictor for the butterfly community through linear regressions. The three-visit scheme was sufficient to retrieve high species similarity and the ecological patterns observed with maximum effort. The two-visit scheme lost a significant amount of information on species composition similarity, but recovered stronger environmental relationships than those observed with the four-visit scheme. The removal of uncommon species did not affect the ecological response of the community, thereby suggesting that common species are driving the spatial patterns of the studied butterflies. Thus, large reductions in costs by reducing sampling effort could be achieved with relatively little loss of information on the species turnover of butterflies and their relationships with the environment. The proposed sampling protocols with reduced effort will allow projects to use their time and financial supply more effectively, showing that cost-effective shortcuts for biodiversity assessments can be useful for conservation, biomonitoring and land use management.