QUANTITATIVELY SAMPLING LAND-SNAIL SPECIES RICHNESS IN MADAGASCAN RAIN-FORESTS

Land-snail species richness in tropical rainforests tends to be high b ut difficult to assess because of low densities and often small shell sizes. We tested three quantitative sampling methods in primary rainfo rests of southeastern Madagascar. Timed searching yielded seven times as many micro-snail species (species that during at least part of thei r life have shells < 5 mm maximum dimension) per person-hour as either litter sampling or soil-plus-litter sampling. The number of species f ound in 20 m x 20 m during three person-hours of searching, however, w as boosted a maximum of 38% by one eight-liter sample each of litter a nd soil-plus-litter. Litter sampling and timed searching both yielded more than 1.5 times the proportion of live-collected species as soil-p lus-litter sampling. Sampling method was unbiased toward 12 of the 20 commonest species, but three large, presumed arboreal species were fav ored by timed searches; two minute, presumed burrowers by soil-plus-li tter sampling; and three minute, cryptically colored species by both l itter and soil-plus-litter sampling. A 1.2-mm sieve caught at least 78 % of the total specimens and passed adults of 7% of species, of which the smallest adult dimension was 1.0 mm. These results suggest that th e best sampling strategy is timed searching for micro-snails, while in cidentally collecting macro-snails and litter-plus-soil for later pick ing of the 5.5-1.2 mm and the 1.2-0.85 mm, dry-sieved fractions. This strategy should be transferable to other tropical-rainforest land-snai l faunas.