<![CDATA[Spiders were sampled from one-hectare tropical rainforest plots in three parks in northern Vietnam. Inventories were based on ecologically structured sampling employing five methods. A series of non-parametric estimators were used to extrapolate the true species richness from the samples for each locality and indicate the magnitude of sampling effort necessary to inventory a variety of protected Southeast Asian tropical forests. We investigated the Beta diversity between sites and explored the distinctness of the communities sampled by the various collecting methods. Our approach takes the incompleteness of our inven-tories into account and estimates the number of unobserved shared species. Rank sample abundance was positively correlated with number of sites observed. However, when sample abundance was scaled by incidence (as an index of de-tection probability), this relationship disappeared. This suggests no difference in the probability that abundant and rare species will be present in different sites even if the detection probability of rare species is low. The three sites differed in their observed and estimated point diversity with the lowest diversity site, Cuc Phuong, also having the least vertically-stratified spider community. The three sites, separated by 150–300 km and differing in vegetation community, eleva-tion, geology, and other attributes, experience an estimated 65–85% turnover in species composition over differences of this magnitude. We discuss the rationale for using the non-parametric estimator approach and caution that estimates can be unreliable when samples contain an insufficient portion of the community.]]>
