Optimizing performance of nonparametric species richness estimators under constrained sampling.

摘要: Understanding the functional relationship between sample size and performance of species richness estimators is necessary to optimize limited sampling resources against estimation error. Nonparametric such as Chao Jackknife demonstrate strong performances, but consensus lacking which estimator performs better under constrained sampling. We explore a method improve scenario. The we propose involves randomly splitting species-abundance data from single into two equally sized samples, using an appropriate incidence-based estimate richness. To test this method, assume lognormal distribution (SAD) with varying coefficients variation (CV), generate samples MCMC simulations, use expected mean-squared error criterion estimators. for Chao, Jackknife, ICE, ACE Between abundance-based sample, split-in-two Chao2 performed best when CV   0.65, given that ratio observed greater than critical value by power function CV respect abundance sampled population. proposed increases substantially more effective rare are in assemblage. also show works qualitatively similarly well SADs log series, geometric negative binomial. application estimating zooplankton communities ballast water. alternative large number individuals increase accuracy estimations; therefore, it wide range resource-limited scenarios ecology.