Estimating the number of species shared by incompletely sampled communities

Yi Zou,Jan Christoph Axmacher

doi:10.1111/ecog.05625

Yi Zou, Jan Christoph Axmacher

Open Access

https://doi.org/10.1111/ecog.05625

Copy DOI

Abstract

There are numerous ways to estimate the true number of species in a community based on incomplete samples. Nonetheless, comparable approaches to estimate the number of species shared between two incompletely sampled communities are scarce. Here, we introduce the ‘total expected species shared' (TESS) measure and provide the R function for its calculation. Based on parametric asymptotic models, TESS provides estimates of the true number of species shared between incompletely sampled communities based on abundance data. We compare TESS results with abundance‐based non‐parametric methods in terms of precision and accuracy, using different simulated sampling scenarios. We further calculate TESS using an empirical dataset, highlighting changes in accuracy and precision with increasing sample size. We also demonstrate how TESS values can be combined with species richness estimators in turnover estimates using traditional β‐diversity indices. Our results show that mean values of TESS reliably approximate the true shared species number for varying sample completeness scenarios, with both accuracy and precision increasing with increasing sample completeness. Overall, we demonstrate the viability of TESS in estimations of the true number of species shared between two incompletely sampled communities. We also stress the importance of a sufficient sample size for the accuracy of estimates – requiring sampling designs that carefully balance sampling effort per site with the number of sampling sites.

Highlights

Introduction βDiversity describes the change in the assemblage composition between different communities (Whittaker 1960) and is commonly measured as the change in species composition between pairs of samples, resulting insimilarity matrices (Tuomisto 2010)
For two samples randomly drawn from the simulated communities, the ‘total expected number of species shared’ (TESS) between control and respective scenario community was generally much closer to the true number of species shared between the two underlying communities than the number of species shared between the two samples (Fig. 2)
total expected species shared’ (TESS) values increased strongly with increasing sample size for very low sample completeness (0.2–0.5, i.e. from 25 to 100 individuals randomly taken from the two baseline communities of approximately 100 000 individuals) and rapidly approached the true number of species shared when sample sizes increased further (Fig. 2)

Summary

Introduction

Diversity describes the change in the assemblage composition between different communities (Whittaker 1960) and is commonly measured as the change in species composition between pairs of samples, resulting in (dis)similarity matrices (Tuomisto 2010) Such (dis)similarity matrices are calculated using a variety of approaches that compare the observed number of species shared by, and unique to, the paired samples (Koleff et al 2003, Baselga 2010). The only available approach has been a non-parametric method that uses frequencies of shared rare species (Chao et al 2000), with subsequent suggested improvements (Pan et al 2009) These calculations form extensions of the abundance-based coverage estimator (ACE, Chao and Lee 1992) that estimates the number of species in a community. Parametric asymptotic models based on curve fitting, which have been used to estimate species richness within a single community (Flather 1996, Rosenzweig et al 2003), have never been explored to estimate shared species between two communities

Methods

Results

Conclusion