Speaker：Prof. Chun-Huo Chiu (Department of Agronomy, NTU)
Topic：Improvement and promotion of Chao’s richness lower bound estimator
Speaker：Prof. Chun-Huo Chiu (Department of Agronomy, NTU)
Date Time：Fri. Dec 16, 2022, 10:40 AM - 11:30 AM
Place： 4F-427, Assembly Building I
Online Seminars- Google Meet
https://meet.google.com/sqy-fzus-oog
https://meet.google.com/sqy-fzus-oog
Abstract
Species richness is the most intuitive and widely used quantitative diversity metric. However, due to resource limitations, investigating all species in the target area and gathering data on their abundance is nearly impossible; hence, observed richness is always underestimated, especially in highly heterogeneous communities or when the sample size is small. Dozens of estimators have been proposed to address the underestimation problem of observed richness. In the literature, methods for estimating the undetected richness of a sample are generally divided into two categories: parametric and nonparametric estimators. Without any assumptions on species detection rates, nonparametric methods demonstrate robust statistical performance and are widely used in ecological studies where Chao’s lower bound estimator is the most common estimator.
However, there are two issues remaining to address for Chao’s richness estimator, which is seriously underestimated in highly heterogeneous communities and not available for integrated data composed of species abundance data and incidence data. In this study, the parametric-based mixture models are used to address the underestimation of Chao's lower bound, and the sampling distribution theory of random sample is used to promote Chao's lower bound to integrated data.
Simulation results indicated that the new estimator reduces the bias presented in traditional nonparametric estimators, presents more robust statistical behavior compared to other parametric estimators, and supplies a more accurate confidence interval among the discussed estimators, especially in an assemblage with high species composition heterogeneity. The proposed new approach is also applied to one real data set to illustrate its application.
However, there are two issues remaining to address for Chao’s richness estimator, which is seriously underestimated in highly heterogeneous communities and not available for integrated data composed of species abundance data and incidence data. In this study, the parametric-based mixture models are used to address the underestimation of Chao's lower bound, and the sampling distribution theory of random sample is used to promote Chao's lower bound to integrated data.
Simulation results indicated that the new estimator reduces the bias presented in traditional nonparametric estimators, presents more robust statistical behavior compared to other parametric estimators, and supplies a more accurate confidence interval among the discussed estimators, especially in an assemblage with high species composition heterogeneity. The proposed new approach is also applied to one real data set to illustrate its application.