Premium
A randomisation program to compare species‐richness values
Author(s) -
RICHARDSON JEAN M. L.,
RICHARDS MIRIAM H.
Publication year - 2008
Publication title -
insect conservation and diversity
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.061
H-Index - 39
eISSN - 1752-4598
pISSN - 1752-458X
DOI - 10.1111/j.1752-4598.2008.00018.x
Subject(s) - species richness , rarefaction (ecology) , biodiversity , statistics , ecology , estimator , sample size determination , sample (material) , species diversity , type i and type ii errors , global biodiversity , sampling (signal processing) , null model , variance (accounting) , null hypothesis , econometrics , statistical hypothesis testing , biology , mathematics , computer science , economics , chemistry , accounting , filter (signal processing) , chromatography , computer vision
. 1 Comparisons of biodiversity estimates among sites or through time are hampered by a focus on using mean and variance estimates for diversity measures. These estimators depend on both sampling effort and on the abundances of organisms in communities, which makes comparison of communities possible only through the use of rarefaction curves that reduce all samples to the lowest sample size. However, comparing species richness among communities does not demand absolute estimates of species richness and statistical tests of similarity among communities are potentially more straightforward. 2 This paper presents a program that uses randomisation methods to robustly test for differences in species richness among samples. Simulated data are used to show that the analysis has acceptable type I error rates and sufficient power to detect violations of the null hypothesis. An analysis of published bee data collected in 4 years shows how both sample size and hierarchical structure in sample type are incorporated into the analysis. 3 The randomisation program is shown to be very robust to the presence of a dominant species, many rare species, and decreased sample size, giving quantitatively similar conclusions under all conditions. This method of testing for differences in biodiversity provides an important tool for researchers working on questions in community ecology and conservation biology.