Document: PAU-3-82-30

Maximum species richness extrapolated from samples: An accuracy assessment of current estimators.

KLAWINSKI, P.*, J.ZIMMERMAN and J.THOMPSON

University of Puerto Rico, San Juan, PR 00936 1

Abstract:
Species richness (S) is becoming increasingly important from a conservation biology perspective (biodiversity inventories), yet our efforts to accurately measure S at different sites and statistically compare these values is still problematic. Prior evaluations of statistical estimators of S have concentrated on their statistical behavior but not on their accuracy. We examined the accuracy 7 indices (Chao1, Chao2, ACE, ICE, Jack1, Jack2, and Bootstrap estimators) for extrapolating maximum species richness (Smax) and compared them to true S for a 16 hectare vegetation dynamics grid in the Luquillo mountains of Puerto Rico. The grid was subsampled at 25m2, 100m2, 400m2 and 1 ha units to examine the effect that different sized plots might have on each of the estimators. By subsampling the larger grid, we could determine whether a series of subsamples of a larger area was capable of accurately estimating Smax of that larger area (analogous to carrying out a diversity inventory). We calculated the various estimators using EstimateS (Colwell 1997). All estimators of Smax exceeded the true Smax after approximately 65% of the total area was sampled. This was true regardless of the dimensions of the individual samples. Individual sample dimensions did, however, affect the performance of the different estimators. Incidence-based estimates (Chao2, ICE, and Jack2) increased with increases in the size of individual samples while one incidence-based estimator (Bootstrap) and the abundance-based estimators (Chao1, ACE, and Jack1) were not affected by the size of individual samples. Overestimates of true Smax varied from 22-70% higher than true species richness. We conclude that current mathematical methods for extrapolating Smax from a set of samples are inadequate and suggest that efforts to develop statistical methods for comparing S among sites (bootstrapping procedures), given similar sampling intensities, might be a more productive avenue of research.

Keywords: species richness, diversity, extrapolation, incidence-based estimator, abundance-based estimator

 Abstracts by Session: Symposia, Oral, Poster
Abstracts Listed by Title/Reference Number
Schedule of Sessions in Chronological Order
Sr. Author and Co-Authors
Information updates, contact source
Snowbird 2000 Program Web Site
Snowbird Page on the ESA Web Site

This abstract is being presented at: 9:45 AM in session:
Oral Session #2: Conservation Ecology.