期刊
FOREST ECOLOGY AND MANAGEMENT
卷 255, 期 3-4, 页码 1002-1010出版社
ELSEVIER
DOI: 10.1016/j.foreco.2007.10.007
关键词
jackknife estimator; species richness estimation; large area forest inventory; species identification
类别
Sample-based large area forest inventories are carried out in many countries specifically to provide information for forest policy and related decision making. In addition, ecologically meaningful information can be derived from the inventory data, including tree species richness. However, sample-based estimation of the total number of species is notoriously difficult. This study explores the strengths and weaknesses of large area forest inventory data in species richness estimation with a focus on the corresponding higher order jackknife estimators. Our test data set comes from a regional forest inventory in Northern Costa Rica. In our study, none of the kth order jackknife estimators reaches an asymptote and most have standard errors increasing with sample size. Inexplicably, the 4th and 5th order jackknives fail as estimators for our data-the estimates are almost identical to and even lower than the observed number of species. Additionally, higher order jackknives show large sampling variation. The 1st and 2nd order jackknives have small variation and thus are more reliable than higher order jackknives. The failure of the higher order jackknives suggests that the bias reduction theories underpinning the generalized jackknives are possibly not suited to the problem of species richness estimation. Care is, therefore, needed when interpreting a jackknife estimate of species richness. Various relevant practical issues in large area forest inventory implementation, particularly in the tropics, are discussed, including the accuracy of tree species identification and the presence of a certain percentage of unidentified tree individuals. These issues, obviously, pose an additional challenge in species richness estimation. (C) 2007 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据