Mao Chang Xuan, Colwell Robert K, Chang Jing
Department of Statistics, University of California, Riverside, California 92521, USA.
Biometrics. 2005 Jun;61(2):433-41. doi: 10.1111/j.1541-0420.2005.00316.x.
As a significant tool in ecological studies, the species accumulation curve or the collector's curve is the graph of the expected number of detected species as a function of sampling effort. The problem of estimating the species accumulation curve based on an empirical data set arising from quadrat sampling is studied in a nonparametric binomial mixture model. It will be shown that estimating the species accumulation curve not only is independent of the unknown number of species but also includes estimating the number of species as a limiting case. For the purpose of interpolation, moment-based estimators, associated with asymptotic confidence intervals, are developed from several points of view. A likelihood-based procedure is developed for the purpose of extrapolation, associated with bootstrap confidence intervals. The proposed methods are illustrated by ecological data sets.
作为生态学研究中的一项重要工具,物种累积曲线或采集者曲线是检测到的物种预期数量随采样努力程度变化的函数图像。基于样方采样产生的经验数据集估计物种累积曲线的问题,在非参数二项混合模型中进行了研究。结果表明,估计物种累积曲线不仅与未知的物种数量无关,而且在极限情况下还包括估计物种数量。出于插值的目的,从多个角度开发了与渐近置信区间相关的基于矩的估计量。为了外推的目的,开发了一种与自助置信区间相关的基于似然的程序。所提出的方法通过生态数据集进行了说明。