Zoology Department, University of Otago, P.O. Box 56, Dunedin, New Zealand.
Ecology. 2011 Dec;92(12):2202-7. doi: 10.1890/11-0332.1.
The recent development of capture-recapture methods for estimating animal population density has focused on passive detection using devices such as traps or automatic cameras. Some species lend themselves more to active searching: a polygonal plot may be searched repeatedly and the locations of detected individuals recorded, or a plot may be searched just once and multiple cues (feces or other sign) identified as belonging to particular individuals. This report presents new likelihood-based spatially explicit capture-recapture (SECR) methods for such data. The methods are shown to be at least as robust in simulations as an equivalent Bayesian analysis, and to have negligible bias and near-nominal confidence interval coverage with parameter values from a lizard data set. It is recommended on the basis of simulation that plots for SECR should be at least as large as the home range of the target species. The R package "secr" may be used to fit the models. The likelihood-based implementation extends the spatially explicit analyses available for search data to include binary data (animal detected or not detected on each occasion) or count data (multiple detections per occasion) from multiple irregular polygons, with or without dependence among polygons. It is also shown how the method may be adapted for detections along a linear transect.
最近，用于估计动物种群密度的捕获-再捕获方法的发展集中在使用陷阱或自动相机等设备进行被动检测上。有些物种更适合主动搜索：可以反复搜索一个多边形样方，并记录检测到的个体的位置，或者只搜索一次样方，并将多个线索（粪便或其他迹象）识别为属于特定个体。本报告介绍了这种数据的新的基于可能性的空间显式捕获-再捕获（SECR）方法。结果表明，这些方法在模拟中至少与等效的贝叶斯分析一样稳健，并且在蜥蜴数据集的参数值下具有可忽略的偏差和接近标称置信区间覆盖。基于模拟，建议 SECR 样方应至少与目标物种的家域一样大。可以使用 R 包“secr”来拟合模型。基于可能性的实现将可用于搜索数据的空间显式分析扩展到包括来自多个不规则多边形的二进制数据（每次是否检测到动物）或计数数据（每次多次检测），以及多边形之间是否存在依赖关系。还展示了如何为线性样带沿线的检测来调整方法。
