Comparison of methods for estimating density of forest songbirds from point counts

Authors

  • Jennifer L. Reidy,

    Corresponding author
    1. Department of Fisheries and Wildlife Sciences, University of Missouri, 302 Anheuser-Busch Natural Resources Building, Columbia, MO 65211, USA
    • Department of Fisheries and Wildlife Sciences, University of Missouri, 302 Anheuser-Busch Natural Resources Building, Columbia, MO 65211, USA.
    Search for more papers by this author
  • Frank R. Thompson III,

    1. United States Department of Agriculture, Forest Service, Northern Research Station, University of Missouri, 202 Anheuser-Busch Natural Resources Building, Columbia, MO 65211, USA
    Search for more papers by this author
  • J. Wesley Bailey

    1. Department of Fisheries and Wildlife Sciences, University of Missouri, 302 Anheuser-Busch Natural Resources Building, Columbia, MO 65211, USA
    Current affiliation:
    1. Forest Wildlife and Populations Research Group, Minnesota Department of Natural Resources, 1201 East Highway 2, Grand Rapids, MN 55744, USA.
    Search for more papers by this author

  • This article is a US Government work and, as such, is in the public domain in the United States of America.

  • Associate Editor: Steven S. Rosenstock.

Abstract

New analytical methods have been promoted for estimating the probability of detection and density of birds from count data but few studies have compared these methods using real data. We compared estimates of detection probability and density from distance and time-removal models and survey protocols based on 5- or 10-min counts and outer radii of 50 or 100 m. We surveyed singing male Acadian flycatchers (Empidonax virescens), cerulean warblers (Dendroica cerulea), Kentucky warblers (Oporornis formosus), Louisiana waterthrushes (Parkesia motacilla), wood thrushes (Hylocichla mustelina), and worm-eating warblers (Helmitheros vermivorum) in bottomland and upland forest across 5 states in the Central Hardwoods Bird Conservation Region during the breeding season in 2007 and 2008. Detection probabilities differed between distance and time-removal models and species detectabilities were affected differently by year, forest type, and state. Density estimates from distance models were generally higher than from time-removal models, resulting from lower detection probabilities estimated by distance models. We found support for individual heterogeneity (modeled as a finite mixture model) in the time-removal models and that 50-m radius counts generated density estimates approximately twice as high as 100-m radius counts. Users should be aware that in addition to estimating different components of detectability, density estimates derived from distance and time-removal models can be affected by survey protocol because some count durations and plot radii may better meet model assumptions than others. The choice of a method may not affect the use of estimates for relative comparisons (e.g., when comparing habitats) but could affect conclusions when used to estimate population size. We recommend careful consideration of assumptions when deciding on point-count protocol and selection of analysis methods. © 2011 The Wildlife Society.

Ancillary