Counting Songbirds That Come and Go: A Better Sampling Model

A gray jay at the top of a spruce. NPS/Kevin Bacher.

Counting birds is harder than it sounds. Biologists often use distance sampling: observers record each bird they detect and how far away it is, and a model turns those records into estimates of how densely birds are spread across the landscape. A 2018 paper in the Journal of Applied Ecology, co-authored by National Park Service scientist Joshua Schmidt, improves that model for mobile birds such as passerines, the perching songbirds.

The problem

• The standard approach assumes birds are spread evenly within each survey plot and captures differences in density only from plot to plot.
• Spatial distance sampling, built on a "thinned point process", uses the exact positions of the birds detected, and so can explain more of the variation in density when birds cluster within a plot.
• But existing spatial models assume every bird is present and available to be counted. Birds that have temporarily left the plot (temporary emigration) bias the estimates of how many there are.

What the authors did

They extended spatial distance sampling to allow for temporary emigration (their TPP model) and compared it, in computer simulations, with an existing temporary-emigration model that assumes birds are spread evenly (TEDS). They then applied both to survey data for two songbird species in Alaska.

What they found

• In simulations the TPP model's estimates were more reliable and more precise, cutting the coefficient of variation of the population estimate by 26%.
• With the Alaska data, TEDS found only weak links between bird numbers and habitat, while TPP found strong links for the same habitat features, evidence that spatial models give much stronger results when density varies within plots.
• For the two species, the population estimates' coefficients of variation were 32% and 4% smaller under TPP.

The authors expect allowing for temporary emigration to become a key part of spatial distance-sampling models, especially for studying changes in the distribution and numbers of highly mobile species such as songbirds.

Sources

• National Park Service, "Sampling Methods for Passerines." https://www.nps.gov/articles/birdsampling.htm
• Mizel, J. D., J. H. Schmidt, and M. S. Lindberg. 2018. Accommodating temporary emigration in spatial distance sampling models. Journal of Applied Ecology 55:1456–1464.
• Words on this page from people and organisations outside the federal government are paraphrased; rewritten in hubnx's own words.