Class Year

2026

Document Type

Student Research Paper

Date of Creation

Spring 2026

Department 1

Environmental Studies

Abstract

Birds’ annual cycles include biologically linked breeding and non-breeding seasons, yet most research and bird atlases focus exclusively on the breeding season. There have been very few winter bird atlas projects in North America, and none have been published for the northeastern United States. Short daylight hours and harsh weather limit winter survey coverage, making species distribution models (SDMs) essential for addressing spatial gaps.

I evaluated the optimal spatial scale for SDMs across bird species differing in body size and tested whether optimal scale varies with body mass. Using 16,967 eBird checklists collected during the first winter survey period (December 2024–February 2025) of the Third Pennsylvania Bird Atlas, along with high-resolution elevation and land-cover data, I fit Random Forest models of encounter rates for 30 bird species at four spatial scales (500 m, 1,000 m, 2,500 m, and 5,000 m).

I did not find a discernable relationship between grain size and model performance or between optimal modeling scale and bird body mass. Despite there being no evident performance gain, the enhanced precision of small-grain models makes them more effective for use in conservation and management as compared to large-grain models. My results will inform analytical approaches and survey design for the Third Pennsylvania Bird Atlas and other ongoing statewide winter atlas efforts.

Comments

Written for ES 460: Individualized Study - Research.

Download

Share

COinS