Effects of weather variation on waterfowl migration: Lessons from a continental‐scale generalizable avian movement and energetics model

Abstract

Abstract We developed a continental energetics‐based model of daily mallard (Anas platyrhynchos) movement during the non‐breeding period (September to May) to predict year‐specific migration and overwinter occurrence. The model approximates movements and stopovers as functions of metabolism and weather, in terms of temperature and frozen precipitation (i.e., snow). The model is a Markov process operating at the population level and is parameterized through a review of literature. We applied the model to 62 years of daily weather data for the non‐breeding period. The average proportion of available habitat decreased as weather severity increased, with mortality decreasing as the proportion of available habitat increased. The most commonly used locations during the course of the non‐breeding period were generally consistent across years, with the most inter‐annual variation present in the overwintering area. Our model revealed that the distribution of mallards on the landscape changed more dramatically when the variation in daily available habitat was greater. The main routes for avian migration in North America were predicted by our simulations: the Atlantic, Mississippi, Central, and Pacific flyways. Our model predicted an average of 77.4% survivorship for the non‐breeding period across all years (range = 76.4%–78.4%), with lowest survivorship during autumn (90.5 ± 1.4%), intermediate survivorship in winter (91.8 ± 0.7%), and greatest survivorship in spring (93.6 ± 1.1%). We provide the parameters necessary for exploration within and among other taxa to leverage the generalizability of this migration model to a broader expanse of bird species, and across a range of climate change and land use/land cover change scenarios.

Keywords

• avian
• energetics
• global climate change
• migration
• predictive modeling
• waterfowl