An agent-based model to quantify energetics, movement and habitat selection of mid-continent mallards in the Mississippi Alluvial Valley

Abstract

The dynamics of wintering waterfowl populations at the landscape scale are the result of complex interactions of environmental, behavioral and energetic drivers. Agent-based models provide a method to directly link these factors in a spatially explicit framework and allow the emergence of patterns from the aggregation of individual agent actions. We adapted the Spatially-explicit Waterbird Agent-based Model Program (SWAMP), originally developed for waterfowl in central California, to simulate a basin-scale population of mallards (A. platyrhynchos) wintering in the Mississippi Alluvial Valley over a four-month period (November–February). Simulated agents move within the landscape, foraging on areas made available based on a probabilistic inundation status, and converting food resources to endogenous energy. The model uses a high-resolution map of eastern Arkansas waterfowl habitats and incorporates a hierarchical habitat selection system that enables mallards to relocate at increasing scales in response to changing food availability. We validated the performance of modeled mallard body condition and behavioral metrics under a range of environmental conditions against expected outcomes derived from empirical data and found that the simulation produced realistic representations of changes in flight distances, energy expenditure, lipid storage, and foraging habitat use in response to depleting food resources over time. We discuss the model's applicability as a tool to quantify waterfowl response to a range of environmental conditions and to evaluate scenarios of landscape composition and configuration in the context of waterfowl population management.