1. Climate change has been identified as one of the most important drivers of wildlife populations. The development of appropriate conservation strategies relies on reliable predictions of population responses to climate change, which require in-depth understanding of the complex relationships between climate and population dynamics through density dependent demographic processes. Integrated population models (IPMs) are a type of modeling approach that unify the analyses of demography and abundance data, providing opportunities to understand and predict population demography and dynamics under climate change. 2. In this study we developed dynamic N-mixture models for large scale population estimates, which became an important component of the IPM we used in data analysis. We then analyzed four decades (1974-2014) of Mallard (Anas platyrhynchos) breeding population survey, band-recovery, and climate data covering a large spatial extent from North American prairies through boreal habitat to Alaska. Our goals were to examine the complex relationships among climate, density dependent processes, waterfowl population demography and dynamics, identify the key demographic parameters that are sensitive to climate change and are influential to population growth, and forecast population responses to climate change. 3. Our results revealed the interactive effects of temperature and density dependent processes on Mallard recruitment and to a less extent apparent survival. We also found that recruitment explained more variance of population growth than apparent survival. We then forecasted a decrease in Mallard breeding population density in the Northern Prairie Potholes and an increase in Mallard breeding population density in the northern part of our study area, indicating potential shifts in Mallard population dynamics under future climate change. 4. Synthesis and applications Different strategies need to be considered across regions to conserve waterfowl populations under climate change. Strategies that facilitate recruitment are essential for high-density populations that are relatively vulnerable to climate change. By contrast, low-density populations are relatively resilient to climate change and their habitats may serve as future climate refugia. Adaptive management is essential for evaluating management consequences. Our modelling framework approach can be easily adapted for other species and thus has wide applications in ecology and conservation.