Populations can vary considerably in their response to environmental fluctuations, and understanding the mechanisms behind this variation is vital for predicting effects of environmental variation and change on population dynamics. Such variation can be caused by spatial differences in how environmental conditions influence key parameters for the species, such as availability of food or breeding grounds. Knowing how these differences are distributed in the landscape allows us to identify areas that we can expect the highest impact of environmental change, and where predictions on population dynamical effects will be most precise. We evaluated how wetland dynamics in the North-American prairies (pond counts; a key parameter for several waterfowl populations) were related to spatial and temporal variation in the environment, as measured by weather variables, primary productivity and phenology derived from annual normalized difference vegetation index (NDVI) curves, and agricultural composition of the landscape. Spatial and temporal variation in pond counts were closely related to these environmental variables. However, correlation strength and predictive ability of these environmental variables on wetland dynamics varied considerably across the study area. This variation was related to landscape characteristics and to the spatial scaling of the wetland dynamics, such that areas with late onset of spring, low spring temperature, high primary productivity, and high proportion of cropland had more predictable and spatially-homogenous dynamics. The success of predicting environmental influences on wetlands from NDVI measures derived from satellite images indicates they will be useful tools for assessing effects of changing landscape and climatic conditions on wetland ecosystems and their wildlife populations.