Temporal shift in density dependence among North American breeding duck populations

Environmental perturbation can have a marked influence on abundance and trend in many animal populations, but information is scant on how numerical change relates to variability in density-dependent and density-independent processes acting on populations. Using breeding population estimates for 10 duck species from a survey area of approximately 2.2 million km2 in central North America (1955-2005), we compared population growth models and related parameters among species and across time. All duck species showed evidence of density-dependent growth, and the best-fit relationship between population growth (r(t)) and population size (N(t)) was linear or convex for all species. Density dependence and associated population parameters were not related to an index of species life history strategy. Reanalysis of segmented (1955-1979, 1980-2005) r(t) time series, where the truncation date coincided with a putative decline in wetland availability on breeding grounds, showed that density-dependent forces were weakened during the latter time segment. Additionally, in later years most populations experienced increased first-order autocorrelation in annual counts, decreased intrinsic growth rate, increased nonlinearity in the relationship between r(t) and N(t), increased equilibrium return time, and increased inter-species synchrony in numbers. Such changes were not closely related to species life history strategy or to shifts in mean population size, average trend, and estimated carrying capacity. We speculate that shifts in breeding duck habitat quality altered historical predator-prey dynamics in the system and thereby underlie observed dynamical changes. The paradoxical finding that population abundance and trend do not reveal shifts in population processes highlights the need to go beyond simple numerical assessment when evaluating population responses to environmental perturbation.