Migratory species can travel tens of thousands of kilometers each year, spending different parts of their annual cycle in geographically distinct locations. Understanding the drivers of population change is vital for conserving migratory species, yet the challenge of collecting data over entire geographic ranges has hindered attempts to identify the processes leading to observed population changes. Here, we use remotely sensed environmental data and count data to investigate the factors driving variability in abundance in two subspecies of a long-distance migratory shorebird, the bar-tailed godwit (Limosa lapponica). We compiled a spatially and temporally explicit dataset of three environmental variables to identify the conditions experienced by each subspecies in each stage of their annual cycle (breeding, non-breeding and staging). We used a Bayesian N-mixture model to analyze 18 years of monthly count data from 21 sites across Australia and New Zealand in relation to the remote sensing datasets. We found that the abundance of one subspecies (L. l. menzbieri) in their non-breeding range was related to climate conditions in breeding grounds, and detected sustained population declines between 1995 and 2012 in both subspecies (L. l. menzbieri, -6.7% and L. l. baueri, -2.1% year-1). To investigate the possible causes of the declines, we quantified changes in habitat extent at 22 migratory staging sites in the Yellow Sea, East Asia, over a 25-year period and found -1.7% and -1.2% year-1 loss of habitat at staging sites used by L. l. menzbieri and L. l baueri, respectively. Our results highlight the need to identify environmental and anthropogenic drivers of population change across all stages of migration to allow the formulation of effective conservation strategies across entire migratory ranges.