Uncovering the demographic basis of population fluctuations is a central goal of population biology. This is particularly challenging for spatially structured populations, which require disentangling synchrony in demographic rates from coupling via immigration. In this study, we fit a stage-structured metapopulation model to a 29-year times series of threespine stickleback abundance in the heterogeneous and productive Lake Myvatn, Iceland. The lake comprises two basins (North and South) connected by a channel through which the stickleback disperse. The model includes time-varying demographic rates, allowing us to assess the potential contributions of recruitment and survival, spatial coupling via immigration, and demographic transience to the population’s large fluctuations in abundance. Our analyses indicate that recruitment was only modestly synchronized between the two basins, whereas survival probabilities of adults were more strongly synchronized, contributing to cyclic fluctuations in the lake-wide population size with a period of approximately six years. The analyses further show that the two basins are coupled through immigration, with the North Basin subsidizing the South Basin and playing a dominant role in driving the lake-wide dynamics. Our results show that cyclic fluctuations of a metapopulation can be explained in terms of the combined effects of synchronized demographic rates and spatial coupling.

 The raw data are catch per unit effort (CPUE) of threespine stickleback from surveys conducted from 1991-2020 in Lake Myvatn, Iceland. Stickleback were sampled in June and August of every year at 8 locations within the lake by setting 5 minnow traps for two 12h sessions. The data are expressed as total number of fish captured, effort as number of traps set across both 12h sessions (usually 10), and CPUE calculated by dividing the total number of fish captured by the effort. We have also included a separate file with GPS coordinates for the sampling locations. Note that in 2016 only the CPUE data are available due to incompleteness in the raw data records. These data were used to fit and analyze a demographic model. The corresponding scripts are being uploaded separately to Zenodo.