Demographic correlations are pervasive in wildlife populations and can represent important secondary drivers of population growth. Empirical evidence suggests that correlations are in general positive for long-lived species, but little is known about the degree of variation among spatially segregated populations of the same species in relation to environmental conditions. We assessed the relative importance of two cross-season correlations in survival and productivity, for three Atlantic puffin (Fratercula arctica) populations with contrasting population trajectories and non-overlapping year-round distributions. The two correlations reflected either a relationship between adult survival prior to breeding on productivity or a relationship between productivity and adult survival in the subsequent year. Demographic rates and their correlations were estimated with an integrated population model, and their respective contributions to variation in population growth were calculated using a transient life table response experiment. For all three populations, demographic correlations were positive at both time lags, although their strength differed. Given the difference in year-round distributions of these populations, this variation in the strength of population-level demographic correlations points to environmental conditions as an important driver of demographic variation through life-history constraints. Consequently, the contributions of variances and correlations in demographic rates to population growth rates differed among puffin populations, which has implications for – particularly small – populations’ viability under environmental change as positive correlations tend to reduce the stochastic population growth rate.

Mark-recapture histories of adult puffins were available for 699 individuals at Isle of May (1984–2019), 572 individuals at Røst (1990–2019), and 952 individuals at Hornøya (1990–2019). Breeding puffins were caught and marked with individually coded colour-rings or a unique colour ring combination. Birds were captured either in the nest burrow (Isle of May, Hornøya), with noose traps (Hornøya) or in mist nets erected on the colony surface (Røst). Visual resightings of ringed birds were conducted in subsequent years, predominantly in the areas where puffins had been ringed. Productivity data consisted of annual numbers of fledged chicks (F_t) from a sample of monitored pairs (E_t) that made a breeding attempt (defined as egg laid on Isle of May and Hornøya and egg hatched on Røst). Island-wide population counts (C_t) of adult breeding pairs were conducted at each colony. For Røst and Hornøya, the number of breeding pairs (based on the number of apparently occupied burrows) was up-scaled from counts in study plots made every year during the study period (1990–2019) (see Anker-Nilssen & Røstad 1993 for methodological details for Røst). At the Isle of May, total population counts of occupied burrows were made in 1984, 1989, 1992, 1998, 2003, 2008, 2013, and 2017.