1. Migratory species form an important component of biodiversity; they link ecosystems across the globe, but are increasingly threatened by global environmental change. Understanding and mitigating threats requires knowledge of how demographic processes operate throughout the annual cycle, but this can be difficult to achieve when breeding and non-breeding grounds are widely separated.

2. Our goal is to quantify the importance of variability in survival during the breeding and non-breeding seasons in determining variation in annual survival using a single population and, more broadly, the extent to which annual survival across species reflects variation in probability of surviving the migratory period.

3. We use a 25 year dataset in which individuals of a long-distance migratory bird, the alpine swift Tachymarptis melba, were captured towards the beginning and end of each breeding season to estimate age- and season-specific survival probabilities and incorporate explicit estimation of the correlations in survival between age-classes and seasons.

4. Monthly survival was higher during the breeding period than during the rest of the year and strongly affected by conditions in the breeding season; effects that remained apparent in the following non-breeding season, but not subsequently. Recruitment of juveniles was dependent on the timing of breeding, being higher if egg-laying commenced before the median date, and substantially lower if not.

5. Across migratory bird species, variation in annual survival largely reflects variation in the probability of surviving the migratory period. Using a double-capture approach, even within a single season, provides valuable insights into the demography of migratory species, which will help understand the extent and impacts of the threats they face in a changing world.

The Alpine Swift data are a a mark-recapture data set based on two capture occassions each breeding season. The data are presented in m-array format - i.e. the number of birds released at occasion y (rows) recaptured at occasion x (columns). Annotated R code is provided to undertake the analysis.

The cross-species data (season.dat.rda) are derived from a literature search as described in the text.

Capture dates vary between years, thus the survival periods vary slightly. Some capture occasions are missing, the R code imputes dates for these. Birds are caught either as chicks, or full grown birds. Thus the 'adult' matrix includes a small number of unknown immature birds. The readme.txt file indicates what each file contains and the run_models.R file imports the data and does the analysis calling the various the JAGS model definitions.