2.2 Encounter data
We used mark-recapture data from 899 adult penguins, 387 at Robben Island and 512 at Stony Point (Table A1). Encounter data were collected over the breeding season (March to October); survival rates consequently refer to the nonbreeding season (November to February). We assigned each penguin a state both during initial marking and each subsequent encounter. Individuals were considered breeders if they were attending chicks or eggs in a nest site or guarding a nest site during nest inspections. Additionally, if they were encountered in the colony via the ground reader ≥ 6 times over a minimum of 12 days and a maximum of 120 days but not confirmed as breeders during nest inspections, they were considered breeders based on knowledge of African penguin breeding biology (Williams & Cooper 1984). To ensure that any birds encountered as nonbreeders were definitely skipping reproduction and were not younger birds yet to breed for the first time, we included an additional state of ‘prebreeder’, assigned to any birds marked/encountered as nonbreeders that had not yet been encountered breeding within their encounter histories.
Earlier tagging of African penguins used stainless steel flipper bands (e.g., Sherley et al. 2014). However, these have now been phased out due to potential data bias, both through human error (e.g., incorrectly reading tags) and potential deleterious effects on penguin survival and behaviour (e.g., Dugger et al. 2006; Dann et al. 2014). Within this dataset, four double marked (flipper band from previous tagging, and transponder from this study) individuals were removed to reduce any potential bias. Additionally, movement between colonies is rare in breeding African penguins (Sherley et al.2014); here, one individual encountered breeding at >1 colony was removed from the dataset to improve the reliability of model estimates by reducing the need for more complex models incorporating inter-colony movement (as this was not the focus of our study).