Reproductive performance typically improves with age, reaching a plateau at middle age and subsequently declining in older age classes (senescing individuals). Three potential non-exclusive mechanisms can explain the improvement in reproductive performance with age: (1) selection (poor quality individuals are removed from the population with increasing age), (2) constraint (individual efficiency increases through experience) and (3) restraint (reproductive investment increases with age as the residual reproductive value decreases). While all three mechanisms received strong empirical support, few studies have aimed at teasing apart those hypotheses and understanding their underlying functioning. In little penguins (Eudyptula minor), we used a 19-year longitudinal dataset on breeding and foraging of more than 450 individuals to investigate the effect of age on breeding success. We separated within- from among-individual age-effects using state-of-the-art statistical methods (within-subject centering and population change decomposition). We then assessed whether within-individual changes in breeding resulted from ontogenetic changes in foraging performances, breeding phenology or access to mates and nest sites. Fidelity and assortative pairing explained the high correlation in male and female ages within a pair. Breeding performances followed a typical bell-shaped curve with performance increasing up to 8 years-old, before reaching a plateau and subsequently declining after age 16. Both selection and within-individual processes occurred, although within-individual changes dominated differences in age-dependent breeding success. The selective appearance had almost no effect (apart from ages 2 to 3), and selective disappearance mostly affected changes at old ages (above 16), although they were also responsible for the slight increase in reproductive performances from ages 5 to 8. Focusing on within-individual changes, birds exhibited higher performances at middle ages, with birds foraging better, laying earlier and changing partner and nest less often. Their reproductive investment did not vary with age for females and slightly decreased for males. This supports the constraint hypothesis but not the restraint one. Finally, the increase in breeding performances at young ages was explained by the age-related increase in foraging performances during chick-rearing and advancement of laying. In contrast, reproductive senescence was defined by a general decrease in bird performances.

This data comes from a long-term monitoring of little penguin breeding at Summerland Peninsula on the western tip of Phillip Island (33° 31'S, 145° 09'E), Victoria, Australia. This study covers 19 breeding seasons from 2000 to 2018 (inclusive), where each breading season spans from August to March the year after (the austral summer). Sex was determined by bill measurements when the birds were first found in the colony as adults, while age was known from bird tagging as chicks. All penguins included in this study nested in 100 artificial burrows (nest boxes) established behind the Penguin Parade® area. The artificial nest boxes were monitored three times a week during the breeding season to assess breeding phenology and success of each pair. Individuals with transponders were detected in the nest using a purpose-built handheld transponder reader (Kean Electronics, Australia). This monitoring ensured a recapture probability close to one and informed us about laying date, the number of clutches per season (1 to 3), as well as the number of fledglings per breeding event (0 to 2). Chicks were considered fledged when they had all their adult feathers and were older than 40 days when last seen. Fledging success was estimated as the number of chicks that survived to fledging per breeding event. Attendance data were recorded by an Automatic Penguin Monitoring System (APMS) located along the main path into the Penguin Parade study site, and were used to calculate three foraging parameters: foraging trip duration, change in adult body mass, and chick meal size. The APMS consists of an electronic transponder reader and a weighing platform that records the transponder identification, mass, and time of a penguin crossing the weighbridge. Foraging trip duration was calculated as the number of days at sea between colony visits. Foraging duration was measured in days, as little penguins depart before sunrise and return after sunset. Short trips were defined as those lasting less than or equal to 2 days (<2 days) whereas long trips lasted more than 2 days. Two mass parameters were determined for each trip: parental body mass change and chick meal size. Adult body mass change was the amount of mass change per foraging trip and was calculated as the difference in body mass between when a penguin goes out and when it returns from the foraging trip. Chick meal size was the amount of food given to the chick(s) and was assumed to be the mass difference between the penguin arriving at the colony after the foraging trip and returning once again to sea. Chick meal size could solely be estimated during post-guard, when little penguins only spend a few hours in the colony, the mass loss while in the colony assumed to be due to feeding their chicks rather than their metabolism.

Note that the Automatic Penguin Monitoring System records started in June 2001, and foraging variables are thus absent in the 2000 season. APMS is calibrated weekly to account for tare drifts, and deviance is considered for weight assessment. However, automatic weighing can sometimes provide inconsistent data due to multiple birds crossing at once. Only individual masses ranging between 700 and 1700g were considered, based on field observation of body mass range. Only trips that lasted between 1 and 16 days were considered for incubation and post-guard, and trips up to 3 days in guard (longer trips were classed as false negatives or abandonment and end of breeding). Based on the manual weighing of individuals in the colony, changes in body mass outside of the 0 to 600g range during chick-rearing (guard and post guard) and -75 to 500g range during incubation were considered inconsistent and discarded. Only chick meal sizes smaller than 550g were included in data analysis following previous direct observations.