1. Life history trade-off theory predicts that current reproduction can negatively affect survival and future reproduction. Few studies have assessed breeding costs for males of polygynous species compared to females, despite substantial variation in breeding success among individual males (e.g. subordinate cf. dominant breeders). Specifically, differentiating between the cost of attending breeding seasons, and the additional cost of successfully securing and mating females is lacking.
2. We investigated whether trade-offs are present in the highly polygynous male southern elephant seal (Mirounga leonina) using 34-years of individual-level data. We compare age-specific survival, recruitment and future breeding success probabilities of pre-breeders (males yet to recruit) and breeders (subordinate and dominant social ranks) using multievent models.
3. Pre-breeders and breeders of overlapping ages had similar survival probabilities, suggesting that there was no attendance cost for early recruits. In addition, the probability of recruiting as a dominant breeder never exceeded recruitment probability as a subordinate breeder of the same age. Therefore, older pre-breeders that delayed attendance costs generally did not improve their breeding success (probability of being dominant) at recruitment more than younger recruits. Rather, recruitment age may be a function of individual quality, with lower quality individuals requiring more time to socially mature.
4. When comparing subordinate and dominant breeders, we found clear evidence for survival senescence, with subordinate breeders having a higher baseline mortality. In contrast, age-specific future breeding success (probability of being dominant at t+1) increased with age, with dominant breeders maintaining higher subsequent breeding success than subordinate breeders.
5. The opposite trends in survival and future breeding success for both subordinate and dominant breeders may indicate a lifetime, population-level trade-off. However, we found no evidence to suggest that being a dominant breeder consecutively (and having a higher accumulated breeding cost) accelerated the rate of senescence when compared to individuals that were previously subordinate.
6. Thus, males experienced actuarial senescence regardless of social rank, with dominant (and possibly high quality) breeders showing a reduced trade-off between survival and future breeding success. We make several novel contributions to understanding polygynous male life histories and southern elephant seal demography.

All elephant seal pups weaned at Marion Island from 1983 to 2009 were uniquely marked with two plastic livestock tags applied to the hind flippers (n = 27 cohorts, 6245 male pups; Pistorius et al., 2011). Pups were tagged in the inner interdigital webbing from 1983 to 1999 and in the outer interdigital webbing from 2000 to 2009. From 1983 to 2016, all beaches where elephant seals bred were surveyed every seven days during the breeding season and every ten days outside of the breeding season. In total, 58 177 observations of marked male elephant seals were made over 34 years. A seal year began at the start of every breeding season and ended before the following breeding season.

When a male elephant seal was resighted, one of nine events could be assigned to it: not seen (0); seen as a pre-breeder with two tags (1); seen as a pre-breeder with one tag (2); seen as a subordinate breeder with two tags (3); seen as a subordinate breeder with one tag (4); seen as a dominant breeder with two tags (5); seen as a dominant breeder with one tag (6); seen with an unknown breeding state and two tags (7); and seen with an unknown breeding state and one tag (8). Dominant breeders were identified as being associated with one or more females with pup(s), which they defended and mated (male often positioned within the harem). Subordinate breeders were either on the periphery of harems or on beaches not occupied by females. We assigned animals to events 7 and 8 when an individual was seen as both subordinate and dominant several times during the same breeding season. In such cases (n = 43), uncertainty existed about the true breeding state that the individual occupied during the breeding season. On only nine occasions were individuals that had previously participated in a breeding season not seen in a future breeding season, given that the individual was known to be alive through subsequent encounters. These nine individuals were also classified as having an unknown breeding status during absent years.

An individual could occupy one of nine possible states during each seal year (Pradel, 2005). The states were: pre-breeder with two tags (PB2, has not previously participated in a breeding season); pre-breeder with one tag (PB1); pre-breeder alive elsewhere with two tags (PBAE2, temporarily emigrated and last seen with two tags); pre-breeder alive elsewhere with one tag (PBAE1); subordinate breeder with two tags (SB2); subordinate breeder with one tag (SB1); dominant breeder with two tags (DB2); dominant breeder with one tag (DB1); and dead (D, an absorbing state representing death and permanent emigration). Pre-breeders alive elsewhere were hidden states constructed to account for temporary emigration from the study population (Schaub, Gimenez, Schmidt, & Pradel 2004). Goodness-of-fit tests indicated that this behaviour was prevalent among pre-breeders, but not breeders. Introducing an unobservable state improved estimates of survival and detection probabilities of pre-breeders.

Data were formatted into a multievent encounter history matrix to be analysed in E-SURGE 2.1.4 (Choquet, Rouan, & Pradel 2009). Rows represent the encounter history of individuals and columns represent time steps (i.e. 34 seal years). The last column identifies the position of tag placement on each seal (IIT - inner-interdigital, OIT - outer-interdigital). The encounter history of each individual is numbered using event codes (see above).