Methods

Samples of supermoms

We selected a subsample of 64 females (supermoms), all of whom were observed breeding at least 10 times in their lifetimes. All were tagged between 1982 and 2002, and all their breeding observations were between 1987 and 2017 (we have excluded all 2018 observations). Most of the supermoms (56) were born at Año Nuevo, but 6 were born at other colonies and 2 had unknown birthplace (those tagged as yearlings). To generate a comparable group of control animals, extracted all females born over 1982-2002 who had breeding observations at Año Nuevo at least once during 1987-2017. Not including the supermoms, this included 908 generic females whose age was known, and 716 of those were born at Año Nuevo.

For calculations of lifetime probabilities of breeding, it was necessary to focus on cohorts of animals born at Año Nuevo, because we often do not know exactly how many females were tagged at other colonies. The 1982-2002 cohorts from Año Nuevo included 4645 females tagged soon after birth. It is from that sample that 56 lived to become supermoms, and 716 became generic moms, observed as breeding adult females at least once but < 10 times.


[Note: An additional 24 females were observed breeding on 10 or more occasions but were unknown age because they were tagged as adults.]
[Note: An additional 1070 generic females were observed, those whose age was not known but had at least one but < 10 breeding observations during 1987-2017.]
N_females pups_produced
23 10
24 11
6 12
1 13
4 14
4 15
1 16
1 17

Table Extra (not for paper). Number of supermoms producing 10-17 pups in their lifetimes.


Weaning success

During a single season, a female's record was considered adequate if she was observed on at least four days beyond six days after her first sighting. Since females give birth six days after arriving, on average, this was intended to insure that there were four observations of a female beyond her likely parturition. If a female was with exactly one pup on at least three-quarters of her post-parturition sightings, she was deemed to have raised her pup until weaning. If she was seen with a pup less often, she was deemed to have lost her pup, who was then orphaned and most likely died soon after. The criterion of four observations post-parturition is arbitrary, but it eliminates females with poor observations without inparting any obvious bias relative to weaning success. It meant that 50.4% of all female breeding records (one record meaning one breeding season) were deemed inadequate and not used to estimate weaning success. The proportion weaned was modeled as a function of mother's age with the Von Bertanlanffy asymptotic function, using a binomial error and a Bayesian procedure for sampling parameters (see Condit et al. 2007 for details of the Bayesian method).


Pup sex ratio

The sex of pups raised by all females was observed opportunistically, including after weaning when the pup was weighed (when sex was always recorded). Since pup sex was rarely observed in pups soon after birth, and was often recorded after weaning, we consider this closest to the weaning sex ratio, not the birth the sex ratio. The proportion of male pups was modeled as a function of mother's age using the Von Bertanlanffy formula, exactly as for weaning success.


Body mass at weaning

Pups of known-age females were weighed whenever possible a few days after weaning. Weight at weaning was back-extrapolated assuming daily weight loss of 660 g. Crocker and Le Boeuf (2005) described details of weighing procedures. Generic mothers and supermothers were divided into five age categories (3-7, 8 and above years), and mean weanling mass was calculated in each category, along with standard deviation. Ages 8 and above were merged because weanling mass reached an asymptote by then. Confidence limits per group were calculated as 1.96 times the SD divided by square root of sample size above and below the mean.


Site fidelity

The location of groups, or harems, of breeding females was highly consistent from year to year. There was a single such group on Año Nuevo Island and usually 18 groups on the mainland, though the latter varied over the 20 years of the study as some harems divided as the colony expanded. The location of each harem was established precisely by extracting coordinates from a Google map in 2017. Sand dunes and vegetation shifted over the decades of the project, but the exact location of harems in every year was reasonably established to within about 20 m. Each female's breeding position in a year was taken as the coordinates of the centroid of the harem she used in that year. Movements between years were then calculated from successive coordinates, using only cases where a female was seen often in both years, based on the criteria defined above for establishing weaning success. Since the distance moved by females was highly non-normal (many zeroes but some distances up to 1.5 km), we analyzed movements using a cutoff of 150 m, which is the approximate size of the larger harems. Site fidelity was defined when successive breeding locations were < 150 m apart.



Results

Chance of becoming a supermom

Only a fraction of the females in the total known-age sample of Año-Nuevo-born females, 56/4645 = 1.21%, became supermoms and gave birth to pups 10 or more times in their lifetimes, while 772 or 16.62% became breeders at all (including those supermoms). Of those who became breeders, 56/772 = 7.25 % became supermoms.


Birthing pattern

The supermoms produced 728 pups, a mean of 11.4 pups in each of their lifetimes, while the generic females produced a mean of 3 pups each. The maximum number of pups produced by any female was 17 (female GH883), and the most consecutive years in which a female produced a pup was 16 (female GM583).


Weaning success

In generic mothers, the percent of pups successfully weaned rose from 55% at age three to 90% in older females, a highly significant increase. In supermoms, however, wean success was high at all ages, and significantly higher than generic mothers at ages 3-5 (Fig. WS).

plot of chunk weansuccess

Fig. WS. Wean success versus mother's age, generic moms (black) and supermoms (red). Wean success is the proportion of pups born that are raised successfully to weaning by the mother. The curves are fits to the Von Bertanlanffy model, with vertical bars giving 95% credible intervals based on the Bayesian parameter search. Points are the observed proportion weaned at each age.


Sex ratio at weaning

Young females had a sex ratio among successfully weaned pups < 40% male, but this rose to > 55% in older females. The increase was statistically significant. Super moms did not differ in this regard, having the lowest proportion males when young (Fig. SR).

plot of chunk sexratio

Fig. SR. Pup sex ratio (at weaning) versus mother's age, generic moms (black) and supermoms (red). The curves are fits to the Von Bertanlanffy model, with vertical bars giving 95% credible intervals based on the Bayesian parameter search. Points are the observed proportion males at each mother's age. Sex ratio increased significantly with mother's age in generic females, but in supermoms the increase in sex ratio with age did not reach statistical significance. Regardless, the sex ratio did not differ signficantly between super and generic moms at any age.


Body mass at weaning

Mothers at age 3 weaned pups < 100 kg, but wean mass increased with mother's age, reaching an asymptote of 134 kg above age 8. Supermoms were no different from generic moms in these respects (Table WW).

mom_age wean_wt_other_mom N_other wean_wt_super_mom N_super
3 94.3 (86.7,101.9) 30 107.4 (84.4,130.4) 5
4 110.9 (108.0,113.8) 157 113.4 (102.4,124.4) 22
5 121.0 (117.9,124.1) 155 123.6 (116.6,130.6) 19
6 128.9 (125.8,132.0) 114 126.7 (117.9,135.5) 27
7 131.9 (128.3,135.5) 82 128.2 (120.6,135.8) 13
8 and above 134.3 (132.0,136.6) 209 132.5 (129.2,135.8) 120

Table WW. Mean mass at weaning per mother\'s age class, with 95% confidence limits in parentheses. The columns N give sample sized of weanlings weighed in each age group. At no age was the difference between generic and super moms statistically significant.



Site fidelity

The distance moved between successive breeding attempts was highly skewed. The median distance was 117 m, but 11.5% moved > 1 km while 22.7% remained at exactly the same location (N = 829 observed movements). Old females were faithful to a harem, moving < 150 m, more often than young females, but there was no difference between super versus generic mothers in site fidelity (Fig. M).

plot of chunk movement

Fig. M. Proportion of females faithful to a breeding location in two successive breeding attempts, generic moms (black) and supermoms (red). Site fidelity was defined as any movement < 150 m. The curves are fits to the Von Bertanlanffy model, with vertical bars giving 95% credible intervals based on the Bayesian parameter search. Points are the observed proportion males at each mother's age. In generic mothers, the proportion faithful increased significantly with age, but super moms did not differ from generic moms at any age.



Discussion