Discussion

Comparison of Reproductive Effort between the Sexes

This study reveals important similarities and differences in the magnitude and timing of reproductive effort between male and female northern elephant seals. The most striking similarity is in the energetic index of RE, adults of both sexes losing slightly more than one-third of their body mass, on average, while on the breeding rookery. This finding suggests that a common physiological mechanism, perhaps related to changing body composition or increasing protein catabolism (Groscolas 1986; Cherel, Robin, and Le Maho 1988), may underlie the termination of the breeding fast in both sexes.

There is no a priori reason why male and female RE should be equivalent, since the factors affecting reproductive success differ so widely between the sexes in elephant seals. In males, mating success achieved for a given effort depends on the number, competitive ability, and RE of other males on the rookery (Trivers 1972; Warner 1980). Intense male intrasexual com-

petition for mates has been a strong selective force favoring continued growth and delayed maturity, thereby molding the level and developmental timing of mating effort. Consequently, young male elephant seals invest little in reproduction and much in increased growth during the first two or three years that same-age females are pupping, a common phenomenon in polygynous species (Selander 1972; Clutton-Brock, Guinness, and Albon 1982). Compared to males, competitive social interactions exert less influence on female RS (see Reiter, Panken, and Le Boeuf 1981) and, therefore, on the magnitude and ontogeny of maternal effort. Natural selection favors an investment pattern in females that optimizes the opposing effects of energy transfer to the pup on maternal and offspring fitness, relationships that are affected by fasting physiology, foraging ecology, thermoregulatory requirements at sea, and other ecological factors.

Data on mass change for four other phocid species allow comparison of RE between the sexes. The mean percentage of body mass lost over lactation is in the range of 30 to 40% for females of all five species (table 10.7). Breeding mass loss was similar for the two sexes in both species of elephant seals but higher for females than males in gray, Weddell, and harbor seals (table 10.7). The energetic cost of reproduction in the sexually monomorphic harp seal, Phoca groenlandica , appears to be much greater for females than males based on changes in blubber thickness over the breeding season (Sergeant 1973). These data suggest that only in the most polygynous seal species does male mating effort approach the level of female parental effort in terms of energetic costs.

Energy allocation to reproduction (i.e., breeding activity, gonads, gametes, and young) is far greater in female golden-mantled ground squirrels, Spermophilus saturatus , than in males (Kenagy 1987; Kenagy, Sharbaugh, and Nagy 1989), and this is probably the case for most mammals. But is this the best measure of RE? Our use of relative mass loss as an index of RE in fasting pinnipeds included energy devoted to maintenance metabolism while breeding, and so it was not a measure of reproductive energy expenditure per se. Nevertheless, since elephant seals and some other pinniped species must fast in order to breed (see Riedman 1990), it is reasonable to include this maintenance metabolic expenditure as part of the total energy cost of reproduction. Because the mating strategy of males in polygynous species often entails reduced food intake during a period of high activity, males may deplete their energy reserves to a greater extent during the mating period than females do during pregnancy and lactation (e.g., Leader-Williams and Ricketts 1981; Michener and Locklear 1990), even though actual reproductive energy expenditure may be much higher in females. Thus, we consider percent mass loss (an index of percent energy loss) while breeding to be a more appropriate measure of RE in fasting seals than energy allocation to strictly reproductive functions.

Differences in the pattern of reproductive effort between male and female elephant seals are related to their divergent reproductive strategies, as follows:

1. Rate of investment in reproduction, measured as mass-specific daily mass loss, was over three times higher for females than males. This has the effect of reducing the metabolic overhead costs of females relative to males. About 40% of the energy lost by females during lactation is utilized for maintenance and activity metabolism, the rest being transferred to the pup as milk (Costa et al. 1986). On the con-

trary, most of the energy males expend during the breeding season is probably allocated to maintenance metabolism since 85 to 95% of their time is spent resting (Sandegren 1976; Deutsch 1990).

2. Duration of RE was three times longer for males than females, reflecting the importance of early male arrival to obtaining high rank and of a lengthy stay to maximize chances for copulation. A consequence of this longer effort is that males had two fewer months than females to recover lost energy stores and to gain additional mass before the next breeding season. Of course, females invest energy in gestation for eight months of the year, which is not reflected in these comparisons (see Deutsch, Haley, and Le Boeuf 1990).

3. Variation in RE was greater among males than females. Male mating effort was quite flexible, as evidenced by the more than twofold variation in duration of the breeding fast and by great variability in measures of the intensity of effort, such as the proportion of time spent competing near harems (Deutsch 1990) and mass-specific rate of mass loss (Deutsch, Haley, and Le Boeuf 1990). Consequently, some young subordinate males invested considerably less than any female did, and the RE of some high-ranking males soared well above the maximum for females. In contrast, three observations indicate that flexibility in adjusting maternal effort during lactation is limited by energetic considerations. First, mothers that invested at a higher relative rate weaned their pups sooner, suggesting the existence of a cost ceiling. Second, females that lose their pups and therefore invest relatively little energy in lactation remain on the rookery longer than those who rear pups successfully (Reiter, Panken, and Le Boeuf 1981). Third, a 3-year-old mother ("Bat") who nursed two pups simultaneously lost mass at a rate similar to other females of her size (observed = 6.35 kg/day; expected = 6.25 kg/day), and the pups gained mass at a combined rate similar to single pups (observed = 3.46 kg/day; expected = 3.62 kg/day). The minimum RE of parous females would have been lower than 30% mass loss if we had included those that lost their pups and did not nurse. But since 90 to 95% of females pupping on the Año Nuevo mainland are successful in raising their pups to normal weaning age (B. Le Boeuf, unpubl. data), including unsuccessful mothers in our sample would not have altered the above finding.

This sex difference in the variance in RE mirrors the great difference in the variance in reproductive success between the sexes (Le Boeuf and Reiter 1988). Females must invest some minimum RE to produce a healthy offspring, whereas males that invest even less effort have some chance of siring a pup. Extremely high levels of maternal effort (over 40% of body mass

lost) probably benefit a pup relatively little compared to the future costs incurred by the mother, but alpha males can raise their current RS substantially through continued high mating effort. Thus, the sex difference in RE variation reflects the fundamentally different means by which male and female elephant seals maximize their RS.

Reproduction can be a risky activity, sometimes even leading to death (e.g., Wilkinson and Shank 1976; Brunton 1986), so a complete sex comparison must also examine this facet of RE. Frequency of external wounds, the measure of risk used in this study, was many times greater in males than females, reflecting the violent competition among males for high rank and access to females. Similar sex differences in frequency of injury have been found in other polygynous mammals (e.g., Smith 1966; Michener and Locklear 1990). We note, however, that the observed injury rate for females may be an underestimate because they were sampled in harems, but they are most likely to get injured during departure from the rookery. The risk of receiving a lethal injury during agonistic or sexual interactions is relatively small for both sexes at Año Nuevo; female mortality due to male-inflicted injuries (0.1% per year; Le Boeuf and Mesnick 1991) appears to be about twice as high as that for males (Deutsch 1990; C. Deutsch and B. Le Boeuf, unpubl. data). Female elephant seals probably incur a higher incidence of fatal injuries than males because of the large sex difference in mass, the large canines of males, the bites to the female's neck during typical sexual behavior, and the emaciated condition of the female on departure, when she is exposed to mating attempts by aggressive and sexually motivated peripheral bulls (Mesnick and Le Boeuf 1991).

Differences in life history parameters between the sexes may be due to differences in the timing and magnitude of RE. In Richardson's ground squirrels, S. richardsonii , for example, males deplete their fat stores to a greater degree during the mating period than females do during lactation, and this is associated with higher male mortality (Michener and Locklear 1990). The authors conclude that the costs of mating effort for males are greater than the costs of parental effort for females in this species. This pattern is taken to the extreme by the marsupial shrew, Antechinus stuartii , in which all males die soon after the mating season, whereas females are iteroparous (Lee, Wooley, and Braithwaite 1982). In northern elephant seals, the survivorship curves are roughly similar for both sexes (see fig. 22.2 in Le Boeuf and Reiter 1988), although longevity is somewhat greater for females. This is consistent with the similar magnitude of RE for the sexes found in this study.

Age-specific Reproductive Effort and Fitness Consequences

Variation in the energetic component of RE was not associated with age for males 6 years and older or for parous females. This finding, based on mass

loss, is corroborated by complementary data showing no correlation between mass-specific energy expenditure and milk energy output of lactating females and age (Crocker 1992; D. Crocker, unpubl. data) and no age-related variation in male time-activity budgets (Deutsch 1990). Unfortunately, very old females, with the lowest reproductive value and the highest predicted RE, were not included in our sample, but a follow-up study (D. Crocker, unpubl. data) found that three 11- to 12-year-old mothers lost a similar proportion of their body mass over lactation (range = 31–34%) as younger mothers. W. J. Sydeman et al. (1991) suggested that age-specific increases in female aggressiveness, success in dominance interactions, and weaning success could be attributed to increasing maternal effort with age. The energetic data do not support this interpretation; instead we propose that the increase in size and experience with age accounts for the higher dominance status and weaning success of older mothers. Even though the proportion of resources allocated to reproduction did not vary with age, it is still possible that younger, smaller seals incurred a greater risk by depleting their fat reserves to a lower, perhaps marginal, level than larger animals, as R. E. A. Stewart (1986) found for female harp seals. This was apparently not the case in elephant seals, however, since females weaned their pups on reaching 21 to 25% body fat, regardless of maternal age, parturition mass, or initial body composition (Crocker 1992; D. Costa and B. Le Boeuf, unpubl. data). Female investment in offspring was in direct proportion to maternal mass at parturition. This is similar to the pattern of maternal investment in southern elephant seals (McCann, Fedak, and Harwood 1989; Fedak et al., this volume) but unlike harp seals, in which small, young mothers produce weanlings of the same size and lose mass at the same rate as larger, older mothers (Stewart 1986; Kovacs, Lavigne, and Innes 1991).

The fitness costs of reproduction (e.g., reduced survival) should be positively related to the proximate energetic costs (i.e., RE), though probably in a nonlinear fashion, and this relationship should be manifested in the age-specific patterns of RE and mortality. This expectation is partially supported by the fact that age-specific mortality rates rise sharply upon the initiation of reproduction—at primiparity for females (age 3–4; Reiter and Le Boeuf 1991) and at puberty for males (age 5; Clinton, this volume). For females, this survival cost is associated with lactation, since mortality remains low during the year of their first pregnancy. Similarly, Clutton-Brock, Albon, and Guinness (1989) concluded that the fitness costs of reproduction in red deer, Cervus elaphus , are attributable to the energetic demands of lactation and not to gestation.

Two lines of evidence led us to expect the RE of reproductively active but physically immature "subadults" (i.e., 3- to 5-year-old females and 5- to 7-year-old males) to be less than that of physically mature animals. First, RS increases with age in both sexes, especially from puberty to physical

maturity, due largely to the effects of intrasexual competition (see references given on p. 173). Young small females obtain a lower benefit (in terms of weaning success and weanling mass) for the same proximate cost (relative mass loss) as older, larger mothers. Even more striking is the fact that subadult males obtain little immediate benefit for the same mass-specific cost as low-ranking adult males; only 6 to 15% of 6- to 7-year-olds copulate (Clinton 1990). Second, a given RE seems to impose a greater strain on the younger breeders of both sexes, manifested as higher fitness costs. Females giving birth at age 3 suffer reduced survivorship and future fecundity relative to those delaying reproduction to age 4 or 5 (Huber 1987; Reiter and Le Boeuf 1991; but see Sydeman and Nur, this volume). Current mating success is negatively correlated with subsequent survival and fecundity for young males (6–8 years old) but not for older males (Clinton and Le Boeuf 1993; Clinton, this volume). Also, C. R. Cox (1983) found that 6-year-old males that spend a greater amount of time on the breeding beach suffer higher mortality than those exhibiting lower effort. The higher fitness costs incurred by young breeders may result from the simultaneous allocation of energy to growth and reproduction, leaving less available for storage or critical maintenance functions.

Given that benefits of reproductive investment increase and fitness costs decrease with age during the period between sexual and physical maturity, selection should favor a relatively low level of RE for "subadult" animals. This probably explains why most females delay primiparity beyond age 3 (Huber 1987; Reiter and Le Boeuf 1991) and why pubescent males (4–5 years old) invest little time or energy in mating attempts (Clinton 1990; Deutsch 1990), despite the theoretical benefits of early breeding (Sibly and Calow 1986). Still, the pattern of constant mean RE with age for parous females and for males older than 5 years is perplexing. Theoretically, the optimal level of investment at each age is determined by the age-specific functions relating RE to fitness benefits (i.e., current RS) and to fitness costs (i.e., reduction in residual reproductive value). These functions are extremely difficult to determine empirically, and few data are available (Charlesworth 1980). It is noteworthy, however, that a constant age-specific RE is theoretically plausible, at least among adults, as stated by B. Charlesworth (ibid. 246): "It is not difficult, in principle, to find cases in which an optimal life-history exists which has constant reproductive effort among the adult age-classes, and hence constant adult survival and fecundity."

One could reasonably argue that a small, young female must expend a relatively large proportion of her energy stores during lactation to produce a weanling that is large enough to survive well (see Fedak et al., this volume). The point at which survival from weaning to age 2 appears to drop off is at a weaning mass of only 80 to 90 kg (Le Boeuf, Morris, and

Reiter, this volume). A hypothetical young mother of 400 kg would produce a 34-kg newborn (from fig. 10.3) and nurse it for 23 to 26 days with a mass transfer efficiency of 63% (from table 10.3). This female would lose 105 kg, or only 26% of her parturition mass, to produce a 100-kg weanling. Clearly, young mothers invest substantially more (36% of body mass) than is required to produce a viable pup. So this argument can probably be rejected; but the future fecundity of offspring may be enhanced by maternal investment above what is necessary for offspring survival. We offer three other nonexclusive explanations that may account for the lack of age-related variation in RE.

One explanation is that the seals are able to tolerate moderate energy deficiencies up to some threshold level, and only when that threshold is exceeded does a cost of reproduction become apparent (Tuomi, Hakala, and Haukioja 1983). Large fluctuations in body mass (e.g., loss of one-third mass) are a normal part of the elephant seal's annual cycle (Costa et al. 1986; McCann, Fedak, and Harwood 1989; Deutsch, Haley, and Le Boeuf 1990; Kretzmann, Costa, and Le Boeuf 1993; Worthy et al. 1992) and may entail little or no fitness costs in healthy animals. While costs associated with additional increments of investment above a critical level are likely to escalate (Burley 1988), the benefits should approach an asymptote, particularly for females. In fact, the survival benefit of large weanling mass is equivocal (Le Boeuf, Morris, and Reiter, this volume). Crocker (1992) postulated a physiological mechanism, based on changing body composition and rate of protein catabolism, that may account for a threshold relationship between maternal investment and fitness costs. Circumstantial evidence for such a threshold phenomenon includes the fact that female body fat at the end of lactation lies in a narrow range, despite the large variation in initial fat stores, in lactation duration, in absolute rates of mass loss, and in rates of lean and adipose tissue loss (Crocker 1992; see also Gales and Burton 1987). An animal exceeding the hypothesized threshold level of RE might be faced with the choice of either depleting its blubber layer beyond what is necessary for thermoregulation at sea or oxidizing muscle and other lean tissue. While this cost-based explanation is plausible, the trade-off between reproduction and growth (e.g., Robinson and Doyle 1985; Green and Rothstein 1991) would seem difficult to avoid, and so attempting to breed as a growing subadult could reduce size and, hence, fecundity as an adult (Clinton 1990; Haley, Deutsch, and Le Boeuf, in press). Furthermore, the evidence of a trade-off between reproduction and survival for young animals also argues strongly for the existence of substantial benefits, either immediate or delayed, to early reproductive attempts.

The second possibility draws on the idea that in expanding populations the fitness benefit of early-born offspring is greater than that of late-born young, thus favoring early maturity and declining RE with age (Lewontin

1965; Charlesworth and León 1976). This may partially explain why females mature at a relatively young age in M. angustirostris and in exploited populations of M. leonina (Carrick et al. 1962), although the costs of early reproduction would appear to outweigh the benefits under the prevailing demographic conditions (Huber 1987; Reiter and Le Boeuf 1991). The hypothesized effect seems too small, however, to compensate for the extremely low mating success of subadult males, unless we have grossly underestimated their success. Who mates with virgin females and where this occurs is poorly known (Huber et al. 1991) and could potentially affect our interpretation. Overall, we think that this can be a partial explanation at best.

The final explanation for the unexpectedly high RE among physically immature seals is that experience gained when young improves future reproductive performance (Caro 1988; Robinson 1988). The benefits of early maternal experience to subsequent weaning success have been documented in elephant seals (Reiter, Panken, and Le Boeuf 1981; Sydeman et al. 1991). C. J. Deutsch (1990) presents circumstantial evidence for the importance of fighting and sexual experience to males; this includes the observation that subadult males engage in frequent mock-fighting activity in contexts in which no immediate reproductive benefit is possible (e.g., during the molt period). Even male weanlings exhibit behaviors during social interactions that resemble the fighting behavior of adult males (Rasa 1971; Reiter, Stinson, and Le Boeuf 1978), suggesting that early social experience plays an important role in reproducing successfully later in life (Fagen 1981). This raises the interesting question of whether such social play among juvenile males should be considered a form of early RE with delayed benefits that has evolved via sexual selection on adult males.

In conclusion, we found that the magnitude of reproductive effort in northern elephant seals, as estimated by relative mass loss, was similar for males and females but that the temporal patterning of effort differed between the sexes both within a breeding season and over development. The similarity in RE between the sexes was unexpected, given the large size dimorphism, divergent reproductive strategies, and the generally held view that female mammals invest more in reproduction than males. RE did not vary with age in males (older than 5 years) or parous females, despite an increase in reproductive success and a decline in the fitness costs of breeding with age for both sexes. Future studies should investigate whether the substantial interindividual variation in RE is related to the amount of energy stores available at the start of breeding, thus potentially linking foraging success at sea with reproductive success on land.