|Home | About | Journals | Submit | Contact Us | Français|
Large mammals in seasonal environments have a pattern of high-reproductive synchrony in spring, but how the timing of reproduction affects resource allocation decisions at different stages of the reproductive cycle remains largely unexplored. By manipulating the timing of conception in reindeer (Rangifer tarandus), we tested how the timing of conception affected sex ratio, gestation length and weight development of mother and offspring. Females that conceived at their first ovulation within the rut had a 60.5% probability of producing a male; in contrast, females that conceived a cycle later had a 31.3% probability of producing a male. Late conceiving females had gestation times that were 10 days shorter and the calves were 0.6kg (9.2%) lighter at birth and 7.4kg (14.7%) lighter in autumn. Over the year, female weight changes was similar between the groups suggesting reindeer follow a bet-hedging strategy; reducing the quality of this year's offspring to ensure their own future reproduction and survival. Harvesting is often selective leading to skewed sex ratios and age structure, which may influence the timing of reproduction due to females hesitation to mate with young males. Whenever this hesitation is strong enough to increase the frequency of recycling, harvesting is likely to have profound life history consequences.
The Trivers–Willard model (TWM; Trivers & Willard 1973) of optimal sex ratio adjustment predicts that, in polygynous species, mothers in poor condition should produce offspring sex ratios biased towards females, with the converse true for mothers in good condition (review in Hewison & Gaillard 1999). Because large mothers produce large offspring (Weladji et al. 2003; Loison et al. 2004), in species where mass is a measure of condition, large mothers should produce more males than females. Factors other than mother's condition, however, can influence offspring size: the timing of conception and birth in ungulates is associated with offspring condition (Bunnell 1982; Festa-Bianchet 1988; Côté & Festa-Bianchet 2001). These differences at birth can persist throughout life (review in Lindström 1999), with late-born offspring typically smaller in autumn than those born earlier in the year (Eloranta & Nieminen 1986; Holand et al. 2003). Consequently, the TWM can be used to predict that females conceiving late should produce more females, as in highly dimorphic and polygynous species, birth size is less correlated with fitness in females than in males (Kruuk et al. 1999).
Among placental mammals a widely accepted theory is that gestation lengths vary little intra-specifically (Kiltie 1982, 1988), and that the timing and synchrony of births occurs by adjustment of the timing of ovulation (Langvatn et al. 2004). Late ovulation is, therefore, expected to provide the next year's offspring with a poor start in life, as late-born offspring are smaller in autumn and thus more likely to die during their first winter (Hogg et al. 1992; Langvatn et al. 2004). Mothers allocating extra resources to offspring that then achieve large body sizes typically increase their offspring's chances of surviving harsh winters (review in Gaillard et al. 2000). However, prolonging offspring investment too much into the autumn may lower the mother's own performance and hence her ability to ovulate sufficiently early during the next reproductive cycle (Hogg et al. 1992).
All deer species except roe deer Capreolus capreolus (Sempere et al. 1998) have repeated oestrus within a breeding season with a mean interval between two consecutive oestruses of about 18 days (Guinness et al. 1971; Putman 1988). However, the life history consequences of late conception in synchronously breeding species has not to date been investigated using an experimental approach. Using a research herd of semi-domestic reindeer—the species used by Trivers & Willard (1973) to illustrate their adaptive sex ratio theory—we experimentally manipulated conception date. Fertile adult males were introduced to one-half of the herd, while the other half of the herd was exposed to males that had undergone vasectomies. Mating then occurred. The two classes were then reunited and females that had not conceived had an opportunity to conceive in a later cycle. We tested the consequences of this manipulation of the timing of conception on sex ratio, gestation length and weight development of mother and offspring over the reproductive cycle.
The study was carried out at the Kaamanen Experimental Reindeer Station in Finland (69° N, 27° E; Holand et al. 2003; Mysterud et al. 2003). The total area of 43.8km2 is fenced and sub-divided into several smaller enclosures. This study was conducted in the northwest (Lauluvaara, 13.8km2) and southeast section (Sinioivi, 15km2). Birch (Betula spp.) and pine (Pinus sylvéstris) forests with numerous lakes and bogs dominate the habitat in these areas. The habitat composition is similar in the two enclosures. The experimental herd is given supplementary feeding during winter only (Holand et al. 2003).
During the rut of 2003, we manipulated the timing of conception by exposing half of the female herd (n=51) to four vasectomized males (one 1.5 year old male, one 2.5 year old male and two 3.5 years old) kept in the Lauluvaara enclosure, while keeping four intact males (one 1.5 year old male, one 2.5 year old male and two 3.5 years old) together with the other half of the female herd (n=49) in the Sinioivi enclosure. The female segment was kept similar with respect to age and weight structure between treatments. After the first ovulation cycle, the female herds were joined in Sinioivi together with the four intact males. Thus, the females reovulating were inseminated by the same males that fertilized half of the female herd in first ovulation. We confirmed this with paternity analyses (details on methodology in Røed et al. 2002). During the rut, we followed the herd by direct observation to determine date of copulation. As this is very difficult in the large enclosures, sample sizes involving copulation dates are smaller than the number of females present.
Females are all aged and individually marked and were weighed (to the nearest kilogram) before (11 September) and after (7 November) the rutting season of 2003. Additionally, individuals were weighed every month throughout the winter, the last time before calving being 7 April 2004 when they were confined to a calving paddock (about 50ha). The calving area was searched every morning and evening and newborn calves were caught by hand after the mother had licked them dry, weighed with a steelyard to the nearest 0.1kg, ear-marked with numbered ear-tags and sexed. All animals, both calves and their mothers, were weighted on the 11th June before being released into their larger summer enclosure. The animals were again rounded up on 13 September 2004 and weighed.
DNA marker analysis revealed that 13 females were fertilized by two of the assumed vasectomized males in Lauluvaara during first ovulation. This failure in the sterilization procedure led to 13 calves of the ‘late conceiving’ herd actually conceiving in the early conception period, all of which we excluded from analyses (explaining the slightly smaller sample size of the late conceiving herd).
We used a combination of generalized linear models (GLM, McCullagh & Nelder 1989), generalized linear mixed-effects models (LME and GLMM, Crawley 2003), and generalized additive models (GAM, Hastie & Tibshirani 1990) to estimate the effect of conception period (early versus late) on variation in performance of reindeer. To all models, we added both female age and weight to avoid possible bias, as these factors have been reported to be important for female performance in other studies of deer (e.g. Langvatn et al. 2004). For the sex ratio analysis, we tried both female weights before rut and close to the time of conception (i.e. using weight at a later date for late conceiving females). The effect of female age on reproduction is typically nonlinear (peaking at prime-age, Gaillard et al. 1998), and we tried both a second and a third order term for age. However, the way female age was modelled did not affect the factor of interest (conception period). Whenever we considered repeated measurement of the same individual, we used LME with female identity as a random factor to account for non-independency of observations. In the model of sex ratio variation, the first alternative would be to assume a binomial distribution. Any violation of the assumption of independence will lead to unaccounted heterogeneity, what is termed overdispersion indicating a poor model fit. Overdispersion is assessed by statistical significance of the residual deviance (i.e. the goodness-of-fit statistic) of the fitted model. As we found evidence of overdispersion in initial modelling attempts assuming a binomial error distribution, we turned to a quasi-likelihood test that need not correspond to any particular distribution, and which can be used to combine any available link and variance function (Venables & Ripley 1994). There was no evidence of overdispersion when using a quasi-likelihood link (, p=0.437), i.e. the model fitted the data appropriately. We also ran a GLMM fitting male ID as a random term (using library ‘MASS’ in R) to control for the possibility of individual males consistently produced biased sex ratio.
As predicted, conception date had a strong influence on offspring sex ratio. Females that conceived at their first ovulation within the rut had a 60.5% probability of producing a male, while females that conceived a cycle later had a 31.3% probability of producing a male (tables 1b and and2).2). There was only a tendency for heavy mothers (weighed before rut; on 11 September) to produce more males (table 1), this result was similar when using weight closer to time of conception: 11 September for early breeders; 22 October for late breeders). Females were on average in poorer condition during the second ovulation cycle (80.0kg on 11 September versus 77.9kg on 7 November LME, p<0.001; no difference between early and late breeders; p=0.30; and no interaction between ‘weight’ and ‘conception period’, p=0.78). The interaction between female age and conception period (T=1.231, p=0.222) was far from significant when added to the sex ratio model. An overview of male paternities in the first and second conception period is given in table 3. One younger male was more involved in the last rut and produced a very female-biased sex ratio. When including father ID as a random term, the effect of conception period on sex ratios was still significant (GLMM; Z=−2.233, p=0.029), but not when adding also father age as a fixed effect (Z=−0.883, p=0.381).
Female reindeer conceiving late shortened their gestation by approximately 10 days (figure 1a, n=18, r2=0.85). Mean date of birth for early and late breeders was, respectively, 19 May and 30 May, and thus more synchronous than expected from conception times.
Late conceiving females produced lighter offspring (figure 1b, tables 1c and and2,2, n=67, r2=0.41). Average weight of offspring born from early conceiving females was 6.4kg (for a 5yr old female weighing 80kg), as compared to offspring of late conceiving females that weighed only 5.8kg (table 1c, no interaction ‘sex’ and ‘time of breeding’; p=0.65). This means that on a given date in early summer (11 June 2004, figure 1c), the weight difference was considerable: calves from early and late breeders weighing, respectively, 15.6 versus 10.37kg (a 34% difference).
Late conceiving females lost less weight from September to June (mainly gestation period, early lactation period for early breeders). However, late breeders increased reproductive effort during late summer (post natal period, mainly lactation period). Even with this increased energy expenditure, the absolute difference in the weight of calves between early and late conceiving females was similar during autumn (13 September 2004; 50.4kg versus 43.0kg, tables 1e and and2,2, n=60, r2=0.59). Over the year, female weight change was similar (tables 1e and and2)2) in both early and late breeders.
The tactic of not breeding or failure to breed was rare among adult females; 1 out of 45 in the early breeding enclosure and 2 out of 31 in the late breeding enclosure did not breed.
Our study provides the first evidence based on an experimental manipulation that the cycle in which conception occurs influences the life history of an iteroparous mammal.
Late conceiving females produced a more female-biased offspring sex ratio, as may be expected from theoretical arguments (see §1). Recent evidence based on meta-analyses suggests that studies using female condition at conception often report sex ratio variation consistent with the TWM (Cameron 2004; Sheldon & West 2004). Females, through an unknown mechanism (see discussion in Cameron 2004) are, therefore, presumably able to assess their own condition and the consequences of conceiving at different times. Females were indeed on average in poorer condition during the second ovulation cycle, but individual female weight was not a significant predictor of offspring sex (though the estimate was in the predicted direction). It has also been suggested that if individual males produce biased X- or Y-chromosome ejaculates, this may lead to biased sex ratios (DeYoung et al. 2004). In white-tailed deer (Odocoileus virginianus), individual males did not produce biased ejaculates nor did ratio of X- and Y-sperm differ depending on collection period (DeYoung et al. 2004). However, several studies indicate that younger males are more active during the later part of the rut (Prothero et al. 1979; Noyes et al. 1996), and thus that male age may potentially affect sex ratio if young males produce more X-sperm or via sperm selection in the uteri. In moose (Alces alces), male age was found to be a significant predictor of offspring sex ratio, with prime-aged males siring more male offspring (Sæther et al. 2004). Our data was not decisive in this respect; as there was a tendency for younger males to be more involved in the last rut and siring more female offspring (table 3). Clearly, further replication is required before the generality of this result can be determined with confidence.
While the exact physiological mechanisms involved in sex ratio variation remain elusive, evolutionary models such as the TWM are intended to assess the pattern of variation in sex ratio rather than the mechanism. Thus, the selection pressure is the more important for the TWM and other evolutionary-based models. A common weakness with much of the work on sex ratios is indeed the failure to link patterns of variation with selection. The critical issue of understanding the direction of the shift depends on knowledge of which sex gains relatively more by being large (Charnov et al. 1981; Leimar 1996), and in our case equivalent to being born early (table 2). There is considerable evidence that birth size is more important for future survival (Loison et al. 1999b) and reproductive success of males than females (Clutton-Brock et al. 1988). For red deer on Rum
birthweight was a significant determinant of total lifetime reproductive success in males, with heavier-born males being more successful than lighter ones. In contrast, birthweight did not affect female reproductive success
Reindeer and red deer have a similar life history and a similar degree of sexual body size dimorphism (Weckerly 1998; Loison et al. 1999a), and both species have been shown to fulfill the underlying assumptions of the TWM (Hewison & Gaillard 1999). There is thus ample evidence that males benefit more from being born early than females in highly polygynous, sexually dimorphic ungulates (Hewison & Gaillard 1999), and thus that the selection pressure behind the observed sex ratio variation is as expected in reindeer. Therefore, a beneficial strategy for females may be to reduce the reproductive costs of late conception by producing a female calf rather than a male calf.
Our study provides new insight into how iteroparous mammals in seasonal environments solve the trade-off between investment in current year offspring versus future reproduction and survival. Comparative studies indicate that adult survival in mammals varies little between years while juvenile survival can vary substantially (Gaillard & Yoccoz 2003). The contrasting responses of juvenile and prime-age survival to environmental variation in ungulates is considered to originate from the way females partition resources between offspring production and self maintenance, and could be viewed as a bet-hedging strategy (Gaillard & Yoccoz 2003). Our results provide support for this because female weight changes over the year were similar between the groups (table 2), while quality of offspring differed between female groups (table 2). Reindeer thus follow a bet-hedging strategy, and can both reduce the quality (weight) and potential cost (sex) of this year's offspring to ensure their own future reproduction and survival.
Our manipulation further revealed phenotypic plasticity in gestation length. Female reindeer conceiving late increased their probabilities of gaining sufficient condition to breed next year by shortening their gestation by approximately 10 days (figure 1a). That females can reduce gestation length by up to ~6 days has previously been shown for bison (Bison bison) cows in good condition (Berger 1992). Following dry summers, pronghorn antelope (Antilocapra americana) also prolonged gestation with between one and 10 days depending on age (Byers & Hogg 1995). Lactation is the most energy demanding part of reproduction (Clutton-Brock et al. 1989), and it is regarded as very important to have high plant quality available during this stage. It is, thus, likely that the reduction in gestation length was also a way to ensure better use of plant quality during periods of intense lactation. Nevertheless, there was a penalty of such short gestation length in terms of lighter offspring (table 2, figure 1b,c). Late born offspring also experienced a shorter growth period before winter, attained a lower body mass at the onset of winter and, therefore, had a decreased expectation of survival compared to early born calves (Festa-Bianchet et al. 1997). As argued above, this effect is typically stronger in males than females in sexually size dimorphic species (Kruuk et al. 1999; Loison et al. 1999b). Late conceiving females lost less weight from September to June (mainly gestation), probably because early breeders had experienced a longer period of lactation at this time. However, late breeders increased reproductive effort (as evidenced from weight changes) during late summer (post natal period), when they were also lactating. Even with this increased energy expenditure, the absolute difference in the weight of calves between early and late conceiving females was similar during autumn.
Virtually all populations of large herbivorous mammals in the temperate and sub-arctic zones are exposed to human manipulation through harvesting and poaching. Further, harvesting is usually sex- and/or age-biased affecting the remaining age and sex structure of the population (Langvatn & Loison 1999). Such a change influences the dynamics of the rut, since females will not mate with a young male if a larger, more dominant male is available (Komers et al. 1999, review in Mysterud et al. 2002). A skew in sex ratio and/or in the male age structure may thus result in delayed and less synchronous conception reflected in the calving season (Mysterud et al. 2002; Holand et al. 2003; Sæther et al. 2003), although we found that the females may try to compensate through shortening gestation length. Under natural conditions in our reindeer population (i.e. in the years before the experiment), only 4.5% of the females recycled (Holand et al. 2003). We still lack good quantification of if and how much harvesting increases recycling rates. A key question is whether the hesitation to mate with young males is strong enough to increase the frequency of recycling. If recycling becomes more frequent due to harvesting, there may be opposing selection on the sex ratio to balance the overall population sex ratio. Our study underlines the potential strong effect on the life history if sex ratios become as skewed as that currently is observed in heavily harvested populations.
We gratefully acknowledge the financial support of the Research Council of Norway (NFR) and the Reindeer Husbandry Administration to O.H. and K.R. and of NFR to A.M. (YFF Project). Mika Tervonen and his crew at the Kaamanen Experimental Reindeer Station are acknowledged for their contribution with the collection of the data from reindeer. We would like to thank Dr Timo Soveri for vasectomies of reindeer, and Dr Jean-Michel Gaillard, Dr Jouko Kumpula and two anonymous referees for valuable comments to the manuscript.