- Top of page
- Materials and methods
1. The Trivers–Willard model of optimal sex ratios predicts that in polygynous species mothers in better condition should produce more male than female offspring. However, empirical support for this hypothesis in mammals and especially ungulates has been equivocal. This may be because the fitness of mothers has been defined in different ways, reflecting morphological, physiological or behavioural measures of condition. In addition, factors other than maternal condition can influence a mother’s fitness. Given that recent studies of wild ungulates have demonstrated the importance of the timing of conception and birth on offspring fitness, litters conceived at different stages of the rut might be expected to exhibit differences in types and embryonic sex ratio.
2. Based on a 6-year survey of the reproductive tracts of female moose harvested in Estonia, we investigated the effect of conception date on the types of litters produced and on the foetal sex ratio.
3. There was a clear relationship between conception date and litter characteristics. Overall, earlier conceived litters were more likely than those conceived late to contain multiple embryos and a high proportion of males. However, while foetal sex ratio varied nonlinearly with conception date in yearlings and subadults, no relationship was found in adults.
4. We conclude that female moose adjust foetal sex ratio and litter type/size depending on their age and the date of conception, and that these adjustments are in accordance with the Trivers–Willard hypothesis if females that conceive earlier are in better condition.
- Top of page
- Materials and methods
Trivers & Willard (1973) suggested that in polygynous mammals, where the fighting ability of males determines their access to females, the condition and hence reproductive success of males reflects the ability of their mothers to invest in them during the pre- and postnatal period. Consequently, mothers in good condition should produce an offspring sex ratio biased towards males (Trivers & Willard 1973). However, the degree to which this theoretical model is supported by empirical data from various taxa, including mammals, has been much debated (see references in Sheldon & West 2004).
First, a difficulty in testing the Trivers–Willard hypothesis in mammals is that theoretical work has indicated that the relationship between maternal condition and offspring sex ratio may be influenced by the characteristics of both male and female offspring (Leimar 1996). Specifically, if mothers can effectively transmit their condition to daughters – for instance, through inheritance of territories or social rank – this may outweigh the benefit of producing well-conditioned sons.
Second, the relationship between sex ratio and maternal condition may depend upon the chosen measure of condition. For example, a recent meta-analysis suggested that morphological and physiological measures collected post-conception give a poorer index of a mother’s ability to invest in offspring than evidence of behavioural dominance (Sheldon & West 2004). Hence, there is a need to identify appropriate measures of maternal fitness from among those traits that can be easily obtained from a large sample of animals in a range of environments.
In contrast to reindeer and reed deer (Cervus elaphus L.), the litter size of moose (Alces alces L.) is not fixed: females usually produce one or two offspring (but occasionally more) during a single gestation period. In moose, the twinning rate (the average frequency of twins per female) is considered to be one of the most reliable indicators of reproductive ability (Solberg et al. 1999), which in turn has been found to be closely related to the age (Solberg et al. 1999) and body condition of females (Sæther & Haagenrud 1985; Sand 1996). To apply the Trivers–Willard hypothesis to species with variable litter sizes and modest cost differences between the production of male and female offspring, Williams (1979) proposed a model whereby females should change their litter size and sex ratio in relation to their own condition in the following sequence: ♀ < ♂ < ♀♀ < ♀♂ < ♂♂ < ♀♀♀ < ♀♀♂ and so on as their condition improves. However, this sequence of litter types can be different, when the cost difference between the embryonic sexes is large (Cassinello & Gomendio 1996) or if there is strong maternal transmission of condition to daughters (Leimar 1996).
In this study, we test the effect of conception date on the type of litters produced by female moose in Estonia. As an earlier conception date can be beneficial in terms of offspring fitness, we hypothesize that females that conceive earlier should produce a relatively higher proportion of expensive litters (male-biased litters with a higher twinning rate), compared with females that conceive later. In addition to Williams’s model (see above), we test other possible orderings of litter combinations, assuming that the relative costs of males and females stay the same in all litter types (see Materials and methods). Further, we look for overall differences in foetal sex ratios during different stages of the conception period. We also examine the effects of maternal age, body size, regional differences and population density on these parameters.
- Top of page
- Materials and methods
Conception date significantly affected the type of litter conceived (Table 1). In agreement with Williams’s (1979) model (the cheapest litter with one female embryo and the most expensive with two male embryos), females that conceived earlier had a higher probability of producing more expensive litters than those conceived later in the season (Table 1). Other possible orderings of litter types were not significantly related to the conception date (Table 1). Female age and region were also significant predictors of litter type (Table 1). Among younger age groups (1·5- to 3·5-year-old individuals), the ‘cheapest’ litter with one female embryo was the most frequent litter type (36–51%), whereas in prime-aged females (4·5–7·5 years old) male–female twins were most frequent (39%), and in older females (≥8·5 years old) female–female twins were most frequent (28%, Table 2). Among regions, the mean litter value increased from the western to the south-eastern and northern areas (mean regional values 2·30, 2·60 and 2·98 respectively).
Table 1. The influence of conception date on the type of litter conceived by female moose in Estonia during 1999–2004 (N = 248; ordinal logistic regression, F – female, M – male; four possible models are presented)
|Factor||D.f.||Wald Stat||P-value||Parameter estimate||SE of estimate|
| Conception date||1||5·97||0·015||−0·21||0·09|
| Age||4||29·57||<0·001|| || |
| Region||2||11·52||0·003|| || |
| Year||5||2·21||0·8|| || |
| Conception date||1||2·37||0·12||−0·12||0·08|
| Age||4||19·64||<0·001|| || |
| Region||2||7·88||0·019|| || |
| Year||5||3·12||0·68|| || |
| Conception date||1||2·62||0·10||−0·13||0·08|
| Age||4||20·93||<0·001|| || |
| Region||2||5·33||0·07|| || |
| Year||5||3·25||0·66|| || |
| Conception date||1||1·32||0·25||−0·09||0·08|
| Age||4||10·85||0·028|| || |
| Region||2||4·02||0·13|| || |
| Year||5||3·87||0·57|| || |
Table 2. Overview of twinning rates, the proportion of male embryos and the proportional distribution of different types of litter categories among different age groups in our study
|Age group (years)||N||% of twins||% of male embryos||Litter categories|
|1·5||41||15·6||33·3||21 (51)||13 (32)||4 (10)||3 (7)||0 (0)|
|2·5||64||39·1||39·3||24 (37)||15 (23)||10 (16)||10 (16)||5 (8)|
|3·5||50||46·0||38·4||18 (36)||9 (18)||10 (20)||7 (14)||6 (12)|
|4·5–7·5||57||66·7||38·9||12 (21)||7 (12)||12 (21)||22 (39)||4 (7)|
|≥8·5||36||75·0||50·8||2 (6)||7 (19)||10 (28)||9 (25)||8 (22)|
|Unknown||41||41·5||39·7||13 (32)||11 (27)||8 (19)||6 (15)||3 (7)|
|Total||289||47·4||40·1||90 (31)||62 (21)||54 (19)||57 (20)||26 (9)|
In agreement with Williams’s model, conception date also predicted the twinning rate. The results of the binomial logistic regression indicated that females that conceived earlier had more twins (conception date: Wald Stat = 5·28, d.f. = 1, P = 0·021, Estimate ± SE: −0·28 ± 0·12), as did older females (age: Wald Stat = 18·93, d.f. = 4, P < 0·001). There were also regional (mean number of embryos per female: 1·60, 1·57 and 1·36, respectively, in northern, south-eastern and western regions), but not inter-annual differences in the twinning rate (region: Wald Stat = 9·80, d.f. = 2, P = 0·007; year: Wald Stat = 0·35, d.f. = 5, P = 0·6).
The overall foetal sex ratio (pooled data of all studied females) differed significantly from 1 : 1 ratio (1·5 female per one male, N = 437; χ2 = 8·74; P = 0·0031), only 40·1% of all detected embryos were males (Table 2).
Foetal sex ratio varied nonlinearly with the progress of the conception period, with the proportion of male embryos highest in the first stage (Table 3). Moreover, a nearly significant age by conception date interaction emerged (Table 3). Further analysis revealed that in the case of yearlings and subadults, the best-fit model included the third degree polynomial relationship between the sex ratio and conception date (Fig. 3, Table 3), whereas no linear or nonlinear relationship was found in adults (Table 3). The body size index was unrelated to litter type and foetal sex ratio and was omitted from all final models.
Table 3. The influence of conception date on the foetal sex ratio in female moose in Estonia during 1999–2004 (ordinal logistic regression, categories of foetal sex ratio per female: 1 – one or two female embryos, 2 – one male and one female embryo, 3 – one or two male embryos)
|Factor||D.f.||Wald Stat|| P-value||Parameter estimate||SE of estimate|
|Age||4||6·06||0·19|| || |
|Year||5||1·60||0·9|| || |
|Age × conception date||4||8·86||0·066|| || |
|1·5–3·5 Years old|
| Conception date||1||4·63||0·031||4·02||1·87|
| Conception date2||1||6·23||0·012||−0·66||0·27|
| Conception date3||1||7·20||0·007||0·032||0·012|
| Year||5||2·11||0·83|| || |
|>3·5 Years old|
| Conception date||1||2·03||0·15||0·22||0·12|
| Year||5||1·81||0·87|| || |
Figure 3. Temporal changes in the proportion of male embryos produced by yearlings and subadults (1·5–3·5 years old). Least-square relationship is given (without accounting for year).
Download figure to PowerPoint
Owing to the nonlinear relationship between sex ratio and conception date, we also used the alternative approach of dividing the conception period into three consecutive stages and examining sex ratio separately within each (see Materials and methods). The overall foetal sex ratio (all females included) appeared to differ significantly between the different stages. In the first stage, the number of female and male embryos was almost equal (0·9 females per male, respectively, N = 159; χ2 = 0·38; P = 0·54). During the subsequent stages, the foetal sex ratio was strongly skewed towards females: 2·5 : 1 in the second stage (N = 145; χ2 = 13·47; P = 0·0002); and 1·8 : 1 in the third stage (N = 133; χ2 = 5·23; P = 0·022).
The foetal sex ratio of yearlings and subadults differed considerably between the different conception stages (Fig. 3). During the first stage, the embryonic sex ratio (0·7 females per one male) did not differ significantly from the balanced (1 : 1) sex ratio (N = 78; χ2 = 1·27; P = 0·26). The embryonic sex ratio was most strongly female biased (3·2 : 1) in the second stage of conception (N = 71; χ2 = 10·40; P = 0·0013), and still moderately female biased (2·5 : 1) in the third stage (N = 69; χ2 = 6·42; P = 0·011).
In adults (prime-aged and older females), no significant differences from the balanced ratio were found in foetal sex ratios during any of the conception stages: 1·6 : 1 in the first stage (N = 52; χ2 = 1·4; P = 0·24); 1·4 : 1 in the second stage (N = 52; χ2 = 0·62; P = 0·43); and 1·1 : 1 in the third stage (N = 57; χ2 = 0·08; P = 0·78).
- Top of page
- Materials and methods
According to Williams’s (1979) model, the females in best condition should produce two sons; those in slightly poorer condition should produce one son and one daughter, then two daughters, and then a son, whereas females in the poorest condition should produce a daughter. We demonstrate that other female characteristics, such as the date of conception, are also good predictors of the ordering of litter types: earlier conceived females more frequently produced expensive litters (male-biased litters with a high twinning rate) in comparison with later conceived females. However, it remains to be established whether the mechanism of sex allocation is directly related to conception date per se, or whether the effect of conception date is mediated by maternal condition.
Why should females that conceive early bias their foetal sex ratio in favour of males? It has been shown that an early start in life can be beneficial in terms of offspring fitness – higher body mass at birth and in the first autumn (Holand et al. 2006), a higher probability of survival (Festa-Bianchet 1988; Langvatn et al. 2004) and a higher lifetime reproductive success (Kruuk et al. 1999). Given that the effect of birth size on future survival (Loison et al. 1999) and reproductive success (Clutton-Brock et al. 1988; Kruuk et al. 1999) is more pronounced in males (due to sexual selection) than in females, sons that are born earlier are relatively more successful than earlier-born daughters.
The negative effects of delayed calving might be reduced to some extent if extra resources could be allocated to offspring during the lactation period or if the period of maternal care could be prolonged (review in Gaillard et al. 2000). However, an increase in investment during lactation, which is considered to be the most energy demanding period in the reproductive cycle (Clutton-Brock 1991), would probably further reduce the fertility and probability of survival of the mother (review in Mysterud, Coulson & Stenseth 2002). Hence, fitness returns to mothers can be increased not only by adjusting the foetal sex ratio, but also by advancing conception as a strategy to minimize the postpartum costs of reproduction.
It is noteworthy that the effect of conception date on sex ratio (more male embryos in the first stage of conception) was significant only among yearlings (1·5 years old) and subadults (2·5 and 3·5 years old), but not among prime-aged and older adults. This finding is somewhat contrary to our initial predictions. However, the costs of reproduction are indeed considered to be the highest among younger individuals (Clutton-Brock 1991) because young animals are more constrained by a trade-off between reproduction and future body growth (Sand 1996; Landete-Castillejos et al. 2004) in comparison with fully grown individuals. Accordingly, pre-weaning maternal care has been found to increase as the age of the mother increases (Weladji et al. 2003). This suggests that prime-aged and older females may be able to compensate for giving birth to their offspring late in the season by increasing their maternal care with relatively little cost, compared with younger individuals. Hence, it is likely that the effect of advancement of conception (and hence birth) is more important among younger age groups. Second, we cannot exclude the possibility that foetal sex ratio and the type of litter produced by prime-aged and older females remained relatively constant throughout the conception period because such females mated with higher-than-average quality males.
Besides conception date, female age was also linked to the type of litter conceived. This finding concurs with the many previous studies that have shown a strong effect of female age on reproductive effort, with prime-aged and older females producing the highest number of offspring (Sand 1996; Solberg et al. 1999) and the highest rate of males calves (Côté & Festa-Bianchet 2001,Landete-Castillejos et al. 2004; Säde 2004). Hence, if age is indicative of condition, then these observations are in agreement with the Trivers–Willard theory. However, in our study, female age itself was not a significant predictor of foetal sex ratio, suggesting that a complex relationship exists between litter type and sex ratio in species that produce a variable litter size.
Although the majority of the moose cows (83%) in our study were conceived during a period of 4 weeks, from mid-September up to mid-October, the whole conception period was extended over a period of 12 weeks (taking all years together). It has been shown for moose that a long conception period may reflect a predominance of cows in the population (Lent 1974; Sigouin et al. 1995; Bubenik 1998). Indeed, the sex ratio in the Estonian moose population has been female biased during the last 20 years, varying from 1·3 to 1·6 females per male (Tõnisson 2005). During the last decade, the mean age of males and females has declined sharply in the Estonian moose population due to the high reproductive rates that have accompanied the increasing population size, and due to human exploitation. We propose that the skew in the overall embryonic sex ratio and the long duration of the conception period in Estonian moose population are caused by a deficit of prime-aged males. In this respect, it has been shown that young males produce less male offspring (Sæther et al. 2004), they enter the rut later compared with older males (Clutton-Brock, Rose & Guiness 1997; Yoccoz et al. 2002), and females mate with them only when the prime-aged dominant males are not present (Komers, Birgersson & Ekvall 1999). The deficit of high-quality males is probably most obvious during the peak period of the conception period when the number of females in heat is highest. This may explain the extremely biased foetal sex ratio among younger females (three females per male) that conceived during the peak conception phase (median week) in our study.
We also found significant regional effects on the type and size of litters produced. These differences can most likely be explained by local variations in population density. While litter sizes increased from western towards south-eastern and northern areas, regional densities showed the opposite trend: 0·45 ind. km−2 in western, 0·39 ind. km−2 in south-eastern and 0·37 ind. km−2 in northern areas (Veeroja et al. 2008). Hence, the negative effects of increased density may be explained by increased competition for food, as has previously been reported from several ungulates in different geographical regions (Weladji & Holand 2003). It is also possible that slight regional differences in climatic conditions (e.g. the more maritime climate in western areas) may have had some influence.
In conclusion, we established that foetal sex ratio and litter investment vary in relation to conception date and the age of the mother. However, as the positive effect of advanced conception on foetal sex ratio was evident only in yearlings (1·5 years old) and subadults (2·5 and 3·5 years old), it is likely that prime-aged and older females adopt more flexible reproductive strategies by being more selective in their choice of mates and by investing more into postnatal maternal care.