Genetic variation and inbreeding
The level of genetic variation in the small and isolated moose population on Vega was much lower in allelic richness (AR mean = 3.7, SD = 1.1) and heterozygosity (HE mean = 0.50, SD = 0.14) than recorded in the mainland Norwegian population (in the same 15 microsatellite loci, Haanes et al. 2011: AR mean = 7.4, SD = 2.5, HE mean = 0.66, SD = 0.13). This suggests strong genetic drift, as was expected from the few founders and subsequently low population size. However, albeit a low genetic variation, genetic parentage assignment was significant in most individuals. Interestingly, assignments with strict confidence corresponded very well with observed social maternities (95%). Also assignments with relaxed confidence corresponded with social maternities (83%) and after exclusion of accepted twin mothers, assignments with relaxed confidence corresponded even better with social maternities (92%). This high agreement between genetic and socially determined maternities provides credibility to previous ecological investigations at Vega (e.g., Sæther et al. 2003, 2004; Solberg et al. 2007).
In accordance with the low population size, the level of inbreeding calculated from the pedigree (f) was high. The level of inbreeding was reduced following immigration but shortly after increased again (Fig. 2). The variance in f was also high, as can be expected when inbred populations contain immigrants and their descendants (Reid et al. 2006). The relatedness structure subsequent to immigration also explains the correlation in f between offspring and each parental sex (Reid et al. 2006; Reid and Keller 2010). A low variation in f is often reported from wild populations (Grueber et al. 2008, 2011), which may explain the often reported weak or absent correlations between inbreeding and heterozygosity (Slate et al. 2004; Chapman et al. 2009). With a limited number of loci, random segregation in each locus may also have an effect (Slate et al. 2004; Hill and Weir 2011). We detected a negative correlation between f and MLH, indicating a sufficient mean and variance in f compared to the number of loci. The correlation in homozygosity across loci suggests a genome-wide effect (Szulkin et al. 2010), indicating that variation in MLH was due to inbreeding.
Effects of inbreeding on fitness-related traits
We found negative effects of inbreeding on three fitness-related traits (Tables 1-3 and Fig. 3): (1) a later date of birth was associated with high f-values in calves, (2) calf body mass was negatively related to calf f and increased with mother MLH, and (3) twinning rates were lower for cows with higher f-values. Inbreeding may operate on different life history stages (Szulkin et al. 2007; Grueber et al. 2010) and the maintained relationships between inbreeding and both calf body mass and twin rate after accounting for preceding life history traits suggest that inbreeding has a separate effect on these traits. Surprisingly, we did not find any inbreeding effects on female age-specific lifetime reproductive success, but this was probably because of few individuals with data on asLRS and hence low statistical power. By comparison, significant effects of inbreeding have been found on the lifetime reproductive success in other ungulates (Slate et al. 2000) as well as in juvenile survival whereas none or only small effects have been found on date of birth and juvenile body mass (Overall et al. 2005; Dunn et al. 2011; Walling et al. 2011). Indeed, inbreeding depression seems to be stronger in traits that are closely related to fitness (De Rose and Roff 1999; Wright et al. 2008) and should therefore be expected in survival and reproduction parameters.
The later date of birth for inbred calves may have two explanations: (1) that conception occurs later in the rut for inbred than for more outbred calves and (2) that inbreeding involves a longer gestation period. Variation in conception date can occur as a result of varying cow condition at the onset of rut (Garel et al. 2009), or as a consequence of low availability of high-quality males (Mysterud et al. 2002). In female moose, fecundity, age at first reproduction, and twinning rate depend on body mass, which is an important life history trait in moose (Sæther and Haagenrud 1983; Solberg et al. 2008). Given the strong effect of juvenile body mass on adult body mass (Solberg et al. 2004, 2008), it is likely that the negative effect of inbreeding on calf body mass (Fig. 3B) is maintained into adulthood. Therefore, as the date of birth was unaffected by the level of inbreeding in mothers and fathers, we find it unlikely that inbreeding effects on cow conditions or mate choice caused the later birth dates of inbred calves. More likely, variation in gestation length can explain some variation in birth date, for example, if inbred fetuses have slower growth. Schwarts et al. (1988) reported that moose delayed birth date by 2 weeks following starvation while reindeer can shorten gestation by 2 weeks in response to delayed conception (Holand et al. 2006). Hence, there seems to be some flexibility in the length of the gestation period of ungulates.
One benefit of early birth is that calves have longer access to high-quality forage, which can have profound effects on body growth and fitness in large herbivores (e.g., White 1983). Moreover, as cold springs involve better forage and faster moose growth (Herfindal et al. 2006a), mothers may allocate more energy to fetuses in cold than warm springs. This could enable an earlier birth, as was found in outbred but not in inbred calves. Possibly, inbred calves are less able to take the advantage of such variation in mother's foraging conditions, making them more inclined to be born after a fixed gestation period than are outbred calves.
The lower body mass in autumn and winter of inbred calves may be explained by lower birth weight and/or lower weight gain after birth. Previous studies have found calf body mass to be related to birth date (Sæther et al. 2004), which may explain why birth date was included in the highest ranking model of inbreeding effects on calf body mass. Moreover, ungulate birth weight is also related to the time of birth (Coulson et al. 2003; Dunn et al. 2011; Walling et al. 2011). Early birth may thus affect body mass in moose by providing the calf with a longer period of access to fresh vegetation, or earlier born calves may simply have been born with higher body mass. However, as the effect of inbreeding was maintained even after accounting for date of birth, we believe that variation in calf body mass was at least partly related to different weight gain during spring and summer. This relationship was however affected by calves also being smaller when born by mothers with low heterozygosity (MLH). Hence, inbreeding effects on calves may also be affected by maternal effects, for example by the level of resources mothers allocate to the calf during gestation or lactation.
The lower twinning rate of inbred cows may be because inbreeding affects the body condition of cows, which is known to affect reproductive performance in moose (Sæther and Haagenrud 1983; Sæther and Andersen 1996; Solberg et al. 2008). Accordingly, the inbreeding effects on twinning rate could simply be the outcome of the inbreeding effects on birth date and calf mass, because individual variation in juvenile body mass is negatively related to birth date (Sæther et al. 2004) and is maintained into adulthood (Solberg et al. 2004, 2008). However, inbreeding can also affect fertility directly, either through sperm quality (Salisbury and Baker 1966) or female ovulation rate (Falconer and Roberts 1960; Doney and Smith 1968). At Vega, the twinning rate was weakly positively related to the calf mass of the mother and negatively related to her birth date, which supports the hypothesis that inbreeding effects on the twinning rate operate through the body conditions of the mother. However, there was still an effect of inbreeding after accounting for these relationships, indicating that inbreeding also has a separate effect on fertility. Indeed, similar independent effects of inbreeding on separate life history traits have also been reported in other studies (Szulkin et al. 2007; Grueber et al. 2010).
Because of population fragmentation and decline (Parmesan 2006; IPCC 2007), the potential effects of inbreeding on population growth and viability receive increasing concern (Keller and Waller 2002; Bijlsma and Loeschcke 2012; Pekkala et al. 2012). In ungulates, inbreeding depression has been documented in a few small or fragmented populations (e.g., red deer; Slate et al. 2000; Walling et al. 2011; Soay sheep; Coltman et al. 1999). Low genetic variation and jaw deformities were reported in a small and isolated red deer population (n = 50, Zachos et al. 2007), and increased genetic drift and inbreeding was found in small isolated populations of mountain goats (Oreamnos americanus, Ortego et al. 2011), alpine ibex (Capra ibex, Biebach and Keller 2010) and pronghorn (Dunn et al. 2011). Here, we report strong genetic drift and inbreeding depression in fitness-related traits within the small and isolated moose population on Vega. The inbreeding depression was expected to become more pronounced with increasing environmental stress (Keller and Waller 2002), but only variation in moose birth date was better explained when including climate variables. Early life history traits seem to be more sensitive to environmental stress than later appearing traits (e.g., Gaillard et al. 2000) and for that reason fetus growth and length of gestation may be more affected in harsh conditions. The high body growth, fecundity, and calf recruitment rates at Vega suggest that the island provides favorable living conditions for moose (Sæther et al. 2007), and this may explain why the climatic conditions are of little importance for the observed inbreeding effects. However, such effects could become more apparent at higher population densities or if the environment changes. The population has been harvested since 1989, but still the Vega population has so far been above the Norwegian average in body mass and reproduction (Solberg et al. 2011). Such high performance could also indicate that detrimental alleles not yet have accumulated or become fixated to any large extent.