Introduction

There is increasing evidence of both density-dependent and density-independent factors, as well as their interaction, affecting life history parameters, and thus population dynamics of northern ungulates (see reviews by Putman et al. 1996; Sæther 1997; Post and Stenseth 1999; Gaillard et al. 2000; Stenseth et al. 2002; Weladji et al. 2002). While density-independent effects (being either direct or indirect) are highly variable in terms of directions or trends (see review by Weladji et al. 2002), studies of density dependence in mammals suggest that increasing population density has an adverse impact on most life history parameters (Fowler 1987; Weladji 2003). However, most studies have focused on body mass, reproductive performance, survival and more recently sex ratio (see review by Weladji et al. 2002: Tables 1, 2).

Secondary sexual traits such as the horns and antlers of ungulates that are particularly costly to grow and maintain (Andersson 1994) year after year have received little attention in this context. Yet, antler length or horn length can be easily measured in the field from both live and dead individuals, and requires less handling than, for example, live body weights. Importantly, antler and horn growth are dependent on and reflect the condition of the animal during the year of growth. To our knowledge, only one study has analyzed the influence of density and climate on antler length (Schmidt et al. 2001); and a few others have assessed the influence of environmental factors (e.g. Giacometti et al. 2002 and von Hardenberg et al. 2004) or resource availability (Festa-Bianchet et al. 2004) on horn growth.

It has been argued that individuals that develop larger antlers may have a better ability to buffer developmental stress than individuals that develop smaller antlers (Markusson and Folstad 1997), and antler length is functionally related to body mass. Because antler size can be related to food availability, as there are large mineral costs as well as energetic costs associated with growing them (e.g. Suttie and Kay 1982; Ullrey 1982), it is likely that factors affecting body mass variation, such as density and climate, will also affect antler length. Indeed, Schmidt et al. (2001) recently showed that density and June temperature best explained variation in antler length in yearling red deer (Cervus elaphus) on the Isle of Rum, Scotland. We therefore predict that environmental conditions might also affect antler length variation in reindeer (Rangifer tarandus), even after correcting for the allometric effect of body mass.

Rangifer are unique among Cervids, with both sexes carrying antlers that begin to grow within weeks after birth (Blake et al. 1998). Antlers presumably provide a selective advantage for female reindeer in intraspecific competition, primarily during foraging, and especially during winter (Skogland 1989), allowing them to have higher dominance status (Holand et al. 2004) than antlerless males during winter and spring. Indeed, males shed their antlers after the rut in October when they use their antlers in competition for mates, whereas reproductive females shed their antlers around calving. Antlers in female reindeer can also be employed in defending newborn calves from predators (Suttie and Webster 1998). It is therefore important in an evolutionary perspective to investigate sex differences in the response of antler development to environmental variation.

We address these questions using 8 years of data from 5,123 reindeer calves of known body mass, sex and antler length in Østre-Namdal (Norway). More specifically, we (1) investigated cohort-specific variation in antler length of both sexes, (2) assessed the allometric relationship between body mass and antler length, (3) tested the hypothesis that antler length is influenced by density and climate, and (4) discussed sex-specific resource allocation (antler growth vs. body growth) and response to environmental variation (antler length). We used a relative measure of antler length, that is, corrected for the allometric effect of body mass, when assessing the influence of climate and density.

Materials and methods

Reindeer data

We used data from a semi-domestic reindeer population in Østre Namdal reindeer-grazing area, Nord-Trøndelag County, West coast of Central Norway. The local Reindeer Husbandry Administration Office in collaboration with local reindeer herders collected these data during a project aiming at investigating condition-dependent calf mortality (Ansgar Kosmo, unpublished data). The data are available in the records of the regional reindeer administration office of Snåsa, Nord-Trøndelag, and is described in details elsewhere (Weladji and Holand 2003a, b). Previous analyses tested the effect of winter weather and density on calf live weight and growth (Weladji and Holand 2003a), and on calf sex ratio (Weladji and Holand 2003b), but did not investigate influence of weather and density on antler length. Our dataset included 2,503 male and 2,620 female calves with known antler length, live weight and date of measurement (see Fig. 1 for sample size by year and sex). This is a managed population as all semi-domesticated reindeer population throughout Norway, Sweden, Finland and Russia as they belong to the reindeer herders. However, the density varies greatly between years due to environmental stochasticity (climate) and predators influencing calf survival. This environmental unpredictability in addition to the between year variation in the herders success to gather the herd and hence slaughter according to their plans, imply an adaptive harvesting strategy where density varies greatly.

Fig. 1
figure 1

Inter-annual variation in mean antler length adjusted for the date of measurement (with error bars) for male (filled circles and solid lines) and female (open circles and dotted lines) reindeer calves in Østre-Namdal Norway, during 1979 through 1986. Sample size for each year in sequential order: males, 95, 194, 58, 240, 275, 429, 650 and 562; females, 98, 209, 60, 250, 304, 421, 684 and 594

Full size image

Following the calving period in the middle of May, reindeer are gathered for calf marking (ear tagging), and live body weight, antler length and sex of calves were recorded, in addition to ‘weighing dates’ or ‘dates of antler length measurement’ (hereafter referred to as ‘date’). Recorded dates were mainly from mid-July. Antler length is the mean length (cm) of the left and right antler. In order to obtain an index for density, the population size (estimated from a yearly census at the end of March during a round up) was divided by total available range (Reindriftsforvaltningen 1998). During the study period, the density varied between 0.27 animal/km2 and 0.57 animal/km2 (mean=0.40, SD=0.10), with no temporal trend.

Climatic data

Based on previous work relating weather to northern ungulate phenotypic traits (review in Weladji et al. 2002), climatic variables affecting fitness, energetic requirement, and plant phenology were selected (Langvatn et al. 1996). We used mean snowfall from December to March as a winter condition index, and monthly mean temperature and precipitation from May to June as our early summer indices (e.g. Hewison and Gaillard 2001). We considered the winter index in the wintering area, and the summer indices in the summer area as reindeer display migratory patterns within the study area. Values are calculated from Namdalseid weather station (~80 m asl) for the winter area and from Nordli-Holand station (~433 m asl) for the summer area. The meteorological data were obtained from the Norwegian Meteorological Institute, Oslo.

Data analysis

We used linear models (GLM procedure; SAS 1999) to test the effects of year (cohort effect in this case), and sex as well as their interaction, on the variation in antler length, while including ‘date’ of measurement, the only continuous independent variable in the models, to account for temporal changes in antler length. From these models, we also generated least square means (lsmeans) of antler length by year and sex (Fig. 1). We compared inter-cohort variability between lsmeans body mass and antler length using the Levene’s test that fit any continuous distribution (Levene 1960). Since antler size and body mass are functionally related (see also Fig. 2), we assessed their allometric relationship by applying allometric regression (y=bx a; or log–log regression) separately for male and female, between antler length (the response) and body mass (the predictor) of calves. As we found evidence of an allometric relationship (see Results section), we then used the residuals of each individual from the preceding models when testing the effect of environmental conditions as a measure of relative antler length (RAL), that is, antler length corrected for the allometric effect of body mass.

Fig. 2
figure 2

Scattergram displaying the relationship between live weight (adjusted for the weighing date) and antler length (adjusted for the date of measurement) for male (filled circles and solid lines) and female (open circles and dotted lines) reindeer

Full size image

Environmental effect on antler length was analyzed by mixed linear models with both fixed and random effects (Littell et al. 1996) using the Mixed procedure in SAS (1999), version 8 with a 95% level of significance. Because of repeated sampling within a year, ‘year’ was fitted as a random effect in the model (see e.g. Weladji and Holand 2003a). The following fixed and continuous independent variables were included in the model: ‘date’, density, winter snowfall, May–June temperature and precipitation. Different models were performed for each sex. There was no correlation between the climatic variables (May–June temperature and May–June precipitation: r=0.22, P=0.6; May–June temperature and winter snowfall: r=−0.16, P=0.7; May–June precipitation and winter snowfall: r=−0.22, P=0.6). We used the parameter estimates of the models (i.e. comparison of the ‘effect size’) to discuss the eventual sex-specific effect of weather on the RAL.

Results

Cohort and sex-specific variation, and allometric relationship

Antler length varied significantly among cohorts (F 7,5106=27.71, P<0.001; Fig. 1) and between sex (F 1,5106=726.47, P<0.001; Fig. 1). The interaction between sex and year was also significant (P<0.001), suggesting a gender specific pattern of variation in antler length. Males had on average 3.8 cm (SE=0.22) longer antlers than female calves. Adjusted yearly mean of body mass and antler length showed similar inter-cohort variability (Levene’s test statistic=0.02, P=0.9), also reflected by the high correlation between the two phenotypic traits (male: r=0.89, P=0.003; female: r=0.86, P=0.007; Fig. 2). Accordingly, we found the relationship between antler length and body mass to be allometric [i.e. allometric exponent a expected to be 3 as it relates a volume (here body mass with power of 3) to a length (here antler length with a power of 1); Huxley 1932] for both male (2.72±0.04, df=2501, P<0.001, R 2=0.70) and female (3.11±0.05, df=2614, P<0.001, R 2=0.60) calves. RAL also varied significantly between years for both male (F 7,2494=18.05, P<0.001) and female (F 7,2607=21.24, P<0.001) calves.

Effect of environmental condition

Decreasing early summer temperature (male: F 1, 2496=18.20, P<0.001; female: F 1, 2609=18.26, P<0.001; Table 1, Fig. 3a), and increasing early summer precipitation (male: F 1, 2496=3.73, P=0.05; female: F 1, 2609=6.04, P=0.014; Table 1, Fig. 3b) had a positive effect on RAL. Density-dependence was expressed through a positive effect of density on RAL (male: F 1, 2496=5.25, P=0.02; female: F 1, 2610=8.79, P=0.003; Table 1). Although not significantly different, the effects of environmental variation (i.e. early summer weather and density) on RAL were, on average, higher on females than on males (parameter estimates and standard errors in Table 1). The effect of ‘date’ was positive and significant, whereas the effect of winter snowfall was not significant (Table 1).

Table 1 Parameter estimates and standard errors (SE) from the mixed model describing the response of the relative measure of antler length of males and females to density and weather variables. Significant parameters (P≤0.05) are from the model excluding non-significant terms (reduced model). Denominator degrees of freedom were 2,496 for males and 2,609 for females
Full size table
Fig. 3
figure 3

Variation in mean relative antler length (RAL) (adjusted for calf sex and the date of measurement) for male (filled circles and solid lines) and female (open circles and dotted lines) reindeer calves in relation to (a) May–June precipitation (mm) and (b) May–June temperature (°C)

Full size image

Discussion

Antler lengths of calves were highly variable among cohorts. The variability in antler length was induced by the combined effect of density-dependent and density-independent factors experienced by cohorts in utero and during their postnatal period of early summer. Indeed, even after the allometric effect of body mass was removed, variation in RAL was sensitive to environmental conditions. Increasing population density while calves were in utero, increasing May–June precipitation and decreasing May–June temperature all positively influenced the relative measure of antler length for both male and female calves. Several other studies have shown effects of environmental conditions on inter-cohort variation in life history traits and population dynamics of large herbivores (see review by Putman et al. 1996; Sæther 1997; Gaillard et al. 2000; Weladji et al. 2002; Stenseth et al. 2002), including reindeer (Weladji 2003).

As expected in Rangifer, a highly sexual size dimorphic species, male calves had longer antlers (e.g. Wika 1980; Blake et al. 1998) than females. Overall and for both sexes, live weight and antler length were positively correlated and showed similar intercohort variability. Indeed, both phenotypic traits showed a strong allometric relationship. Because male calves are heavier (Weladji and Holand 2003a, b) and have longer antlers than females (this study) in this population, and as sexual dimorphism in antler length was largely explained by the difference in body mass, RAL did not differ between sexes. This supports the view that body size is one of the most important life history traits (Calder 1996).

That the effect of density on RAL was positive might seem surprising, but this may also reflect differential tactics of individuals towards various phenotypic traits like body mass and antler length. Indeed, a negative effect of increasing population density on live body mass of this population have been shown (Weladji and Holand 2003a), and was attributed to increased intra-specific, winter competition resulting in reduction of the per capita food availability while in utero (through the mother by affecting foetal growth: see e.g. Skogland 1984 for Rangifer) and after birth (neonatal resource limitation) (see e.g. Reimers et al. 1983 and Skogland 1983 for Rangifer). Given the high correlation between live weight and antler length in this population, the effect of density on antler length is also expected to be negative. Indeed, reduced resource availability has been reported to negatively affect antler growth or length (Wika 1980; Scribner et al. 1989; Schmidt et al. 2001), as well as horn growth (Festa-Bianchet et al. 2004). The reported positive effect of density on RAL could be caused by the removal of the allometric effect of body mass on antler length. Indeed, live weight was, largely, a linear function of population density (Weladji and Holand 2003a) and increased population density would certainly lead to increased competition. That the relative allocation of resource towards antler growth (in comparison to body growth as measured by the RAL) is higher at high density suggests a real investment into antler, leading us to argue that antlers are important for both male and female calves. Longer antler is reflected in higher social rank in this population (see Holand et al. 2004), beneficial in both intrasexual and intersexual competition.

The effect of early summer weather variables on RAL could partially be attributed to variation in forage quality and quantity. Increased precipitation is expected to favor plant growth, that is, increased biomass, while temperature changes have potentially more influence on the forage quality. That both precipitation and temperature were important predictors of variation in RAL suggests that a combination of both quality and quantity of the forage determine the relative growth of antlers in summer. Indeed, appropriate proportion of nutrients, water, light and heat are necessary for optimal plant growth (Jurïsson and Raave 1984). We found increased temperature to significantly reduce RAL, most likely through its influence on the phenological development of forage plants. This is in accordance with recent findings by Schmidt et al. (2001) for red deer. Several other studies have shown a relationship between ungulate condition and the conditions of the summer range (Reimers et al. 1983; Sæther et al. 1996; Hjeljord and Histøl 1999), with cool early summers being favorable for weight gain (Langvatn et al. 1996) and antler growth (Schmidt et al. 2001; this study). During cool summer weather, lignification of the vegetation is delayed; thereby extending the period when high quality forage is available by increasing the crude protein:dry matter ratio (Jonasson et al. 1986; Laine and Henttonen 1987). Another potential, but not mutually exclusive, explanation is an increase in parasitic insect harassment towards reindeer, and especially calves, with increasing temperature and moisture. Insect harassment would have a similar effect through decreasing time spent feeding (reduced intake) and increasing an individual’s energy expenditure (Colman et al. 2002; Weladji et al. 2003), thus decreasing growth (of e.g. antlers). Calves are especially susceptible to the negative effects of insect harassment through direct losses, but also indirectly through a potential decrease in their mother’s lactation and allocated suckling time.

Antler length inter-cohort variability confirms the potential for a cohort effect caused by physical condition experienced by pregnant females. Antler size is important for breeding success for males and winter and spring foraging success for females (Prichard et al. 1999). Moreover, it has been shown that the conditions under which individuals are born may have important long-term consequences for the cohort’s life history (Albon et al. 1992; Forchhammer et al. 2001; Gaillard et al. 2003 for a review on deer; see also Solberg et al. 2004). These suggest that the conditions experienced by female reindeer during pregnancy, and thereby their offspring, could persist throughout a calf’s life span (see e.g. Adams and Dale 1998), as we show for antler length in reindeer (see electronic appendix S1).

The allometry between antler length and body mass was gender specific, being positive for females (i.e. antler length increased proportionally faster than body mass; a>3) and negative for males (i.e. body mass increased proportionally faster than antler length; a<3). Indeed, compared to male calves, female calves likely prioritize antler growth over body mass, as shown by their significantly higher allometric exponent (especially in poor years with reduced resources). This is likely caused by male calves having higher body mass threshold for winter survival than females, and may therefore have a more risky allocation strategy (Loison et al. 1999; see also Clutton-Brock et al. 1982). Our study therefore demonstrates significant sexual dimorphism in allometry. This suggests that female calves may be directing proportionally more resources to antler growth, as compared to other nutritional costs related phenotypic traits such as body mass. They may thereby enhance their competitive ability for food (in winter/spring), and when sexually mature secure their access to dominant males through intrasexual competition, and/or their attractiveness to males, be more successful in intersexual competition and, lastly be able to provide security to their offspring after calving. Indeed, in semi-domestic reindeer, antlers provide females a social advantage during rut, winter and calving (Espmark 1964, 1971) and, thus, have an important social function (Reimers 1993; Holand et al. 2004). Female reindeer need larger antlers to aggress and repel young bulls during the rut, and to secure access to feeding craters during winter for themselves and their offspring (Reimers 1993). This suggest a sexual difference in the resource allocation strategy for growth of various life history traits and for reproduction (see LeBlanc et al. 2001). Accordingly, Festa-Bianchet et al. (2004) found that as resources become scarce, young bighorn rams allocate an increasing proportion of those resources to body growth rather than to horn growth, thereby increasing their probability of survival. Solberg et al. (2004) recently reported sex-specific difference in the magnitude of the effects of environmental conditions at birth on adult body mass, being more influential on males than females. Timing of antler growth differs between reindeer sexes (see Blake et al. 1998), so any environmental influence on condition/growth that operates seasonally might also influence the sexes differently for this trait (see Solberg et al. 2004 for evidence on body mass). Our results lend support to the hypothesis that environmental variation influences the extent of sexual dimorphism (Post et al. 1999) in antler size. Indeed, the cohort variation in antler length was gender specific.

Finally, our analysis suggests that antler length is a phenotypic trait as good as live body mass for signalling individual quality for male and female reindeer, and that antler length is hence an indicator of viability (Kokko 1997; Ditchkoff et al. 2001). This has already been proposed for several large herbivore populations (reindeer: Markusson and Folstad 1997; red deer: Schmidt et al. 2001; white-tailed deer Odocoileus virginianus: Ullrey 1982; Rasmussen 1985; Ditchkoff et al. 2001; moose Alces alces: Solberg and Sæther 1993). Horn size has also been proposed to be an ‘honest’ signal of individual quality (Festa-Bianchet et al. 2004; von Hardenberg et al. 2004). Because factors affecting antler growth and length are important for breeding success (Prichard et al. 1999), because cohort effects are widespread in deer populations (Gaillard et al. 2003), and because intra-specific competition favor individuals with large antlers (e.g. Clutton-Brock 1982; Barrette and Vandal 1985), we suggest that antler length, which is easily recorded in the field, could be used when selecting superior animals for future reproduction. Indeed, we also found antler length to increase with increasing age, and show that on average, individuals with long antlers as a calf also have long antlers as an adult (see Fig. S1 in electronic appendix S1). This has important implications for the semi-domestic reindeer husbandry and slaughtering practices, as well as wild reindeer population assessments and management decisions. Trophy hunters should be aware that removals of superior animals might have undesirable evolutionary consequences (see Coltman et al. 2003).