1. It has been hypothesized that a balanced adult sex ratio is necessary for the full participation of ungulate females in reproduction and therefore high productivity. We tested this general hypothesis by combining two complementary approaches.
2. First, using telemetry (n = 60) and annual aerial censuses between 1995 and 1998, we compared two moose Alces alces populations in Quebec, Canada, one non-harvested and the other subject to intensive sport harvesting from the end of September to mid-October. We tested the following predictions for the harvested population: (i) females increase movements and home ranges during the mating period; (ii) the mating system is modified, with the appearance of groups of one male and many females; (iii) subadult males participate in reproduction; (iv) the mating period extends over two to three oestrus cycles; (v) the calving period extends over several months; and (vi) productivity declines.
3. Daily movements and home range sizes during the mating period did not differ between harvested and non-harvested populations. Most groups observed were male–female pairs. Subadult males (1·5–2·5 years old) were only observed with females in the harvested population. Mating and calving periods did not differ between populations. The proportion of females that gave birth and the number of calves produced were also comparable in the two populations.
4. Secondly, we also assessed the existence of a relationship between population productivity and percentage of males in various management units of the province of Quebec that were characterized by a wide range in sex ratios. Contrary to prediction (vi), the number of calves per 100 adult females was not related to the percentage of adult males in the population.
5. The participation of young adult males (subadults) in reproduction in our harvested population may have compensated for the lower percentage of adult males, and thus productivity was unaffected. We therefore reject the hypothesis that intensive harvesting, at least at the level we observed, affects reproduction and population productivity.
6. As there are some uncertainties regarding the long-term effects of high hunting pressure, however, managers should favour sex ratios close to levels observed in non-harvested populations.
Ginsberg & Milner-Gulland (1994) suggested that an unbalanced sex ratio could reduce conception rates and population growth. These conclusions were, however, based on indirect evidence reviewed mostly from African ungulates and modelling, and none of the studies actually tested the effects of a skewed sex ratio on the reproduction of the animals considered.
Due to high hunting pressure, moose in Quebec represent a suitable species to asses the combined effects of skewed sex ratio, low densities and rarity of old males on the reproductive process. The sport harvest rate averages 20% in most accessible zones, whereas other causes of mortality (accidents, predation, diseases, etc.) represent only an additional 4–5% (Ministère du Loisir, de la Chasse et de la Pêche (MLCP) 1993). As a result of intensive hunting, Quebec's moose population dropped to about 67 500 animals in the early 1990s from an estimated 80 000 in the 1980s (Courtois et al. 1994). In most zones, densities have reached 1 moose 10 km−2 in recent years, which is well below the habitat's carrying capacity, estimated at over 20 moose 10 km−2 (Crête 1989).
Using telemetry and annual aerial surveys, we compared two moose populations between January 1995 and February 1998, one in an area without no harvesting and the other intensively harvested. At the same time, we assessed the relationship between population productivity and percentage of males in various management units of the province of Quebec that were characterized by a wide range of sex ratios. We tested the general hypothesis that intensive moose harvesting in Quebec leads to changes in this species' reproductive processes. We tested six specific predictions to determine if the expected changes occurred in the harvested population: (i) females increase movements and home range sizes during the breeding season; (ii) the mating system is modified, with fewer pairs and the appearance of groups composed of one male and several females; (iii) young adult males (subadults, 1·5–2·5-year-olds) take part in reproduction; (iv) the breeding period extends over two or three oestrus cycles; (v) the calving period covers several months; and (vi) productivity decreases.
The study areas were located in Quebec, Canada, one in the Jacques-Cartier Conservation Park (JCCP), and the other in two controlled harvest zones (ZEC), ZEC Batiscan-Neilson and ZEC de la Rivière-Blanche. JCCP and the ZEC were located approximately 45 km north-west of Quebec City (Fig. 1).
JCCP occupies a vast plateau at approximately 800–1000 m a.s.l., dominated by undulating hills and marked by deep valleys that are occasionally very steep (Breton, Courtois & Beaumont 1998). Valleys are dominated by sugar maple Acer saccharum Marsh. and yellow birch Betula alleghaniensis Britton stands. The neighbouring plateaux are covered with softwood forests, the main species being balsam fir Abies balsamea (L.) Mill. and black spruce Picea mariana (Mill.) B.S.P. mixed with paper birch Betula papyrifera Marsh. This 670-km2 area has not been logged since 1981. However, logging was relatively intensive up to that time so there are numerous stands of young trees.
The ZEC cover an area of 942 km2 and topographical, weather and vegetation conditions are similar to those of JCCP, particularly in the portion where high moose densities are found. The plateaux, however, are at a lower altitude (600–700 m a.s.l.), and logging operations are still carried out. Forests are dominated by mixed stands and dotted with yellow and paper birch, cut-overs and softwoods.
In both study areas, a major spruce budworm Choristoneura fumiferana Clemens epidemic raged in both study areas 10–15 years ago (Breton, Courtois & Beaumont 1998). Consequently, the study areas now support young forests, a situation that is favourable to moose.
Sport hunting has been prohibited in JCCP since its creation in 1981. In contrast, hunting has always been very intensive in the ZEC, with an estimated annual harvest rate varying from 15% to 20% between 1990 and 1995 (Breton, Courtois & Beaumont 1998). The 16-day firearm hunting season (95% of the harvest) generally begins after the rut, on the second Saturday of October, but a 14-day archery season occurs during the rut, immediately prior to the firearm season. In 1994 and 1995 the harvesting of females was prohibited; a limited number of special licences were issued in 1996 and 1997, to limit the harvest rate of females to 10% (MLCP 1993).
Wolf Canis lupus Linné and black bear Ursus americanus Pallas are present in both areas. The predation rate on calves is not known, but causes of mortality other than hunting do not have a strong effect on the adult segment of moose populations in these areas (MLCP 1993).
Annual aerial surveys were conducted in both study areas from 1995 to 1998, between mid-January and the end of February, when animals are most visible (Crête, Taylor & Jordan 1981). The two study areas were surveyed by helicopter (Bell 206B or Astar 350) using north–south transects located 500 m apart, at an altitude of approximately 100 m and a speed of 160 km h−1. The transect method made it possible to delineate moose track networks directly onto 1 : 50 000 topographic maps during the survey. When it became evident that a specific track network was completely mapped (no new tracks), the helicopter left the transect to fly over the track network at low speed and altitude to search intensively for the presence of moose. We classified animals by age and sex class on the basis of size and shape of antlers for antlered moose (males lose their antlers in winter), or by vulva patch, colour of nose bridge, body size or size and shape of the bell for antlerless moose (Crête & Goudreault 1980; Timmermann 1993). The total number of animals in the study areas (and moose density) was corrected based on the proportion of marked animals found using the method described by Crête et al. (1986). Four age classes were used for males (calves; 1·5–2·5-year-olds; 3·5–4·5-year-olds; and > 5·5-year-olds) and three classes for females (calves; 1·5-year-olds; and > 2·5-year-olds). It is possible that a certain degree of overlap existed among age classes but we did not quantify this potential bias.
We estimated productivity of females as the number of calves per 100 adult females in the autumn (prior to the hunting season) for the different management units and subunits (reserves and ZEC) of the province of Quebec. We compared this to the number of males per 100 adult females based on aerial surveys conducted the following winter (St-Onge, Courtois & Banville 1994, 1995, 1998a,b). We also recorded the number of moose taken during the harvest. As part of a monitoring programme, moose abundance by sex and age class (adults or calves) and population structure (percentage males in adults and number of adult males and calves per 100 adult females) are estimated every 5–7 years in Quebec (Courtois & Lamontagne 1997). Data collected between 1994 and 1998 were utilized in this analysis because a selective harvest was introduced in most hunting zones in 1994, yielding a wide range in sex ratio. The census data covered almost the entire province of Quebec. To obtain estimates of numbers of adult males and calves per 100 adult females in the autumn prior to hunting, we corrected the data obtained from the aerial surveys conducted the following winter by adding the harvest data in each management unit and subunit.
Moose were radio-collared immediately after the aerial survey. Track networks were numbered and then randomly selected to determine which animals would be collared. We immobilized the chosen animals using a syringe fired from a capture gun from the helicopter (extra long range projector rifle; 3–10 cc syringe). In 1995, the animals' body mass was estimated visually, allowing us to determine the quantity of sedative (xylazine; Miles Canada Inc., Etobicoke, Ontario, Canada) required. We later injected RX821002A (Ultrafine Chemicals, Manchester Science Park, Manchester, UK), an intravenous antidote (Delvaux et al. 1999), to neutralize the effects of xylazine. In 1996 and 1997, 3 mg of carfentanil (Wildlife Laboratories Inc., Fort Collins, CO) mixed with 50–150 mg of xylazine, was employed to immobilize the animals. The antidote was naltrexone (Wildlife Laboratories Inc.) and the same dosages were used for both males and females. No capture-related mortality occurred in 1995. One yearling male and one adult female died from pneumonia and capture myopathy, respectively, in 1996 (Delvaux et al. 1999).
Of the 60 moose that were immobilized, 40 were fitted with Lotek LMRT-4 transmitters (Lotek Engineering Inc., Newmarket, Ontario, Canada) mounted on home-made collars (Jolicoeur & Beaumont 1986), whereas the other 20 received a Lotek Global Positioning System (GPS) collar. The transmitters were equipped with a movement sensor that increased the transmission frequency after 4 h of total immobility, thereby allowing us to identify dead animals. We also placed a numbered colour-coded ear tag on each radio-collared animal. Thirty-five animals were fitted with a collar in the non-harvested area (eight males and seven females in 1995; seven males and four females in 1996; and nine females in 1997) and 25 in the harvested area (eight males and eight females in 1995; and five males and four females in 1996).
Sixty-five per cent of telemetry locations were made from a fixed-wing aircraft (Cessna 206 or 185), whereas the others were made from a helicopter (Bell 206B). Two Yagi-type directional antennas (one on each side of the aircraft) with an ATS receiver (ATS, Isanti, MN) were used to locate each animal. When a signal was detected a switch was activated to determine the origin of the signal. The aircraft then flew in the direction indicated, subsequently flying in concentric circles or in an increasingly tighter grid pattern as the aircraft approached the animal. In all cases we used a GPS unit and recorded the date of the location and the Universal Transverse Mercator (UTM) coordinates. During the autumn (between 18 September and 9 November 1995 and between 16 September and 7 November 1996), we tracked animals daily, weather permitting, in an aeroplane or a helicopter. During the autumn of 1997, we located moose only six times. We noted the class age (young or adult in all cases; occasionally according to the ages defined above) and sex of each moose accompanying a radio-collared animal.
During calving periods (from 19 May to 9 June 1995, 15 May to 6 June 1996, and 21 May to 9 June 1997) we only used a helicopter every 3–4 days. Radio-collared females were observed directly on each flight, which allowed us to estimate dates of calving. For the rest of the year, a fixed-wing aircraft was used, on average every 3 weeks, to determine if the animals were still alive.
For each of the winters considered, we used a log-linear model to compare sex ratio and age structure between the two populations (Zar 1984).
Telemetry data were analysed according to four distinct periods (winter from 10 November to 14 May; spring from 15 May to 10 June; summer from 11 June to 14 September; and autumn from 15 September to 9 November) that correspond to the various phases of the moose's reproductive cycle (Sigouin, Ouellet & Courtois 1997). Calving occurs during spring, followed by raising of young animals during summer. The autumn season includes mating, while gestation takes place in winter. In 1997 animal locations were sporadically monitored for all seasons but spring. As a result calving was the only period analysed that year.
We calculated the minimum daily distances travelled (daily movements) by the animals. These daily movements correspond to the linear distance covered by the animals between two successive locations, divided by the time (in days) separating the two successive locations (Courtois & Crête 1988). The Calhome software program (Kie, Baldwin & Evans 1996) allowed us to calculate annual and autumn home range sizes. Home range sizes were first estimated using the minimum convex polygon (100% and 90% of the points, PC 100% and PC 90%, respectively; Eddy 1977) and then using the adaptive kernel method (Worton 1989) with 70% of the locations (K 70%).
For the two methods used, we calculated Spearman's correlations on the annual and autumn data to determine if there was a relationship between home range sizes and number of locations. We detected no significant correlation (r < 0·40, P > 0·05, Nannual= 29 and Nautumn= 25) in 1995. However, a significant correlation was noted (r > 0·30, P < 0·05, Nannual= 41 and Nautumn= 35) for the two estimators and the two periods in 1996. Consequently, for further analysis we used only those animals located 15 times or more, as recommended by Courtois, Labonté & Ouellet (1998).
We used a multifactor anova to compare daily movements and home ranges. We transformed the data to normalize the residues and homogenize the variance [daily movement, log (y + 0·1); home range: annual PC 100%, (y + 1)1/2; autumn PC 100% and annual K 70%: (y)1/3; autumn PC 90% and autumn K 70%: log (y + 1); annual PC 90%: log (y + 1)3/2].
For daily movements, we simultaneously tested the influence of study area, year of study, sex of animals, period of year (only when all the annual data were analysed) and individual animal using anova. In the case of home ranges, we analysed the effects of study area, sex, year of study and individual animal, both for annual and autumn data. All potential interactions among these factors were also tested and a Bonferroni test was used a posteriori.
We used the Mann–Whitney test to compare calving dates in the two study areas for each year, whereas the Kruskal–Wallis test provided a comparison among years. The log-linear model was used to test for productivity (number of calves per 100 adult females). We also used this test to compare the two study areas for frequency of groups of different sizes in order to compare the mating system of each population and to compare which age classes participate in reproduction.
Finally, we compared productivity (the number of calves per 100 females) for different management units and structured territories (reserves and ZEC) of the entire province of Quebec, according to the number of males per 100 females using ancova. The location of each site (south or north) with regard to the St Lawrence River was considered as a covariate because it has been suggested that productivity is greater in management units and subunits located south of the St Lawrence River (MLCP 1993), notably because wolves are absent.
A total of 2493 telemetry locations (mean ± SE; 41·5 ± 3·9 locations per animal, n = 60) was made during the study. The number of locations varied among years (881, 1196 and 416 in 1995, 1996 and 1997, respectively; Table 1), with more locations in the first 2 years of the study and females being slightly more intensively monitored than males, particularly during calving periods.
Table 1. Average number of moose locations recorded by sex (±SE) and year (1995–97) in harvested and non-harvested moose populations in central Quebec
We were unable to monitor some radio-collared moose over the entire study period. Of the 60 radio-collared animals, only 37 were still alive on 28 November 1997. During the study, six males and four females died in the non-harvested population, as opposed to nine males and four females in the harvested population. In the harvested population, sport hunting (85%, 11 out of 13) was the main cause of mortality of radio-collared moose, whereas animals died of natural causes (90%) in the non-harvested area.
Characteristics of study populations
Moose density was three times higher in the non-harvested than in the harvested population (Table 2). The ratio of males per 100 females among adults (excluding calves) was lower in the harvested than in the non-harvested population (G = 17·29, P < 0·0001, d.f. = 1, n = 984), with no significant difference recorded among years (G = 1·37, P = 0·50, d.f. = 2, n = 984). Males were younger in the harvested population (Table 3) (G = 7·03, d.f. = 2, P = 0·03, n = 189); no difference was recorded among years (G = 2·90, P = 0·23, d.f. = 2, n = 189). The percentage of > 5·5-year-old males varied between 35% and 55% in the non-harvested population, whereas it was between 8% and 22% in the harvested population (Table 3). The age structure of females, however, did not differ between populations (G = 0·01, P = 0·91, d.f. = 1, n = 335) and across years (G = 4·20, P = 0·12, d.f. = 2, n = 335). Despite a mean annual harvest of 16% (13%, 11%, 25% and 14% in 1994, 1995, 1996 and 1997, respectively) in the harvested population, mean annual growth in both areas was comparable (23·7% vs. 24·0%; Table 2). Emigration and immigration due to natal dispersal of 1-year-old moose may have been different in these two populations because of harvesting (Labontéet al. 1998).
Table 2. Autumn characteristics of harvested and non-harvested moose populations studied from 1994 to 1997 in central Quebec
Visibility rates: 0·86 in 1995 and 1996; 0·75 in 1997 and 1998 (number of marked animals observed/total number of marked moose).
Adult males: 2·5-year-olds and older.
(a) Non-harvested population (670 km2)
(b) Harvested population (942 km2)
Table 3. Percentage of moose in winter by sex and age classes from 1995 to 1998 in harvested and non-harvested populations in central Quebec, based on aerial surveys
Male age class (years)
Female age class (years)
Daily movements and home ranges
Daily movements of moose (Fig. 2) did not differ significantly between the two study areas (F = 2·13, P = 0·1454). There were no significant interactions between study area and any of the factors considered. However, movements differed significantly according to period of year (F = 57·87, P = 0·0001) and to sex of the animals by period (F = 7·01, P = 0·0001). For both years, moose of both sexes moved more in the autumn than in any other period of the year (3·30 < t < 13·95, P < 0·01). In addition, daily movements of males were significantly higher than those of the females at that time of year (t = 4·05, P = 0·0001). Movements did not differ significantly between sexes (0·07 < t < 1·06, P > 0·05) for any other time period. Daily movements did not vary significantly between the two autumns considered (t = 1·23, P = 0·22), whereas movements did differ between the two summers (t = 4·50, P = 0·0001) and winters (t = 2·79, P = 0·005). Finally, we found significant interindividual differences, independent of sex and study area, whether the analysis was done on an annual basis (F = 2·54, P = 0·0001) or restricted to autumn data (F = 2·99, d.f. = 47, P < 0·0001).
Independent of the estimator used or the time period considered (annual or autumn) (Fig. 3), there was no significant difference in home range sizes between study areas (0·003 < F < 3·11, P > 0·05). Similarly, in most cases no significant difference existed in home range sizes between sexes (2·11 < F < 3·74, P > 0·05). However, males had larger home ranges in autumn based on the minimum convex polygon method (100%: F = 4·83, P = 0·03; 90%: F = 4·48, P = 0·04).
In the harvested population, young males (1·5–2·5 years old) were frequently observed with females in the autumn (Table 4), whereas this was exceptional in the non-harvested population, where the old (> 5·5-year-old) males were more frequently observed with females (G = 13·11, P = 0·001, d.f. = 2, n = 187).
Table 4. Percentage of adult moose of each age and sex classes observed with a radio-collared moose during the autumn of 1995 and 1996 in harvested and non-harvested populations in central Quebec
Male age class (years)
Female age class (years)
1·5 – 2·5
3·5 – 4·5
(a) Non-harvested population
(b) Harvested population
During the autumn, most observed groups included only one adult of each sex. Group sizes comprising individuals of both sexes (Table 5) did not differ significantly between the two study areas (G = 3·24, P = 0·36, d.f. = 3, n = 175) and years (G = 1·85, P = 0·40, d.f. = 2, n = 175). The groups that comprised more than two animals had a larger proportion of females than males (59% of groups had a majority of females, 35% of groups had a majority of males and 6% of groups had equal numbers of males and females), both in the harvested and non-harvested populations.
Table 5. Group sizes of potentially reproducing individuals including at least one radio-collared moose (excluding calves) recorded during the breeding period of 1995 and 1996, in harvested and non-harvested populations in central Quebec
The distribution of pairings during autumn appeared to differ between the 2 years of study and between the two areas (Fig. 4). In 1995, a peak in the pairing period was observed in the non-harvested population, between approximately 20 September and 5 October (Fig. 4a). In the harvested population, however, a second peak was noted beginning on approximately 21 October and up to early November. Thus in the harvested population, pairings took place over a period of time that could correspond to two consecutive oestrus cycles. In contrast, in 1996 (Fig. 4b) the trends in the two areas appeared more comparable, with a single peak in the pairings occurring between approximately 16 September and 8 October. We were unable to determine if a second peak occurred due to the absence of data between 31 October and 4 November 1996, because of inclement weather conditions.
The percentage of radio-collared females accompanied by a male (Fig. 5) also showed one major pairing period in both years between approximately mid-September and 10 October (Fig. 5a,b). Pairing of females with animals of the opposite sex continued at a lower frequency up to early November. It is possible that a second peak in reproduction thus occurred in 1995 for the harvested population. The lack of data between 31 October and 4 November did not allow us to assess whether a second bout of pairing also occurred in 1996.
In 1995, only one radio-collared female of the non-harvested population was not observed with a male during the breeding season. This female, however, gave birth to a calf the following spring (29 May). In 1996, two radio-collared females of the harvested population and one radio-collared male of the non-harvested population were not observed with a sexual partner. As these moose were rarely observed (four, three and two locations, respectively), they may have mated nevertheless. Neither of the two females, one of which was over 15 years old, was observed with calves at heel during the following spring.
Calving period and productivity
Calving dates did not differ significantly between the harvested and non-harvested populations [U(0·05; 5; 5) = 17·00 in 1995, U(0·05; 6; 8) = 47·50 in 1996 and U(0·05; 4; 12) = 28·50 in 1997; P =0·10]. All calving occurred within a 22-day period between 18 May and 8 June (only one calf was born after 3 June). In addition, the calving dates did not differ significantly among years (Hc = 0·24, P = 0·89).
Of the 32 radio-collared females, four of 20 females in the non-harvested population and one in the 12 females of the harvested population were never observed with a calf. Three females were first radio-collared in 1997, and the other two were 14 and 16 years old in spring 1995. Advanced age may have prevented these last two females from reproducing. Two other females that were radio-collared in 1995 (one in each area) did not give birth that year. The female in the non-harvested population died prior to the 1996 calving season, and the one from the harvested population was shot in 1996. The other radio-collared females all gave birth, in equivalent proportions in both areas for each year (P = 0·57, n = 15 for 1995; P = 0·48, n = 20 for 1996; and P > 0·99, n = 25 for 1997). The proportion of females having one, two or no calf did not differ between populations (G = 1·20, P = 0·55, d.f. = 2, n = 60) nor across years (G = 3·95, P = 0·41, d.f. = 4, n = 60). On average, 50% of the radio-collared females (n = 16) had twins in the non-harvested population compared with 34% in the harvested population (n = 9). Similar proportions of females in the harvested (36%, n = 11) and non-harvested (35%, n = 10) populations had only one calf. Further, the number of calves per 100 females (Table 2) did not differ between areas [χ(0·05;3)2 = 2·002, n = 404, P = 0·57), suggesting that productivity was comparable between the two populations.
Aerial survey information from various management units or subunits censused between 1994 and 1998 across the entire range of moose in the province of Quebec was compared using ancova (Fig. 6). There was no significant relationship between number of calves per 100 adult females and number of adult males per 100 adult females in the autumn (i.e. just prior to the hunting season) in these areas (F = 0·84, P = 0·37, d.f. = 1, n = 30). The covariate (i.e. management units or subunits located south or north of the St Lawrence River) was also not significant (F < 0·01, P = 0·96, d.f. = 1, n = 30).
The aim of this study was to determine whether intensive harvesting of moose leads to changes in reproduction, ultimately reducing population productivity. To achieve this goal, we wanted to compare a moose population that was subject to high hunting pressure with an expanding population that was not harvested. At the start of the project, the population in the harvested area showed an unbalanced sex ratio in favour of females. Adult males represented approximately 30% of the adult population in the autumn, whereas the population in the non-harvested area had a balanced sex ratio among adults. Following implementation of a selective harvest in 1994, the percentage of males among adults further declined in the hunted population, an outcome noted in all populations subject to this harvesting method in Quebec (Courtois & Lamontagne 1999). In addition, densities remained approximately three times lower in the harvested population than in the non-harvested population for the duration of the study. The males were also younger in the harvested population than in the non-harvested population. Hence the premise was met of comparing an area unaffected by hunting with one affected by this factor.
Movements and home ranges
Numerous studies have shown that male moose are more mobile than females during the breeding period and that daily movements of animals of both sexes increase during this period of the year (Lent 1974; LeResche 1974; Hauge & Keith 1981; Lynch & Morgantini 1984; Cederlund, Sandegren & Larsson 1987; Courtois & Crête 1988; Garner & Porter 1990; Cederlund & Sand 1994). Our results are consistent with these observations for both populations studied. Some authors, however, believe that females become much more mobile when densities are low and the sex ratio is skewed in their favour (Houston 1968; Lent 1974; LeResche 1974; Courtois & Crête 1988). According to a study in Alaska on a population with very few males (10–20 males 100−1 females), female moose became more mobile during the rut (LeResche 1974), in sharp contrast with what was observed in Wyoming where there was a sex ratio of 81 males 100−1 females (Houston 1968). In Grimsö, Sweden, Cederlund & Okarma (1988) found no evidence of increased female movements during the mating period in a high density population (from 2·2 to 1·6 moose km−2) with a sex ratio of 50 males 100−1 females. These authors hypothesized that female activity should depend on density of reproductive animals and that, even if the sex ratio was skewed, the density was sufficient for females to avoid travelling long distances to find partners. Female movements therefore could be correlated with the distance to potential sexual partners (Courtois & Crête 1988; Cederlund & Sand 1994; Courtois, Labonté & Ouellet 1998). In our study, the females of the harvested population did not move more than those of the non-harvested population, despite low densities. No significant difference was detected between females in the two areas, even though average movements were slightly farther in the non-harvested area than in the harvested area in the autumn (Fig. 2b).
Comparison of our results with those of other studies is difficult because the periods under consideration vary and very few studies have focused on autumn home ranges. Some authors believe that home range sizes vary according to the composition of the population (Houston 1968; Lent 1974; LeResche 1974) and, in particular, that large female home ranges are associated with a female-biased sex ratio (Lent 1974; Courtois & Crête 1988). Home range sizes of moose in both Lynch & Morgantini's (1984) and our study did not increase during the breeding season, despite a female-biased sex ratio.
Among cervids, the mating system is usually polygynous with a male mating with several females during the same breeding season (Bubenik 1985). Among some cervids such as roe deer Capreolus capreolus Linné and several subspecies of forest moose (Alces alces americana Clinton, A. a. andersoni Peterson, A. a. cameloides), the female attracts the male and the male moves from one female to another during the breeding season. For other cervids (Cervus elaphus Linné, Dama dama Linné, Odocoileus virginianus Zimmermann, O. hemionus Rafinesque, Rangifer tarandus tarandus Linné, R. t. caribou, A. gigas, etc.), the male attracts a variable number of females at one site and attempts to mate them in succession (Chesser et al. 1982). According to Van Ballenberghe & Miquelle (1996), low densities and highly skewed sex ratios in moose could lead to the formation of groups made up of one reproductive male and several females. In our study, the animals in the harvested population continued to form pairs, despite the low density and the preponderance of females. Moose reproductive strategy does not seem to be influenced by population density or sex ratio (Bubenik 1985). It may, however, be influenced by physical characteristics of the habitat, as groups of female moose are found exclusively in open environments (Lent 1974).
Formation of potentially reproductive pairs was mainly observed between approximately 16 September and 10 October, which corresponds fairly closely to the average dates usually observed throughout the species range, namely between 22 September and 8 October (Sigouin, Ouellet & Courtois 1997). Couples were, however, observed throughout the autumn period (from 15 September to 9 November) in both study areas. This corresponds to a second oestrus cycle, because two oestrus cycles are separated by some 22–28 days (Schwartz & Hundertmark 1993). Further, most females were seen with potential partners on several occasions, even after 20 October. Accordingly, the presence of a male accompanying a female during the breeding period does not necessarily mean that the female is mating with the male or, if the female has indeed mated, she seems to tolerate the presence of other males later in the autumn. It has been suggested, however, that females with calves at heel prefer to isolate themselves from conspecifics (Altman 1959; Courtois & Crête 1988).
Females in the two populations showed no significant difference in calving dates. Birthing occurred within a 22-day period (18 May to 8 June). Sigouin, Ouellet & Courtois (1997) compiled data from 18 studies of moose populations around the world and showed that most calving takes place between 15 May and 8 June, as observed here. The oestrus cycle of females was not completely synchronized but rather staggered by a few days from one female to the next, as seen the taiga and boreal forest (Bubenik 1985, 1987). Subsequent calving may also be extended because the gestation period following conception during the first or second oestrus cycle is invariable (Schwartz, Hundertmark & Becker 1994). Given that all calving took place over a short period of time (no birth was observed in late June or in July), perhaps all radio-collared females mated during their first oestrus cycle, regardless of whether the animals were in the harvested or non-harvested population. We did not observe an extension of the calving period (Markgren 1969; Bubenik 1987; Ballard, Whitman & Reed 1991; Crichton 1992; Gaillard et al. 1993) in the harvested population. If females bred during a second oestrus cycle, the subsequent births would have been later and calves would have been smaller the following autumn, as shown by Schwartz, Hundertmark & Becker (1994) with captive animals. Only a weak relationship between calf size in the autumn and the number of males per 100 females was observed across Quebec hunting zones (Taquet et al. 1999). Furthermore, our results indicate that productivity is not related to adult sex ratio in various hunting zones of Quebec.
Productivity, or the number of calves born, was comparable in the two populations, and proportions of singletons or twins were also similar. It has been shown that female body reserves play an important role in reproduction (Edwards & Ritcey 1958; Pimlott 1959; Simkin 1965; Bunnell 1987; Folk & Klimstra 1991; Boer 1992; Gaillard et al. 1993; Schwartz & Hundertmark 1993; Noyes et al. 1996), with female moose between 4 and 12 years of age generally being more productive (Crichton 1988; Claveau & Courtois 1992; Heard et al. 1997). Both study populations were increasing, which suggests that numbers were below the carrying capacity. Aerial surveys revealed that the number of calves per female increased in the harvested population during the 4 years of this study, while adult sex ratio became increasingly female-biased. This may be due to an increase in the mean age of females. No significant age difference was found in the female segment of the populations between the two areas, but this may be due to the fact that it is impossible to differentiate animals that are > 2·5 years old. Age differences could have a considerable impact because number of calves produced annually by females increases considerably between 2·5 and 4·5 years of age (Claveau & Courtois 1992).
Based on the predictions considered in this study we must reject the general hypothesis that intensive harvesting leads to changes in the reproductive process in moose, at least within the levels observed. Indeed, despite a skewed sex ratio in favour of females and a low density in the harvested population, the active participation of young males in reproduction appeared to allow the majority of females to reproduce. Females probably bred during their first oestrus cycle, as births occurred at the same time as in the non-harvested population. Females found mates without the need to increase movements or to extend autumn home ranges. Thomson (1991) concluded that the reproductive behaviour of moose was flexible enough to compensate for relatively large variations in population densities, sex ratios and age structure of females. According to his study, a sex ratio in favour of females does not obviously reduce recruitment, an observation also made in our study. In addition, Ozoga & Verme (1975) and Williamson (1993), who studied white-tailed deer Odocoileus virginianus in eight ecological regions of Texas with varying hunting pressures, suggested that populations dominated by subadults have reproduction levels comparable to those of populations with normal age class distributions of males.
Although high productivity is observed, the participation of subadult males in reproduction could lead to reduced genetic variability in the population over the medium or long term (Hartl et al. 1991; Wilton 1992, 1995) because the relative contribution of large sized males may decrease, especially if there is little immigration. However, this does not seem to be the case in the areas we studied, as the genetic composition of these two populations is not significantly different (M.A. Cronin, J.C. Patton, R. Courtois & M. Crête, unpublished data). Natal dispersal of distances of up to 100 km permits some genetic mixing between populations subject to very different hunting pressures (Labontéet al. 1998). However, dispersal occurs mainly over 5–10 km (Labontéet al. 1998), and populations may become independent over the long term (Cederlund, Sandegren & Larsson 1987). In addition, it is only since 1981 that moose of JCCP have not been hunted, which may represent too short a time period for any genetic changes to appear. Other studies, however, have shown that geographical isolation and low genetic variability are not necessarily linked in moose populations (Hundertmark, Johns & Smith 1992; Broders et al. 1997).
It is possible that deleterious side-effects may appear in males that commence sexual activity at 1·5 years of age. According to inWilton 1995), young males that participate in reproduction produce high hormone levels, which leads to an early mineralization of the epiphyses. Males would then be smaller in size and less capable of reproducing than males that first reproduce at an older age. That does not seem to be the case in the populations we studied, at least for the time period considered. Moreover, Jorgenson et al. (1997) have suggested that in hunted bighorn sheep populations Ovis canadensis Shaw, subadult animals that took part in reproduction had lower survival rates than those that reproduced only at an adult age.
Finally, Ginsberg (1991) suggested that the impact of hunting varies according to ecological and behavioural characteristics specific to each harvested species. The negative effects of hunting would be particularly evident among species for which the breeding season is synchronized. Small disturbances could reduce the conception and growth rates of such populations subject to hunting (Ginsberg & Milner-Gulland 1994). For numerous ungulates, dominant males have the largest spermatozoid reserves and mate with the most females (Bubenik 1985). In moose, males are more mobile and less vigilant during the breeding period than at other times of year. They are thus vulnerable to hunting (Crête, Taylor & Jordan 1981; Wilton 1992, 1995), which could lead to a lack of reproductive males should hunting seasons and mating periods coincide. In contrast, when the hunting season occurs after the breeding period, the negative impacts would be very much attenuated (Ginsberg & Milner-Gulland 1994). A simulation model by Gruver, Guynn & Jacobson (1984) suggested that when the hunting period is delayed, an increased proportion of white-tailed deer hinds conceives during the first oestrus cycle. The main rutting period does not vary significantly among moose populations throughout its distribution range (Sigouin, Ouellet & Courtois 1995). Sigouin, Ouellet & Courtois (1995) suggested postponing the opening of the hunting season to after the second week of October because reproduction peaks between 5 and 10 October (Claveau & Courtois 1992). Because the main hunting period occurred after the breeding season in this study, productivity was not affected and females probably reproduced during their first oestrus cycle.
Recommendations for management
On the basis of scientific literature, we had predicted that females would increase daily movements and home range sizes during the breeding season, that the mating system would be modified, that subadults would take part in reproduction, that the breeding and calving periods would be extended, and that productivity would be lower as a result of harvesting. Aside from the fact that subadult males did take part in reproduction in the harvested population, the other predictions were not supported. The harvested population was composed of < 30% adult males in the autumn, which is a highly skewed adult sex ratio for taiga habitats. We therefore conclude that moose have a certain adaptability with respect to the breeding process, allowing this species to maintain a high productivity despite intensive harvesting.
Two studies (Crête, Taylor & Jordan 1981; Bubenik 1987) suggested that the adult sex ratio of moose should be as close as possible to parity during the breeding period in order to optimize reproduction. Courtois (1991) then hypothesized that approximately 40% males would be required at the time of the rut (30% in winter). In the case of our low-density study, it appears that 30% males in the autumn was sufficient to ensure mating of all females in the population, noting that the hunting season followed completion of the breeding season. Males mated with one or more females before being harvested. As there are some uncertainties regarding the long-term effects of high hunting pressure on moose, we recommend that management scenarios should be adopted to favour sex ratios close to levels found in non-harvested populations.
This project was made possible due to the close collaboration of many people. Daniel Banville, Jean-Luc Brisebois and Jean-Guy Frenette of the Service de l'aménagement et de l'exploitation de la Faune de la région 03 (saef 03) and Michel Crête of the Service de la Faune Terrestre (sft) provided necessary background information in the planning stages. Aldée Beaumont, Hugues Delvaux and Jean-Guy Frenette participated in aerial surveys and moose collaring operations. Robert Patenaude of the Quebec City zoo provided invaluable experience in handling wild animals and knowledge of immobilizing agents. Alain Caron helped in data analysis. The first author received financial support from the Société de la Faune et des Parcs du Quebec, Université du Quebec à Rimouski as well as from Forêt Modèle Bas-Saint-Laurent. Finally, a special thanks to Michel Crête, Charles C. Scwartz, Marco Festa-Bianchet, Wendy King and three anonymous referees who provided useful comments on earlier versions of this manuscript.
Received 29 May 1999; revision received 15 February 2000