Effects of food availability and density on Red Squirrel (Sciurus vulgaris) reproduction.
In many vertebrate species reproductive rates are inversely related to density, while in others no density-dependent effects on reproduction are found (Sinclair 1989). Furthermore, in some species, clutch size is density dependent in some populations but not in others, and even in the same population the inverse relationship between density and clutch size was found only temporarily (e.g., Great Tits, Parus major, Perrins and McCleery 1989, Perrins 1990, Dhondt et al. 1992). These results suggest that long-term studies are needed to investigate density-dependent effects, since in short-term studies stochastic variation may render detection difficult (Hassell 1987, Hassell et al. 1989, Dhondt et al. 1992).
Few studies have compared the relative importance of food availability vs. density dependence in determining annual reproductive rate, or have tried to understand the mechanisms by which density-dependent reproduction arises. Lack (1954) suggested that as density increases, each individual suffers reduced fecundity (Lack 1954). This idea was opposed by Andrewartha and Birch (1954), who stated that in heterogeneous habitats density dependence is the result of a lower proportion of individuals living in favorable places at higher densities. This view has recently gained considerable support (e.g. Lomnicki 1980, Begon 1984, Clutton-Brock et al. 1987), and experimental work on different Great Tit populations produced evidence that density-dependent fecundity occurs because at high density more poor-quality sites (with small clutches) are occupied (Dhondt et al. 1992).
In this paper, we describe the changes in numbers over a 9-yr period (1984-1993) and annual and seasonal variations in reproductive rate in two populations of Eurasian red squirrels (Sciurus vulgaris L.), one from a coniferous woodland, the other from a deciduous woodland. We investigate the relative importance of food supply and (male and female) density on reproductive parameters, and the mechanisms that cause some reproductive parameters to be density dependent. We then use these results to test the hypothesis of Lack vs. the Andrewartha and Birch hypothesis. The hypothesis of Lack (1954) predicts that: (1) increased breeding failure at high density will occur in all females, also territorial ones, and thus should be independent of the number of floaters or of territory quality; and/or (2) litter size of successfully reproducing females will be density dependent.
If the Andrewartha and Birch (1954) hypothesis is correct, we predict for red squirrels that: (1) only females in poor territories will suffer reduced fecundity at high density; (2) breeding failure at high density occurs mainly in those females that do not occupy good territories (floaters or territorial females in poor territories); and (3) litter size of successfully reproducing females is not density dependent.
Red squirrels are well suited for investigating effects of home range (territory) quality on density-dependent reproductive rate because: (1) the reproductive performance of all females can be monitored throughout the breeding season, which allows us to determine when breeding failure occurs (Wauters and Dhondt 1989, 1990); (2) the habitat is heterogeneous at the level of the female home-range, and home-range quality changes between years (Wauters and Dhondt 1989, 1992); and (3) measuring tree-seed production within a home range gives an accurate estimate of home-range quality (Wauters and Dhondt 1992, Wauters et al. 1995).
We used two study plots of 30 ha each, which were part of larger forests. Area C (212 ha, North Belgium, 51 [degrees] 08 [minutes] N, 4 [degrees] 3 [minutes] E) was mainly coniferous woodland, dominated by Scots pine (Pinus sylvestris) and Corsican pine (Pinus nigra). On the south, the study plot is bordered by meadows and a road that were never crossed by resident squirrels. Farmland and less-suitable patches of wood occurred to the west. Along the north and east sides of the study plot suitable habitat continues. Area D (150 ha, North Belgium, 51 [degrees] 16 [minutes] N, 4 [degrees] 29 [minutes] E) was mainly deciduous woodland dominated by mature oak (Quercus robur) and beech (Fagus sylvatica) with some chestnut (Castanea sativa). A 1-ha plot is covered with Scots pine. It is bordered by houses and a road to the north and by meadows and a motorway along the south and southwest, which were never crossed by residents. Along the east side of the study plot the forest continues (Wauters and Dhondt 1993).
Trapping was carried out bimonthly, for at least 5 d, from October 1984 to January 1993 (deciduous area) and January 1994 (coniferous area). Wooden box traps and Tomahawk "squirrel" traps were placed on the ground or against the tree trunk in a 70 x 70 m grid and baited with sunflower seeds and hazelnuts. During the breeding season (March-mid-April, and July-mid-August) traps were set during 1 wk of every 2-wk period and all females were trapped repeatedly. Every squirrel was individually marked, using small pieces of colored wires inserted through the ear or numbered metal ear tags (type 1003 S National Band and Tag Company USA), and its sex and age were determined (Wauters and Dhondt 1993). Female red squirrels are in oestrus for [approximately equal to]24 h per breeding cycle, but postoestrus (from oestrus to parturition, 38-42 d, Gurnell 1987) can easily be recognized. Young are weaned between 10 and 12 wk old (Wauters and Dhondt 1989). A female's reproductive status was scored as: (1) anoestrus (vulva small, no longitudinal opening) or oestrus (or postoestrus, vulva partly or strongly swollen with longitudinal opening) early in the breeding season; and (2) lactating (nipples large, milk excretion can be stimulated) or not lactating (nipples small or invisible) later on. Between 80 (1985-1988) and 100% (1989-1993) of lactating females were fitted with radio collars (Biotrack, UK) to locate breeding nests and determine whether loss of complete litters occurred. Breeding failure could occur: (1) before parturition, when an oestrous female was never found lactating; or (2) after parturition (loss of whole litter), when, prior to the earliest possible date of weaning, a previously lactating female shifted nest site and no young were found after checking the nest, and/or when the nipples were no longer swollen. Litter size at weaning was determined by trapping and observing young (Wauters et al. 1993, 1994b). Of 9 litters (5 spring, 4 summer) that were checked during the 1st wk after birth, litter size averaged 3.0 [+ or -] 0.9 young (range 2-4; t test for differences between seasons t = 0.75, df = 7, P = 0.52).
In red squirrels not all resident females managed to defend core areas against other females (intrasexual territoriality, Wauters and Dhondt 1992, Wauters et al. 1995), and only those with territories reproduced (Wauters et al. 1995). Therefore, we recorded a female's territorial status in each breeding season. Those with exclusive core areas are called territorial females, others are called "floaters," since they often moved over large parts of the study plot (Wauters and Dhondt 1992).
The size of the seed crop of Scots and Corsican pine, oak, and beech was used as a measure of food abundance. Food abundance was calculated by counting fallen seeds (including the remains of food items consumed before they had fallen) on 1-[m.sup.2] plots. We had an average of one counting plot per 0.5 ha. Overall food abundance was expressed in [10.sup.4] kJ/ha, using the energy content of Scots pine seeds, acorns, and beechnuts (Grodzinski and Sawicka-Kapusta 1970). To estimate territory quality per year for each female, mean food abundance of all counting plots within a female's core area was calculated (Wauters and Dhondt 1992). Since seed crops fluctuated annually [ILLUSTRATION FOR FIGURE 1 OMITTED], food abundance on a territory changed with year. To classify a territory as good vs. poor in each breeding season we used data on the energy content per seed species, and a squirrel's feeding behavior and daily energy consumption (Wauters et al. 1992). A territory was classified as good when it contained [greater than]30 cones/[m.sup.2] (area C), [greater than]30 pinecones or beechnuts/[m.sup.2], and/or [greater than]10 acorns/[m.sup.2] (area D); when fewer tree seeds were available, it was classified as poor.
Tree seeds are the main food resources for red squirrels (Gurnell 1987, Wauters et al. 1992). Seeds of Scots and Corsican pine and spruce mature from June to September, those of oak and beech from July to September. Red squirrels start consuming the seeds of freshly-formed pinecones from the second half of June onwards (Wauters and Dhondt 1987). From July to September they also eat partly-matured acorns and beechnuts (Wauters et al. 1992). Matured pine seeds, acorns and beechnuts, recovered from the ground or from trees, form the bulk of the squirrel's diet until the following spring (February-April). Hence, food resources differ for spring (born in March, weaned in May) and summer litters (born in July, weaned in September) from the same calendar year. Females lactating in the summer feed on that year's seed crop; those lactating in the spring feed on the previous year's seed crop.
TABLE 1. The different k values involved in calculating total losses (K) over a breeding season using a key-factor analysis. K = [k.sub.1] + [k.sub.2] + [k.sub.3] + [k.sub.4] (Dempster 1975).
[N.sub.0] maximum potential natality (3 young/female x number of adult females present at the start of breeding season)
[k.sub.i] ln [N.sub.0] - ln [N.sub.1] (females not entering oestrus)
[N.sub.1] fecundity (3 young/female x number of oestrus females)
[k.sub.2] ln [N.sub.1] - ln [N.sub.2] (breeding failure before birth)
[N.sub.2] maximum birth rate (3 young/female x number of lactating females)
[k.sub.3] ln [N.sub.2] - ln [N.sub.3] (nest loss by breeding females)
[N.sub.3] maximum weaning rate (3 young/female x number of successfully reproducing females)
[k.sub.4] ln [N.sub.3] - ln [N.sub.4] (nestling mortality)
[N.sub.4] total number of weaned young
Since analysis of reproduction involves different parameters that contribute to the net reproductive rate in the population, and these parameters might be affected by different environmental factors, data were log-transformed and values of each parameter were summed to calculate the total reproductive output from one breeding season to the next, using a key-factor analysis (Harcourt 1971, Dempster 1975, Begon and Mortimer 1986). The sequence of parameters included in the analyses is shown in Table 1. The k values ([k.sub.i]) represent relative strengths of various mortality factors as contributors to the total rate of mortality (K) over a reproductive period. A mortality factor is important in determining population changes when its regression coefficient with K is close to unity, and the largest regression coefficient will be associated with the key factor causing population change (Podoler and Rodgers 1975). Density dependence of each key factor can be examined by calculating the regression coefficient (and determination coefficient) of [k.sub.i] on their appropriate ln of initial densities (ln [N.sub.i-1] in Table 1, Begon and Mortimer 1986, but see Varley and Gradwell 1968).
Data on spring and summer reproduction were collected between 1985-1992 (area D) and between 1985-1993 (area C). For each study area, we tested for effects of food abundance and male and female density (independent variables) on [k.sub.i] (dependent variable) with a multiple ANCOVA model. Seasonal effects were evaluated by adding them as a class in the model (PROC GLM, SAS Institute 1989). If there was no significant effect of season, a multiple linear regression model (stepwise backward procedure in PROC REG, SAS Institute 1989) was used to select a final model that included only significant (P [less than] 0.05) variables. If season had a significant effect, separate regression models were calculated for spring and summer breeding.
A squirrel year
We tested whether spring and summer reproduction were affected by different seed crops by comparing the correlation coefficients of food abundance with the reproductive rate of males and females in two situations. Reproductive rate per sex was expressed as the number of locally weaned males (or females) per resident male (or female). In situation 1, food abundance in year t was correlated with spring and summer reproduction in year t + 1 (n = 34, males r = 0.12, P = 0.50; females r = 0.17, P = 0.34) In situation 2, food abundance in year t was correlated with summer reproduction in year t and with spring reproduction in year t+1 (n = 34, males r = 0.50, P = 0.0025; females r = 0.44, P = 0.0094). Correlation coefficients were significant only in situation 2. Kenward and Holm (1993) also found that summer litters benefited less than spring litters from the seed crop of the previous fall. Therefore, we define a "squirrel year" from July (calendar year t) to June (calendar year t + 1). In all analyses investigating effects of food abundance, the size of the seed crop in year t is related to parameters of summer reproduction of the same year and of next year's spring reproduction.
Squirrel numbers fluctuated within and between years, and fluctuations were stronger in the deciduous forest [ILLUSTRATION FOR FIGURE 2 OMITTED]. Numbers typically increase in April-May and September-October, as a result of seasonal reproduction and immigration. They decrease from June to August and during winter (November-March) due to emigration and mortality [ILLUSTRATION FOR FIGURE 2 OMITTED]. In area C at least some females reproduced in each breeding season, and both high-density peaks (May and October) occurred in each year [ILLUSTRATION FOR FIGURE 2A OMITTED]. In area D, however, no litters were produced in spring 1985 and the summer of 1991. Consequently, densities decreased strongly from May 1991 (2.2 squirrels/ha) to March 1992 (1.0/ha, [ILLUSTRATION FOR FIGURE 2B OMITTED]). Also in the deciduous forest, numbers continued to increase over the autumn-winter of 1990-1991 and the following spring [ILLUSTRATION FOR FIGURE 2B OMITTED] due to successful autumn and winter reproduction by five females and large-scale immigration as a response to a high mast crop of beech [ILLUSTRATION FOR FIGURE 1B OMITTED]. From July 1990 to May 1991, density doubled. Finally, in the pine forest, the decrease in numbers between October 1992 and October 1993 [ILLUSTRATION FOR FIGURE 2A OMITTED] coincided with the arrival of a pair of Goshawk, Accipiter gentilis, nesting in the study area. The Goshawks took at least three juvenile and eight adult squirrels between early winter and the end of July (L. A. Wauters, personal observation). Goshawks had previously been observed predating squirrels in Germany (Stubbe and Stubbe 1987).
In the coniferous woodland, the regression coefficients of the different [k.sub.i] with K was largest for [k.sub.3] and [k.sub.2], much smaller for [k.sub.4], and very small for [k.sub.1] (Table 2, [ILLUSTRATION FOR FIGURE 3A OMITTED]). One of the key factors causing changes in population size, [k.sub.2], increased with female density (Table 3). In a regression model, variation in male and female density explained 55% of variation in [k.sub.2] (selected regression model: [k.sub.2] = -0.6 + 0.09 x female density - 0.03 male density; [R.sup.2] = 0.55; P = 0.003; partial significance: female density P = 0.004; male density P = 0.029). Neither food abundance nor male or female density at the onset of the breeding season were correlated with [k.sub.3], or with [k.sub.1] (Table 3). There was an effect of season on [k.sub.4] (Table 3). During spring breeding, pre-weaning mortality of young correlated negatively with female density and with food abundance (selected regression model: [k.sub.4] = 1.4 - 0.04 x female density - 0.00032 x food abundance; [R.sup.2] = 0.74; P = 0.019; partial significance: female density P = 0.036; food abundance P = 0.009). During summer, none of the variables had a significant effect on [k.sub.4].
In the deciduous woodland, the regression coefficient of the different [k.sub.1] with K was largest for [k.sub.3] and [k.sub.2], much smaller for [k.sup.1], and very small for [k.sub.4] (Table 2, [ILLUSTRATION FOR FIGURE 3B OMITTED]). One of the key factors causing changes in population size, [k.sub.3], increased with female density and decreased [TABULAR DATA FOR TABLE 2 OMITTED] when food abundance was high (selected regression model: [k.sub.3] = 0.3 + 0.08 x female density - 0.00006 x food abundance; [R.sup.2] = 0.65; P = 0.001; partial significance: female density P = 0.002; food abundance P = 0.002). Only food abundance had a significant effect on [k.sub.2] (selected regression model: [k.sub.2] = 0.8 0.00003 x food abundance; [R.sup.2] = 0.21; P = 0.076). Finally, fewer females entered oestrus when male densities were high and food abundance was low (selected regression model: [k.sub.1] = 0.2 + 0.022 x male density 0.000019 x food abundance; [R.sup.2] = 0.52; P = 0.008; partial significance: male density P = 0.025; food abundance P = 0.015). There were no significant effects of spring vs. summer reproduction on any of the [k.sub.i] (Table 3).
In the coniferous woodland, only [k.sub.2] and [k.sub.4] are significantly related (P [less than] 0.10) with the ln of the initial density (Table 2). For [k.sub.2] the slope is positive, indicating density dependence, for [k.sub.4], it is negative, suggesting inverse density dependence [ILLUSTRATION FOR FIGURE 4A, C OMITTED]. The latter is probably an artefact of the decrease of pre-weaning mortality of young squirrels from successful litters with increasing food abundance (Table 3, [ILLUSTRATION FOR FIGURE 5A OMITTED]).
In the deciduous woodland, a positive slope of [k.sup.1] on ln of its initial density suggested density dependence (Table 2, [ILLUSTRATION FOR FIGURE 4B OMITTED]). The negative slopes of [k.sub.2] and [k.sub.3] on In of their initial densities suggested inverse density dependence (Table 2, [ILLUSTRATION FOR FIGURE 4D OMITTED]). On both k values, however, food abundance had a much stronger effect than density (Table 3), and the apparent inverse density dependence is probably an artefact of both breeding failure before birth ([k.sub.2]) and litter loss by lactating females ([k.sub.3]) decreasing with increasing food abundance [ILLUSTRATION FOR FIGURE 5B, C OMITTED].
Mechanism of density dependence
In the coniferous woodland, 38-60% of females were in poor territories in years with a seed crop smaller than the median, compared to 0-27% in years with a seed crop larger than the median. In the deciduous woodland, 0-100% and 0-20% of females occupied poor territories in years with a seed crop that was smaller and larger, respectively, than the median. The proportion of females in poor territories was not correlated with female density (coniferous woodland: r = 0.35, n = 18, P = 0.16; deciduous woodland r = 0.073, n = 16, P = 0.79) but was negatively correlated with [TABULAR DATA FOR TABLE 3 OMITTED] food abundance (coniferous woodland: r = -0.94, n = 18, P = 0.0001; deciduous woodland: r = -0.62, n = 16, P = 0.01). The number of floater females present in March or July, which varied from 0 to 5 in the coniferous and from 0 to 8 in the deciduous woodland, was strongly related to female density in both breeding seasons ([ILLUSTRATION FOR FIGURE 6A OMITTED]; coniferous woodland: ANCOVA, season P = 0.72; female density P = 0.0001; deciduous woodland ANCOVA, season P = 0.92; female density P = 0.0001). In the coniferous and the deciduous woodland 75 and 79%, respectively, of variation in the number of females in the spring (March) and summer (July) population were explained by variation in number of floaters.
Overall, significantly more females in good than in poor territories weaned young (coniferous area: good, 71 out of 177 cases, 40%, poor, 2 out of 81, 2.5%; deciduous area: good, 56 out of 175 cases, 32%, poor, 1 out of 49, 2%; three-way G test: factor area G = 2.20, df = 1, P [greater than] 0.1; factor territory G = 72.3, df = 1, P [less than] 0.001).
In the coniferous woodland, density dependence occurred in [k.sub.2]. We tested whether density dependence was caused by: (1) more females inhabiting poor territories, or (2) more females without a territory (floaters) at high density, by calculating an ANCOVA with [k.sub.2] as dependent variable, season as class, and the number of floaters and the proportion of females in poor territories as independent variables. For both spring and summer litters [k.sub.2] increased with the number of floaters (ANCOVA model: season [F.sub.1,14] = 1.73, P = 0.21; floaters [F.sub.1,14] = 7.95, P = 0.014), while the proportion of females in poor territories did not explain any additional variation in [k.sub.2] ([F.sub.1,14] = 1.25, P = 0.28). Variation in the number of floaters explained 44% of the variation in [k.sub.2] ([ILLUSTRATION FOR FIGURE 6B OMITTED], linear regression [F.sub.1,16] = 12.6, P = 0.0026). In the deciduous woodland, [k.sup.1] increased when more females occupied poor territories (ANCOVA model: females in poor territories [F.sub.1,12] = 13.0, P = 0.004; floaters [F.sub.1,12] = 0.11, P = 0.74; season [F.sub.1,12] = 0.44, P = 0.52) ([ILLUSTRATION FOR FIGURE 6C OMITTED], [R.sup.2] = 0.51). Also in the deciduous woodland, annual and seasonal variation in [k.sub.3] positively correlated with the proportion of females in poor territories (ANCOVA model: [F.sub.1,12] = 14.6, P = 0.002) and with the number of floaters ([F.sup.1,12] = 13.6, P = 0.003; season [F.sup.1,12] = 4.05, P = 0.067). Variation in the number of floaters and in the proportion of females in poor territories explained 70% of the variation in [k.sup.1].
Squirrel dynamics and habitat type
Both total densities and densities of adult females fluctuated more strongly in the deciduous than in the coniferous woodland; red squirrel numbers were lower in oak-beech forest than in mixed pine forest when food supplies were poor but higher after a mast year of beech [ILLUSTRATION FOR FIGURE 2 OMITTED]. Also, the reproductive rate did not differ significantly between woodlands, and K [ILLUSTRATION FOR FIGURE 3 OMITTED] was highest in some years in the deciduous forest, in others in the pine forest. Our data, therefore, contradict the suggestion that deciduous woodland is poorer red squirrel habitat (Kenward and Holm 1993). The latter authors based their statement on densities found in an island population (Furzey Island: 1.3 females/ha and [approximately equal to]6 squirrels/ha [Kenward and Holm 1993]), which are certainly not representative for red squirrel populations throughout Europe (Bianchi 1987, Stubbe and Stubbe 1987, Andren and Lemnell 1992, Wauters et al. 1994a). It is well documented that island populations of rodents often have much higher densities than nearby mainland populations (e.g., Tamarin 1977). We therefore conclude that pine forests are not necessarily better habitats than mixed deciduous woods, but that differences in food availability and spacing behavior (Wauters et al. 1992, Wauters and Dhondt 1992) result in different average densities and quite different dynamics.
Food abundance and density dependence
Of the different k values, [k.sub.2] and [k.sub.3] were most strongly related to total K, indicating that breeding failure of oestrous females, before or after parturition, was the key factor causing most of the annual and seasonal variation in reproductive rate.
In the coniferous woodland, [k.sub.2] was correlated with the natural logarithm of its initial density. Hence, variation in the proportion of oestrous females giving birth was (female) density dependent. The size of the seed crop of pines was only correlated with [k.sub.4]; the mortality of young in litters of successful breeders increased when pinecone production was poor. However, [k.sub.4] was less strongly related to total mortality of young (K) than [k.sub.2], suggesting that in red squirrel populations in pine forests, density-dependent litter loss has a stronger effect on annual and seasonal variation in reproductive rate than food-related variation in litter size. Also for territorial American red squirrels (Tamiasciurus hudsonicus) in mixed spuce forests, supplemental food did not affect reproductive parameters, but densities increased due to higher recruitment of juvenile immigrants (Klenner and Krebs 1991). Effects of food on immigration rate in Eurasian red squirrels are discussed elsewhere (L. A. Wauters, unpublished manuscript).
Litter size of successful breeders not being density dependent, either in the coniferous or the deciduous woodland, disagrees with the second prediction of Lack's (1954) hypothesis but supports the Andrewartha and Birch (1954) hypothesis.
In the deciduous woodland only [k.sup.1] was correlated with the natural logarithm of its initial density, indicating that the proportion of adult females entering oestrus was density dependent. Variation in [k.sub.1] was mainly explained by variation in the proportion of females occupying poor territories; when more females held home ranges with poor seed crops, fewer entered oestrus, probably because of poor body condition (low body mass) when food abundance was low (Wauters and Dhondt 1989). Hence, reduced fecundity at high density occurred mainly in females living in territories with poor food supplies, as predicted by the Andrewartha and Birch hypothesis.
However, [k.sup.1] was less strongly related to total K than [k.sub.2] and [k.sub.3] (breeding failure of oestrous females), and thus had a relatively small effect on annual and seasonal variation in reproductive rate. Although [k.sub.3], representing mortality caused by breeding failure of lactating females, was not significantly correlated with the natural logarithm of its initial density, [k.sup.3] did increase with the number of females, adults, and subadults in the population at the start of the breeding season. Hence, at high density when more females are in the population causing more overlap between home ranges of neighboring females (Wauters and Dhondt 1992), a higher proportion of lactating females lost their entire litter than when female density (and female-female home-range overlap) is low. At the same time, [k.sub.3], and also [k.sub.2], decreased with increasing food supply. Thus, rich seed crops of beech, oak, and to a lesser extent Scots pine resulted in few oestrous females failing to wean offspring. We therefore conclude that in the deciduous woodland variation in food abundance is the main factor causing annual and seasonal variation in the reproductive rate of female red squirrels. Also in British oak-hazel woods, the number of juveniles per adult female (total reproductive rate) red squirrel (with hazel crop) and grey squirrel (Sciurus carolinensis) (with oak crop) increased the year following a rich seed crop, and was negatively correlated with adult female density (Kenward and Holm 1993).
Of all key factors, only [k.sup.1] in the deciduous woodland significantly correlated with male density. These data, therefore, provide evidence that males and females represent two partly separated units within the population, as was already suggested by differences between male and female spacing behavior and by the importance of intrasexual, but not of intersexual competition in shaping dispersal behavior (Wauters and Dhondt 1992, 1993).
The mechanism of density dependence
The proportion of oestrous females giving birth to a litter (in the pine forest) and the proportion of lactating females weaning young (deciduous forests) decreased with female density. Hence, in both cases, any mechanism causing density dependence should be related to the female's social behavior. Dhondt et al. (1992) predicted that, in territorial species, density dependence will be found if (1) habitat quality differs between territories; and (2) density varies in such a way that animals use the poorer territories in some years but not in others. Both these conditions are met for female red squirrels: (1) their territories differ in the amount of food they contain, and within a given territory food availability changes over the years (Wauters and Dhondt 1992, Wauters et al. 1995); and (2) in some years, no females occupied poor quality territories, while in others up to 60% of adult females held poor quality territories. In the spring of 1985, following a crop failure of both oak and beech, all territories in the deciduous woodland were of poor quality and none of the females produced a litter.
However, in the coniferous woodland, [k.sub.2] was not correlated with the variation in the number of females occupying poor territories, but increased when there were more females not holding territories (floaters) in the population. Many floaters were subadults, and their numbers were not used in the calculations of key factors. Therefore, the number of floaters can be regarded as an independent variable. Adult floaters never produced offspring (Wauters and Dhondt 1992), although they sometimes entered oestrus. Thus, an increase in the proportion of females failing to give birth when many floaters are present was a direct effect of female density causing density-dependent mortality. In the deciduous woodland, both an increase in the number of females occupying poor territories and in the number of floaters caused [k.sub.3] to increase. Floaters might also interact indirectly with territorial females in two ways: (1) overlapping with some of the territorial females, floaters might increase competition for food, causing weight loss in breeding females during lactation, which increases the probability of breeding failure (Wauters and Dhondt 1989); and (2) floaters might find litter nests and kill the young while the mother went out foraging. We have no evidence for infanticide in red squirrels, but since territory acquisition, especially in pine forests, can only occur when a vacant territory is available (Wauters and Dhondt 1993, Wauters et al. 1995), female offspring of territorial females and floaters are potential competitors. Hence, at high density, most oestrous females that failed to produce offspring were floaters, or were those that occupied territories with poor food supplies.
We therefore conclude that the mechanism by which density dependence arises in red squirrel reproduction supports the predictions derived from the hypothesis of Andrewartha and Birch (1954); density-dependent breeding failure was the result of a lower proportion of females living in good territories at high density. However, a second mechanism also operated: a direct density effect of more nonbreeding floaters that were present in the population at high female density.
We thank the families Stoelen, Bittebier, and Van Havre and the city of Antwerp for allowing us to work on their estates. Constructive criticism by S. Dobson and T. S. Risch greatly helped to improve the manuscript. The work was funded by a concerted action from the Belgium Ministry of Education. L. Lens holds a doctoral fellowship from the National Fund for Scientific Research, Belgium.
Andren, H., and P. A. Lemnell. 1992. Population fluctuations and habitat selection in the Eurasian red squirrel Sciurus vulgaris. Ecography 15:303-307.
Andrewartha, H. G., and L. C. Birch. 1954. The distribution and abundance of animals. University of Chicago Press, Chicago, Illinois, USA.
Begon, M. 1984. Density and individual fitness: asymmetric competition. Pages 175-194 in B. Shorrocks, editor. Evolutionary ecology. Symposium 23 of the British Ecological Society. Blackwell Scientific, Oxford, UK.
Begon, M., and M. Mortimer. 1986. Population ecology. A unified study of animals and plants. Second edition. Blackwell Scientific, Oxford, UK.
Bianchi, M. G. 1987. Reintroduzione dello scoiattolo (Sciurus vulgaris L. 1758) nel Parco Naturale delle Groane: valutazione ambientale e tecniche di rilascio. Thesis, Universita' di Milano, Milano, Italy.
Clutton-Brock, T. H., M. Major, S. D. Albon, and F. E. Guinness. 1987. Early development and population dynamics in red deer. I. Density-dependent effects on juvenile survival. Journal of Animal Ecology 56:53-67.
Dempster, J. P. 1975. Animal population ecology. Academic Press, London, UK.
Dhondt, A. A., B. Kempenaers, and F. Adriaensen. 1992. Density-dependent clutch size caused by habitat heterogeneity. Journal of Animal Ecology 61:643-648.
Grodzinski, W., and K. Sawicka-Kapusta. 1970. Energy values of tree-seeds eaten by small mammals. Oikos 21:52-58.
Gurnell, J. 1987. The natural history of squirrels. Christopher Helm, London, UK.
Harcourt, D. G. 1971. Population dynamics of Leptinotarsa decemlineata (Say) in eastern Ontario. III. Major population processes. Canadian Entomologist 103:1049-1061.
Hassell, M. P. 1987. Detecting regulation in patchily distributed animal populations. Journal of Animal Ecology 56:705-713.
Hassell, M. P., J. Latto, and R. M. May. 1989. Seeing the wood for the trees: detecting density dependence from existing life-table studies. Journal of Animal Ecology 58:883-892.
Kenward, R. E., and J. L. Holm. 1993. On the replacement of the red squirrel in Britain: a phytotoxic explanation. Proceedings of the Royal Society, London, Series B 251:187-194.
Klenner, W., and C. J. Krebs. 1991. Red squirrel population dynamics. I. The effect of supplemental food on demography. Journal of Animal Ecology 60:961-978.
Lack, D. 1954. The natural regulation of animal numbers. Clarendon, Oxford, UK.
Lomnicki, A. 1980. Regulation of population density due to individual differences and patchy environment. Oikos 35:185-193.
Perrins, C. M. 1990. Factors affecting clutch size in great and blue tits. Pages 121-130 in J. Blondel, editor. Population biology of passerine birds. NATO Advanced Study Institutes Series, Volume G 24. Springer-Verlag, Berlin, Germany.
Perrins, C. M., and R. H. McCleery. 1989. Laying dates and clutch size in the great tit. Wilson Bulletin 101:236-253.
Podoler, H., and D. Rodgers. 1975. A new method for the identification of key factors from life-table data. Journal of Animal Ecology 44:85-114.
SAS Institute. 1989. SAS/STAT user's guide. Version 6. Fourth edition. SAS Institute, Cary, North Carolina, USA.
Sinclair, A. R. E. 1989. Population regulation in animals. Pages 197-241 in J. M. Cherrett, editor. Ecological concepts. Blackwell Scientific, Oxford, UK.
Stubbe, M., and A. Stubbe. 1987. Die Eichhornchenpopulation des Hakelwaldes im Nordharzvorland der DDR. Hercynia 24(1):1-10.
Tamarin, R. H. 1977. Dispersal in island and mainland voles. Ecology 58:1044-1054.
Varley, G. C., and G. R. Gradwell. 1968. Population models for the winter moth. Pages 132-142 in T. R. E. Southwood, editor. Insect abundance. Blackwell Scientific, Oxford, UK.
Wauters, L., L. Bijnens, and A. A. Dhondt. 1993. Body mass at weaning and local recruitment in the red squirrel. Journal of Animal Ecology 62:280-286.
Wauters, L., and A. A. Dhondt. 1987. Activity budget and foraging behaviour of the red squirrel (Sciurus vulgaris, Linnaeus 1758) in a coniferous habitat. Zeitschrift fur Saugetierkunde 52:341-353.
Wauters, L., and A. A. Dhondt. 1989. Body weight, longevity and reproductive success in red squirrels (Sciurus vulgaris). Journal of Animal Ecology 58:637-651.
Wauters, L., and A. A. Dhondt. 1990. Red squirrel (Sciurus vulgaris Linnaeus, 1758) population dynamics in different habitats. Zeitschrift fur Saugetierkunde 55:161-175.
Wauters, L., and A. A. Dhondt. 1992. Spacing behaviour of red squirrels, Sciurus vulgaris: variation between habitats and the sexes. Animal Behaviour 43:297-311.
Wauters, L., and A. A. Dhondt. 1993. Immigration pattern and success in red squirrels. Behavioral Ecology and Sociobiology 33:159-167.
Wauters, L., Y. Hutchinson, D. T. Parkin, and A. A. Dhondt. 1994a. The effects of habitat fragmentation on demography and on the loss of genetic variation in the red squirrel. Proceedings of the Royal Society, London, Series B 255:107-111.
Wauters, L. A., L. Lens, and A. A. Dhondt. 1995. Variation in territory fidelity and territory shifts among red squirrel, Sciurus vulgaris, females. Animal Behaviour 49:187-193.
Wauters, L., E. Matthysen, and A. A. Dhondt. 1994b. Survival and lifetime reproductive success in dispersing and resident red squirrels. Behavioral Ecology and Sociobiology 34:197-201.
Wauters, L., C. Swinnen, and A. A. Dhondt. 1992. Activity budget and foraging behaviour of red squirrels (Sciurus vulgaris) in coniferous and deciduous habitats. Journal of Zoology, London 227:71-86.
|Printer friendly Cite/link Email Feedback|
|Author:||Wauters, Luc A.; Lens, Luc|
|Date:||Dec 1, 1995|
|Previous Article:||Critical thresholds in species' responses to landscape structure.|
|Next Article:||Food hoarding by Merriam's kangaroo rats: a test of alternative hypotheses.|