Population biology of stoats Mustela erminea and weasels Mustela nivalis on game estates in Great Britain


Robbie A. McDonald, Game Conservancy Trust, The Gillett, Forest in Teesdale, Barnard Castle, Durham DL12 0HA, UK (fax +44 1833 622343; e-mail rmcdonald@gct.org.uk).


  • 1British gamekeepers commonly trap and shoot stoats and weasels in order to increase the abundance of game. We provide details of the population biology of 822 stoats and 458 weasels collected on 25 game estates and use simple population models to assess the effects of culling.
  • 2Seventy-one per cent of stoats and 94% of weasels were trapped, while 26% of stoats and 5% of weasels were shot. While trapped samples exhibited typically male-biased sex ratios, the sex ratio of shot stoats was even. Eight of 305 female stoats and six of 77 female weasels were visibly pregnant, with mean litters of 9·0 and 6·2 embryos, respectively. Median ages at death were 11·6 and 8·0 months for male and female stoats, respectively, and 9·3 and 9·2 months for male and female weasels. Male and female stoats, but not male and female weasels, had significantly different rates of survival.
  • 3Model weasel populations continued to increase (λ = 1·35) despite culling as a result of high productivity when sufficient food was available. Model stoat populations declined slightly (λ = 0·95), probably as a result of concerted culling effort when young stoats were dependent on maternal survival. This suggests that persistence of culled stoat populations may depend on immigration.
  • 4To reduce stoat populations without affecting the survival of dependent juveniles, culling effort could be focused on trapping females in late winter and shooting females in early spring, where landscape and climate permit. For control of weasel populations, trapping effort should be, and in practice often is, focused on late spring, following a period of high natural mortality.
  • 5High rates of immigration mean that culling by gamekeepers will not ordinarily lead to any long-term decline in actual stoat and weasel populations. We suggest that measures taken to enhance immigration will improve the long-term status of stoats and weasels in regions where their conservation is desirable, and whilst this persists the impact of culling will be short-lived and local.


Predator control is an integral part of game management in Britain (Tapper 1992). Stoats Mustela erminea L. and weasels M. nivalis L. are widely culled by gamekeepers aiming to increase game bird populations (McDonald & Murphy 1999). Numbers of stoats and weasels culled in Britain have been in decline since the mid-1970s (Tapper 1992), leading to suggestions that actual populations may also be in decline (Harris et al. 1995; Macdonald, Mace & Rushton 1998). To a large extent, declines may be attributed to bias in the records arising from changes in culling effort (McDonald & Harris 1999). Secondary exposure to rodenticides is commonplace and may also be contributing to a pattern of decline (McDonald et al. 1998). Changes in prey availability are thought not to have a role, as both species appear to have benefited from increasing rabbit Oryctolagus cuniculus L. populations following recovery from myxomatosis (McDonald, Webbon & Harris 2000). Because legal protection is often implemented for declining species, it is necessary to assess whether culling may have a role in bringing about a decline in either species. Furthermore, although culling is undertaken to enhance game populations, so far there has been no assessment of the relative merits of different techniques for stoat and weasel control. To make these assessments, information is required on culling practice and the demography of culled populations.

Although both species are generally short-lived and are relatively prolific breeders, the population biology of stoats differs from that of weasels in several significant ways (King 1989). Following mating in stoats, fertilized blastocysts undergo a period of delayed implantation lasting 9–12 months. Thus, stoats mate one year and give birth the next and can only have one litter a year. In contrast, the implantation of blastocysts in weasels is direct and females may have two litters a year when food is plentiful. Young female stoats, but not males, become sexually mature while still naked and helpless in the nest and will mate with adult males at this time. Weasels of both sexes mature at 3–4 months and so females can produce a litter in the year of birth (King 1989).

Tapper (1979 ) and King (1980a ) have analysed large samples of weasels collected on British game estates and shown that populations show marked fluctuations in response to the availability of small rodents. No work has yet been published on the population biology of stoats on British game estates. Nevertheless, live-trapping studies in Europe ( Delattre 1983 ; Erlinge 1983 ; Debrot 1984 ) and studies of culled animals in New Zealand ( King 1981 ; King & Moody 1982 ; King 1983c ; King et al. 1996 ; Powell & King 1997 ) have consistently demonstrated the importance of food supply to stoat demography. Powell & King (1997 ) demonstrated that cohorts of stoats born at different stages of the seed-fall cycle in New Zealand beech Nothofagus spp. forests differed greatly in their fecundity and survival in response to marked fluctuations in small rodent abundance.

No study has yet produced models analysing the aspects of stoat and weasel demography that most strongly affect population growth. Models are most useful when they simulate the behaviour of real populations. These are often the most difficult models to construct or have confidence in, because the natural situation routinely includes environmental and demographic stochasticity and populations generally exhibit some form of density dependence. Defining the effect of these processes is difficult when detailed data are scarce. Nevertheless, it is possible to construct prospective models that provide a general assessment of the degree to which population growth is affected by each vital rate in an analytical and quantitative way (Tuljapurkar & Caswell 1997).

In this study, demographic data derived from stoats and weasels culled on British game estates were used to construct simple population models. These were used to assess the status, defined by the population growth rate, of culled populations of stoats and weasels on British game estates. Models were used to describe the sensitivity of population growth rate to changes in demographic parameters and to provide a preliminary assessment of whether culling by gamekeepers may have a role in any decline in stoat and weasel populations.

Materials and methods

Gamekeepers from 25 estates (Fig. 1) were asked to collect the stoats and weasels trapped or shot in the course of their normal predator control operations. Animals were labelled with the date of death and method of capture and were frozen on site before being collected for further examination. Samples were grouped into four seasons according to the date of death: spring, March–May; summer, June–August; autumn, September–November; winter, December–February We examined patterns in the composition of the samples by hierarchical log linear analysis using spss for Windows (Norusis 1994; Andersen 1997). The interactions in a three-way sex × season × method of capture model were examined for stoats and weasels separately.

Figure 1.

Locations of the 25 game estates collecting stoat and weasel samples.

During post-mortem examination, the reproductive condition of females was noted if they were visibly pregnant, had recently given birth or were lactating. The uteri and ovaries of visibly pregnant females were extracted and the number of uterine swellings or embryos was counted. To determine the approximate age of the animals, skulls from both sexes and the bacula from males were extracted and cleaned using sodium perborate solution (McDonald & Vaughan 1999). In combination with other specimens from the same estates, the animals were classified as < 1 year old or > 1 year old by the date–skull–baculum method described by King (1980b, 1991). They were then aged to the nearest month based on the difference between date of death and a median birth date of 1 April for stoats (King 1991) and 1 June for weasels (King 1980b). In order to estimate the maximum age commonly attained by stoats and weasels in Britain, lower canines were extracted from a subsample of 40 stoats and 38 weasels that were more than 1 year old. The teeth were sectioned and cementum layers counted (Grue & King 1984; King 1991) by Matson's Laboratory (Milltown, MT, USA).

To estimate the survivorship rates of stoat and weasel populations, the approach of Powell & King (1997) was followed. They assumed that the distribution of ages among the trapped animals was the same as in the living population (Caughley 1977; King 1989). Trapping stoats becomes selective when trap spacing exceeds 800 m (King 1980c; King & McMillan 1982). Gamekeepers on the sampled estates set traps much closer than this, so it was assumed that trap spacing was not acting as a selective factor. It was assumed that all cohorts in the collection were sampled with equal intensity and the figures for all cohorts were pooled. Survival and mortality rates were calculated using Caughley's (1977) method 3, which does not require knowledge of the rate of increase and does not assume a stable age structure (Caughley 1977; Krebs 1989). Although this method requires individuals to be marked at birth, this was fulfilled by the assumption that they shared a common birth date. Three-monthly survival rates, the product of which is the annual survival rate, were calculated to reflect the pattern of mortality induced by seasonal variation in culling. Three-month periods started from the median birth dates and so differed between the species.

the population projection model

A comprehensive review of the use of matrix models (Leslie 1945, 1948) is provided by Tuljapurkar & Caswell (1997). Therefore, only an outline of the principles involved and the methods used is provided here. The models only considered females and so assumed that real sex ratios in the field were even. Changes in the size and structure of an age-structured population N between times t and t + 1 can be expressed as:

Nt +1  =  ANt

where A is a population projection matrix and N is a vector describing the age-structured population.


where for age class i, Fi is the productivity of independent female young, Pi is the survival and ni is the size of the age class. Note that this is an extended form of the Leslie matrix because there is a non-zero value at P3 that assumes that survival is similar for all age classes where i > 2. An illustration of age-structured populations of stoats and weasels is given in Fig. 2. Repeatedly multiplying N by A results in a series of vectors that differ by a scalar factor equivalent to the population growth rate λ, where λ=er and r is the per capita rate of increase. Successively multiplying N by A also results in a stable age-structure proportional to the vector w and in a reproductive value vector v that represents the contribution of each age class to population size (Tuljapurkar & Caswell 1997). The sensitivity of λ to any element of A can be calculated analytically by:

Figure 2.

Illustration of the basic life cycle of stoats and weasels corresponding to the matrix model. F = age-specific productivity. P = age-specific survival rate. Note that for weasels F2 and F3 are augmented by the productivity of the young weasels born in the first litter of the year.


where sij is the sensitivity of λ to the cell aij, vi is the ith element of v and wj is the jth element of w. The sensitivity of λ to changes in any vital rate that underlies a cell, or multiple cells, can be calculated in a similar way. The proportional sensitivity, or elasticity, of each cell, i.e. the proportional change in λ caused by proportional change in the cell, can be calculated by:


where eij is the elasticity of λ to the cell aij. The elasticity has practical significance to a conservation or management project by showing where treatments might be applied in order to effect the largest changes on λ. The elasticity of the vital rates underlying any cell can also be calculated but these do not sum to one and so cannot be regarded as proportional contributions to λ. Clear examples of this modelling approach are provided by Crouse, Crowder & Caswell (1987), Lande (1988), Brault & Caswell (1993) and partly by Smith & Trout (1994).

For this preliminary analysis, there was no demographic stochasticity, rates were not density dependent and model populations were closed. Because a deterministic approach was adopted, projected growth rates may be somewhat higher than in a stochastic model (Nations & Boyce 1997). Powell & King (1997) demonstrated a density-dependent reduction of juvenile survival when stoat population density increased during seed-fall years in New Zealand beech forest. However, density dependence was not incorporated into this model because of the lack of information about varying rates of recruitment in Britain. Probably the most significant limitation was that the model populations were closed, but this was supportable in view of the main aim of determining growth rates and the sensitivity of growth rate to vital rates.

In each model, there were three age classes and it was assumed that when animals survived beyond the first year of life, subsequent survival and fertility rates were constant, i.e. P2 = P3 and F2 = F3. Because no difference was observed in the survival rates of male and female weasels, these were combined. Because of differences in the seasonal pattern of survival of male and female stoats, only the survival rates for female stoats were used. In order to estimate the annual survival rates for animals more than 1 year old, it was assumed that the maximum age observed from cementum analysis represented the upper 95% limit of the age distribution of individuals surviving the first year of life. So:


where i > 1 and A is the maximum age observed by cementum analysis. The maximum age observed for a stoat in this sample was 4 years, so P2 and P3 were set to 0·37. No weasels older than 2 years were observed in this sample, so P2 and P3 were set to 0·05. The age-specific annual survival rate Pi is the product of the survival rates Pij for the ith age class in each of the four 3-month periods j, starting from the median birth date. So, for example, for first year stoats:

P1  =  P11  *  P12  *  P13  *  P14

where P11 represents the survival rate of new-born stoats during the period 1 April to 30 June. The values of Pij when i= 1 are taken directly from the results (Table 6), although this may overestimate survival. When i > 1, Pij is calculated numerically from dx assuming that the numbers dying are distributed among subsequent age classes, with constant annual survival.

Table 6.  Vital rates used in the projection matrix model for stoat and weasel populations
Seasonal survivalP110·860·88
Probability of a late litterL1·0

Productivity is controlled by a range of factors (King 1989). In stoats, fecundity is relatively constant and does not vary with age but with food supplies in the year during which mating takes place, prior to delayed implantation (King 1981, 1983a; Powell & King 1997). There is, however, no relationship between fecundity and productivity the following year (King 1981). In terms of productivity, the main response to food supplies takes place in the year during which implantation and births take place, in the form of highly variable levels of intrauterine mortality (King 1981, 1989). In weasels, fecundity and fertility are set in the same year, again in response to food supplies (King 1989). Counts of embryos in the uterus take into account intrauterine mortality of blastocysts, but not subsequent resorption of embryos, so they incorporate part, if not all, of the difference between fecundity and fertility. Fecundity was not measured in this collection and so embryo counts were taken as the measure of fertility for both stoats and weasels. Pre-independence mortality was not measured directly, but is at least partly affected by the survival of the mother during the period of juvenile dependence. This is particularly relevant in mustelid populations where breeding females are culled more readily after giving birth than before (see below). There are no data available to determine rates of natural pre-independence mortality, and so adult survival rates in the period during which juveniles are dependent on their mothers have been used as an index. Killing behaviour is not fully developed until approximately 12 weeks in stoats (King 1983b) and 7 weeks in weasels (Heidt, Petersen & Kirkland 1968; Sheffield & King 1994) and it was assumed that survival of juveniles up to these ages is directly related to maternal survival. Stoat productivity rates take into account the survival rates P21 and P31 over the period 1 April to 30 June. In weasels, the first litter is produced towards the end of the last 3-month period of the first year of life, from 1 March to 31 May. Therefore, the productivity of adults in the first litter of the year has already taken account of postpartum adult mortality. In contrast, the component of total productivity arising from second litters born to adults and the first litters of weasels in their first year of life takes into account the survival rates P21 and P11 in the period 1 June to 31 August, scaled to a 7-week period. Initially, the probability L of there being a late breeding phase in weasels was taken to be 1, and the sensitivity of the population growth rate to this measure was examined. Fi is the productivity of the age class i and Ferti is the fertility of age class i expressed as the number of female embryos, which is half of the total number of embryos, assuming an equal sex ratio at birth. For both species, F1 is set to zero to avoid confusion with overlapping birth pulses. To summarize, for stoats:

F2  =  F3  =  Fert2  *  P21

and for weasels:


Fertility rates for the model are taken from the embryo counts given in Table 5, which summarizes the findings of a number of studies of stoat and weasel populations in Europe and New Zealand, and which include the counts from the current collection. A summary of all the vital rates used in the model is given in Table 6. Matrix calculations were conducted using the program ulm (Legendre & Clobert 1995).

Table 5.  Summary of embryo counts from stoat and weasel pregnancies
SpeciesMean number of embryosRangeNumber of pregnancies observedLocation of studyReference
  • *

    These records are of births to captive animals.

  • These records are cited in King (1971, 1975 ).

Stoats9·07–10 8Great BritainThis study
 8.86–1313New ZealandKing & Moody (1982)
 9·0 1Great BritainEast & Lockie (1965 ) *
 7·56–9 2IrelandFairley (1971 )
 9·06–1312Great BritainDeanesly (1935 )
Mean8·8 34  
Weasels6·24–9 6Great BritainThis study
 5·421Great BritainTapper (1979 )
 5·64–712Great BritainKing (1980a )
 4·72–7 3Great BritainEast & Lockie (1965 ) *
 6·0 1Great BritainEast & Lockie (1964 ) *
 6·46–7 5Great BritainDeanesly (1944 )
 6·312Great BritainD. Stephen, unpublished data*†
 5·310GermanyF. Frank, unpublished data*†
 4·5 4New ZealandHartman (1964 ) *
 5·24–814PolandJedrzejewska (1987 )
Mean5·6 88  


Between October 1995 and November 1997, 822 stoats and 458 weasels were collected. Sex ratios were biased towards males for stoats (1·69 : 1) and weasels (4·95 : 1). Animals were culled throughout the year (Table 1), although spring culls accounted for 41% of stoats and 43% of weasels. The importance of trapping and shooting differed between species (Table 2). For stoats, 71% were trapped and 26% were shot, whereas for weasels 94% were trapped and 5% were shot (χ2 test with Yates’ correction; χ2 = 325·0, d.f. = 1, P < 0·001). Only 2·4% of stoats and 0·7% of weasels were killed by other means, including road kills, cats, dogs and unexplained deaths.

Table 1.  Composition of the sample of stoats and weasels by sex and season. Percentages of the annual sample for which date of death was known are shown in parentheses
SpeciesSexSpringSummerAutumnWinterDate knownDate unknownTotal samples
StoatMale232 (52) 88 (20) 63 (14) 67 (15)450 64514
 Female 61 (23)113 (42) 47 (18) 47 (18)268 37305
 Unknown   2    2  1  3
 All293 (41)203 (28)110 (15)114 (16)720102822
WeaselMale144 (41) 79 (23) 86 (25) 39 (11)348 33381
 Female 37 (53)  7 (10) 18 (26)  8 (11) 70  7 77
 All181 (43) 86 (21)104 (25) 47 (11)418 40458
Table 2.  Composition of the sample of stoats and weasels by sex and method of capture. Percentages of the annual sample for which the method of capture was known are shown in parentheses
SpeciesSexTrappedShotOtherMethod knownMethod unknownTotal
StoatMale348 (76) 98 (22) 9 (2)455 59514
 Female164 (62) 92 (35) 7 (3)263 42305
 Unknown  1  1  2  1  3
 All513 (71)190 (26)17 (2)720102822
WeaselMale333 (94)  18 (5) 2 (1)353 28381
 Female 64 (94)   3 (4) 1 (1) 68  9 77
 All397 (94)  21 (5) 3 (1)421 37458

For stoats, the prevalence of the sexes varied with season (log linear analysis, sex × season, likelihood-ratio (LR) χ2 = 56·3, d.f. = 3, P < 0·001). Female stoats were taken less frequently than expected in spring and more frequently in summer. Susceptibility to method of culling varied between the sexes (sex × method, LR χ2 = 5·8, d.f. = 1, P < 0·05). A greater proportion of female stoats was shot than males, but the sex ratio of shot stoats did not differ from even (G-test with Yates’ correction, G= 0·13, d.f. = 1, P > 0·05). The importance of culling method for stoats varied between seasons (season × method, LR χ2 = 25·6, d.f. = 3, P < 0·001). Shooting accounted for a greater proportion of the sample of stoats than expected in summer. The removal of the interaction, sex × season × method did not detract from the fit of the log linear model for stoats. The composition of the weasel sample was significantly, but independently, affected by the terms sex (LR χ2 = 196·2, d.f. = 1, P < 0·001), season (LR χ2 = 77·6, d.f. = 3, P < 0·001) and method (LR χ2 = 413·9, d.f. = 1, P < 0·001). More male weasels were caught than females, spring captures were more prevalent than other seasons and trapping was more important than shooting. No interaction terms significantly affected the fit of the model describing the composition of the weasel sample.

Only eight of 305 (2·6%) female stoats and six of 77 (7·8%) female weasels were visibly pregnant (Table 3). The mean number of implanted embryos observed was 9·0 in stoats and 6·2 in weasels. In stoats, active pregnancies were observed between 9 March and 4 April, postpartum uteri between 14 March and 19 April and lactating females between 16 March and 9 July. In weasels, pregnancies were observed between 25 April and 5 October, postpartum uteri between 5 April and 13 October and lactating females between 2 May and 27 August. Of the six weasel pregnancies, five were in females born the previous calendar year. One female weasel that was pregnant in October was breeding in the year of its birth.

Table 3.  Pregnancies observed in stoats and weasels
SpeciesRegionNumber of embryosDateYear
StoatSouth England 99 March1997
 South England 916 March1996
 South England 916 March1997
 North England 917 March1997
 South England 74 April1996
 South England10Unknown1996
 South England10Unknown1996
 South England 9Unknown1997
 Mean 9·0  
WeaselSouth England 925 April1996
 South England 426 April1997
 South England 630 May1996
 Scotland 612 May1997
 South England 55 October1997
 South England 7Unknown1997
 Mean 6·2  

Age at death was estimated for 717 (87%) stoats (449 males and 268 females) and 403 (88%) weasels (344 males and 59 females). The remaining samples were not aged because date of death was unknown or ageing characteristics were unclear. The maximum age attained in the 40 stoats and 38 weasels for which cementum analysis was conducted was 4 and 2 years, respectively. The median ages for male and female stoats were 11·6 months (95% confidence interval 11·3–11·8 months) and 8·0 months (7·4–8·9 months), respectively. For weasels the median ages of males and females were 9·3 months (8·6–9·7 months) and 9·2 months (6·3–10·4 months).

The finite rate of survival px in the first year of life for male stoats was 0·41, for female stoats was 0·26, for male weasels was 0·15 and for female weasels was 0·03. Patterns of survival and mortality varied between 3-month periods (Table 4). There was a significant difference between the sexes in the cumulative rate of survival lx of stoats (Kolmogorov–Smirnov test, D268,449 = 0·24, P < 0·001) but not weasels (D59,344 = 0·11, NS). Finite rates of mortality qx are highest around March and April, the time at which gamekeepers make the greatest trapping effort (McDonald & Harris 1999).

Table 4.  Approximate seasonal life tables for male and female stoats and weasels on game estates. Survival between birth and 3 months of age is overestimated, because samples from gamekeepers do not include animals too young to leave the nest. Survival of animals more than 1 year old is underestimated, because several year classes are compressed into a single age class. Seasons are based on 3-month periods starting from the median birth date of 1 April for stoats and 1 June for weasels. nx= number alive at start of age class x ; lx = proportion surviving at start of age class x ; dx= number dying between age class x and x +  1; qx = finite rate of mortality; px = finite rate of survival
SpeciesSexSeasonAge x (months)nxdxlxqxpx
StoatsMalesApr.–Jun. 3449441·000·100·90
  Jul.–Sep. 6405540·900·130·87
  Oct.–Dec. 9351460·780·130·87
 FemalesApr.–Jun. 3268381·000·140·86
  Jul.–Sep. 6230620·860·270·73
  Oct.–Dec. 9168510·630·300·70
WeaselsMalesJun.–Aug. 3344531·000·150·85
  Sep.–Nov. 6291710·850·240·76
  Dec.–Feb. 9220360·640·160·84
 FemalesJun.–Aug. 35951·000·080·92
  Sep.–Nov. 654160·920·300·70
  Dec.–Feb. 93860·640·160·84

population models

Matrix models yielded population growth rates of λs = 0·95 and λw = 1·35 corresponding to rates of increase of rs = –0·05 and rw = 0·30 (Table 7). The age structure vectors w of stoats and weasels (Table 7) are similar to the actual age structures observed in this and other collections of trapped samples (King 1989) and these vectors were adopted for further analyses of the model. The matrix element that contributed most to population growth rate in both species was the survival of the first year class P1. P1 was less important to λs than it was to λw. The importance of the productivity of second-year weasels F2 was almost equal to the contribution of P1, while for stoats the importance of productivity was even among second-year and older animals. In stoats, the rate that made the greatest contribution to λs was P21, the survival of second-year animals in the period April to June, when their first litters are born (Table 8). In weasels, the greatest contributions to λw were made by P14, the survival of first-year weasels from March to May when the first litters are born, and P11, the survival of weasels born in the first litter of the year from June to August (Table 8).

Table 7.  Summary of the population projection matrix analyses. A is the population projection matrix, λ is the population growth rate, w is the stable age distribution vectors and v is the reproductive value vector for stoats and weasels. S and E are the sensitivity and elasticity matrices derived from the population projection matrices. These matrices show the sensitivity and elasticity of λ s and λ w to changes in the cells of the population projection matrices As and Aw . Therefore, only the sensitivities for non-zero matrix elements are shown
inline imageinline image
λs = 0·953λw = 1·350
inline imageinline image
vs  = [12·2 43·9 43·9] vw  = [5·1 47·4 47·4]
inline imageinline image
inline imageinline image
Table 8.  The sensitivity and elasticity of population growth rate λ to changes in vital rates. Pij is the survival rate of age class i in the 3-month period j . Ferti is the fertility rate of age class i and L is the probability of weasels having a late litter
Parameter typeParameterStoatsWeasels
Probability of a late litterL0·510·38

To evaluate the effect of large perturbations in sensitive parameters, the rates P21 for stoats and P11 andP14 for weasels were varied and new values for λs and λw were calculated. The probability of weasels having a late litter L varies with food supply (King 1980a) and is also the factor about which the fewest data are available. In the initial model L was set to 1·0, so to gauge the sensitivity of the model to variation or inaccuracy in L, the perturbation analysis was also conducted for L. Only slight changes to spring survival in second-year stoats P21 were required in order to bring about an increasing population (Fig. 3). P21 only had to increase from 0·48 to 0·54 in order for the population to stabilize, i.e. λs = 1·0. In order to effect a decline in breeding weasel populations, i.e. λw < 1·0, the survival rate of new-born weasels in the first 3 months of life, P11 would have to be reduced from 0·88 to less than 0·56 (Fig. 4). If the effect of trapping was removed, and survival in this period was 1·00, λw would increase slightly to 1·47. Small changes in P14 resulted in proportionally larger changes in λw than changes in P11. A reduction of P14 from 0·27 to less than 0·15 was required in order to produce a declining population, whereas an increase of survival in this period to 1·00 resulted in λw of 2·49 (Fig. 4). Reducing the probability of weasels having a late litter L, i.e. a second litter for adults and a first litter for young weasels, to less than 1·0 did not have a proportionally great effect on population growth rate (Fig. 5). A reduction of L from 1·0 to less than 0·40 was required in order to bring about a decline in trapped populations of weasels.

Figure 3.

The effects of changes in the seasonal survival rate P21 on stoat population growth rate. The horizontal dashed line marks the baseline run of the model (λ = 0·953).

Figure 4.

The effects of changes in the seasonal survival rates P11 (dotted line) and P14 (solid line) on weasel population growth rate. The horizontal dashed line marks the baseline run of the model (λ = 1·350).

Figure 5.

The effects of changes in the probability of a late breeding phase in weasels L on weasel population growth rate. The horizontal dashed line marks the baseline run of the model (λ = 1·350).


This and other collections of culled mustelids are biased by seasonal changes in culling effort and variation between the species and sexes and among seasons in susceptibility to different control methods. Such biases in samples of Mustela have been accounted for as an effect of trap density (King 1975, 1980a,c). Buskirk & Lindstedt (1980) suggested that behavioural differences between the sexes of several mustelid species could create a biased sex ratio independently of trap spacing. This may be the case for stoats, which show a sex bias in susceptibility to trapping but not shooting, and varying susceptibility to both methods with season. The sex ratio of this sample of weasels was more skewed towards males than comparably large collections (Deanesly 1935; King 1975, 1980c). A pronounced bias towards males may be a symptom of a declining population (Lockie 1966; Erlinge 1974; Sheffield & King 1994). but when other studies are considered there is no trend in the sex ratio of British collections of stoats or weasels with time (Table 9). Even if there had been a trend, this may have been the result of a general tendency for trapping effort to decline in the period 1960–97 and for effort to be concentrated in periods that result in few female captures (McDonald & Harris 1999). Records of changes in sex ratio can only be a useful factor in gauging the status of stoat and weasel populations as long as effort is recorded (McDonald & Harris 1999).

Table 9.  The sex ratio (male : female) of samples of stoats and weasels collected by gamekeepers in Great Britain
SpeciesMalesFemalesSex ratioYears of collectionReference
  • *

    This sample includes those described by King (1975, 1977 ).

  • This record is based on a questionnaire of gamekeepers.

Weasel381774·951995–97This study
 70125·831971–72Moors (1975 )
 3391162·921968–72King (1980a ) *
 65183·611966Walker (1972 )
 78136·001960–63Day (1963 )
 3261262·591931–35Deanesly (1944 )
Stoat5143051·691995–97This study
 85302·831960–63Day (1963 )
 3922481·581930–34Deanesly (1935 )
 390523061·691930–34Flintoff (1935 )

The embryo counts and age structures observed in this collection were comparable to those recorded in other studies of stoats (Fog 1969; van Soest & van Bree 1970; Jensen 1978; King & McMillan 1982; Erlinge 1983; Murphy & Bradfield 1992; Murphy & Dowding 1995; King et al. 1996) and weasels (Fog 1969; Jensen 1978; King 1980a) in temperate zones. However, as Caughley (1977) points out, differences in age structure are the least sensitive indicator of differences in population status, as growth rates may vary widely while age structure remains constant. Because of the difficulty of capturing pregnant female weasels and the difficulty of determining their age and exact birth dates it is not possible to determine the frequency of late litters in weasels accurately. Jefferies & Pendlebury (1968) suggested that two litters in weasels were the norm, but there is no evidence of this from the game bag records they used (King 1980a). Stenseth (1985) assumed that weasels had two litters in a model of stoat and weasel life history strategies; again this is not supported by field observations. Without information on food supply (King 1980a, 1989) and a much greater sample of pregnant females, no further interpretation of the probability of late litters is possible.

In contrast to Powell & King (1997), the exact age of most of the stoats or weasels was not determined by cementum analysis and this should be a priority in improving measurement of adult survival rates in Britain. This would also permit the distinction of the multiple cohorts that make up the total sample, and would allow an assessment of the variation between cohorts in survival rates (Powell & King 1997). The seasonal survival rates on game estates were not constant. The season of highest mortality in these samples is spring, because of the greater effort that gamekeepers make in this season. Spring may also be the season in which natural mortality is highest, given comparatively low food availability and higher energetic demands at this time of year (King 1980a). If this is the case, then culling may not result in adult mortality that is additive to natural rates (King 1980a, 1989). However, animals that are near to starvation are unlikely to invest substantial amounts of energy into reproduction (King 1989) and thus culling in spring is likely to have an effect that is additive to natural mortality by reducing productivity.

To our knowledge, these are the first estimates of growth rates of model stoat and weasel populations on game estates. They should be considered in the light of a scarcity of data on several aspects of stoat and weasel demography, particularly those for which demographic sensitivity is highest. Because the sensitivity of a demographic parameter is a quantitative index of the effect that changes in the parameter value will have on growth rate, it also reflects the relative importance of accuracy in measuring that rate. In these models, survival rates are derived from trapped samples and sources of bias, for example differences between the sexes, are considered explicitly. Mean measures of fertility are inherently biased towards years in which population fertility rates are highest. Therefore, model growth rates may be more applicable to years in which food supply is greatest. While the probability of second litters is a significant parameter in weasel populations, it is also the factor about which the least is known. Methods of determining the age and date of birth of juvenile weasels would be useful in evaluating the accuracy of the probability rates adopted in the model. In addition to improving parameter estimates, a significant development of this modelling exercise would be to take immigration and emigration into account. However, such an exercise would require much better information about stoat and weasel spatial behaviour, which is poorly known in Britain, and could only be provided by a labour-intensive field study (Byrom 2002).

Population projection models suggested that culling by keepers does not result in a long-term decline in weasel populations. Even though a large proportion of the weasel population may be culled over the year and even with low survival between March and May, those that are left can reproduce sufficiently quickly that the population still increases. The importance of high productivity to tolerance of trapping was highlighted by King & Moors (1979), King (1980a) and Tapper, Green & Rands (1982) and is confirmed quantitatively by this study. In contrast to weasels, gamekeepers’ culling effort produced a slight decline in model stoat populations. Because of the high sensitivity of growth rate to spring survival, a small degree of error in estimating these rates could result in a growth rate estimate that is not significantly different from 1·0 (Lande 1988) or, conversely, is lower than calculated. Equally, changes in the intensity of trapping in the spring when juveniles are dependent on their mothers will have a substantial effect on population growth. It was assumed that juveniles would not survive at all before they could kill prey for themselves and that they would survive normally thereafter. In reality, the probability of independent survival after this period will increase gradually rather than there being a distinct cut-off point, and this should be considered in further detail by observation of free-living litters.

Actual stoat and weasel populations will rarely exhibit the long-term growth rates predicted by these models. The main reason for this is not likely to be inaccuracy in estimating the model parameters, but the fact that real populations are not closed and immigrants form a large part of the total number of animals caught. Both stoats and weasels, particularly males, are known to disperse over large distances (Erlinge 1977; King & McMillan 1982; Robitaille & Raymond 1995) and dispersing individuals can rapidly replace trapped or shot animals. In the long term, trapping records are augmented by high numbers of immigrants and dispersers, particularly in years with high food availability. While locally breeding weasel populations will increase despite control efforts, stoat populations are likely to be dependent on the immigration of females in order to sustain their numbers and prevent extinction. In the field, a distinction should be drawn between locally breeding populations and areas in which stoats are constantly present but not necessarily breeding and which may be sustained by male immigration. This distinction could be made on a large scale if the sex of all animals caught by gamekeepers was recorded (McDonald & Harris 1999). Whereas rates of productivity, which are affected by the survival of adult females in spring, have a significant role in making stoat populations ‘resistant’ to trapping (King & Moors 1979; King & McMillan 1982; Tapper et al. 1982) this study underlines the likely importance of immigration (King & McMillan 1982).

In terms of game bird protection, females are clearly critical for maintaining population growth rate in both species, and so population control will be realized most effectively by focusing on periods or methods that most efficiently remove females from the population. In the case of adult female stoats, which are hard to trap in the spring while they are pregnant, shooting in late winter and early spring will contribute disproportionately to the efficacy of control measures. Because females are hard to trap in the most sensitive period, trapping females in the winter, during delayed implantation but prior to active pregnancy, will be important in reducing population productivity in the spring. Trapping in early summer also appears effective in reducing stoat population growth, because the survival of dependent juveniles is reduced during this period. Control measures in late summer and autumn, following the independence of both young stoats and game birds, appear to be of comparatively little utility in reducing stoat populations during the game bird nesting period. For weasels, shooting contributed little to culled samples, but trapping throughout the spring and summer, focused on the early period between March and May when first litters are born, assisted in reducing population growth. None the less, exponential growth was maintained in the face of concerted culling effort during these periods.

In terms of the apparent decline in mustelid populations in Great Britain, we echo the conclusion of earlier investigators (King & Moors 1979; King 1980a; Tapper et al. 1982) that the culling regimes currently employed by gamekeepers do not have the capacity to cause a long-term decline in weasel populations, under average conditions of survival and fertility. Furthermore, culling is not likely to bring about long-term population declines in stoat populations, except where immigration is limited or reduced, perhaps by habitat degeneration. As a result, protection of stoats and weasels from trapping and shooting by gamekeepers does not appear justified on conservation grounds alone. Given that the only apparent circumstance under which stoats might decline if culled is when immigration rates are low, habitat management that maintains recruitment on and off game estates and facilitates successful dispersal and immigration could be a more effective target for conservation measures than legal protection.


This work was supported by a scholarship from The Wingate Foundation to Robbie McDonald. We are grateful to The Dulverton Trust for financial support, to the many gamekeepers who assisted by collecting stoats and weasels, to Charlotte Webbon, Sue Holwell and Ellen Green for assisting with sample preparation, to Gary Matson of Matson's Laboratory for conducting cementum analyses, and to Nigel Barlow, Huw Griffiths, Andrew Hoodless, Carolyn King, Adrian Seymour, Graham Smith, Stephen Tapper and an anonymous referee for advice and comments. The map was produced using dmap.