Resource partitioning among British and Irish mustelids

Authors

  • Robbie A. McDonald

    Corresponding author
    1. The Game Conservancy Trust, The Gillett, Forest-in-Teesdale, Barnard Castle DL12 0HA, UK
      Robbie A. McDonald, Tel.: 01833 622208. Fax: 01833 622343. E-mail: rmcdonald@gct.org.uk
    Search for more papers by this author

Robbie A. McDonald, Tel.: 01833 622208. Fax: 01833 622343. E-mail: rmcdonald@gct.org.uk

summary

  • 1 Seven species of mustelid carnivore live in the British Isles: weasel, stoat, mink, polecat, pine marten, badger and otter. Recent studies have hypothesized that coexistence of these species is facilitated by partitioning of resources according to prey size, particularly that of mammalian prey. This hypothesis has been supported by evidence of character displacement derived from even size ratios in skull length and canine diameter.
  • 2 To test whether this hypothesis is supported by empirical data, 98 studies of the diet of mustelids living in Great Britain and Ireland were analysed. Two main predictions were tested; that larger males ate larger prey than females and that larger species ate larger prey than smaller species.
  • 3 Male mustelids ate larger prey than females but there was no relationship between predator size and prey size, either for all species or when largely vermivorous badgers and piscivorous otters were excluded. There was no difference in dietary niche breadth between the sexes. Dietary niche breadth increased with body size in the assemblage excluding otters and badgers. The dietary niches of mustelids were partitioned along several axes, none of which was clearly related to prey size.
  • 4 The dietary niches of the five species living in Ireland (stoat, mink, pine marten, otter and badger) were more similar to one another in Ireland than in Great Britain and there was no difference in niche breadth between Irish and British mustelids. There was no evidence of competitive release in the diets of Irish mustelids.
  • 5 Resource partitioning according to prey size is apparent between the sexes within species, and this is probably correlated to sexual selection for size dimorphism. Empirical data do not, however, support a hypothesis of resource partitioning according to prey size between species. Interspecific aggression provides an alternative hypothesis explaining character displacement among mustelids.

Introduction

Competition for limited food resources is a major driving force of evolution within assemblages of related and morphologically similar species (MacArthur & Levins 1967). Analyses of the evolutionary effects of competition have often focused on morphological differences between competing species. Such differences are often accentuated in the region of sympatry, a phenomenon termed character displacement (Brown & Wilson 1956). This is related to Hutchinson’s (1959) theory of limiting similarity, a corollary of which was that for species to coexist there must be minimum ratios between them in the size of their trophic apparatus. The commonplace detection of even size ratios has been widely taken as evidence of the evolutionary importance of competition. None the less, the unqualified interpretation of size ratios has been severely criticized mainly because of a lack of statistical rigour (Strong, Szyka & Simberloff 1979; Roth 1981; Simberloff & Boecklen 1981; Ranta, Laurila & Elmberg 1994). Recently, the development of statistically robust tests (Tonkyn & Cole 1986) has led to something of a revival in size ratio analysis in community ecology (Dayan et al. 1989, 1990; 1992; Dayan & Simberloff 1994, 1996, 1998; Jones 1997).

Mammalian carnivores, particularly mustelids living in the British Isles, have had a significant role in the development of character displacement theory (Dayan & Simberloff 1996, 1998). Indeed, Hutchinson (1959) illustrated his case with a description of differences in the size of stoats (for species names and authorities see Table 1) between Great Britain and Ireland. In Ireland, stoats were reputedly smaller than in Britain and Hutchinson (1959) believed this to be a consequence of competitive release arising from the absence from Ireland of the smaller weasel. Subsequent analyses have shown Hutchinson’s argument to be based on incomplete data (King & Moors 1979; Fairley 1981; King 1989, 1991). Nevertheless, his work has become a classic example of competitive release (Williamson 1972) and has been a cornerstone of several other investigations of mustelid assemblages. In common with Hutchinson (1959), several authors have forwarded the proposition that coexistence of several species of mustelid is mediated principally by selection of differently sized prey (Rosenzweig 1966; McNab 1971; Simms 1979; Dayan et al. 1989; Dayan & Simberloff 1994). Holmes (1987) and Holmes & Powell (1994) countered propositions of resource partitioning in the evolution of sexual dimorphism by finding that the carnassials of north American mustelids were less dimorphic than were the rest of their bodies. Meanwhile, other authors studying a range of morphological characters have not found evidence of character displacement among mustelids. Instead, they have explained size variation and sexual dimorphism in this group with hypotheses based on biogeography and sexual selection (Ralls & Harvey 1985), interspecific aggression (Rosenzweig 1966; Erlinge & Sandell 1988) or from combinations of these factors with stochastic variation in habitat and prey availability (Rosenzweig 1968; King & Moors 1979; Powell & Zielinski 1983; King 1989; Powell & King 1997). In this paper, I test whether empirical data on the diets of mustelids living in the British Isles support a hypothesis of resource partitioning according to prey size, established in particular by the work of Dayan & Simberloff (1994).

Table 1.  The mass of mammalian prey species and the mass, skull length and canine diameter of mustelids
Common nameSpeciesMass/gSkull length/mmCanine diameter/mmSource
  1. Sources: 1Fairley 1972; 2Fairley 1980; 3Fairley 1981; 4Corbet & Harris 1991; 5Lynch & O’Sullivan 1993; 6Dayan & Simberloff 1994, 7Lynch et al. 1996; 8Sadlier 1999; 9A. Kitchener, unpublished data; 10R. McDonald, unpublished data. *This figure takes into account an unknown proportion of smaller juveniles. †This figure is based on the weights of skinned animals provided by Fairley (1972) and recalculated according to the formula provided by Jefferies (1986).

Harvest mouseMicromys minutus Pallas     6  4
ShrewsSorex pygmaeus L.    10  4
 S. araneus L.    
Water shrewNeomys fodens Pennant    15  4
House mouseMus domesticus L.    17  4
Young small mammal     18  4
Wood mouseApodemus sylvaticus L.    19  4
Hazel dormouseMuscardinus avellanarius L.    20  4
Unidentified small rodent     20  4
Bank voleClethrionomys glareolus Schreber    24  4
Unidentified insectivore     30  4
Unidentified vole     30  4
Field voleMicrotus agrestis L.    35  4
MolesTalpa europaea L.    98  4
Water voleArvicola terrestris L.   290  4
Common ratRattus norvegicus Berkenhout   500  4
SquirrelsSciurus carolinensis Gmelin   550  4
 S. vulgaris L.    
Lagomorphs (rabbits and hares)Oryctolagus cuniculus L. *800  4
 Lepus europaeus Pallas    
 L. timidus L.    
WeaselMustela nivalis L.    
 Great Britain     97 36·61·616, 10
StoatMustela erminea L.    
 Great Britain    305 46·92·256, 10
 Ireland    208 44·62·073, 6
MinkMustela vison Schreber    
 Great Britain    885 63·13·334, 6
 Ireland  1 002 63·73·502, 6
PolecatMustela putorius L.    
 Great Britain  1 113 63·83·444, 6
Pine martenMartes martes L.    
 Great Britain  1 600 82·14·584, 6
 Ireland  1 500 81·94·274, 6
OtterLutra lutra L.    
 Great Britain  8 500111·55·914, 7, 9
 Ireland 7 340113·95·911, 5, 9
BadgerMeles meles L.    
 Great Britain 10 850126·77·474, 6
 Ireland  9 900126·27·564, 6, 8

The British Isles consists of two main islands. The larger, Great Britain, includes England, Scotland and Wales. Ireland includes Northern Ireland and the Republic of Ireland. Seven species of mustelid live in the British Isles. In order of increasing size (Table 1): weasel, stoat, polecat, pine marten, otter and badger are native to Great Britain. Feral American minks are larger than stoats and slightly smaller than polecats but are not native and are a recent addition to the assemblage, following escapes and releases from fur farms (Dunstone 1993). Of these seven species, only five (stoat, mink, pine marten, otter and badger) live in Ireland. It should be noted that throughout this paper, ‘weasel’ is used in the British context to mean the species Mustela nivalis rather than the North American context where ‘weasel’ is applied to the three species M. nivalis, M. erminea and M. frenata.

Dayan & Simberloff (1994) defined the mustelids of the British Isles as ‘a guild on the basis of limb morphology, which reflects foraging behaviour’ (Simberloff & Dayan 1991; Dayan et al. 1992). Using museum specimens, they measured skull size (condylobasal length) and the size of the apparatus mustelids used to kill their mammalian prey (the maximum diameter of the upper canine) (Dayan et al. 1989). They omitted otters since they held there was virtually no ecological overlap or potential for competition with the native terrestrial mustelids. They conducted their analyses both with and without badgers on the grounds that their largely vermivorous diet may have weakened selective pressure on canine size. All mustelids exhibit pronounced sexual size dimorphism (Moors 1980) and so Dayan & Simberloff (1994) considered males and females to be functionally distinct ‘morphospecies’ and accorded similar importance to intraspecific and interspecific competition in structuring the assemblage. In Great Britain, they found equal size ratios in skull lengths and canine diameters and concluded that this was evidence of community-wide character displacement. In Ireland, they found that ratios for canine size were even but those for skull lengths were not. They identified differences in the sizes of British and Irish mustelids and this led to their confirmation of Hutchinson’s (1959) suggestion that ecological release had taken place in Ireland.

The morphological analyses of Dayan & Simberloff (1994) strongly supported resource partitioning according to prey size, particularly mammalian prey size, among mustelids in the British Isles. What do empirical studies of the diets of these animals contribute to their argument? If displacement of a trophic character is to be regarded as an evolutionary symptom of competition, evidence for size-related differences in predator diets must be provided. Dayan & Simberloff (1994, 1998) recognized this need. They predicted that males of any species would take larger mammalian prey than females, and that the average prey size for each species would be smaller than that for larger species. These two predictions form the basis of the investigations described in this paper. A relationship between carnivore size and prey size is well established (Gittleman 1985), but this should be examined closely in the specific assemblage under investigation. The data on mustelid diet provided by Dayan & Simberloff (1994) were limited and in several cases did not refer to British or Irish studies (Brugge 1977; Storch, Lindström & De Jounge 1990). This is important because the body sizes of mustelids living in continental Europe and the variety and size of available prey differ markedly from Great Britain and Ireland (Moors 1980; Erlinge 1986; King 1989). Therefore, in this case, as in other character displacement studies (Ralls & Harvey 1985), only local dietary information can be used as supporting evidence. Fortunately, the diet of mustelids in the British Isles is well described and several major studies have been published since Dayan & Simberloff (1994).

In this paper, empirical data on the diets of mustelids living in Great Britain and Ireland have been synthesized to test whether they support a hypothesis of resource partitioning according to the size of mammalian prey. In considering mustelid diet in Britain and Ireland separately throughout, the hypothesis of competitive release among Irish mustelids compared to British mustelids has also been tested. An assessment has been made of whether the seven species living in the British Isles can validly be described, not just as a taxonomic assemblage (Jaksic 1981), but as a guild when defined as an ‘arena of intense interspecific competition’ (Pianka 1980). In effect, this is a post hoc assessment of Dayan & Simberloff’s (1994) omission of otters from all and badgers from some of their analyses. This is in accord with the counsel of Simberloff & Dayan (1991) that in defining a guild, the exclusion of sympatric related biota should be justified, where the grounds for exclusion are not self-evident and consist of intuition or simple taxonomy.

Materials and methods

Data on the diets of all mustelids living in Great Britain and Ireland were collated from the literature. Several studies reported diets from more than one site and in these cases each site was considered a separate sample. The diets of males and females were recorded separately where possible. The reviewed studies used several techniques to investigate diet, including direct observation and analysis of faeces and gut contents. A range of methods, including frequency of occurrence and estimated weight or volume intake, was then used to describe diet. The raw data do not permit recalculation into a common currency, neither would this be desirable, given differences in foraging ecology within the group. Nevertheless, for smaller mustelids most authors used frequency of occurrence, since guts or faeces generally contained only one item. For larger species, percentage volume was used most often as this took account of the importance of small but common prey, such as invertebrates and fruit. Several studies were not included since they did not provide sufficient detail (Neal 1988), focused on one prey group (Fairley 1984; Kruuk et al. 1993; Carss et al. 1998) or used data published elsewhere (King 1977; Thompson 1978; Pringle 1982).

Two main analyses were conducted. The first examined the relationship between predator size and prey size at a fine scale and followed the approach of Jones (1997). A weighted-mean prey mass was calculated for each sample by multiplying the mean mass of both sexes of each species of mammalian prey (Table 1) by the prevalence of that species as a proportion of all mammalian prey items. Non-mammalian prey was excluded since Dayan & Simberloff (1994) argued that natural selection acted on canine diameter because of its specific role in killing mammalian prey. Carrion and mustelid hairs from grooming were also excluded. For comparison between species, a mean prey mass was calculated and compared to the species’ mean body weight, skull length and canine diameter (Table 1). For comparison between sexes, a paired comparison was made within samples between the mass of prey eaten by males and females. Following Dayan & Simberloff (1994), this analysis excluded otters and was conducted with and without badgers.

For the second analysis of dietary niche relationships at a broader scale, prey items were grouped according to taxonomy, size and requirements for capture (Rosenzweig 1966). Small mammals included shrews, moles, voles, dormice and mice. Medium-sized mammals included rabbits, water voles, rats, squirrels and hares. Large mammals were taken almost entirely as carrion but included sheep Ovis aries L. and deer Capreolus capreolus L. and Cervus elaphus L. Further categories were birds, birds’ eggs, fish, herpetofauna, earthworms, other invertebrates and fruit and vegetables. Levins’s (1968) measure of niche breadth (B) was calculated and standardized across prey categories (BA) (Krebs 1989). To examine resource partitioning among species, a principal component analysis (PCA) was conducted on the prevalence of each food category in the diet of British mustelids (Holmes, Bonney & Pacala 1979). The PCA was conducted on the covariance matrix in order to preserve the relative magnitude of each prey group. The Euclidean distances between the group centroids of each species were calculated as a measure of the differentiation of their dietary niches (Holmes et al. 1979). These distances were used to construct a community dendrogram by hierarchical clustering with the average linkage method (Ryan & Joiner 1994). To test whether the dietary niches of each species were dispersed more widely in Ireland than in Britain, i.e. whether there was evidence of competitive release, PC scores for the Irish samples were calculated by using the loadings from the PCA of the British samples. The Euclidean distances between the group centroids of the Irish samples were then calculated and compared to the British samples. These analyses considered the sexes together and were conducted on the whole assemblage and then repeated omitting otters and badgers. Minitab (Ryan & Joiner 1994) was used for statistical analysis.

Results

Ninety-eight samples of mustelid diet were analysed (Tables 2–5, summarized in Table 6). Mass of mammalian prey was not related to predator size (Fig. 1). For the British assemblage excluding otters and badgers, there was no significant relationship between the mass of mammalian prey and predator mass (r = –0·21, n = 5, P = 0·73), skull length (r = 0·20, n = 5, P = 0·75) or canine diameter (r = –0·18, n = 5, P = 0·75). For the British assemblage excluding otters but including badgers, there was again no significant relationship between the mass of mammalian prey and predator mass (r = 0·30, n = 6, P = 0·56), skull length (r = 0·20, n = 6, P = 0·71) or canine diameter (r = –0·18, n = 6, P = 0·71). For the Irish assemblage excluding otters, the mean prey mass eaten by badgers was greater than the other three species. The mean mass of prey eaten by Irish martens was, however, less than that eaten by smaller Irish stoats and minks. The mass of prey eaten by Irish stoats was less than British stoats, but greater for Irish minks and martens than for British minks and martens (Fig. 1).

Table 2.  The diets of (a) weasels (b) stoats and (c) polecats in Great Britain and Ireland
SourceLocationYearTimenTypeMethodMassSMMMLMUMBEFHWIFVOBA
  1. 1 This study includes an unknown number of samples from Northern Ireland. NS, not stated in original source, or not in an appropriate format for inclusion in this review. Location: GB, Great Britain; I, Ireland. Time: Y, year-round; Sp, spring; Su, Summer; Au, Autumn; Wi, Winter; 1–5, months of year, e.g. January–May. Type: F, faecal analysis; G, gut contents analysis; O, direct observations; R, prey remains. Method: F, frequency of occurrence; EWI, estimated weight intake; DW, dry weight; V, volume or percentage bulk estimate. Mass: weighted mean mass of mammalian prey (Jones 1997). Prey categories: SM, small mammals; MM, medium-sized mammals; LM, large mammals taken as carrion; UM, unidentified mammals; B, bird; E, birds’ eggs; F, fish; H, herpetofauna, i.e. reptiles and amphibians; W, earthworms Lumbricidae; I, invertebrates other than earthworms; FV, fruit and vegetable material eaten deliberately and not ingested incidentally; O, other food items. BA; Levins’s (1968) standardized measure of niche breadth.

(a)Day (1963, 1968)GBNational1960–63Y115F,GF217·759% 20%  14%     2%  5%  0·13
 Walker (1972) Hertfordshire1966Sp, Au 43GF130·271% 18%  11%       0·07
 Potts & Vickerman (1974) Sussex1968–72Y186GF141·769% 12%  15% 2%     3%0·08
 Moors (1975) Aberdeenshire1971–73Y264FF148·771% 14%  13% 2%      0·08
 Moors (1975) Aberdeenshire1971–723–6 82GF188·565% 17%  17% 2%      0·10
 Tapper (1976) Sussex1971–745–6151GF115·561%  8%  27% 4%      0·11
 King (1977) NorthumberlandNSNS 54GF323·351% 33%  16%       0·14
 Howes (1977) YorkshiremanyNS 40OF149·350% 10%  28%  13%    0·17
 Tapper (1979) Sussex1971–77Y687GF200·660% 17%  18% 2%  1%   2%  0·12
 King (1980) Oxfordshire1968–70Y344FF 29·968%< 1%  13% 5%    9%  5%  0·10
 Armitage (1980) Yorkshire1975–78NS  7OF500·0  14%  86%       0·03
 Pounds (1981) Aberdeenshire1976–79Y204F,GF229·168% 23%   9%       0·08
 McDonald, Webbon & Harris (2000) National1995–97Y458GF236·668% 26% 1% 4% 1%  1%    0·08
(b)Day (1963, 1968)1GBNational1960–63Y115F,GF427·128% 33%  33%     1%  6%  0·22
 Hewson & Healing (1971) Aberdeenshire1969NS 16OF800·0  31%  69%       0·07
 Howes (1977) YorkshiremanyY 60OF591·9 5% 38%  52% 5%     0·12
 Armitage (1980) Yorkshire1975–78Y  5OF 86·080% 20%          0·04
 Pounds (1981) Aberdeenshire1976–79Y 97F,GF386·644% 38%  18%       0·15
 Pringle (1982) Sussex1971–76Y282GF617·417% 57% 1%18% 8%  < 1%   0·14
 McDonald et al. (2000) National1995–97Y789GF644·016% 65%  14% 5%   < 1%  0·10
 Fairley (1971)INorthern Ireland1963–66Y 24GF406·929% 7%  64%       0·09
 Fairley (1971) Northern Ireland1968–70Y 58GF174·0 7% 67%  13%13%      0·10
 Sleeman (1992) IrelandNSY138GF701·735% 46%  19%       0·15
(c)Walton (1968)GBWalesNSNS 38GFNS   33%13%  25%  23% 7%0·28
 Blandford (1986) WalesNSNS251FF304·446% 27%  1%16%   8%   2%  0·20
 Birks & Kitchener (1999) Western England1990sNS 83GF712·8 8% 71%   9% 1% 8%  2%   0·08
Table 3.  The diets of (a) minks and (b) pine martens in Great Britain and Ireland. For a key to the abbreviations used, see Table 2
SourceLocationYearTimenTypeMethodMassSMMMLMUMBEFHWIFVOBA
  • 1

    Some of the figures in this study are estimated from graphs.

  • 2

    2 These samples are piles of faeces.

(a)Day & Linn (1972)GBEngland and Wales1964–68Y 273GF460·512%25%     1%36%    1%   14% 3%  8%  0·29
 Akande (1972) Scotland1967–68Y  88F,GF150·325% 7%    1% 21%    46%     0·19
 Cuthbert (1979) ScotlandNSY 722FF165·9 6% 4%     6% 7%    67% 2%  9%  0·10
 Cuthbert (1979) ScotlandNSNS  78FF432·1 4% 6%     4%16%    28%  42%  0·23
 Chanin & Linn (1980) Devon1972–73Y 475FF269·718%11% < 1%11%    54% 3% 2% 1%  0·17
 Chanin & Linn (1980) Devon1972–73Y  57FF603·1 3%13%  30%    54%  1%   0·14
 Chanin & Linn (1980) Dorset1972–73Y 153FF220·518%13%  24%    35% 1% 8% 3%  0·31
 Jenkins & Harper (1980) Aberdeenshire1975–78Y 185FF652·112%51%     3%23%       8% 2%    0·18
 Jenkins & Harper (1980) Aberdeenshire1977–78Y  59FF655·315%68%     4% 7%       6%     0·09
 Wise, Linn & Kennedy (1981) Devon1975–76Y 513FV459·912%17%  37%    32% 1% 1%   0·23
 Wise et al. (1981) Devon1975–76Y 448FV584·115%43%   5%    25%10% 2%   0·23
 Birks & Dunstone (1984) Coastal Scotland1981–82Y  96RF776·8 1%60%    1% 38%       0·09
 Dunstone & Birks (1987) Coastal Scotland1980–83Y2043FF697·6 5%35%< 1%< 1%11%    29%  19%  0·26
 Clode & Macdonald (1995) Coastal Scotland1991–93Su 160FF511·6 3% 5%    4%  8%    46%  34%  0·18
 Strachan et al. (1998) Southern England1994Y 863FDW380·4 7%29%  44%    18%     2%0·20
 Ferreras & Macdonald (1999) Southern England19963–8 115FDW442·524%28%  23%    18%   6%  0·30
 Fairley (1980)INorthern Ireland1975–76Year  46GF745·5 44%  12%    12%  20%  0·24
 Ward, Smal & Fairley (1986) Ireland1983–85Year2510FF622·5 1% 4% < 1%18%    17% 8% 50%  2%0·19
(b)Lockie (1961)GBRoss-shire1956–58Y 337FF 34·236%    9%    1%   4%24%18% 9%0·30
 Velander (1985) Inverness-shire1982–83Y 375FF 52·929% 1%    3% 14%< 1%  1% 34%17% 0·27
 Shaw & Livingstone (1992) Galloway1988–89NS  12F2F312·232%18%  23%   9%  9% 9% 0·34
 Balharry (1993) Ross-shire (Kinlochewe)1988–89Y 723FEWI212·024% 7%   34% 13%< 1%  4% 15% 3% 0·33
 Balharry (1993) Ross-shire (Strathglass)1989–90Y 581FEWI 33·547%    30% 12%    1%  7%  3% 1% 0·19
 Gurnell et al. (1994)1 Ross-shire1984–85Y 240FF 39·742% 1%    2%  8%     1%10% 1%28% 8% 0·23
 Bright & Smithson (1997)1 Dumfriesshire1995–96Y 100FEWI181·423% 6%   20% 22%     9% 6%14%0·43
 Halliwell (1997) Ross-shire1993–95Y1938FF101·329% 3%    5% 16%    2%  1% 24% 9%10%0·39
 Putman (2000) Highland (Morvern)1996–98Y 174FFNS  < 1%   27% 9% < 1% 1% 8%34%21% 0·28
 Fairley & O’Gorman (1974)IClare1970SuNSF2F 19·020%   20%  20% 40%  0·23
 Fairley (1975) Clare1974NS  23F2F301·413% 9%   7%   4%29%22%16% 0·39
 Warner & O’Sullivan (1982) Clare1973–77NS 609FF387·2 6% 6%   1%18%    2%  2%25%23%17% 0·40
Table 4.  The diet of badgers in Great Britain and Ireland. For a key to the abbreviations used, see Table 2
SourceLocationYearTimenTypeMethodMassSMMMLMUMBEFHWIFVOBA
Middleton (1935)GBOxfordshire1935Su   1GF282·7   62%   31%        8%  0·10
Kruuk (1978) Oxfordshire1974Au  39FFNS     5%   57%17%21% 0·14
Kruuk (1978) OxfordshireNSNS   8GFNS        88% 13% 0·03
Kruuk & Parish (1981) Scotland1975–78Y2159FV654·2    2%    9%1%  3% < 1%    2%58% 7%18%< 1%0·14
Harris (1984) Avon1977–80Y2376F,GV411·7< 1%< 1%   2% < 1%< 1%18%20%35%   24%0·26
Kruuk & Parish (1985) Scotland1975–82Y3419FV669·7    1%    5%3%  2%    2%48% 7% 32%0·17
Hofer (1988) Oxfordshire1982–835–11NSFDWNS        58% 6%27%    9%0·12
Skinner & Skinner (1988) Essex1984–86Y 315FVNS   5% 4%1%  16%23%48%   2%0·19
Shepherdson, Roper & Lüps (1990) Sussex1984–85Y 192FVNS   1%    43%10%45% 0·14
Fairley (1967)IN. Ireland1963–6412–4  33F,GF800·0     6%  20%3%  37%23%11% 0·28
Table 5.  The diet of otters in Great Britain and Ireland. For a key to the abbreviations used, see Table 2
SourceLocationYearTimenTypeMethodSMMMLMUMBEFHWIFVOBA
  • 1

    Some of the figures in this study are estimated from graphs.

Stephens (1957)GBNational1952–54Y 110F,GF    4%    5%     1%   11% 69%    8%     3%  0·09
Weir & Bannister (1973) Norfolk1969–70Y1200FF       2%    3% 89%      7%  0·02
Webb (1975) Somerset1972–74Y 858FF   < 1%    4% 71%    7%    18%  0·08
Weir & Bannister (1977 ) Norfolk1973–74Y1060FF       2%    2% 93%    2%     2%  0·01
Watson (1978) Shetland1976–77Y 437FF    < 1% 92%      8%  0·02
Jenkins, Walker & McCowan (1979)1 Aberdeenshire1975–76Y1018FF    1%    2%     3%    5% 77%   12%    0·06
Cuthbert (1979) ScotlandNSNS  47FF      16%    4% 73%    7%0·07
Jenkins & Harper (1980) Aberdeenshire (Dinnet)1975–78Y1135FF    1%    3%     1%    3% 80%   12%    0·05
Jenkins & Harper (1980) Aberdeenshire (Dee)1975–78Y1253FF    1%    4%< 1%    1%    1% 88%    4%    0·03
Mason & Macdonald (1980) Ross-shire1978–79Wi  50FF        2% 80%     18%  0·04
Wise et al. (1981) Devon1974–76NS1547FV< 1%    1%      5% 93%< 1%     1%  0·02
Wise et al. (1981) Devon1974–76NS 675FV    2%    4%     1%    2% 82%    5%     4%  0·04
Chanin (1981) Devon (Teign)1972–73Y 253FF    1%    3%      1% 92%    1%2%< 1%  0·02
Chanin (1981) Devon (Slapton)1972–735–10 389FF< 1%    1% < 1%    7% 91%< 1%    0·02
Herfst (1984) Shetland19826–8 107FF        1% 94%      4%  0·01
Weber (1990) Aberdeenshire19872–12 919FF       3%    2% 73%   21%     2%  0·07
Carss, Kruuk & Conroy (1990) Aberdeenshire1989–90Wi 324FF       2%  96%    3%    0·01
Kruuk & Moorhouse (1990) Shetland1983–87Y2031OF < 1%    97%      3%  0·01
Roche et al. (1995) Hertfordshire1992–94Y  51FF       3%    2% 92%    3%    0·02
Clode & Macdonald (1995) Scotland1991–93Su  77FF     7%      7% 68%     18%  0·09
Fairley (1972)IGalway1969–71Y  33GF        2% 67%   13%    18%  0·09
Fairley & Wilson (1972) Londonderry1970Au  88FF        5% 88%    7%    0·02
Gormally & Fairley (1982) Mayo (Furnace)1980–818–1 440FF        1% 86%    1%    12%  0·03
Gormally & Fairley (1982) Mayo (Feeagh)1980–818–1 266FF        1% 84%    4%    11%  0·03
McFadden & Fairley (1984) Galway and Mayo19817–12 781FF < 1% < 1%    5% 66%   13%    16%  0·10
Fairley & McCarthy (1985) Kerry19855–6  89FF        2% 98%     0·00
Murphy & Fairley (1985a) Galway1982–83Y1026FF    < 1% 81%< 1%    18%  0·04
Murphy & Fairley (1985b) Western Ireland1982–83Wi 394FF    < 1% 76%    2%    22%  0·05
Fairley & Murdoch (1989) Kerry19876 174FF        1% 98%    1%    0·00
Kyne, Smal & Fairley (1989) Midlands1983–85Y2349FF        2% 48%   15%    35%  0·15
Breathnach & Fairley (1993) Galway1991–92Y1709FF < 1% < 1%    1% 58%    9%    28% 4%0·13
Tangney & Fairley (1994) Connemara1993Su 316FF< 1%       2% 78%   12%     8%  0·05
Table 6.  Summary of the diets of British and Irish mustelids. Weasels and polecats are absent from Ireland. n is the number of samples reviewed. For a key to the abbreviations used, see Table 2
SpeciesLocationnMassSMMMLMUMBEFHWIFVOBA
WeaselGB13200·9   59%    16% < 1%21%     1%        1%       1% 1% < 1%0·13
StoatGB 7507·6   27%    40% < 1%29%     2%     1% < 1% 1%        1%0·19
 I 3427·5   23%    40%  32%     4%      0·19
MinkGB16466·4   11%    26%< 1%     1%21%< 1%   30%       1%       1% 8% < 1%0·32
 I 2684·0< 1%    24% < 1%15%    15%   10% 35%        1%0·30
PolecatGB 3508·6   18%    33%    11%13% < 1%   14%       1% 8%        2%0·38
Pine martenGB 9120·9   29%       4%  11%     3%14%     1%< 1%       4%       1%20%10%       4%0·44
 I 3235·9   13%       5% < 1%15%     1%        9%  18%29%11% 0·43
BadgerGB 9504·5     7%       5%< 1%     1% 2%< 1%< 1%< 1%  43%11%27%       4%0·24
 I 1800·0      6%  20%     3%    37%23%11% 0·28
OtterGB20590·3     1%     1%< 1%     2% 3%    84%       4%< 1% 4% < 1%0·04
 I12536·7< 1%< 1% < 1% 2%    77%       6% 14% < 1%0·05
Figure 1.

Relationship between weighted mean mass of mammalian prey and canine diameter of terrestrial mustelids in Great Britain and Ireland. British samples are marked with a square, Irish samples with a triangle.

Separate details of the diets of male and female mustelids were provided in 11 samples (Table 7). The mean mass of mammalian prey taken by males was consistently greater than by females (t = 5·93, n = 10, P < 0·001). Sex-related differences in diet were examined statistically in 16 studies (Table 8). In six studies, differences between males and females were statistically significant, but in only five of these were the differences related to prey size. Kruuk & Moorhouse (1990) found the lengths of fish eaten by male otters were greater than female otters. They also found that female otters took differently sized prey depending on whether they were feeding cubs. The differences observed between otters of varying status were, however, not significant when variation in prey availability in their home ranges was taken into account (Kruuk & Moorhouse 1990).

Table 7.  Differences in the diets of male and female mustelids. For a key to the abbreviations used, see Table 2
SpeciesSourceLocationYearTimeTypeMethodSexnMassSMMMLMUMBEFHWIFVOBA
  • 1

    This study includes an unknown number of samples from Northern Ireland.

WeaselDay (1963, 1968)GB1960–63YF,GFM 96255·654%24%  16%   1%   5%  0·15
       F 19 73·076% 5%  10%   5%   5%  0·06
WeaselPringle (1982)Sussex1971–76YGFM146240·747%10%    1%15% 1%      0·27
       F 66208·051% 4%    1% 8% 1% 1%    0·24
WeaselMcDonald et al. (2000; unpublished data)GB1995–97YGFM381163·166%26%    1% 5% 1% 1%    0·09
       F 77 83·477%23%          0·05
StoatDay (1963, 1968)1GB1960–63YF,GFM 85443·927%34%  32%   1%   5%  0·22
       F 30379·929%29%  36%       7%  0·22
StoatPringle (1982)Sussex1971–76YGFM138718·714%61%    1%18% 6%      0·12
       F114550·522%52%    1%14%10%  1%   0·17
StoatMcDonald et al. (2000; unpublished data)GB1995–97YGFM503488·5 8%73%  12% 6%   < 1%  0·07
       F286287·126%55%  16% 4%      0·14
StoatSleeman (1992)IrelandNSYGFM 58652·225%53%  23%       0·14
       F 34568·353%35%  12%       0·13
PolecatBlandford (1986)WalesNSNSFFM337330·742%30%    1%14%  9%    4%  0·21
       F251273·750%24% < 1%18%  8%    0·17
PolecatBirks & Kitchener (1999)Western England1990sNSGFM 53743·1 5%77%   7%  9%2%   0·06
       F 30652·113%61%  13%  3%6%3%   0·13
MinkBirks & Dunstone (1985)Coastal Scotland1980–83YFFM379731·5 4%49% < 1% 7% 20%   19%  0·19
       F645608·0 5%17% < 1%11% 38%   29%  0·24
Pine martenBalharry (1993)Ross-shire1988–90YFEWIM 62NS28% 46% 17%  7%    1%2% 0·20
       F 64NS25% 41% 27%  3%    1%2% 0·20
Table 8.  Statistical tests of the differences in the diets of male and female mustelids
SpeciesSourcenTestP Conclusion
MF
  1. NS = no significant difference; *P < 0·05;**P < 0·01; ***P < 0·001. LLM = Log-linear modelling, PIO = Pianka’s Index of Niche Overlap.

WeaselDay (1963, 1968) 96 19χ2NSM = FNo difference in diet. Non- significant tendency for males to take more lagomorphs than females
WeaselKing (1977)394118χ2*M < FSmall rodents
     NSM = FLagomorphs and birds
WeaselTapper (1979)199G*M > FLagomorphs in spring only
     *M < FMicrotus in spring only
     NSM = FAll groups in summer and winter
WeaselKing (1980)158 23Sign*M > FMicrotus
     NSM = FClethrionomys, Apodemus, birds, eggs or unidentified rodents
WeaselPringle (1982)146 66χ2NSM = FNo difference in diet. Non- significant tendency for males to take more lagomorphs than females
WeaselMcDonald et al. (2000)381 77LLMNSM = FNo difference in diet
StoatDay (1963, 1968) 80 35χ2NSM = FNo difference in diet
StoatPounds (1981)χ2NSM = FNo difference in diet
StoatPringle (1982)138114χ2NSM = FNo difference in diet
StoatSleeman (1992) 58 34χ2**M < FShrews
     NSM = FAll other groups
StoatMcDonald et al. (2000)503286LLM***M < FSmall rodents
     ***M > FLagomorphs
     ***M > FPasserines
     NSM = FBirds’ eggs and game birds
PolecatBlandford (1986)337251PIOM = FNeither sex showed selection in respect of prey size (p. 296)
PolecatBirks & Kitchener (1999) 53 30χ2NSM = FNo difference in prevalence of rabbits, other mammals, birds or amphibians
MinkBirks & Dunstone (1985)379645χ2***M < FFish
     ***M < FCrustaceans
     ***M > FLagomorphs
     *M < FBirds
     NSM = FSmall mammals
Pine martenBalharry (1993) 62 64χ2NSM = FNo difference in diet
OtterKruuk & Moorhouse (1990)510660χ2NSM = FNo difference in prey size or species composition between males and females with cubs, when differences in prey availability taken into account

Dietary niche breadth (BA) increased with body size in the assemblage when otters and badgers were excluded (rs = 0·76, n = 56, P < 0·001), but not when otters and badgers were included (rs = –0·18, n = 98, P = 0·08). There was no consistent difference in dietary niche breadth between male and female mustelids (t = –0·17, n = 11, P = 0·87). There was no significant difference in dietary niche breadth between British and Irish mustelids (t = –0·39, n = 5, P = 0·72).

Seventy-seven British samples were included in the initial PCA (Table 9). The species formed groups in multivariate space (Fig. 2) and otters and badgers were distinguished easily from the rest of the assemblage along the axes described by PC1 and PC2, respectively. There were no apparent trends in the first three components that were related to predator size. The second PCA omitted badgers and otters and was based on 48 samples. The first three components explained a similar proportion of the total variance (Table 9). From the initial analysis including all species, Euclidean distances between dietary niches were not related to predator size (Table 10). This was reflected in the community dendrogram (Fig. 3). Otters were clearly an outlying species since their diet was more distant (0·91) from the rest of the assemblage than the average distance between all species pairs (0·58). The diet of the smallest mustelid, weasel, was most similar to that of one of the larger species, pine marten. Stoats, which are closest in size to weasels, were most similar in their diet to polecats and then to minks. The Euclidean distances among members of the Irish assemblage were significantly lower than among members of the British assemblage (W = 4·0, n = 10, P < 0·05). In the second analysis, omitting otters and badgers, the Euclidean distances were similar to the first analysis (Table 10). There was no significant difference between the two analyses in the distances between species pairs in Britain (t = – 0·41, n = 10, P = 0·69) and in the dendrogram, weasels were again most similar to pine martens and stoats were most similar to polecats and then minks (Fig. 4).

Table 9.  Results of the principal component analysis on the composition of the diets of mustelids in Great Britain. For a key to the abbreviations used, see Table 2
 FactorWhole assemblageOmitting otters and badgers
 PC1PC2PC3PC1PC2PC3
Eigenvalue   0·170  0·055  0·041  0·078  0·047 0·027
Proportion of variance explained   0·545  0·176  0·132  0·419  0·252  0·145
Cumulative variance explained   0·545  0·721  0·853 0·419 0·671 0·815
Factor loadingsSM  0·400–0·617–0·549–0·875–0·045  0·097
 MM  0·211–0·210  0·619 0·193–0·800 0·409
 LM  0·020  0·007–0·035–0·021 0·069–0·079
 UM–0·008  0·045 0·002 0·043  0·077–0·045
 B  0·140–0·073  0·488 0·265–0·212–0·689
 E  0·010–0·011–0·001–0·015–0·015 0·003
 F–0·877–0·288–0·046 0·345 0·422  0·566
 H–0·015–0·003–0·002  0·006  0·027–0·021
 W  0·054  0·571–0·203–0·002  0·023–0·003
 I  0·007  0·179–0·104  0·072  0·338–0·103
 FV  0·047  0·352–0·144–0·008  0·085–0·089
 O  0·010 0·049–0·025–0·003 0·030–0·045
Figure 2.

Projection of 77 samples of the diets of seven species of mustelid in Great Britain along the first three principal component axes. Axes 1, 2 and 3 account for 55, 18 and 13% of the total variance respectively.

Table 10.  Euclidean distances between the group centroids from the first three components of the PCA of dietary composition of British and Irish mustelids. Details of British mustelids are in Roman characters and Irish mustelids in italics. Weasels and polecats are absent from Ireland. The Euclidean distance is a measure of the degree of differentiation between the species’ dietary niches. As it increases the similarity between their diets decreases. Distances in the bottom left part of the matrix were derived from the first PCA including otters and badgers. Distances in the top right part of the matrix were derived from the second PCA omitting otters and badgers
 WeaselStoatMinkPolecatPine martenBadger
Weasel*0·3960·5720·3890·320*
Stoat0·392*0·3740·1490·429*
   0·420 0·267 
Mink0·5620·372*0·2910·404*
  0·364  0·441 
Polecat0·3940·1550·239*0·338*
Pine marten0·3250·3910·4170·270**
  0·4880·304   
Badger0·7420·7050·6390·5590·426*
  0·6000·368 0·193 
Otter1·0470·9960·6330·9070·8970·985
  0·9070·624 0·7740·827
Figure 3.

Community dendrogram of the Euclidean distances between the dietary niches of the mustelids living in Great Britain. The dashed line indicates the mean Euclidean distance (0·58) between all combinations of species pairs (n = 21) in the assemblage.

Figure 4.

Community dendrogram of the Euclidean distances between the dietary niches of the mustelids living in Great Britain, omitting otters and badgers. The dashed line indicates the mean Euclidean distance (0·36) between all combinations of species pairs (n = 10) in the assemblage.

Discussion

The work of Dayan & Simberloff (1994) has lent substantial weight to Brown & Wilson’s (1956) notion of community-wide character displacement and the importance of competition in structuring assemblages of related species. Because of their work, resource partitioning according to prey size has become the established hypothesis explaining the coexistence of mustelid assemblages in the British Isles and elsewhere. This analysis of the dietary relationships among British mustelids has confirmed that otters should not be considered part of the same guild as the other six species (Dayan & Simberloff 1994), when adopting Pianka’s (1980) definition of an arena of intense interspecific competition. The omission of badgers from certain of Dayan & Simberloff’s (1994) analyses was also justified by the distance between this species’ dietary niche and those of the other mustelids.

As predicted by Dayan & Simberloff (1994), larger male mustelids consistently took larger prey than females. Outside the British Isles, larger male mustelids have often been observed to eat larger prey than females (Erlinge 1975, 1979; Simms 1979). Previous considerations of such intraspecific differences have supported the resource partitioning hypothesis by following the intuitively appealing notion that dimorphism had arisen primarily as a mechanism for avoiding competition between the sexes (Simms 1979). In contrast, Holmes & Powell (1994), following Holmes (1987), found that the carnassials of male and female mustelids were less dimorphic in size than were the rest of their bodies and concluded that resource partitioning could not have been the driving force in the evolution of size dimorphism. By constructing a model of the optimal body sizes of male and female mustelids, Sandell (1989) concluded that size dimorphism had arisen from differing energetic requirements during the breeding season. Gittleman & Van Valkenburgh (1997) found that the degree of sexual size dimorphism in carnivore canines was best explained by breeding system and not dietary traits and therefore concluded that dimorphism was the result of sexual selection. A sexual selection hypothesis is particularly convincing because male mustelids are always larger than females and this would not be predicted by theories based on prey selection (Gittleman & Van Valkenburgh 1997). Indeed, Dayan & Simberloff (1994) themselves did not explicitly claim that sexual size dimorphism was driven primarily by prey size partitioning, but that it might be an additional mechanism of coexistence. Thus, this study and the bulk of previous work accords with Erlinge (1979) and Moors (1980) that differences in diet between male and female mustelids are a secondary consequence of size dimorphism rather than a primary cause of it.

In contrast to the predictions of Dayan & Simberloff (1994), prey size did not increase with predator size and there was no evidence of resource partitioning according to prey size among species. While niche breadth increased with body size (Gittleman 1985), increasing diversity was not explicable solely by increases in the importance of larger mammalian prey. Pine martens ate larger amounts of fruit and invertebrates as well as carrion than smaller mustelids, while polecats and mink diversified their diet into herpetofauna and fish. Dayan & Simberloff (1994) cited a study conducted in the Netherlands (Brugge 1977) in support of their assertion that larger mustelids took larger prey. Nevertheless, a full examination of the British studies they cited (Day 1968; Moors 1975; King 1980; Blandford 1986, 1987; Blandford & Walton 1991) has revealed that these local studies do not support their assertion. The average prey sizes of British stoats, polecats, minks and badgers are similar, yet these predators vary greatly in size and large pine martens take much smaller prey than all of them. As Dayan & Simberloff (1994) are right to point out, pine martens living in continental Europe do eat hares and squirrels (Nyholm 1970; Pulliainen & Hiekkinen 1980). Such medium-sized mammalian prey is, however, rarely eaten by pine martens in Britain or Ireland, even when available in the habitats they occupy (Wray 1992; Hulbert, Iason & Racey 1996; Halliwell 1997). While large mammals make up a substantial proportion of pine marten diet in Britain, these are taken almost entirely as carrion. The empirical data analysed here do not therefore support the established hypothesis of resource partitioning according to prey size as an explanation for even size ratios in the canines of British mustelids.

Irish mustelids exhibit greater variation in canine size and greater size dimorphism than British mustelids. Dayan & Simberloff (1994) took this as evidence of competitive release. In contrast to predictions of competitive release and despite differences in the community of prey species (Corbet & Harris 1991), the diets of Irish mustelids were similar to British mustelids. The Euclidean distances among dietary niches were smaller for Irish mustelids than for British mustelids and the dietary niches of Irish mustelids were no broader than for British samples. These facts suggest that the character release detected by Dayan & Simberloff (1994) was not related to changes in diet. None the less, differences in prey size for Irish stoats and minks, but not pine martens, did correspond to differences in canine size and further studies of mustelid diet in Ireland would be invaluable.

Precise data on the size of individual animals killed by mustelids are required before the possibility of resource partitioning by selecting individuals of varying size within prey species can be ruled out (Kruuk, Conroy & Moorhouse 1987). Gathering these data would be challenging since mustelids often capture their prey underground and remains do not generally permit calculation of prey size. Equally, this analysis may not have fully reflected the extent of niche differentiation among truly sympatric mustelids, since most of the reviewed studies were based on one species and were drawn from throughout the British Isles. Notwithstanding the important issue of scale in defining sympatry and allopatry, a major addition to the current debate would be made by a study of the morphology and prey choice of mustelids living on a single site (Jones 1997; Jones & Barmuta 1998).

The question remains, if resources are not partitioned according to prey size, why do mustelids exhibit character displacement for canine size? Pimm & Gittleman (1990) have pointed out that the process giving rise to even size ratios in carnivore canines must be insensitive to variability in diet, but not to the presence of other carnivores. Interspecific aggression and intraguild predation fit this criterion (Palomares & Caro 1999) and have been considered an obvious alternative hypothesis (Dayan et al. 1989; Dayan & Simberloff 1998). Whitehead & Walde (1993) considered that canine size was an honest signal of threat and demonstrated that a signalling role for canines could lead to even size ratios among morphospecies. Dayan & Simberloff (1994) discounted the role of canines in threat displays since such displays were rarely observed in the Mustelidae. Gossow (1970), Poole (1967), Erlinge (1977) and numerous other studies reviewed by Pruitt & Burghardt (1977) have, however, described threat displays in mustelids that included baring canines. Canines are also involved in biting during conflict between mustelids (Poole 1967; Lüps & Roper 1988). Rosenzweig (1966) did not discount a hypothesis of interspecific aggression and colleagues have highlighted the importance of interference in facilitating the coexistence of weasels and stoats (King & Moors 1979; Erlinge & Sandell 1988). Curiously, Sidorovich, Kruuk & Macdonald (1999) demonstrated an increase in the size of native polecats and European minks Mustela lutreola L. in Eastern Europe following the arrival of the larger feral American minks, which have simultaneously decreased in size. Sidorovich et al. (1999) suggested that direct aggression between species was leading to enhanced survival of larger individuals of smaller species. Alternatively, Abrams (1986) has suggested that evolutionary ‘races’ do not always lead to intuitive endpoints. In this case, competing mustelids may find stable strategies not by resource partitioning via scaling of trophic apparatus, but by character convergence (Grant 1972) or they may exhibit no obvious response to the presence of competitors at all. Thus, as suggested by this analysis, patterns in carnivore dentition or body size may not be a dependable index of resource partitioning. While field studies of interactions between carnivores are particularly challenging, examination of alternative hypotheses to resource partitioning is clearly necessary for explaining character displacement among mustelids.

Acknowledgements

This work was supported by a Royal Society Postdoctoral Fellowship, hosted by the Centre for Biodiversity and Ecology Research, University of Waikato, New Zealand and a Royal Society Conference Grant to attend the European Congress of Mammalogy in Jyväskylä, Finland. I am grateful to Philip Baker, Johnny Birks, Paul Bright, Paul Chanin, James Fairley, Elizabeth Halliwell, Carolyn King, Roger Powell, Rory Putman, Linda Sadlier, Dan Simberloff and Nancy Vaughan for reprints, access to original data and advice and to Andrew Kitchener for measurements of otter canines. The manuscript was greatly improved by the incisive comments of the anonymous referees.

Ancillary