Intraspecific variation in habitat availability among ectothermic animals near their climatic limits and their centres of range



1. In a modelling exercise, the quantity and distribution of habitat patches within a heathland biotope for four ectothermic heathland animals (silver-studded blue butterfly [Plebejus argus], a red ant [Myrmica sabuleti], heath grasshopper [Chorthippus vagans] and sand lizard [Lacerta agilis]) were compared in space and time assuming two climates: that experienced at the northern edge of the species’ ranges and that 300–400 km further south, where mean summer temperatures are 2–3 °C warmer.

2. Habitats both at the northern edge of their ranges and 300–400 km further south for the four species were defined qualitatively from existing sources and then expressed quantitatively in terms of the attributes recorded in the Dorset Heathland Survey. The Survey was then used as a GIS to map the occurrence of the habitats of the four species under two climates and a decade apart.

3. The model predicts that an increase of 2–3 °C can result in a large increase in the area of habitat available to these north temperate species, that the length of time that individual patches of successional habitat may be occupied increases and that the distance between habitat patches within the biotope decreases.

4. The warmer conditions should result in a more stable metapopulation structure for P. argus, with fewer metapopulations existing in the landscape but each, on average, containing a greater number of larger and more stable constituent populations.

5. These predictions are of significance to ectothermic species which currently live at the northern limits of their ranges in the British Isles. The reverse effect is likely for species at the southern limits of their ranges. Conservationists who wish to maintain the status quo may be able to reduce some effects of these changes by appropriate habitat management.


A promising way of predicting how species’ populations may respond to climate warming in the Holarctic is to examine how the same organisms function at lower altitudes or further south in their ranges, where current climates are warmer. Here, we compare how the quantity and distribution of the habitats of four species of ectothermic animal differ in space and time across a heathland landscape under the climates that prevail near their northern edges of range and 300–400 km south, where mean spring and summer temperatures are 2–3 °C higher. Both these habitat parameters are key determinants of the size and persistence of animal populations in a region (Dempster & McLean 1998). The quantity (and quality) of habitat determines the carrying capacity of a site for a species or the levels at which populations of interacting species equilibrate, while the juxtaposition and continuity of habitat patches of different sizes influence the extent to which species can track the occurrence of new patches of habitat, and can persist as metapopulations in a landscape.

Thomas (1991, 1993, 1995) has suggested that small differences in mean temperatures may have large effects on the availability of habitat for ectotherms in temperate regions because cold-blooded animals generally need to raise their body temperatures above a threshold before activity or development becomes possible (Logan et al. 1976) and are often unable to influence this by thermoregulation during certain life-stages, or can do so to only a limited extent within the range of microclimates available in their vicinity (e.g. Willott 1997). As a result of this constraint, many invertebrates occupy narrower niches within their biotopes towards their northern edges of range, characterized by the need for their basic resources to coexist with substantially warmer microclimates than are typical for that latitude or altitude (Thomas 1983b, 1991, 1993; Cherrill & Brown 1990, 1992; Thomas & Morris 1994). For example, the subterranean young stages of the ant Myrmica sabuleti have a long growing season and require frequent daytime temperatures of 20–25 °C at the soil-surface between early April and October (Elmes & Wardlaw 1983). To obtain this exposure, M. sabuleti is mainly restricted to south-facing slopes in the UK, where the local climate is warmest, and within these slopes to patches where the sward is short (< 5 cm tall) or sparse, causing the soil to become several degrees hotter by day than under tall or dense vegetation (Thomas 1983b, 1993). But this habitat is too hot and dry for M. sabuleti in lowland central-southern France, where the macroclimate in summer is c. 3 °C warmer: instead, it occupies terrain of any aspect other than south-facing, and is further restricted to swards that are tall enough (> 20 cm) to provide shade from the summer sun. Intermediate niches are occupied by M. sabuleti at intermediate latitudes or altitudes (Thomas et al. 1998).

Intraspecific comparisons of how core and edge-of-range populations function are especially pertinent for the British fauna and those of similar latitudes across the Holarctic because many north temperate ectotherms reach their northern limits of range within this zone. About 80% of British butterfly species reach their range edge somewhere along the 5 °C gradient in mean spring-summer temperatures that exist between south England and north Scotland, and many nationally endangered or notable species are confined to the extreme south (Shirt 1987; Thomas & Morris 1994; Thomas 1995). Predicting how northern edge-of-range populations may change if the climate becomes warmer is important because, if rainfall remains similar, they might be expected to spread locally onto land with different aspects and to derive their resources from different seral stages; they also form the vanguard from which species may expand north.

To date, comparisons of edge-of-range and core populations of British butterflies indicate that the former experience large erratic ‘random-walk’ annual fluctuations compared with southern populations of the same species, which tend to oscillate more closely around a mean (Thomas, Moss & Pollard 1994). Core populations also tend to exist at slightly higher average densities (Thomaset al., in press). In addition, Thomas (1991) estimated the area of habitat available for the Silver-studded Blue Butterfly (Plebejus argus L.) in a small subset of the Dorset heathlands using definitions of its habitat made in the north and nearer the centre of its range. Here, we extend this approach to predict how the number, size and continuity of habitat patches changes for three insects and one lizard in space and time across the whole of the Dorset heathlands if the identical set of biotope islands were first exposed to the climate tolerated by each species at its northern limit and then to spring-summer temperatures that are 2–3 °C warmer. In the case of P. argus, whose vagility and population structure in Britain are well described (Thomas & Harrison 1992; Thomas, Thomas & Warren 1992; Lewis et al. 1997), we also predict how its metapopulations may vary under the two different local climates. In all analyses we assume that the community of dominant heathland plants remains stable under the range of climates considered for each species, including the status of the main larval foodplants. This is reasonable, in the light of studies of heathland floral composition on similar soils further south in Europe, where summer temperatures are higher than temperature elevations considered here (Webb 1986). Another prerequisite is to distinguish between the distribution and abundance of the islands of a biotope—in this case heathland—in a landscape (Fig. 1) and the distribution of the patches of habitat of each individual heathland species within this biotope (Thomas 1991; Webb & Thomas 1994). Thus for a hypothetical insect species x which requires Calluna vulgaris L. (Hull) as a larval foodplant, the entire archipelago of heathland biotopes might represent the distribution of its habitat. However, useable patches of habitat may be few and far between within heathland if species x requires an additional or narrower resource, such as young foodplants growing in a warm local climate: some heathland biotope islands may contain no patches with these attributes; others may contain several patches, each supporting a discrete population if species x has low mobility, as is the case with all species considered in this paper. For each population of this species, the nearest neighbouring population may live elsewhere on the same heathland island, or one or several heaths away, with the intervening land being equally inhospitable—from species x's viewpoint—whether it consists of unexploitable heathland biotope, farmland or other land between individual heaths (Webb & Thomas 1994). Using individual species’ habitat patches rather than biotope islands as the geographical unit can lead to very different estimates of the extent of habitat availability and connectivity, and has been used to identify more precise area/isolation patterns in the persistence of species’ populations than had previously been possible (Thomas 1991; Thomas et al. 1992; Webb & Thomas 1994; Thomas & Hanski 1997).

Figure 1.

. Distribution of all heathland in Dorset, UK, with the south coast shown as a solid line. The box (dotted line) represents the subset of squares used to illustrate the availability of habitat for Plebejus argus on individual heaths in Fig. 3. Each square is 200 m × 200 m (4 ha).

Dorset heathland (Fig. 1) was chosen for this exercise because of the availability of a unique database (The Dorset Heathland Survey) describing sufficient different attributes in every subarea for intra- and interspecific comparisons of habitats to be made (Webb & Haskins 1980; Webb 1990). Furthermore, because two surveys were made, a decade apart, of every attribute of every patch of heathland, we can compare changes in habitat availability in time as well as in space under different scenarios of local climate.

Materials and methods


The extent to which aspect and vegetation-cover modify local temperature, especially at ground level, has been described for many other biotopes (e.g. Perring 1960; Smith 1980; Thomas 1983b, 1993; Mitchell 1992; Pigott & Pigott 1993). Similar patterns were recorded beneath two and four seral stages of heathland by Delany (1953) and Barclay-Eastrup (1971), respectively, but neither included aspect as a variable. To confirm that both processes function in combination on heathland, we simultaneously measured soil-surface temperatures using thermocouples placed under heather stands of various heights on 30 March on one north-facing slope (aspect 340 °, slope 15 °) and on one south-facing slope (aspect 170 °, slope 25 °).


The entire heathlands of Dorset (Fig. 1) were surveyed in 1978 and 1987 using 200 m × 200 m square recording units referenced to the National Grid. The Heathland biotope islands were distributed over 3110 grid squares in 1978; the same squares were also used for the 1987 survey (Webb & Haskins 1980; Webb 1990). For each grid square, 184 attributes were recorded, representing vegetation composition and structure (100 attributes), land-use (48 attributes), and topography and physical characteristics (36 attributes). Heathland succession was classified into one of four recognized seral stages: pioneer, building, mature and degenerate phases (Watt 1955), while aspect was measured to the nearest 45° bearing. Estimates of the area of each heathland and associated heathland vegetation type, and the amalgamation of the 200 m × 200 m squares into heathland fragments, were calculated by the methods given in Chapman, Clarke & Webb (1989). For clarity, the illustrated worked example for P. argus is restricted to a subset of the squares (Figs 1, 3), although the statistics (Table 2) have been generated for all 3110 squares.

Figure 3.

. Estimated distribution of habitat suitable for P. argus (▪) in a subset of Dorset heathlands in 1978 and 1987, assuming the landscape is situated at the north edge of P. argus’ range and c. 400 km south where mean summer temperatures are 2–3 °C warmer. The heathland biotope is shown in grey and represents the distribution of the larval foodplants of P. argus.

Table 2.  . Comparisons of estimated habitat available in 1978 and 1987 for four species if the same landscape of Dorset heathland were situated near their northern limits or more centrally in their ranges, where summer temperatures are 2–3 °C warmer Thumbnail image of

The habitat patches of each individual species within heaths were defined using the following rules. All adjoining squares (both adjacent and on the diagonal) that contained some of the habitat defined for the species (Table 1) were assumed to belong to the same habitat patch, except when adjoining squares had already been defined as being within different heathland fragments. The area of a species’ habitat within each habitat patch was calculated directly: for P. argus to a resolution of 0·2 ha per 200 m × 200 m square; for Myrmica sabuleti Meinert, Chorthippus vagans (Eversmann) and Lacerta agilis L., the areas of heathland-type that were unsuitable within each 200 m × 200 m square (i.e. the proportions of humid heath, peatland and wet heath) were subtracted from the total area of heathland in each habitat patch. Distances between habitat patches were measured as a straight line between the closest points of the nearest squares of each habitat patch.

Table 1.  . Definitions of the habitats of ectothermic species near their north range limits and near their centres of range Thumbnail image of


Two definitions of habitat requirement were obtained from the literature (Table 1) for each of the four heathland species considered: P. argus (Silver-studded Blue Butterfly), C. vagans (Heath Grasshopper), M. sabuleti (a Red Ant) and L. agilis (Sand Lizard). The first described each species’ habitat near the north edge of its range; the second was described nearer the centre of each species’ range, where spring and summer temperatures are 2–3 °C warmer (Table 1). Each set of attributes listed in Table 1 was specified in the Dorset heathland GIS database, to generate the number, size and spatial distribution of the habitat patches available for each species in 1978 and 1987 under the two local climates considered.

All the species studied are known to be comparatively sedentary (Nicholson 1980; Thomas & Harrison 1992; Clarke et al. 1997), and any break in the distribution of a species’ habitat of one grid square or more (i.e. > 200 m) was assumed to mark the boundary of a separate population. That is not to say that a few individuals do not disperse further; rather that birth and death rates within each continuous patch of habitat exceed emigration and immigration between patches by a sufficient amount for each patch to be demographically distinct. Longer range dispersal across heathlands by P. argus has been studied in particular detail: for it, distances of 50–100 m of unsuitable vegetation are considered sufficient to isolate one population from another, and distances of 1 km between suitable patches may a sufficient barrier to isolate one metapopulation from another (Thomas & Harrison 1992; Lewis et al. 1997). In a separate exercise using these criteria, we grouped the patches of habitat available to P. argus in the landscape to predict the potential metapopulation structure for this species if the Dorset heathlands experienced the climates that pertain at its edge and centre of range.



Temperatures were 4–5 °C warmer on a slope that faced due south compared with north, and were 3–8 °C warmer under 1–3 cm tall (pioneer phase) heath compared with > 10 cm (mature) heath (Fig. 2). The combined effect is to expose a ground-dwelling ectotherm inhabiting pioneer heath on a south-facing slope to temperatures > 10 °C warmer in spring than if it lived under mature heath with a north aspect. These effects of aspect and shading on soil surface temperatures are similar to those reported from other biotopes.

Figure 2.

. Temperatures recorded at the soil surface beneath heathland vegetation of varying height, on heathland slopes with north- and south-facing aspects.


Remarkable differences were predicted in both years for all four species if the identical landscape were situated near their edge of range or c. 300–400 km further south in their core range (Table 2). These were particularly marked for P. argus and M. sabuleti, whose lengthy immature stages need the warmth found beneath early successional heath to survive near their northern limits. In contrast, the juveniles and adults of L. agilis and C. vagans—like other lizards (Wagner & Gleeson 1997) and grasshoppers (Willott 1997)—can thermoregulate to a considerable extent in any successional stage of heathland vegetation, but require bare sand to coexist with a warm aspect for their egg stages (Table 1). Warm sand is most abundant in pioneer heathland, but is not confined to it, and frequently coexists with building-phase and degenerate heath, as well as on paths, banks, etc. The differences for the four scenarios for P. argus (edge and central range in 1978 and 1987) are illustrated for a subset of heathlands (Fig. 3). The very small number and size of habitat patches available to ‘northern’ populations of P. argus and M. sabuleti in 1987 is particularly striking, reflecting the fact that not only are warm aspects comparatively scarce in this landscape but that very little of the heath growing on them was maintained in early successional stages. In contrast, fires during the extreme drought of 1976 resulted in a comparative abundance of early successional stages—and hence warm microclimates—at ground level on these heaths in 1978. The folly of failing to maintain on 95% of heaths (Table 2, column 5) even a single patch of habitat for many of the thermophilous invertebrate species for which this biotope was once famed is discussed elsewhere (Webb & Thomas 1994) and is beyond the scope of this paper. Suffice it to note that we estimate that, for P. argus, the same heathland landscape (Fig. 1) would provide four to 20 times more habitat patches, each on average of two to four times the size, and on average one-half to one-eighth the distance apart, if it were situated under a climate 2–3 °C warmer (equivalent to being 300–400 km further south) than at the range edge for this species in Britain. Moreover, for both P. argus and M. sabuleti, there was very little temporal continuity of ‘edge-of-range habitat’ in the same patch of heathland between 1978 and 1987 (none in the case of M. sabuleti: Table 2, column 11). In order to track new habitat patches that arose near their range edges, P. argus and M. sabuleti would, on average, have to disperse and colonize habitat 22 km and 48 km, respectively, from the nearest previous patch, compared to mean distances of only 0·4 km and 0·3 km, respectively, under the warmer climate. The latter distances are within the known colonization ranges of each insect whereas the former represent almost insuperable obstacles over so short a time span (Thomas 1991; Thomas & Harrison 1992; Clarke et al. 1997; Lewis et al. 1997; Thomaset al., in press; J. A. Thomas, unpublished data).

Differences for C. vagans and L. agilis are substantial but less extreme. Although the number of habitat patches predicted as being available for each species at different latitudes and dates are similar (Table 2, column 4), predicted average patch size (column 6) is two (C. vagans) or five times (L. agilis) larger under the southern climate, and the distance required to track new habitat is substantially less (column 12).


The isolation of individual habitat patches predicted for P. argus if Dorset experienced the climates of its edge or centre of range is reflected in a very different metapopulation structure predicted for the two scenarios. Figure 4 shows all patches of suitable habitat in the landscape in 1978–1987, linked (in stipple) by gaps that are within the 1 km colonization range of this butterfly during the 5 years (in the north) or 10 years (south) that an individual patch remains suitable. We predict a potential maximum of 33 metapopulations of P. argus, of which only 14 contain more than one constituent population under a northern climate, but only 19 metapopulations if the same landscape experienced a 2–3 °C warmer climate, with each containing an average of nine constituent populations and three times the total breeding area per metapopulation (Table 3). In theory (Hanski & Gilpin 1997), the former situation represents an unstable situation for a species that lives in a metapopulation structure approximately midway between the extremes represented by Levins (classical) and Boorman–Levitt (mainland-island) metapopulations. This predicted instability is supported by another analysis based on empirical patterns demonstrating the effect of area and isolation on the probability of habitat patches being occupied by P. argus in a British landscape (Thomas et al. 1992): from left to right, the diagonal lines in Fig. 5 represent the 90%, 50% and 10% probabilities of patches of varying size and isolation being occupied by P. argus (from Thomas et al. 1992); superimposed on these are the sizes and distances apart of all predicted habitat patches of P. argus in the Dorset heathland landscape, in each survey year and under each scenario of regional climate. Clearly, under the southern climate, the majority (85% in 1978, 88% in 1987) of patches are within the zone representing a > 90% probability of occupancy by this butterfly. In contrast, 28% of the patches predicted as existing in this landscape for P. argus at its edge of range have a < 10% probability of being occupied in 1978, and under the conditions of 1987 only three of the nine habitat patches present have a reasonable probability of being used.

Figure 4.

. Predicted metapopulation structure of P. argus in the same landscape in 1978–1987 at (a) its northern edge of range and (b) c. 400 km further south.

Table 3.  Estimated metapopulation structure if the Dorset heathland landscape was situated at the edge of P. argus’ range or c. 400 km south where the summer climate is 2–3 °C warmer Thumbnail image of
Figure 5.

. Area/isolation graphs for P. argus on Dorset heathland in 1978 and 1987 under two scenarios of regional climate. The zone above (top left) each diagonal line represents the 90% (dotted line), 50% (solid) and 10% (dashed) probabilities of a habitat patch being occupied by P. argus, based on empirical occupancy patterns described by Thomas et al. (1992). ●, the predicted size and isolation of habitat patches for P. argus in Dorset under the four situations.


Our results suggest that small increases in mean temperatures near current northern range limits may lead to large increases in the area of habitat available to north-temperate species on individual sites and across landscapes; that the length of time that individual patches of successional habitat may be occupied will also increase; and that the distances between isolated habitat patches will decrease. It is important to note that the niche occupied by some species (e.g. M. sabuleti) not only broadens but also alters between its northern limit and centre of range, and that some situations, like south-facing aspects, that are suitable under cool conditions may become unsuitable under warm ones (Table 1).

For species which live in metapopulations, the changed population structure under a slightly warmer climate is likely to result in populations that are spatially more stable, as well as larger, in the current northern halves of their ranges. Moreover, we may have underestimated this change if individuals become more mobile under warm conditions. At a finer scale, empirical studies (Thomas et al. 1994) show that edge-of-range populations of some butterfly species experience large erratic annual fluctuations in abundance compared with southern populations on individual sites; one explanation is that, at range edges, comparatively small annual variations in the weather may result in considerable differences in the amount (and quality) of habitat being available per square metre of ground.

Our simulations compared only the habitat available to core and edge-of-range populations of four ectotherms in the northern halves of their ranges, where current differences in spring-summer temperatures are similar to the increases predicted for many temperate regions in the Holarctic by the year 2050 (Houghton et al. 1995). Additional habitat definitions made at intermediate latitudes and altitudes in the northern half of M. sabuleti's range indicate a gradual broadening of its fundamental niche from cooler to warmer climates (Thomas et al. 1998). This suggests that the availability of habitat may also change gradually as climates warm between the extremes described in Table 2.

Predictions also exclude effects of other factors that may change alongside climate warming, such as enhanced CO2 concentrations or higher soil water deficits, although the latter is to some extent encompassed by the empirical descriptions of species’ habitats made near current centres and edges of range. Nevertheless, the magnitude of change predicted as a result of higher temperatures is so great that it is likely to have important implications for ectotherms in the British Isles, where a great many species currently reach their northern boundaries (Thomas & Morris 1994; Thomas 1995). Indeed, Thomas (1993) argues that the historical diversity of ectotherms at these latitudes is unnaturally high, because many ground-dwelling thermophilous species were not forced to track south when summer climates cooled by c. 2 °C about 5000 years ago, but were able to persist beyond their ‘natural’ limits in the warm, very early successional habitats that had been created extensively within grassland, woodland and heathland by prehistoric agriculture and silviculture. Thomas (1993) argues that extinction was postponed for about 5000 years for these species, until 20th century methods of agriculture and forestry resulted in a shift towards late successional stages predominating in most semi-natural biotopes. Whatever the merits of this hypothesis, we predict that so long as rainfall and land-use remain similar, most of the British invertebrate fauna will be more abundant in mid- to late-seral stages under a warmer climate, and that the recent declines of many endangered thermophilous species will be reversed.

We were unable to model the changes in habitat available to northern temperate ectotherms in the southern halves of their current ranges following climate warming, owing to a lack of precise definitions of intraspecific variation in the niches occupied in these zones. However, we would expect the same processes to operate in reverse, although soil moisture deficit rather than excessive temperature may be a more important constraint on what constitutes suitable habitat at southern range limits (Pigott & Pigott 1993) and foodplant populations may also show greater change. Whatever the mechanism, we predict that the montane and Boreal invertebrates which currently reach their south limits in Britain, and temperate European species which do so around latitudes 35–45 °N, are likely to suffer the triple blow of their habitat patches becoming smaller, more ephemeral, and more isolated in the southern halves of their ranges if the climate warms.

Although the current exercise was restricted to animals, we suspect that similar patterns of change will occur among temperate plants, because they also depend on local climate for their temperatures and they, like ectotherms (Turner et al. 1987; Dennis 1993), frequently have northern range boundaries that are correlated with spring and summer isotherms. The dramatic effects described in Table 2 are most likely to be experienced by small plants inhabiting low strata that can be shaded by larger species. However, even among trees, some of these patterns may apply, owing to the predominance of populations on south- and north-facing aspects, respectively, near their northern and southern edges of range (Pigott 1991; Pigott & Pigott 1993).

Finally, conservationists who wish to maintain the status quo may have some scope to diminish these predicted changes, at least for ground-dwelling ectotherms which currently occupy early successional stages. Although food quality may also differ between foodplants growing in early and mid-successional stages of vegetation, there may be scope for reducing the warming effect by maintaining more vegetation in later successional stages (Webb & Thomas 1994). Selecting conservation sites that have heterogeneous terrain—including north-facing aspects—may also enhance the persistence of biodiversity. These arguments apply equally to species in other biotopes, such as the current inhabitants of early seral stages in calcareous grassland and woodland (Thomas 1993).


We thank J.M. Bullock and M.O. Hill for their constructive comments. This study was partly funded by NERC's Terrestrial Initiative in Global Environmental Research (TIGER), award no. GST/91/114.