Davis et al. (1998) recently criticised as simplistic the view that an organism's range and abundance are directly determined by its physiological response to climate. Based on laboratory experiments involving three generalist but congeneric species of fruit fly (Drosophila spp.) and a shared parasitoid, they concluded that species interactions (dispersal, competition and enemy–victim interactions) within a community will vary with temperature and that this has important consequences for species distribution and abundance. As a consequence, the range of a single species may differ markedly from that predicted from climatic data alone and interacting species within communities may respond in unforeseen ways. Their experiments, however, were conducted over a range of non-fluctuating temperatures, using various combinations of hosts and parasitoid but avoiding temperature regimes that physiologically excluded selected species. The question I wish to ask therefore is whether their sweeping conclusions are supported by existing field data? Is there not a danger that such conclusions, drawn from limited theoretical studies in the laboratory, may be accorded undue significance compared with extensive field studies where perhaps the emphases have been less strongly directed towards developing general theories?
In many respects this reopens the earlier debate over the role of biotic vs. abiotic factors in determining the distribution and abundance of animals (Andrewartha & Birch 1954; Nicholson 1954). Put simply, this asked whether the distribution limits of species were determined by their response to biotic or abiotic factors, and whether the importance of the different factors and their significance for population density varied across a species’ range.
I and colleagues have, over many years, conducted studies on the distribution of herbivorous insects and soil arthropods at their northern or altitudinal range limits, usually in thermally restricted climates with a low mean temperature and a short growing season, particularly within the Arctic or the mountains of Europe and North America (Hodkinson & Bird 1998; Hodkinson et al. 1998). These are the organisms that are and will be, if current predictions are correct, subject to the greatest and most rapid climate change. Our data suggest that Davis et al. are far too dismissive of the significance of good ecophysiology linked to microclimatic data for understanding species distributions. A wide body of literature appears to reinforce our conclusions (e.g. Kukal, Ayres & Scriber 1991; Porter, Parry & Carter 1991; Drake 1994; Williams & Liebhold 1995; Bryant, Thomas & Bale 1997; Cannon 1998; Virtanen, Neuvonen & Nikula 1998).
There is often a very close and predictable link between climate and distribution, provided one knows ones species, understands its precise ecophysiological limitations, and have a clear appreciation of the temporal and spatial variation in the microclimate (not macroclimate) that it experiences. The response, however, is often individualistic, at the species, rather than the community level. Nevertheless, the characteristic response of individual species is often an excellent sensor of climate change and the species range limits. In fact many species, as they begin to approach their northern or altitudinal range limit, appear to escape from some interactions by entering, for example, predator-free space. Our work includes many examples spanning both parasitoids and predators. For example, the hymenopterous parasitoids of the heather psyllid Strophingia ericae (Curtis) and the arctic aphid Acyrthosiphon svalbardicum Heikinheimo decline to insignificance as their hosts reach their altitudinal or latitudinal limits, respectively (Hodkinson 1973b; Whittaker 1982; Hodkinson et al. 1998). Similarly, predation by syrphid larvae on willow psyllids such as Cacopsylla groenlandica (Šulc) declines rapidly with increasing latitude and the collembolan Onychiurus arcticus (Tullberg) is only predated by the beetle Atheta graminicola (Mulsant & Rey) in the marginally warmer parts of its habitat (Hodkinson 1997; Hodkinson et al. 1998). Similar studies abound in the literature (Whittaker 1971; Randall 1982a; Whittaker & Tribe 1998) and it has been clearly demonstrated how adverse weather may affect the efficacy of parasitoids (Weisser, Vokl & Hassell 1997). Perhaps it could be argued that the host is actively eliminating its parasitoid/predator by evolving a more effective physiology and that this is a temperature-sensitive interaction. However, is this really an outcome that could not be predicted from a precise knowledge of the life-history characteristics and physiologies of the interacting species?
The one point on which we can all agree, however, is that prediction is difficult and generalization across species to communities is nigh impossible. This is primarily because each species has its own individualistic physiology and response to climate, such that detailed work is usually required to discover the climatic constraint that limits distribution. Even among closely related species the response may be quite different. For example, two species of psyllid (Strophingia) live on two heather species, Calluna vulgaris (L.) Hull and Erica cinerea L. with similar altitudinal ranges in the UK (Hodkinson et al. 1999). Both psyllid species have similar tolerances of low temperature but the species on Calluna has a flexible life-history response to temperature and photoperiod that allows it to occupy the continuous range of its host, whereas that on Erica has a less flexible physiology that restricts its distribution to just a limited part of its host range (Hodkinson 1973a,b; Miles, Bale & Hodkinson 1997, 1998). Where congeneric species are potentially in direct competition on the same host, such as the Craspedolepta species on Chamerion angustifolium (L.) J. Holub, the distributional constraints still appear to be related to summer thermal budgets rather than competitive interactions or parasitism/predation (Bird & Hodkinson 1999). Thus, where thermal gradients are steep this can lead to segregation of congeneric species along the gradient, with little to indicate strong competitive interaction where species ranges overlap, such as the three Cacopsylla species on Salix lapponum L. (Hill & Hodkinson 1995; Hill, Hamer & Hodkinson 1998).
Most models of climate change, and for that matter laboratory experimental systems, are too coarse (simplistic?) to incorporate the finer details of microclimate to which many organisms are responding and which can ultimately produce apparently distinct distribution patterns. In environments where temperature (or for that matter water) is limiting, small differences in microtopography within the habitat mosaic can create strong microclimatic differentials over short distances and allow persistent microclimatic refuges to develop (Coulson et al. 1993, 1995). Thus, on Spitsbergen the local distribution of the arctic aphid Acyrthosiphon svalbardicum is restricted to host plants (Dryas octopetala L.) growing in slightly warmer microsites adjacent to the fjord edge and absent from colder sites around the fjord entrance and from further inland, despite the presence of apparently suitable host plants (Strathdee et al. 1993; Strathdee & Bale 1995). Similarly, several thermophilous British species of Lygaeid bug are found only on south-facing chalk escarpments at critical slope angles creating conditions that mimic the thermal environments of southern France, where the bugs are more widespread and common (Judd & Hodkinson 1998). Furthermore, for many soil animals occupying closely similar macroclimates, differences in temperature or humidity microclimates related simply to vegetation cover or surface aspect may be sufficient to produce major apparent differences in species response to temperature (Coulson et al. 1996). Thus, communities of Collembola on Spitsbergen responded differently to the same temperature manipulation at tundra heath and polar semi-desert sites, respectively. For many insects the moisture environment determines their survival response to temperature (e.g. Costanzo et al. 1997; Lindsay, Parson & Thomas 1998) and development rates often differ significantly between fluctuating and constant temperature regimes (Hagstrum & Milliken 1991; Worner 1992).
Rather than make dismissive statements, we need to explore existing data more thoroughly to determine the extent to which range limits and distribution patterns in the field can be fully explained by detailed physiological data linked to microclimatological measurements and to what extent such patterns are disrupted by species interactions. However, this is species-centred, time-consuming, based on fairly routine field studies and consequently scientifically unfashionable.
Ultimately, we are concerned with the dynamic balance between the availability of suitable microenvironments and their occupation, which depends on the dispersive capabilities of the organisms (Hodkinson & Wookey 1998). Many smaller common species of flying insect are a ubiquitous component of the aerial plankton and disperse continually well beyond their established breeding range, suggesting that newly favourable habitats are unlikely to remain undiscovered, even over relatively short time scales (Elton 1925; Glick 1939; White 1970; Papp & Johnson 1979; Ashmole & Ashmole 1988). Even among soil invertebrates there is the apparent paradox that some of the most widely distributed and common organisms on a global scale appear to lack specialist dispersal mechanisms yet appear to disseminate highly effectively. Such ubiquitous species include the mite Opiella nova (Oudemans) and the celebrity nematode Caenorhabditis elegans (Maupas).
Rather than expecting to develop a generalized climatic response theory, perhaps we should recognize that species’ responses vary and attempt to categorize the recognizable patterns. Let us take one simple set of examples: our work suggests that among herbivorous insects we can recognize the following.
1. Species whose distribution appears to be restricted by the distribution of their host-plant: their physiology appears sufficiently robust to allow them potentially to survive and reproduce beyond the climatic range of their host plant. Such species are highly abundant at the very edge of their range (e.g. the heather psyllid Strophingia ericae) (Hodkinson et al. 1999). Under climate change scenarios such species will probably track their host plant distribution.
2. Species that appear to be directly limited by their own physiological requirements, largely independent of their host plant, provided that the host remains in a suitable condition to support development. Such species, including Strophingia cinereae Hodkinson, reach their range limit within the range of their host plant (Hodkinson et al. 1999). Their future distribution is best predicted from a knowledge of their ecophysiological tolerances but will not fall outside the range of the host plant. For example, many psyllid species on alder, birch and rosebay willow-herb reach their northern or upper altitudinal limits before those of their respective host plants (MacLean & Hodkinson 1980).
3. Species whose physiology is closely tuned to that of the host plant to ensure close phenological synchrony. The insect species only occur where such synchrony can be achieved and typically the insect has a more restricted range than that of its host. Examples include many psyllids on willow where larval development must be synchronized to that of the catkin, and the moth Coleophora alticolella Zeller, where synchronization with flower development is obligatory (Hodkinson, Jensen & MacLean 1979; Randall 1982b; Hodkinson 1997). Here prediction of future distribution depends on a knowledge of both insect and plant ecophysiology.
In conclusion, I ask that we consider all the available data and do not draw too general inferences from interesting but narrowly limited laboratory experiments. The continuing successful use of subfossil arthropod fragments to provide testable reconstructions of past climates suggests that the links between distribution and ecophysiology should not be lightly disregarded (Atkinson, Briffa & Coope 1995; Coope & Lemdahl 1996; Elias 1997; Coope et al. 1998). As in most controversies, the real answer probably lies somewhere between the extremes.