Physiological diversity: listening to the large-scale signal
Provide feedback or get help You are viewing our new enhanced HTML article.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Countergradient variation, or the conservation of the rate of a temperature-dependent physiological process in the face of environmental temperature change (Levins 1969; Hazel & Prosser 1974), has been widely documented in ectotherms. Indeed, it is generally accepted as a characteristic especially of the growth rates of several ectotherm taxa (for discussion see Conover & Schultz 1995; Gotthard et al. 2000). However, in the context of metabolic rate variation, the concept is much more polemical. Indeed, whether there is empirical evidence for metabolic cold adaptation (MCA) and whether there are theoretical grounds for considering it an adaptation have been widely debated since the concept was first formulated (Fox 1936; Clarke 1993; Huey & Berrigan 1996; Chown & Gaston 1999; Steffensen 2002).
Apart from the theoretical concerns (Clarke 1993), much of the controversy surrounding MCA has been centred both on methodological problems, that might confound the outcome of investigations (Holeton 1974), and on the extent to which the studies that have been done can be generalized more broadly (for insects, see discussion in Chown & Gaston 1999). While reiterating some of these older arguments, Hodkinson (2003) has added a new dimension to the debate by suggesting that large-scale studies, specifically the one by Addo-Bediako, Chown & Gaston (2002), but clearly implying others (e.g. Scholander et al. 1953; Clarke & Johnston 1999), obfuscate the arguments because they require ‘huge simplification and extrapolation from limited data sets …’. Using a variety of examples, Hodkinson (2003) argues that in arthropods metabolic responses are so subtle in their variation and so specific to species, stages and times of the year that large-scale interspecific comparisons are likely to be hopelessly confounded and therefore add little to understanding of MCA. Here, we first show that Hodkinson's (2003) primary concern about the subtleties of species-specific responses is an old one, which, when taken to its extremes, hampers progress in biology. We then deal with the specific issues he raises.
ten million physiological responses?
Throughout his response to our investigation, Hodkinson (2003) repeatedly emphasizes the considerable variation that can be found in arthropod metabolic responses, concluding, inter alia, that average animals do not exist and that the value of our approach is questionable because metabolic processes are ‘highly subtle’ and ‘species-specific’. Thus, Hodkinson (2003) is of the view that any large-scale interspecific comparison of metabolic rates will be confounded by the bewildering complexity that constitutes arthropod physiological responses. In contrast, we sought to determine whether, among the variation that is associated with all levels below that of the species in the physiological hierarchy (Spicer & Gaston 1999), there are interspecific patterns that will be revealed by averaging out the details in a broad-scale analysis.
Tension between these two approaches, or between those who emphasize the seemingly infinite variety of life and those who seek patterns and rules, is not new to biology (e.g. Feder 1987). Indeed, the existence of, and interplay between, these divergent views has recently been thoroughly explored by Lawton (1992, 1999) in the context of population dynamics and community ecology. Lawton (1992) pointed out that although some ecologists celebrate the complexity of population dynamic patterns (with each species, and perhaps population, being different – characterized as ‘10 million kinds of population dynamics’), this variability amounts to little more than variation around a few major themes. Moreover, while acknowledging the importance of small-scale, mechanistic approaches to understanding the diversity of life that is contingent on local variation, Lawton (1999) argued cogently that if communities are to be understood, studies must be undertaken at broader scales. Such large-scale approaches reduce the importance of the details, making it more straightforward to determine whether there are bold patterns in nature that can be explained by a relatively few underlying rules (Lawton 1999).
It is our view that large-scale studies, of the kind advocated by Lawton (1999), are also critical for understanding physiological variation, its environmental determinants, and the implications thereof for understanding patterns in the distribution and abundance of animals (Spicer & Gaston 1999; Chown, Addo-Bediako & Gaston 2002). Only by adopting this approach is it possible to determine whether there are bold patterns in physiological variation, many of which ecologists presume must exist. For example, the climatic variability hypothesis, a primary explanation for the Rapoport effect (the increase in range size with latitude; Gaston, Blackburn & Spicer 1998) is based on the assumption that the range of species thermal tolerances increases with latitude. Only large-scale analyses can test this hypothesis (Addo-Bediako, Chown & Gaston 2000). Similarly, large-scale studies have revealed that differences in the species richness and range size patterns of the Southern and Northern Hemispheres (Blackburn & Gaston 1996) are also reflected in physiological variables (Lovegrove 2000; Addo-Bediako et al. 2002). This is probably a consequence of the very different abiotic conditions characterizing the hemispheres (Gaston & Chown 1999), which in turn have probably had profound influences on the diversity in each via the physiological and life-history responses of the organisms inhabiting them. Thus, large-scale comparative approaches (see also West, Woodruff & Brown 2002) have much to contribute to evolutionary physiology (Feder, Bennett & Huey 2000), and to the reversal of fortune of environmental physiology, a discipline that in the 1980s was in danger of becoming an anachronism (for further discussion see Feder 1987; Feder & Block 1991; Spicer & Gaston 1999). In consequence, we eschew narrow approaches of the kind adopted by Hodkinson (2003). Rather, we are of the view that a variety of approaches, ranging from mechanistic molecular studies (Storey 2002), through intraspecific comparisons (Gaston & Spicer 1998), to broad-scale comparative analyses, are essential for understanding the responses of organisms to the environment and the implications thereof for biodiversity as a whole.
Having made the case for a pluralistic approach to ecological and evolutionary physiology, which acknowledges the importance of large-scale, macrophysiological studies, the question remains whether the ‘noise’ associated with the variation described by Hodkinson (2003) is sufficient to confound the MCA signal we originally detected, and whether the comparative approach adopted was flawed. To recap, Addo-Bediako et al. (2002) detected three major patterns. First, we demonstrated weak, though significant, support for MCA in an interspecific analysis of the metabolic rates of free-living terrestrial insects from around the globe. Second, we showed that the slope of the rate–temperature relationship tends not to decline at lower environmental temperatures, but rather is invariant in Southern Hemisphere species, and increases with a decline in temperature in Northern Hemisphere species. Third, we showed that much of the variation in metabolic rate is partitioned at higher taxonomic levels, providing some support for a genetic basis for variation in metabolic rate. Importantly, the second and third patterns are broadly supported by other findings in the literature. Variation in physiological and life-history variables tends to be partitioned at higher taxonomic levels in both ectotherms and endotherms (Read & Harvey 1989; Chown et al. 2002). Likewise, interspecific and intraspecific investigations support the idea that the slope of rate temperature curves tends to increase with a decline in temperature in Northern Hemisphere species (Scholander et al. 1953; Gotthard et al. 2000). Therefore, besides the MCA signal, there are bold patterns that emerge from our analysis (Addo-Bediako et al. 2002), which agree with many other studies, and are not therefore artefacts of confounding variation in the data. The question, then, is whether the MCA signal itself is both confounded and based on a flawed protocol.
The second of these issues is the most straightforward to address. Hodkinson (2003) failed to read carefully the description of the methods employed for the comparative analysis. (This was not a meta-analysis as claimed by Hodkinson 2003–Garland & Adolph 1994 and Arnqvist & Wooster 1995 clarify the distinction.) Hodkinson (2003) claims that ‘to compare animals living in different climatic regimes metabolic rates corrected for body mass were standardized to 25 °C by extrapolation from measured data using a Q10 value of 2’. We did no such thing. We used original data, converted to microwatts (not µW g−1) assuming a respiratory quotient of 0·84. We then used a generalized linear (Type III) model to examine the effects of log10 body mass, trial temperature, environmental temperature, experimental method and wing status, and their interaction terms on log10 metabolic rate. Only when investigating collinearity of the independent variables and the partitioning of variance in metabolic rate did we use single data points (i.e. not multiple data points gathered at different temperatures for each species) of metabolic rates at 25 °C or as close to 25 °C as possible. It is only here that we employed a Q10 of 2 to convert those values that were not measured exactly at 25 °C to an approximation for that temperature. Therefore, Hodkinson's (2003) arguments regarding the appropriateness of the Q10 value and its change with temperature are incorrect. Moreover, the range of temperatures characteristic of the data used in the collinearity and variance analyses was small (within 5 degrees of 25 °C), making temperature-associated variation in Q10 (something we are well aware of given its prominence in the literature – see Willmer, Stone & Johnston 1999) again of little relevance.
Hodkinson (2003) describes several sources of variation that he thinks are likely to confound both our analyses and conclusions. First, he takes issue with our use of mean temperatures as measures of environmental temperature, implying that we had not given the use of mean temperatures much thought. In fact, we explicitly discussed the likely problems associated with the use of mean temperature, and also pointed out why we chose not to use some other measure (Addo-Bediako et al. 2002: 335). We have little to add to our previous discussion, which covers the points raised by Hodkinson (2003). However, we do note that we dealt explicitly with changes in the slope of the rate–temperature curve as part of the response of insects to low-temperature environments (Addo-Bediako et al. 2002: 336), which Hodkinson (2003) ignores throughout his discussion of our work.
The likely effects on insect metabolic rates of movement during experiments, and the method of locomotion used by species (flying vs. flightless) have been widely discussed (e.g. Lighton & Fielden 1995; Reinhold 1999). Indeed, contrary to the unsubstantiated statements about recordings of the activity state of animals made by Hodkinson (2003), in most studies of insect standard metabolic rates the authors routinely determine whether the organisms are active, study them at an appropriate point in their diel cycle, and discard data where there is activity (e.g. Bartholomew & Casey 1977; Lighton 1988; Vogt & Appel 1999). Moreover, modern flow-through respirometry systems (or continuously recording calorimeters) enable researchers to select only the lowest rates, thus ensuring sound estimates of the standard metabolic rate of the animals (Lighton & Fielden 1995). In our analyses, we included a variable indicating whether closed or flow-through respirometry systems were used, so taking the likely effects of movement into account. We also included wing-status as a variable in the analysis, so taking the effects of locomotion method into account. Both variables had a significant effect on metabolic rates and the analyses were therefore repeated for each of the four combinations of approaches. In all cases, we found the same results, including a weak, though significant effect of environmental temperature, supporting MCA (Addo-Bediako et al. 2002: Table 2). In any case, much of Hodkinson's (2003) argument about the difficulty of measuring metabolic rates in small arthropods is irrelevant: we did not investigate mites or springtails, for which accurate measurements of standard metabolic rate might well be difficult to obtain, nor did we use 25 °C as a standard temperature (see above).
In our analyses, we also avoided using data from insects that had long been in culture, from immature stages (we included only adults), and from studies where gas concentrations had been manipulated (Addo-Bediako et al. 2002: 333). While we agree that it is valuable to understand variation in metabolic rates associated with stage, age and laboratory culturing (and selection), and with variation in extreme environmental conditions, this variation is irrelevant to our study either because we excluded it, or because we did not seek explicitly to address it. In the latter case, we were quite specific about what we regard as metabolic cold adaptation, and in our study this did not include responses to anoxia, hypercapnia or other stressful conditions such as freezing. Although we do not deny that metabolic rates vary in response to such conditions, our study did not aim to address such variation. In this vein, we are also well aware of the fact that feeding increases metabolic rate (known as specific dynamic action – for an insect example see McEvoy (1984), and for general discussion see Willmer et al. (1999)). Most investigators clarify how their animals have been treated prior to the investigation (usually fasted) (e.g. Bartholomew & Casey 1977; Nylund 1991; Davis, Chown & Scholtz 1999), and we relied explicitly on their knowledge of the insects they were investigating for determining whether the period involved was sufficient to avoid measuring the costs of digestion. Nonetheless, if specific dynamic action were largely responsible for our results we would have to presume that biologists working in the tropics are always scrupulous about fasting their experimental animals, while those working in temperate regions are not, so resulting in higher metabolic rates in animals from the latter. This is plainly absurd. Rather it seems much more likely that metabolic rate variation owing to these other factors adds noise to the analysis, but is unbiased with regard to MCA.
Likewise, we are fully aware that metabolic rate shows both seasonal and geographical variation. Indeed, if it did not, we would have expected much better fits of our models to the data. Not only did we cite some of the examples used by Hodkinson (2003) (e.g. Nylund 1991), but we have reviewed at length the causes of variation in insect metabolic rates (and variation in the physiological characteristics of animals in general) (Chown & Gaston 1999; Spicer & Gaston 1999), and we have previously examined seasonal and geographical variation in metabolic rates ourselves (Davis et al. 2000). However, we are of the view that to understand physiological variation, and especially large-scale variation (which, incidentally, is what the interspecific MCA debate is all about), then it is important to ‘try to see where the woods are, and why, before worrying about the individual trees’ (Lawton 1999).
Science's primary goals are to explain the physical world around us and to generalize these explanations to the extent that predictions can be made (Ziman 1978; Casti 1991). While the degree of success in the search for generalities, and in making useful predictions, varies between the sciences, and between disciplines in biology, whether there is anything interesting to explain and any likelihood of prediction depends fundamentally on whether we concede that there are regular patterns in nature (Feder 1987; Lawton 1999). It is our view that there are regular patterns in physiological variation, that these manifest themselves at a variety of hierarchical levels, and that at a macrophysiological level, documentation of patterns, comparative investigation of their underlying mechanisms, and exploration of their broader implications have much to offer biology. Indeed, without such a large-scale view, explaining why there is interspecific physiological variation, understanding how it evolves, and demonstrating how and why it is relevant to animal life-history responses and variation in biodiversity would be very difficult indeed (see also Ricklefs & Wikelski 2002). Censuring this approach by dwelling on the bewildering complexity of life is not only parochial, but also constitutes a return to case-by-case documentation of the variety of life for its own sake, an approach that will not benefit modern physiological research.
Melodie McGeoch is thanked for comments on an earlier version of this manuscript. S.L.C. is supported by the National Research Foundation of South Africa (GUN 2053570).