Thermal Sensitivity and Evolutionary Responses to Climate Change

One useful way to characterize the effects of temperature or other physical factors on the performance or fitness of an individual organism is in terms of a performance curve, which maps environmental conditions onto physiological or ecological performance (Fig. 2). (For convenience I will focus on temperature effects throughout this section, but the basic ideas should apply to most other physical factors.) Performance initially increases with temperature, reaches some "optimal" temperature for performance, then declines rapidly as it approaches upper lethal levels. Frequently one can characterize the performance curve in terms of three parameters: optimal temperature (Z), the temperature at which performance is maximum; performance breadth (a), the breath or width of the performance curve; and the maximum performance (¡imx), the level of performance at

8 Joel G. Kingsolver ^max

ENVIRONMENTAL TEMPERATURE

Figure 2 Thermal performance curves illustrating the relationship between an individual's performance (assumed directly proportional to fitness) and the environmental temperature it experiences. For each individual, there is an optimal temperature (Z) at which performance is maximized (iCa*) • Performance curves are given for two individuals with identical optimal temperature (Z) and maximal performance (R^,»), but that differ in thermal performance breadth (&). Adapted from Huey and Kingsolver (1989).

ENVIRONMENTAL TEMPERATURE

Figure 2 Thermal performance curves illustrating the relationship between an individual's performance (assumed directly proportional to fitness) and the environmental temperature it experiences. For each individual, there is an optimal temperature (Z) at which performance is maximized (iCa*) • Performance curves are given for two individuals with identical optimal temperature (Z) and maximal performance (R^,»), but that differ in thermal performance breadth (&). Adapted from Huey and Kingsolver (1989).

the optimal temperature (Fig. 2). We shall assume that the measure of performance chosen is directly related to fitness.

One natural question is how do the values of these parameters affect the ecological and evolutionary response to climate change (e.g., climate warming)? Suppose we consider two individuals with identical Zand R^ but that differ in performance breadth cr—that is, one is a thermal "specialist" (small a), the other a thermal "generalist"(large o) (Fig. 2). Suppose the environmental temperature 6 is initially the same as Z, but then 8 increases somewhat. Obviously the reduction in performance (and hence the reduction in fitness) of the thermal generalist is less (Fig. 2). Similarly, a population of thermal generalists will suffer a smaller decline in mean fitness than a population of thermal specialists in the face of a small increase in environmental temperature. Clearly, thermal generalists are at an ecological advantage in the face of climate warming.

But how does thermal performance breadth affect the evolutionary response of a population to sustained, directional climate warming? We have recently examined this question (Huey and Kingsolver, 1993), modifying a quantitative genetic model developed by Lynch and Lande (1993) (see Fig. 3). The performance curve identifies the optimal temperature (Z) and thermal performance breadth (cr) of each individual (Fig. 2). Suppose that optimal temperature Z is now a polygenic trait with some constant phenotypic and genetic variation in the population, but that all individuals

Figure 3 Diagram illustrating the effect of thermal performance breadth on a population's evolutionary response to climate warming. Here f(Z) is the frequency distribution of pheno-typic trait Z, the optimal temperature for performance. In each panel, the solid line represents the change in environmental temperature (0), and the dashed line represents the change in the population mean value of Z with time. As time proceeds, a lag develops between the environmental optimum and the population mean phenotype. For populations with large thermal performance breadths (top), this lag will be greater than for populations with small performance breadths (bottom). From Huey and Kingsolver (1993), Fig. 5.

Figure 3 Diagram illustrating the effect of thermal performance breadth on a population's evolutionary response to climate warming. Here f(Z) is the frequency distribution of pheno-typic trait Z, the optimal temperature for performance. In each panel, the solid line represents the change in environmental temperature (0), and the dashed line represents the change in the population mean value of Z with time. As time proceeds, a lag develops between the environmental optimum and the population mean phenotype. For populations with large thermal performance breadths (top), this lag will be greater than for populations with small performance breadths (bottom). From Huey and Kingsolver (1993), Fig. 5.

in the population (with fixed, constant, effective population size) have identical performance breadth (a) and maximum performance (R,njx). Initially the environmental temperature 8 is at the mean optimal temperature Z of the population; 6 then increases at a constant mean rate, but with some stochastic (random) variation. Given this situation, the mean optimal temperature Z of the population will evolve toward increasingly higher values, but will lag behind the environmental temperature (Fig. 3).

If the rate of climate warming is too rapid, the population's lag will become too great, its mean fitness will approach zero, and extinction will occur. Hence one can identify a critical rate of climate change above which population extinction will quickly occur (Fig. 3).

Using this model, we can address how performance breadth affects the critical rate of climate change that a population can sustain. Consider the simplest case in which the genetic variation in optimal temperature Z in the population is constant with time and independent of performance breadth. The model then predicts that the critical rate of climate change will initially increase with increasing performance breadth, quickly reach a maximal value, and then decline with increasing performance breadth (Fig. 4). Thus the model predicts that populations with intermediate performance breadths will be able to sustain the highest rates of climate change— that populations of thermal generalists are more likely to become extinct in the face of rapid climate change. Stochastic variation in climate decreases the critical rate of change and increases the performance breadth at which the rate is maximal, but does not alter the qualitative result (Fig. 4). Interestingly, these results do not depend on the existence of tradeoffs between specialists and generalists (Huey and Kingsolver, 1993).

What produces this apparently paradoxical result? The key once again is the importance of the intensity of selection. For a population of thermal

Figure 4 Theoretical predictions of the critical rate of climate change (in °C/generation) above which population extinction will occur, as a function of thermal performance breadth (in °C) and the amount of stochastic climatic variation. Parameter values are for the case in which genetic variance is independent of time and of performance breadth. Adapted from Huey and Kingsolver (1993), Fig. 8.

Figure 4 Theoretical predictions of the critical rate of climate change (in °C/generation) above which population extinction will occur, as a function of thermal performance breadth (in °C) and the amount of stochastic climatic variation. Parameter values are for the case in which genetic variance is independent of time and of performance breadth. Adapted from Huey and Kingsolver (1993), Fig. 8.

specialists, the change in climate each generation substantially reduces the population's mean fitness, generating strong selection on the population. Given available genetic variation, this intense selection will result in a substantial evolutionary response to selection and will reduce the lag of the population behind the environment (Fig. 4). For a population of thermal generalists, however, the change in climate each generation has a smaller effect on the population's mean fitness, and thus generates weaker selection. Because of the smaller evolutionary response to this weaker selection, the lag of the population behind the environment may become large: if the lag becomes too great, extinction will occur. Thus thermal generalists may be at an ecological advantage, but at an evolutionary disadvantage, relative to populations of individuals with intermediate thermal breadths (see Huey and Kingsolver, 1993, for further details).

This result is sensitive to assumptions about the determinants of the genetic variation in the population. For example, if genetic variation in optimal temperature results largely from a balance of mutation, selection, and drift, one might expect populations of thermal generalists to possess greater genetic variance in Z than populations of thermal specialists; in this case, thermal specialists no longer can sustain higher critical rates of climate change than generalists (Huey and Kingsolver, 1993). However, Burger and Lynch (1995) have recently extended these models to allow for dynamic evolutionary changes in population size and genetic variance over time in the population. Their simulations suggest that, even when genetic variance is determined by a mutation-selection-drift balance and varies over time, the mean time to extinction will be greatest for populations with intermediate performance breadths.

These models, while clearly simplistic, highlight some of the complexities in predicting the evolutionary responses of populations to progressive climate change. I believe the main lesson from the models is this: the evolutionary responses of populations to climate change cannot be predicted solely based on ecological information, but must include some understanding of the genetics of physiological performance traits, about which we know very little. Is optimal temperature a polygenic trait? Do populations of intermediate performance breadth possess less genetic variation for performance traits than populations of generalists? Will specialists and generalists respond in different ways evolutionarily to abrupt step-changes versus progressive changes in climate? These questions have hardly been asked, much less answered, by physiological ecologists; but the answers may be key to predicting the evolutionary responses to climate change.

0 0

Post a comment