Altitudinal variation in behavioural thermoregulation: local adaptation vs. plasticity in California grasshoppers

Authors


  • Present address: H. Dingle, School of Life Sciences, University of Queensland, Brisbane, Australia.

Dr Joerg Samietz, Agroscope FAW Waedenswil, Swiss Federal Research Station, Schloss, PO Box 185, CH-8820 Waedenswil, Switzerland.
Tel.: +41 1 783 61 93; fax: +41 1 783 64 34;
e-mail: joerg.samietz@faw.admin.ch

Abstract

We investigated the adaptive significance of behavioural thermoregulation in univoltine populations of the grasshopper Melanoplus sanguinipes along an altitudinal gradient in California using laboratory tests of animals raised under different temperatures. Trials consisted of continuous body temperature measurements with semi-implanted microprobes in a test arena, and observation and simultaneous recording of behavioural responses. These responses included mobility, basking and orientation of the body axes (aspect angle) towards a radiation source. Mobility and basking are determined by the altitudinal origin of the parental generation and not by the temperature treatments. With increasing altitude, individuals tend increasingly to raise body temperatures via mobility and increased basking. In contrast, body orientation towards the radiation source is influenced by the temperature treatments but not by the altitude of origin. Individuals experiencing higher temperatures during rearing show a lower tendency to lateral flanking. We conclude that body orientation responses are not adapted locally. In contrast other components of the behavioural syndrome that increase body temperature, such as mobility and basking, are adaptive in response to local selection pressure. The thermoregulatory syndrome of these grasshoppers is an important contribution to life-history adaptations that appropriately match season lengths.

Introduction

Temperature is one of the most important environmental variables influencing organisms because it drives all biological rates and functions. Ectotherms in particular are susceptible to the physiological effects of temperature variation (Cossins & Bowler, 1987). Within the temperature range they experience, most terrestrial ectotherms are adapted to relatively high temperature optima (Willmer et al., 2000). The advantages of high body temperatures and selection to adapt accordingly are driven by enzyme physiology, life-history features and immunology. Many enzymes function most effectively over a relatively narrow range of temperatures above which they quickly lose their effectiveness because they unfold. Although cold-adapted enzymes may be effective at low temperatures, their structures also unfold more readily at higher temperatures than homologous enzymes adapted to warm environments (Somero, 1995). Consequently in environments with relatively high temperature variation, the enzymes should be selected for optimum values in the upper range of the temperature spectrum to achieve greater tolerance and to avoid overheating. In ectotherms higher body temperatures also lead to faster development (Willmer et al., 2000), and variation in development time is one of the essential features of life cycle regulation and plasticity in seasonal habitats (Nylin & Gotthard, 1998). Most univoltine species under cyclical environmental conditions, especially in temperate, arctic or alpine climates, are limited by season length with respect to completion of their life cycles and generally must develop relatively fast (Strathdee & Bale, 1998). Finally, ectothermic animals profit from high body temperatures at or near 40 °C as protection against bacterial and fungal infections (Carruthers et al., 1992; Inglis et al., 1997; Arthurs & Thomas, 2001; Kalsbeek et al., 2001).

Thermoregulatory behaviour is one of the key interactions of ectothermic animals with temperature and is the reason why many ectotherms reach body temperatures higher than the average within their habitat. The resulting behavioural syndromes are especially obvious in reptiles (e.g. Peterson et al., 1993) and insects (Casey, 1981; Heinrich, 1993, 1996). Insect behavioural thermoregulation is well studied and has been discussed as an important characteristic of adaptation (O'Neill & Kemp, 1992; Willott, 1997) or as a response to pathogen infections (Carruthers et al., 1992; Inglis et al., 1997; Kalsbeek et al., 2001). The evolution of insect thermoregulatory behaviour in general has also been discussed with respect to such adaptations. Temperature-influencing colour morphs and thermoregulatory behaviour have been found to be correlated in grasshoppers of the genus Tetrix (Forsman, 2000; Forsman et al., 2002). The intra-population differences and accompanying genetic correlations were discussed as coevolution between colour morph and thermoregulatory behaviour (Forsman et al., 2002). Comparative investigations show that the regulatory performance or behaviour of different species may be different in the same habitat (Kemp, 1986; Willott, 1997; Blanford & Thomas, 2000; Sanborn, 2000). But so far as we know, there are no explicit tests of whether thermoregulatory behaviour within a species is a response to specific selection regimes in local populations. It is still not known whether these behaviours play a role in adaptation to local temperature conditions and in the optimization of body temperature or development rates and other life-history parameters.

The distribution of a species along an altitudinal gradient allows the comparative experimental investigation of its thermoregulatory behaviour under different conditions of selection. For such cross population studies of adaptive significance along gradients of ambient temperatures, altitudinal variation has the advantage over latitudinal gradients because the homogeneity of day lengths and sun angle over the season removes these as possibly confounding variables. Here we report on a comparative study of the thermoregulatory behaviour of a California grasshopper that occurs over a wide altitudinal range.

Grasshoppers of the Melanoplus sanguinipes/devastator complex are distributed within northern California from the Central Valley with very high summer temperatures (daily highs around 40 °C) and a long growing season to high altitudes in the Sierra Nevada with short growing seasons and relatively cool summers (daily highs around 20 °C). In spite of the differences in the length of the growing season, all populations are univoltine. The taxonomy of the complex is confusing (Salser, 2003). Earlier authors classified the group into two species, M. devastator Scudder as confined to the Central Valley of California and a few adjacent drier areas, and M. sanguinipes (Fabricius) with a broad range over most of North America including those areas of California not occupied by M. devastator (Gurney & Brooks, 1959; Strohecker et al., 1968). The molecular and morphological studies of Orr et al. (1994) and Salser (2003) and the presence of extensive hybridization (Orr et al., 1994), however, strongly suggest that the complex is a morphologically and physiologically variable single species. There are also no discontinuities in life history or physiological characteristics that would suggest a species boundary (Dingle et al., 1990; Dingle & Mousseau, 1994). We therefore treat the populations studied in this paper as belonging to one species, M. sanguinipes.

Given the differences in summer temperatures and season lengths as a function of altitude over the range of the species in California, the adaptive significance of variation in thermoregulatory behaviour should be apparent when comparing these univoltine populations from different locations. The hypothesis of adaptive differences in thermoregulation is tested in this paper with animals from five altitudes raised under different temperature conditions in the laboratory as first-generation descendants (F1) of wild-caught animals. The individual trials consisted of continuous body temperature measurements with semi-implanted microprobes in a test arena, and observation and simultaneous recording of the main responses of the behavioural syndrome. These responses consist of (i) locomotion that allows for selection of sites of different temperature, (ii) basking expressed as the seeking of highly radiated areas, and (iii) orientation of the body axes (aspect angle) in relation to a radiation source altering the area of the body subject to radiation (right angle corresponds to flanking, zero angle to facing the source). The performance of up-regulation (increasing the body temperature relative to ambient) was quantified by the difference between the body temperature and the operative environmental temperature (defined in Materials and methods) without active regulation. The latter temperature was estimated for each individual and each step of the behavioural tests by a biophysical model.

Among behaviourally thermoregulating insects, grasshoppers are especially useful for studying the adaptive significance of regulation because the components of the behavioural syndrome are not too complex to be analysed under controlled conditions in the laboratory. They also allow a relatively simple biophysical modelling of operative environmental temperatures, and they are relatively well studied with respect to possible causes of selection for high body temperature, life-history adaptations including the constraints of reproductive season (Dingle et al., 1990; Tatar et al., 1997), and pathogen infections (Carruthers et al., 1992; Inglis et al., 1997; Arthurs & Thomas, 2001). Finally, in comparison with other insect groups displaying thermoregulatory behaviour, such as large-bodied Hymenoptera and Lepidoptera, they are almost exclusively ectotherms (Heinrich, 1993; Willmer et al., 2000).

Materials and methods

Population origins

Five populations of M. sanguinipes were used for the experiments. These came from a range of altitudes on the west slope of the Sierra Nevada of California from the edge of the Central Valley at 250 m to near the Sierra crest at 2650 m. The populations, their geographic coordinates and their altitudes are given in Table 1.

Table 1.  Populations of Melanoplus sanguinipes sampled for this study.
PopulationLatitude/longitudeAltitude (m)
Chico (V)39°45′00′′/121°46′49′′ 250
Bridal Veil Falls (S1)38°46′09′′/120°26′52′′1100
Sacramento Camp (S2)38°48′01′′/120°07′01′′2000
Leek Spring (S3)38°38′18′′/120°14′25′′2500
Frog Lake (S4)38°41′27′′/119°59′14′′2650

Laboratory treatments

About 30 mature females from each site were collected in the field in early September and returned to the laboratory. They subsequently laid eggs in cups of moistened sand. Egg pods were transferred on the day after laying to damp vermiculite in individual capped plastic vials. Eggs were allowed to develop at about 27 °C for a week and then were stored at 4 °C for overwintering where some proportion of the eggs was in diapause (Dingle et al., 1990). In March of the following year, the egg pods were removed from the cold room and randomly assigned to high (day/night: 30/25, 36/10 °C) or low (day/night 25/20, 28/10 °C) temperature conditions in growth chambers at about 40% relative humidity. Hatchlings were raised with their siblings for the duration of the first instar to avoid high mortalities occurring when reared alone at this stage. From the second instar to adulthood all grasshoppers were raised individually in 10 × 10 × 8 cm plastic boxes with screened tops. Fresh organically grown commercial lettuce was supplied daily as a source of food and moisture and was supplemented by wheat bran as necessary. Seventy-eight females from the high temperature treatments and 70 from the low temperature treatments were analysed for their thermoregulatory behaviour.

Tests of thermoregulatory behaviour

An observation cage was designed that allowed simultaneous recording during individual trials of body temperature, exact ambient temperature, and aspects of the thermoregulatory syndrome including mobility, basking/shade seeking and the precise angle of flanking/facing the radiation source. The cage consisted of a transparent plexiglass box (30 × 30 × 45 cm) with an arena sloped to the back of the cage at a 15° angle. A radiation source (white light tungsten-halogen lamps at 500 W, colour temperature 3100 K) at a distance of 30 cm illuminated the cage from the rear at an angle of about 20° above the arena. Vertical stripes of aluminum tape (5 cm wide at a separation of 5 cm) served to provide the cage with areas of different radiation intensity from which the test individual could choose by basking or shade seeking. In this way the parts of the test arena where the insect could walk provided 50% low radiation area (radiation flux 3 W m−2), 35% medium radiation (60 W m−2), and 15% high radiation (300 W m−2).

The test arena within the cage was made of white polyurethane foam (1 cm thick) to reduce the effects of conduction to a negligible level. Holes in the foam of 1 mm diameter arranged in a grid 5 mm apart served to draw air from a regulated fan in a slow laminar flow through the cage from the top down. With this laminar air flow a homogeneous circulation with a minimum convection bias was achieved. The observation cage was placed in a climate controlled chamber (Percival I-36VL, Percival Scientific, Perry, IA, USA) with an observation window in order to apply a gradient of ambient temperatures to the insects during a trial.

Each trial ran over a linear temperature gradient of approximately 7–47 °C within 1 h. In addition to the external control provided by the climate chamber, the ambient air temperature in the arena was controlled by a type-T thermocouple (copper-constantan) mounted 1 cm above the base in a low radiation area. The body temperature of the insect was recorded by a custom made type-T thermocouple microprobe constructed in our laboratory (35 μm copper/constantan wire) implanted from behind just beneath the pronotum. The grasshoppers equipped with probes moved freely in the observation cage and did not show any signs of disturbance by the fine wires nor any cleaning behaviour to remove probes or wires. Through the glass window of the climate chamber the observer registered the behavioural components during the trial using a keyed panel and a dial for body orientation towards the radiation source. The resulting data from the panel were transferred to a data acquisition system and were stored in a computer together with the temperature data at a temporal resolution of 4 Hz. Each individual was used for one trial only.

To achieve the most accurate temperature measurements, the type-T probes and all connectors were made of high precision wire material (Omega Engineering, Stamford, CT, USA) and an appropriate high resolution data acquisition system with an isothermal aluminium plate was used for recording (Data-Shuttle DS-16-8-TC, Omega Engineering). The precision of the temperature measurements was better than 0.1 °C (error <±0.05 °C).

Biophysical model

To assess any active increases in body temperature (Tb) relative to environmental conditions per se, differences in body temperatures were measured relative to temperatures expected in the absence of thermoregulation. To accomplish this, a biophysical model (after Samietz & Köhler, 1998) was applied to simulate the body temperatures of each individual being tested, assuming no active thermoregulation. To approximate these ‘passive’ body temperatures or operative environmental temperatures (Te), the following simplified equation of heat balance (Baumgärtner & Severini, 1987) was used. The equation is based on the assumption that the core temperature of the body is equal to the surface temperature, and the influences of convection and transpiration are ignored.

image(1)

where Ta is the ambient air temperature, ΦR is the radiation flux density (W m−2), a is the absorption coefficient, A is the body cross-sectional area towards the radiation source, S is the external body surface area, b is the body diameter, and k is the thermal conductivity of the air (0.026 W m−1 K−1). Following Chappell & Whitman (1990) and Anderson et al. (1979), the absorption coefficient, a, was set to 0.7 for the grasshoppers under study. The body of a grasshopper female was approximated by a rotational ellipsoid with half the body length as the semi-major axis q. The body cross-sectional area towards the radiation source as a function of the orientation angle α is then:

image(2)

The corresponding derived projection of the semi-major axis is:

image(3)

Passive heat gain under a specified condition of radiation assumes a randomly chosen α between 0 and 90°. The resulting operative environmental temperature, Te, was estimated from eqns 1 and 2 as follows:

image(4)

where Ep*(α) is the expected value of the derived projection p* with respect to α:

image(5)

The probability P(α) is constant due to the required uniform random distribution to describe passive heating. The quotient in eqn 5 is thus:

image(6)

Substituting eqn 3 into the right hand side of eqn 6 leads to:

image(7)

The integral in eqn 7 was numerically solved for each individual grasshopper by applying a 32-point Gauss formula and using the individual values for the semi-major and semi-minor axes of the ellipsoid model (half body length; half maximum body diameter). With the results of eqn 7, eqn 4 served to calculate the operative environmental temperature for each individual in the three different areas of radiation intensity in the cage (see above). Proportional to the areas of the intensities the weighted mean Te in the observation cage at every temperature increment was calculated for each individual.

Model validation

The approach provided here is validated by additionally modelling the each individual body temperature for each time step of the single trials (0.25 s) according to the observed behaviour in relation to the radiation source and the physical properties (ambient temperature, radiation) at the actual recorded position in the observation cage. The integrated difference between modelled and measured body temperature of the grasshoppers is subsequently used as a measure for the validation of the model, i.e. the closer the values to zero, the better the model matches the measured body temperatures. Significant deviations of the integrated differences from zero were tested by one-sample t-tests and Wilcoxon signed rank tests according to the distribution of the data. Possible influences of the altitudinal origin on the model deviation were tested by one-way anova and Kruskal–Wallis tests.

Data analysis

Each of the trials is divided into three phases according to the possibility for a behavioural response and the direction of regulation of body temperature (Fig. 1). Phase 1 extends from the start of the trial until the first active movement of the test animal occurs and does not involve any active behavioural regulation. The upper limit of this phase is defined by the minimum activity temperature above which phase 2 follows as a period of behaviourally increasing the body temperature above the operative environmental temperature and towards an optimum. The upper limit of phase 2 is defined by the maximum voluntarily tolerated temperature that is recorded by the observer when the test animal first avoids further heating during the trial by an active response (Fig. 1). Above this limit phase 3 marks the range of behaviourally regulating the body temperature below the environment.

Figure 1.

Temperature curves of a test run of a grasshopper individual in the observation cage used in this study. The analyses of the temperature data and the behavioural observations are subdivided into three phases with their corresponding processes. These phases are: 1, below the minimum activity temperature without any behavioural response; 2, between the activity minimum and the maximum voluntarily tolerated temperature with up-regulation above the operative environmental temperature; and 3, above the tolerated maximum with down-regulation in relation to the environment.

Phase 2 is relevant for adaptations fostering high body temperature relative to ambient, and defining it is the primary aim of the current investigations. The regulatory performance for this up-regulation phase was quantified by the integrated temperature excess of the body temperature above the operative environmental temperature calculated from the biophysical model. The resolution for the Gauss-integration of measured data over time was 4 Hz.

The influence of altitudinal origin and temperature treatment on regulatory performance, as well as on the components of the behavioural syndrome during phase 2, was analysed by two-way anova. We used this procedure in preference to an alternative analysis of covariance with altitudinal origin as the covariate as we could not exclude a possible noncontinuous effect of altitude. In order to normalize the data, all rates between 0 and 1 (fractions of time spent mobile and basking, flanking score) were arcsine square root-transformed prior to anova. Differences between the groups within the factor combinations were analysed by post-hoc tests applying Fisher's protected least square difference (Fisher's PLSD).

Results

Model validation

Over all phases of the trials the model shows only a marginal overestimation of on average 0.28 °C above the actual body temperatures measured (Fig. 2). This deviation however is not significant in the most relevant phase 2 (t-test, t80 = 0.79, n.s.), in phase 3 (Wilcoxon signed rank test, W69 = 182.5, n.s.) and over the total range of observation (t-test, t80 = 1.13, n.s.). Only in phase 1 below the relevant start of the grasshoppers’ regulation activity the model shows a significant overestimation (Wilcoxon signed rank test, W80 = 804.5, P < 0.001). The altitudinal origin has no influence on the deviation of the modelled temperatures from the temperatures measured during the trials neither over the entire range (one-way anova, F4,138 = 1.91, n.s.) nor for the single phases of analysis (phase 1: Kruskal–Wallis test, H4 = 3.77, n.s.; phase 2: one-way anova, F4,138 =1.86, n.s.; phase 3: Kruskal–Wallis test, H4 = 4.37, n.s.).

Figure 2.

Integrated difference between modelled and measured body temperature of the grasshoppers studied as a measure for validation of the biophysical model (phases as in Fig. 1). Body temperature was modelled for each time step of the individual run (0.25 s) according to the observed behaviour and the physical properties at the position in the observation cage. Box: median, 25th and 75th percentile, dotted line: mean, error bars: 10th and 90th percentile. Significant deviations from zero (*) were tested by one-sample t-tests (total, phase 2) and Wilcoxon signed rank tests (phases 1 and 3) according to the distribution of the data.

Regulatory performance

Both the altitudinal origin of the grasshopper females (factor 1) and the temperature treatment (factor 2) had a significant influence on the integrated difference between body temperature, Tb, and the operative environmental temperature, Te, during the up-regulating phase (phase 2) of the trials (two-way anova; factor 1: F4,138 = 28.8, P < 0.001; factor 2: F1,138 = 11.8, P < 0.001). Thus grasshoppers from higher altitudes and reared at lower temperatures actively increased Tb more during this period (Fig. 3). The analysis also revealed a significant interaction of altitude and rearing temperature, indicating that the temperature treatment acted differently on the performance of the test individuals from different altitudes (two-way anova, interaction factors 1 and 2, F4,138 = 4.3, P < 0.05). Within the low temperature treatment the regulatory performance continuously increases with altitude (Fig. 3). For grasshoppers reared at high temperature, the trend of increasing regulatory performance with altitude is less consistent as shown in Fig. 3. The highest average difference, 6.86 ± SE 0.59 °C, within a group is reached in the individuals from the highest altitude (S4, N = 12) raised under low temperature. The individuals from the same altitude raised under high temperature (N = 16) showed an average temperature excess of 4.55 ± 0.53 °C. The lowest average difference within the group, 0.05 ± 0.42 °C, was reached in the individuals from the lowest altitude raised under low temperature (V, N = 18). The individuals from the same origin raised under high temperature (N = 33) showed an average temperature excess of 0.59 ± 0.45 °C (Fig. 3). The post-hoc analysis by Fisher's PLSD test at the 5% significance level revealed differences between all combinations of low altitudes V and S1 against the high altitudes S2, S3 and S4. There were no significant differences between V and S1 and within S2–S4 at the 0.05 level. With respect to the temperature treatment, the up-regulation performance over all altitudes was 1.46 °C higher in the grasshoppers raised under low temperatures.

Figure 3.

Integrated difference between body temperature and operative environmental temperature without active regulation (biophysical model) as a measure of the thermoregulatory performance (mean ± SE) in the range between minimum activity temperature and maximum voluntarily tolerated temperature (phase 2). Populations are raised in the F1 generation under high and low temperature treatments. Origins are along an altitudinal gradient in California between 100 m (Chico, V) and 2620 m (Frog Lake, S4) as indicated in Table 1.

Behavioural syndrome

Mobility

As a first component of the behavioural syndrome, we analysed mobility (Fig. 4a). This was the time spent actively moving through the test arena to different areas of exposure. The altitudinal origin of grasshopper females (factor 1) had a significant influence on mobility in the up-regulating phase (phase 2) of the trials with greater mobility at higher altitudes, whereas the temperature treatment (factor 2) had no influence (two-way anova; factor 1, F4,138 = 7.1, P < 0.001; factor 2, F1,138 = 1.8, n.s.). There was no significant interaction of the two factors, indicating that the temperature treatment yielded the same pattern of effect across the different altitudes (two-way anova, interaction factors 1 and 2, F4,138 =0.52, n.s.). The highest average mobility is reached by the individuals from the mid-altitude S2 population. The individuals from the lowest altitudes (V, S1) show the lowest levels of spatial activity (Fig. 3). A break between the two low and three high altitude populations is obvious when looking at the average values in Fig. 4a. According to the post-hoc analyses there were no significant differences between the mobility of the individuals from V and S1 and among those from S2, S3 and S4, whereas all the remaining combinations between the low altitude V or S1 vs. the high altitude S2, S3 and S4 reveal significant differences with more activity at the higher altitudes (Fisher's PLSD test, 5% significance level).

Figure 4.

Main components of thermoregulatory behaviour in the range between minimum activity temperature and maximum voluntarily tolerated temperature (phase 2). Populations as in Fig. 2. (a) Mobility and (b) basking are indicated as integrated proportions of time spent performing each behaviour during the tests. (c) Flanking score is plotted as the mean individual quotient of actual body angle (between 0°and 90°) and 90°, i.e. perpendicular orientation with maximum body area exposed to the radiation source.

Basking behaviour

Grasshoppers from higher altitudes spent more time in areas of high radiation in the up-regulating phase (phase 2) than individuals from lower altitudes (two-way anova, factor 1, F4,138 = 17.3, P < 0.001). Temperature treatment had no influence (two-way anova, factor 2, F1,138 = 0.46, n.s.). There was no significant interaction of the two factors (two-way anova, interaction factors 1 and 2, F4,138 = 2.18, P = 0.074). With increasing altitude of origin time spent basking increases continuously in the individuals from the low temperature treatments (Fig. 4b). There was less of a trend at high temperature, but in both treatments the highest mean values are reached in the sample from the highest altitude, while the lowest are recorded from the lowest altitude (Fig. 4b). The post-hoc analyses show that there were no significant differences between the basking behaviour of the individuals from V and S1 and among those from S2, S3 and S4, whereas all remaining combinations between the low altitudes V and S1 against any high altitude population were significant (Fisher's PLSD, 5% significance level).

Flanking behaviour

The orientation of the body axes relative to the radiation source in the up-regulating phase of the trials was not influenced by altitude, but the rearing temperature was found to have a highly significant influence on the flanking behaviour (two-way anova, factor 1, F4,138 = 0.67, n.s.; factor 2, F1,138 = 28.3, P < 0.001). There was no significant interaction of the two factors, indicating that the temperature treatment yielded the same effects across the altitudinal range examined (two-way anova, interaction factors 1 and 2, F = 1.05, n.s.). In each population individuals reared at low temperature displayed a greater tendency for lateral basking (Fig. 4c). There was no significant difference in average values between any of the samples (Fisher's PLSD, 5% significance).

Discussion

Our results and those of Salser (2003) clearly indicate the physiological and life-history adjustments these univoltine grasshoppers have made to altitudinally varying season lengths and prevailing temperatures. In this study, we have demonstrated that grasshoppers from higher altitudes (and therefore that experience shorter seasons) actively increase their body temperatures the most over ambient in the range between the low temperature activity minimum and the maximum tolerated temperature (Figs 1 and 2). They do this by actively moving into positions where they can absorb heat (in the wild this would be in the sunlight) and spending more time basking in these areas. These behavioural responses are adaptations to local selection regimes, with selection favouring up-regulation of body temperature in the colder habitats of higher altitudes. Flanking, by which a grasshopper exposes maximum surface area to a heat source, is not influenced by altitude but is influenced by rearing temperature. Grasshoppers reared in cooler conditions display a greater tendency for flanking independently of altitude, suggesting this is a more general adaptive response not much influenced by local selection. Further experiments and analysis will be necessary to determine the physiological basis for the behavioural differences arising in response to rearing temperatures.

These physiological and behavioural responses of California M. sanguinipes are important in the context of the life histories of these populations (Salser, 2003). High altitude grasshoppers develop faster at all temperatures above the threshold for development and are thus able to complete their life cycles during the compressed season present in the higher reaches of the Sierra Nevada (Dingle et al., 1990; Salser, 2003). This rapid development is preceded by an egg diapause that occurs in the final stage of embryonic development so that there is no hatching delay once snows melt in the spring (Dingle & Mousseau, 1994). Development to adulthood is further accelerated by short days (Salser, 2003). The combination of temperature and photoperiod responses ensures that the adult stage is reached no matter what the altitude or season length. These adaptations are accompanied by two other responses. First, at low altitudes some individuals may insert an extra instar between egg and adult and reach a larger size; within but not among populations larger size in females leads to greater egg production (Salser, 2003). The smaller high altitude grasshoppers can evidently compensate for their smaller size with respect to reproduction, because they do not produce reduced numbers of eggs. Secondly, the more rapidly developing high altitude animals also accelerate senescence (Tatar et al., 1997). What is of particular interest about the results reported here is that they demonstrate that, in addition to developing more rapidly per se at all rearing temperatures, high altitude grasshoppers up-regulate body temperature to higher levels than grasshoppers from lower down and so are capable of accelerating development still further. They are thus likely to be near optimally adapted to the short seasons of higher altitudes. The cost seems to be the foregoing of the ability to produce an extra instar, with its attendant gain in size and reproductive capacity, although the reproductive costs of smaller size have also apparently been mitigated by selection.

Field studies of California M. sanguinipes along an altitudinal gradient had suggested that the degree-days accumulated by high altitude and certain coastal populations were insufficient to allow completion of development to adult eclosion (Dingle et al., 1990). Measurements of surface soil temperatures, influenced by infrared absorption and therefore during the day higher than air temperatures, suggested that grasshoppers would need to remain near the ground surface to accumulate enough degree-days, assuming no thermoregulation. The results reported here show why they are able to accumulate enough degree-days to reach adulthood and reproduce: they are proficient at behavioural thermoregulation. This ability to regulate is certainly a contributor to and may be necessary for the invasion of low temperature and short season habitats.

The fact that terrestrial ectotherms, especially those from cool climates, regulate internal temperature by behavioural means is well established. Many studies confirm that grasshoppers regulate their body temperature above ambient to reach a relatively high thermal optimum (Pepper & Hastings, 1952; Anderson et al., 1979; Chappell, 1983; Gillis & Possai, 1983; Kemp, 1986; Gillis & Smeigh, 1987; Whitman, 1987; Carruthers et al., 1992; Lactin & Johnson, 1996). It is also generally known that developmental processes in grasshoppers are temperature dependent (e.g. Putnam, 1963; Gage et al., 1976; Whitman, 1986; Kemp & Dennis, 1989). Only a few studies, however, have yet tried to link behavioural thermoregulation to developmental rates or reproductive parameters (Begon, 1983; Whitman, 1986, 1988; Carruthers et al., 1992). Still these studies do not clearly separate the passively achieved operative environmental temperature from the heat gain by active thermoregulation. In the genus Tetrix colour morphs of the individuals could be related to specific increase in body temperature while basking and to fitness and life-history consequences (Forsman, 1997, 1999a,b, 2000, 2001; Forsman et al., 2002). These thorough studies also showed that the thermoregulatory behaviour of females correlates well (both phenotypically and genetically) to the colour morph and accompanying trade-offs imply the coevolution of the characters as intra-specific strategies (Forsman et al., 2002). Willott (1997) analysed the thermoregulatory performance of several species of grasshopper in the field and discussed its importance with respect to habitat. The findings were related to life-history parameters of the same species (Willott & Hassall, 1998), and the thermoregulatory ability was discussed as an important influence on habitat use and fitness. Nevertheless, it was not shown yet explicitly that thermoregulatory behaviour evolved as a consequence of local selection pressure to regulate body temperature.

In our study we have not only examined thermoregulatory behaviour but also analysed several components simultaneously. The differentiation of responses along an altitudinal gradient, revealed by ‘common garden’ rearing of grasshoppers, indicates genetically based complex adaptation in response to local selection. Furthermore, in combination with the studies of Dingle et al. (1990), Dingle & Mousseau (1994), Orr (1996), and Salser (2003), this study indicates that thermoregulatory behaviour is an important contributor to life-history adaptations that match grasshopper phenologies to variation in season length.

Acknowledgments

We especially thank Wesley Weathers for allowing us to use one of his environmental chambers and Audrey Chang and Davina Wu for their help with rearing grasshoppers. Günter Köhler is gratefully acknowledged for his support and many fruitful discussions. Mike Ritchie, Anton Stabentheiner and anonymous referees gave valuable comments on the draft paper. J.S. was funded by the Deutsche Forschungsgemeinschaft (Grant SA 770/1-1). Earlier work on grasshopper life histories was supported by grants from the US National Science Foundation to H.D.

Ancillary