Rapid cold-hardening (RCH) is a unique form of phenotypic plasticity which confers survival advantages at low temperature. The fitness costs of RCH are generally poorly elucidated and are important to understanding the evolution of plastic physiology. This study examined whether RCH responses, induced by ecologically relevant diel temperature fluctuations, carry metabolic, survival, or fecundity costs. We predicted that potential costs in RCH would be manifested as differences in metabolic rate, fecundity, or survival in flies which have hardened versus those which have not, or flies that have experienced more RCH events would show greater costs than those which have experienced fewer events. One group of flies cooled to 10°C for 2 h for 11 consecutive days experienced daily RCH (Hardened), whereas the other group exposed to 15°C for the same 2-h period each day formed a Control group. Hardened flies had higher survival at –5°C for 2 h than control flies (69 ± 9% vs. 44 ± 19%, P = 0.04). Hardened flies showed no metabolic or fecundity costs, but had reduced average survival (P = 0.0403). Thus, a major cost to repeated low temperature exposures in Ceratitis capitata is through direct mortality caused by chilling injury, although this appears not to be a direct cost of RCH.
The aim of this study is to examine the relative fitness costs and benefits of the RCH response and fluctuating low temperatures. Specifically, we examined whether ecologically relevant diel temperature fluctuations, which induce typical RCH responses, carry a metabolic, survival, or fecundity cost in the Mediterranean fruit fly, Ceratitis capitata Wiedemann (Diptera: Tephritidae). Previous work has shown that C. capitata has a classic RCH response, similar in magnitude, speed of development of response, and duration of RCH survival benefits to other Diptera (Nyamukondiwa et al. 2010). We predicted that any potential costs in RCH would be manifested as either (1) a difference in metabolic rate, fecundity, or survival in flies that have hardened versus those that have not hardened, or (2) flies that have experienced more hardening events over time would show greater costs than those which have hardened on fewer occasions relative to similar-aged control flies. Specifically, we measured the direct survival benefits of RCH on acute low temperature survival in Hardened versus Control flies. In addition, we investigated whether there is a metabolic cost to RCH. Finally, we investigated the survival and fecundity of flies in Hardened and Control groups under benign conditions to estimate the potential fitness costs associated with RCH.
Materials and Methods
STUDY ANIMALS AND TREATMENT CONDITIONS
Larvae of C. capitata were obtained by shipping in insulated containers from Citrus Research International, Nelspruit, South Africa. The culture conditions are a reasonable reflection of wild populations as it is a large, regularly supplemented population of outbred flies kept in conditions which are similar to ecological conditions of wild populations (see Terblanche et al. 2010). The larvae were randomly split into two treatment groups (Control and Hardened) of equal density and kept in two incubators set at 25°C (±1°C) until eclosion. Relative humidity (RH) was regulated to 76% by means of saturated salt (NaCl) solutions located within each container. Container temperatures and humidity were monitored with temperature and humidity loggers (resolution: 0.5°C, DS1923L-F5, Maxim iButton Logger, Fairbridge Technologies, Essex, UK). Once emerged, the flies in both groups were fed honey–water and sugar and given bananas for oviposition, and maintained at 25°C for at least three days before treatments started. Thereafter, once a day, the temperatures in the incubators were rapidly dropped for 2 h to 15°C at a rate of 0.0833°C/min or 10°C at a rate of 0.1667°C/min for the Control and Hardened groups, respectively. After the 2-h period, the incubators returned to 25°C (Fig. S1). Diel temperature fluctuations continued for 11 consecutive days and flies were exactly the same age within each treatment group at the start of Hardening and Control treatments. However, owing to equipment constraints (a maximum of seven flies per respirometry run) and to ensure that similar-aged flies were used in all respirometry experiments, the trials of Control and Hardened flies were offset by a day. Controlling for age in MR and thermal tolerance is critical since ageing (independent from mass) and flight muscle development can influence MR and thermal tolerance estimates in insects (Bowler and Terblanche 2008; Piiroinen et al. 2010). Thus, day 1 for the Hardened treatment commenced one day prior to the Control group. Only fed male flies were used for metabolic rate estimation to eliminate confounding effects of reproductive status in female flies. The adult flies were weighed before and after each respirometry trial using a calibrated electronic balance (to 0.1 mg; NewClassic MF, MS104S, Mettler-Toledo International Inc., Greifensee, Switzerland). Experiments were repeated in two blocks for each of the main parameters being estimated: metabolic costs and acute survival at –5°C in one block, and survival and fecundity estimated in a separate experimental block.
To determine whether RCH takes place in the 25–10°C (Hardened) group relative to the 25–15°C (Control) group, n = 10 male flies (× 5 replicates, total n = 50) were placed inside a water bath at –5°C for 2 h directly after being exposed to 10°C or 15°C for 2 h in the programmed diel thermal fluctuations. Flies were then returned to 25°C for 24 h (with food and water) and then scored for survival in each of the treatment and control groups. This was repeated after 1, 5, 8, and 11 RCH treatments to verify that typical RCH responses were induced in C. capitata during the experiments.
METABOLIC RATE MEASUREMENTS
Briefly, a flow-through respirometry system was used to determine resting metabolic rates (RMRs) of C. capitata in both the Hardened and Control groups indirectly from the adult flies’ CO2 production (Fig. S1). Groups of seven flies per group per RCH event (i.e., at a given age) were used for MR measurements, that is, n = 7 per control group per RCH event and n = 7 per hardened group per RCH event (total of 56 respirometry recordings). The system is described in detail in the Supporting information.
SURVIVAL AND FECUNDITY ESTIMATES
Flies were kept under the same conditions as previously described for both Control and Hardened groups as above. Here, 400 flies were divided into five replicates at each acclimation temperature per group, that is, five replicates of five females and five males per container for day 1 to 11 in the Hardened group (25–10°C) and five replicates of five females and five males per container for day 1 to 11 in the Control group (25–15°C). Each container was supplied with a whole banana for oviposition and sugar-water was provided for feeding ad libitum. After 11 days, the net survival of flies within each container was scored as well as the summed number of pupae and larvae produced per container. In these experiments, rearing containers were treated as replicates, that is, independent units, but all individuals within a container were pooled for statistical analyses.
Survival data of adult C. capitata following RCH events and exposure to –5°C were investigated using a generalized linear model (GLZ) assuming a binomial distribution and a logit link function in SAS statistical software (Proc Genmod), with corrections for overdispersion (following e.g., Nyamukondiwa et al. 2010).
To assess the effects of RCH treatments on the metabolic rate of C. capitata, we first used a full factorial analysis of variance (ANOVA) with metabolic rate as the dependant variable and acclimation temperature and days of RCH as the categorical variables. In a second model, we investigated the effects of mass on MR using a general linear model (GLM, ANCOVA) with MR as the dependant variable, acclimation temperature and number of RCH events as the categorical variables and mass as the covariate. Key assumptions of ANOVA were tested including normality of data distribution (Shapiro–Wilks) and equality of variance (using Levene's test) followed by a post-hoc Tukey HSD test to identify homogeneous groups. In all cases, RMR data are presented in mlCO2.h−1.
Tests of the effects of treatment temperature and RCH events on survival and fecundity were performed using GLZ in SAS. Fecundity was scored as the number of pupae and larvae produced per five pairs of adult flies after 1, 5, 8, and 11 RCH events. Fecundity analysis was undertaken using a poisson distribution and a log link function with fecundity as the dependent variable, and acclimation temperature and number of RCH events as independent categorical variables. Mortality was scored as the number of dead flies after 1, 5, 8, and 11 days of RCH. Mortality analysis was undertaken using a binomial distribution and a logit link function with proportion mortality as the dependent variable (n dead/ total n), RCH events and treatment temperature as categorical predictor variables. Data are presented as means ± SEM unless otherwise stated and tested by comparison of overlap in 95% confidence intervals among groups.
ACUTE SURVIVAL AFTER EXPOSURE TO –5°C
As expected, the RCH treatment (10°C for 2 h) resulted in a significant improvement in survival of C. capitata exposed to –5°C for 2 h relative to those which did not have a pretreatment at 10°C for 2 h (69 ± 9% vs. 44 ± 19%; df = 1, Wald χ2= 4.21, P = 0.04). However, there was no significant effect of the number of RCH events on survival (df = 2, Wald χ2= 2.49, P = 0.2873), suggesting a lack of acclimation induced by these treatments and a lack of any cumulative effects. In addition, there was no significant effect of the interaction between RCH events and treatment group (Control vs. Hardened) (df = 2, Wald χ2= 1.67, P = 0.4340) indicating that both groups remained similar in basal low temperature tolerance and hardening ability over time (Table S1).
The only variable that significantly affected absolute RMR was the number of hardening events (equivalent to days of treatment), but the acclimation temperature to which the flies were exposed for 2 h per day (10 or 15°C) had no significant effect on RMR measured at 25°C (Table S2). However, mass had a significant effect on RMR (P < 0.005) and we therefore repeated these analyses adjusting for size among days in analysis of covariance (ANCOVA). Once body mass was accounted for, neither the number of days of exposure to hardening conditions, nor the treatment group (Hardened vs. Control) had any significant effect on the mass-specific RMR of the flies recorded at 25°C (Table S2 and Fig. 1). Moreover, the interaction between the number of RCH events and hardening treatment was not significant for either absolute RMR or mass-specific RMR (Table S2).
There was no significant effect of treatment on RMR (Fig. 1A). There was no significant difference in size between the Hardened and the Control flies over the 11-day period (Fig. 1B). However, flies increased in size over the experimental duration (average mass at day 11, Hardened: 6.42 ± 0.15 mg, Control: 5.96 ± 0.17 mg, P < 0.001), probably as a consequence of maturation and adult development, but this pattern is seen in both the Hardened and the Control groups and thus not likely a consequence of the main treatment (Fig. 1B and Table S2, P > 0.05). Mass-specific metabolic rate did not differ among treatment groups (Fig. 1C; P = 0.262).
SURVIVAL AND FECUNDITY ESTIMATES
On average, fewer pupae and larvae were produced in the Hardened group (average number of larvae and pupae produced over 11 days: Hardened: 7.70 ± 1.63; Control: 17.2 ± 3.96). However, fecundity was not significantly affected by the number of RCH events (P = 0.659) or the interaction of acclimation temperature and RCH events (P = 0.082) (Fig. 2A and Table S3). There was considerable variation in the number of pupae and larvae produced at each day and only after 11 RCH events (days) was there an obvious difference in fecundity between Hardened and Control flies (Fig. 2A). Mortality was significantly higher in the Hardened group than the Control group (average mortality across 11 days: Hardened: 4.45 ± 0.63; Control: 2.25 ± 0.39 flies, P < 0.005) and number of RCH events (P < 0.001), but not by their interaction as mortality was scored per RCH event and was not a cumulative count (P = 0.1291) (Fig. 2B and Table S3). After five RCH events, there was already a significant increase in mortality of Hardened compared to Control flies and this trend continued over the course of the 11 RCH events.
There is dramatic variation in the magnitude of RCH responses even among closely related species (e.g., Lee and Denlinger 2010; Nyamukondiwa and Terblanche 2010; Strachan et al. 2011) suggesting that such a response may have evolved differentially among terrestrial insect species. RCH could be associated with freeze-intolerant insects living in temperate environments with unpredictable climates, especially during autumn and spring, where RCH likely enhances survival during sudden cold spells (Sinclair and Chown 2006; Terblanche et al. 2007). In addition, the magnitude of RCH is lower in Drosophila species with greater cold tolerance, suggesting that increased RCH may be a trade-off against basal cold tolerance (Nyamukondiwa et al. 2011). This acute plasticity may also be especially beneficial when the timing of key behaviors (e.g., mating or foraging) coincide with cold events (Shreve et al. 2004; Powell and Bale 2006; Terblanche et al. 2008). Many activities are enhanced by RCH including mating and reproductive success (e.g., Shreve et al. 2004), yet a major benefit of RCH in insects is likely immediately increased survival at lower, potentially lethal temperatures. For example, Sarcophaga crassipalpis flesh flies given 10 min at 0°C increase cold tolerance by 50% at –10°C compared to flies that did not experience this pretreatment (Chen et al. 1987). Indeed, it is these types of direct survival benefits at thermal extremes that are most frequently reported for RCH responses (Lee and Denlinger 2010).
In agreement with previous studies of RCH in these (Nyamukondiwa et al. 2010) and other insects that show RCH responses (e.g., Shreve et al. 2004; Overgaard et al. 2005; Powell and Bale 2005; Nyamukondiwa et al. 2011) C. capitata responds to brief low temperature pretreatments with increased survival relative to a control group not given the low temperature (hardening) pre-treatment. Thus, acute low temperature survival is enhanced, likely through a range of biochemical responses (discussed in Lee and Denlinger 2010) and perhaps involving Hsps (Colinet et al. 2010a,b) and membrane lipid composition changes (Overgaard et al. 2005 reviewed in Lee and Denlinger 2010). Improvements in acute low temperature survival could also be replicated in the present study through the use of diel temperature fluctuations differing principally in their magnitude of severity (although also slightly in rate of temperature change) of low temperature exposure. This allowed us to dissect the potential costs of RCH in greater detail than perhaps a single exposure could and, moreover, for a substantial portion of this flies’ adult (reproductive) life span. Consequently, Hardened flies, having been exposed to 10°C for only 2 h on a single occasion, have greater low temperature survival at –5°C for 2 h compared with the Control flies that experienced only 15°C for 2 h before exposure to –5°C for 2 h. Although not well explored in C. capitata, the RCH responses observed are probably accompanied by improvements in a range of other traits, perhaps including preservation of various mating or reproductive behaviors.
Given the limited effects of treatment or number of RCH events on whole-animal metabolic rate in C. capitata, there is little evidence suggesting RCH is metabolically expensive. Our results therefore support Shreve et al.'s (2004) hypothesis, based on results from Drosophila melanogaster, that RCH is not an energetically expensive process, or at least metabolic costs are not easily detectable at the whole-organism level. The same outcome was seen here in C. capitata assessed over longer experimental periods (i.e., more time in the respirometer) and more RCH events. However, RMR of the Hardened flies is perhaps significantly higher than that of the Control flies after one RCH event at 10°C. This is probably due to some physiological adjustment made after the initial sudden drop from 25°C to 10°C. Overall, RMR adjustments do not seem to be common and, therefore, RCH is probably not particularly energetically demanding.
Despite the lack of an obvious metabolic cost, for RCH to have evolved differentially and not simply be ubiquitous across terrestrial ectotherms, it seems reasonable to expect a fitness cost. Indeed, theoretical models of plastic or evolved low temperature responses specifically require some sort of cost function to be integrated to produce realistic results (e.g., Voituron et al. 2002; Régnière and Bentz 2007). It is possible that the costs involve trade-offs within the organism, which are undetectable at the whole-organism level (e.g., re-allocation of energy resources away from somatic tissue maintenance or enzyme activity). However, the latter costs should therefore be detectable either as fitness cost in terms of survival at benign temperatures, overall reproductive lifespan, or net reproductive output.
Work on the common housefly Musca domestica by Coulson and Bale (1990, 1992) showed a decrease in the number of eggs oviposited by cold-hardened M. domestica, along with a reduced life span of the adults and decreased emergence rate of eggs (Coulson and Bale 1992) suggesting multiple costs, both in terms of reproduction and survival. Shreve et al. (2004) found that cold-hardened D. melanogaster preserved reproductive behaviors, that is, courting and mating. Similarly, our results indicate that although flies exposed to several RCH events pay some minor reproductive penalty, this is likely not a consequence of RCH per se but rather overall reduced average temperature in Hardened flies. This implies that diurnal temperature variation, at least for an extended period of time, perhaps comes at a marginal reproductive cost, but there is great variation among days even within the Control group. A lack of reproductive costs has been reported by Kelty and Lee (1999) in short term exposures (five days) in D. melanogaster. However, the most likely explanation of the fitness cost of the difference in the variable temperatures between treatment groups in C. capitata is the highly significant increase in mortality at benign temperatures. This is probably a consequence of direct, chilling injury as opposed to indirect effects on reduced fecundity, given the manner in which we scored these traits. Given that mortality did not increase over time (i.e., no RCH event × treatment group effect was detected), it seems more likely that this is an effect of reduced temperatures, rather than a cost of RCH. Nevertheless, recent work has shown repeated exposure to chilling events in fluctuating thermal regimes could also result in life-history trade-offs and, ultimately, negative population growth rates (e.g., Marshall and Sinclair 2010). Sarcophaga crassipalpis only shows injury from cold-shock three days after treatment due to damage to the neuromuscular system (Kelty et al. 1996) and thus our speculation of chilling damage is not unfounded (and see Yi and Lee 2003).
Phenotypic plasticity of the kind discussed here (i.e., acute, reversible plasticity), resulting in RCH in adult C. capitata likely comes with both evolutionary fitness benefits and costs, as might be expected for plastic responses more broadly (see discussion in e.g., Gilchrist 1995; Hoffmann 1995; Sultan and Spencer 2002; Angilletta 2009; Kingsolver et al. 2009). Even though the flies do not expend extra metabolic energy on the process, fitness penalties are incurred. The costs involved are not necessarily acclimation or plasticity costs per se, but might be a result of chill injury due to exposure to suboptimal temperatures. Although the flies had hardened at 10°C, which likely minimizes chill injury (Yi and Lee 2003), repeated exposure to fluctuating low temperatures has an obvious detrimental effect on population survival under benign conditions, an effect that was not compensated for (or if so, at least not completely) by the RCH response itself.
In conclusion, it is clear that mortality at benign temperatures, although likely influenced in opposite directions by both low temperature and RCH responses, is a major effect caused by the absolute temperature difference in the two treatments (see Woods and Harrison 2002) and probably carries significant evolutionary fitness costs at the population level. By contrast, metabolic costs, and perhaps also to a lesser extent, reproductive costs, appear relatively insignificant to RCH and diurnally fluctuating low temperatures more generally (and see Terblanche et al. 2010). Moreover, the number of RCH events did not improve C. capitata's ability to survive suggesting limited cumulative effects on plastic responses and a lack of acclimation induced by these treatments. Regardless, cold-hardened flies do not appear to gain major benefits from increased fecundity although they likely offset immediate survival benefits against long-term reproductive penalties. In sum, the present study has identified the fitness costs of fluctuating low temperatures in C. capitata. Population modeling of evolved plastic responses can now incorporate realistic, empirically determined costs into demographic models for determining the evolution of RCH and perhaps also plasticity more broadly, and likely improves forecasting of global invasions and climate change responses for this major agricultural pest. These results are of broad significance to understanding the evolution of rapid, reversible phenotypic plasticity in thermal tolerance traits of terrestrial ectotherms.
Associate Editor: D. Carrier
This project was funded by a NRF Blue Skies Grant BS 2008090800006, and DFPT and CRI funding to JST. An NRF THRIP project to P. Addison and Stellenbosch University Sub-Committee B funding to JST further supported equipment and infrastructure used in this project. We are grateful to two anonymous referees and D. Carrier for constructive comments on an earlier version of this article.