Beneficial effects of a heat wave: higher growth and immune components driven by a higher food intake

Global warming is affecting many ecosystems worldwide. While higher temperatures are often thought to negatively affect the performance of ectotherms, the mild temperature increases predicted by recent global warming scenarios (IPCC, 2013) are often beneficial for temperate organisms (Deutsch et al., 2008; Nilsson-Örtman et al., 2012). However, besides mild increases in mean temperature, global warming will also be characterised by more extreme weather events, such as heat waves. Under global warming, heat waves are predicted to become more frequent, more intense and of longer duration (Meehl and Tebaldi, 2004; Jentsch et al., 2007; IPCC, 2013). Exposure to extreme temperatures typically reduces performance and recent studies suggest that the impact of heat waves may override the beneficial effects of mild increases in mean temperature (reviewed in Lawson et al., 2015; Vázquez et al., 2015). However, this picture may be too simplistic as non-lethal heat waves have also been shown to increase performance (e.g. Adamo and Lovett, 2011; Arambourou and Stoks, 2015). Therefore, to gain more insight into the impact of global warming on ectotherm performance, it is pivotal to predict and understand the impact of heat waves on ectotherm populations. This is especially important as the ability of species to survive global warming may depend largely on their ability to deal with extreme temperatures (Thompson et al., 2013; Vasseur et al., 2014; Ma et al., 2015).

Whether a temperature increase will have positive rather than negative consequences for organisms depends on whether the elevated temperatures surpass the thermal optimum (Angilletta, 2009). Heat wave temperatures typically exceed optimal temperatures and are therefore often associated with mortality and steep declines in performance (Vasseur et al., 2014; Ma et al., 2015). This has, for example, been shown in several insects (Asin and Pons, 2001; Chang et al., 2007; Gillespie et al., 2012; Bauerfeind and Fischer, 2014). Similarly, under extreme high temperatures, reductions in growth rate have been suggested (Lemoine and Burkepile, 2012), a key performance trait frequently studied in thermal research (Schulte et al., 2011) and directly relevant for biotic interactions (Stoks et al., 2017). Yet, several studies that explicitly simulated heat waves or exposed insects to extreme temperatures (6 to 10°C above normal) did not find a decrease in growth rate (e.g. Adamo and Lovett, 2011; Arambourou and Stoks, 2015; Dinh et al., 2016; Klockmann et al., 2016; but see Kingsolver and Woods, 1997).

As growth rate is the result of behaviour (food intake) and physiology (growth efficiency: the efficiency of assimilating and converting ingested food into biomass), heat wave effects on growth rate may be mediated by effects on both components. At high temperatures, both food intake and growth efficiency have been shown to decrease (Heilmayer et al., 2004; Lemoine and Burkepile, 2012), but not in all taxa (e.g. Culler et al., 2014; Schmitz et al., 2016). Furthermore, a higher food intake may not necessarily result in a higher growth rate as an important part of the energy obtained through food intake may not be converted to body mass but to other functions such as metabolic rate (Clarke and Fraser, 2004; Rall et al., 2010; Lemoine and Burkepile, 2012) and investment in energy storage (Kooijman, 1995). Therefore, to obtain a full understanding of the effects of a heat wave on growth rate, it is beneficial to also include metabolic rate and energy storage, in addition to food intake and growth efficiency. For example, at high temperatures, metabolic rate can increase more than food intake (Rall et al., 2010; Lemoine and Burkepile, 2012), resulting in decreased growth (Lemoine and Burkepile, 2012). Intriguingly, the opposite may also occur, and it was recently shown that exposure to a heat wave may reduce metabolic rate (Dinh et al., 2016), reflecting the well-known phenomenon of metabolic depression under high stress levels (Storey, 2015).

Another key trait closely linked to fitness that may be impaired by heat waves is immune function (Roth et al., 2010; Karl et al., 2011; Seppälä and Jokela, 2011; Dittmar et al., 2014; Dinh et al., 2016). Given that investment in immune function is energetically costly (e.g. Siva-Jothy and Thompson, 2002; De Block and Stoks, 2008), the same behavioural and physiological mechanisms driving growth reductions under heat waves may underlie impairment of immune function during heat waves. A suppression of immune function may have important fitness consequences as it reduces pathogen resistance. This is especially important as disease susceptibility, pathogen abundance and virulence are expected to increase under global warming (e.g. Maynard et al., 2015). Yet, importantly, negative effects of heat waves on immune function are not general and some studies detected no or even a positive effect (Adamo and Lovett, 2011; Bauerfeind and Fischer, 2014; Arambourou and Stoks, 2015).

In the current study, we investigated how a heat wave influenced growth rate and immune components in the larvae of two damselfly species. We were particularly interested in how effects of the heat wave on growth rate and immune components were mediated and therefore quantified food intake, growth efficiency, metabolic rate and potential trade-offs with investment in energy storage. Damselfly larvae are important intermediate predators in aquatic food webs that are particularly sensitive to global warming (Hassall and Thompson, 2008). We studied the damselfly Ischnura elegans (Vander Linden 1820) because temperatures of 30°C had no lethal effect and even positively affected growth and physiological traits (immune function, fat content and flight muscle mass) (Arambourou and Stoks, 2015; Arambourou et al., 2017). This suggests that exposure to 30°C is for most performance traits still in the optimal temperature range for this species. This may be because this species is often abundant in small, shallow water bodies (Dijkstra, 2006) that are more affected by heat waves. To start exploring the consistency of responses to heat waves across damselfly species, we also studied larvae of the damselfly Enallagma cyathigerum (Charpentier 1840). While the two species may co-occur, E. cyathigerum prefers larger, deeper water bodies (Dijkstra, 2006). As there are smaller temperature fluctuations in these deeper and cooler water bodies, we hypothesised that E. cyathigerum larvae are more sensitive to heat waves. Understanding the sensitivity of species to extreme temperatures is important for predicting which species will suffer more from global warming (Domisch et al., 2011; Rosset and Oertli, 2011).

In the summer of 2014, we collected mated females of both species. For each species, we randomly selected two populations in Flanders (Belgium), in the core of their distribution (Dijkstra, 2006). Both species were collected in Torfbroek (50°55′32.5″N, 4°32′21.4″E); I. elegans was additionally collected in Oud-Heverlee-Zuid (50°50′30.34″N, 4°39′31.91″E) and E. cyathigerum in Bergerven (51°03′58.9284″N, 5°41′29.9796″E). All populations are located in protected nature areas in Belgium. Females were transported to the laboratory for egg laying. Freshly hatched larvae were first kept in groups to increase survival (De Block and Stoks, 2003). Ten days after hatching, larvae were placed individually in 200 ml plastic cups filled with conditioned tap water (aged tap water with straw and grass). Prior to the experiment, larvae were fed Artemia nauplii ad libitum (mean±1 s.e.m. daily dose 205±54, n=10 daily portions) for 6 days per week. Larvae were checked three times a week for moulting into the penultimate instar. When larvae moulted into the penultimate instar, they entered the heat wave experiment.

To assess the effects of a simulated heat wave, we set up a laboratory experiment where larvae were reared during their last two instars at a water temperature of 18 or 30°C. The temperature of 18°C is the average water temperature during May and June in Belgium (Lake Model Flake 2009, www.flake.igb-berlin.de/index.shtml). During these 2 months, most I. elegans and E. cyathigerum larvae in Belgium are in their penultimate or final instars (based on De Knijf et al., 2006). The growth of these two instars was quantified in this study because they show the largest mass increase. The temperature of 30°C was chosen to reflect a heat wave temperature in Belgium. The Royal Meteorological Institute of Belgium (KMI 2017, www.meteo.be) defines a period consisting of a minimum of 5 consecutive days of at least 25°C, of which at least 3 days are 30°C or higher, as a heat wave. On average, there are 27.9 days per year with air temperatures of 25°C or higher, and 3.9 days with air temperatures of 30°C or higher in Belgium (KMI 2017, www.meteo.be). According to the Lake Flake model (Lake Model Flake 2009, www.flake.igb-berlin.de/index.shtml), using model settings suitable for damselfly larvae (based on Nilsson-Örtman et al., 2012), the maximum daily water temperatures almost never reach 30°C in Belgium. Therefore, long-term exposure to 30°C as in our study can be considered an extreme future heat wave in Belgium, as global warming predicts an increase in the duration of heat waves (Meehl and Tebaldi, 2004; Jentsch et al., 2007; IPCC, 2013).

The heat wave period started 1 day after the larvae moulted into the penultimate instar and ended 1 week after the larvae moulted into their final instar. This period lasted ca. 1 month, corresponding to 20% of the larval stage of the damselfly species studied here. At the start of the heat wave period, we transferred the larvae to a new 100 ml plastic cup filled with aerated conditioned water that was randomly placed in a water bath (2–3 water baths per temperature treatment). To initiate the heat wave, we first placed the larvae at 24°C for 24 h, after which the temperature was increased to 30°C; this was to avoid a shock effect and to more realistically mimic the start of a heat wave. From this point, we fed the larvae 7 days a week with a higher daily dose of Artemia nauplii (mean±1 s.e.m. daily dose 686±28, n=49 daily portions) to meet the higher energy demands of final instar larvae. At the end of the heat wave period, larvae were individually stored in a −80°C freezer for further analysis. For both species, we tested between 23 and 32 larvae per heat wave treatment (total of 104 larvae).

We made daily checks for survival. We quantified traits related to growth (growth rate, food intake, growth efficiency), metabolic rate (activity of the electron transport system, ETS), energy storage (fat content) and investment in immune function [the activity of phenoloxidase (PO) and melanin content]. PO, a key enzyme involved in insect immune function, is part of the prophenoloxidase cascade, which catalyses the production of melanin (González-Santoyo and Córdoba-Aguilar, 2012). Melanin has an important function in invertebrate immunity (Siva-Jothy et al., 2005), as it is deposited around pathogens, thereby cutting the pathogen off from available nutrients and preventing further distribution (Gillespie et al., 1997). The prophenoloxidase cascade also produces several other molecules such as cytotoxic quinones, and reactive oxygen and nitrogen species, which are highly reactive and toxic to pathogens (González-Santoyo and Córdoba-Aguilar, 2012).

We quantified growth rate as the increase in wet mass over the 7 day heat wave period in the final instar of the larvae. Wet mass was measured to the nearest 0.01 mg (Mettler Toledo^® AB135-S, Columbus, OH, USA) after gently blotting the larvae dry with tissue paper. The daily growth rate was calculated as [ln(final wet mass)−ln(initial wet mass)]/7 days (McPeek, 2004). During the heat wave period, we determined for each larva its total food intake and growth efficiency based on McPeek et al. (2001) and Campero et al. (2007). To estimate the total dry mass of food given to a larva, we collected three food portions of Artemia daily and stored these in 70% ethanol. The number of Artemia in these three daily collected food portions did not differ between days (F_1,25=0.50, P=0.49), illustrating food rations were constant through time. As we fed the damselfly larvae daily with freshly hatched Artemia from the same batch of cysts, size differences of Artemia between food portions or days are unlikely and would have been randomised across treatments.

The uneaten food was daily collected for each larva 2 h after feeding. Before the feeding period started, we first transferred the larvae to a new cup with clean conditioned water. This avoided the build-up of detritus and faeces in the rearing cups interfering with the quantification of the amount of uneaten food. Faeces produced during the 2 h feeding period were carefully removed with fine tweezers before we collected the food samples. This way, we avoided collecting any detritus or faeces together with the food remains. To collect the uneaten Artemia, we poured the water from the cup over a sieve (mesh size 64 µm). Then, we rinsed the sieve with 70% ethanol and collected the Artemia and ethanol in plastic vials. At the end of the 7 day period, we pooled the seven daily samples of uneaten food per larva and poured them over a dried and pre-weighed filter paper. These filters were dried at 60°C for at least 48 h before being weighed to the nearest 0.01 mg. The three daily collected food portions were treated in the same way. We calculated the mass of the uneaten Artemia and the daily food portions by subtracting the mass of the pre-weighed filter from the final mass. The total food intake per larva was calculated as the difference between the total dry mass of the given food and the total dry mass of the uneaten food across the 7 day period. The growth efficiency was calculated as the gain in dry mass of a larva divided by its total food intake (McPeek et al., 2001; Campero et al., 2007). To obtain the gain in dry mass, we converted larval wet mass into dry mass using the conversion equation for Enallagma and Ischnura damselfly larvae: dry mass=0.1497×wet mass (McPeek et al., 2001).

For the quantification of metabolic rate, energy storage and investment in immune function, we first homogenised the larvae using a pestle and diluted them 15 times in phosphate-buffered saline (pH 7.4, 50 mmol l⁻¹ PBS) and then centrifuged the sample for 5 min (10,000 g, 4°C). The obtained supernatant was used for the physiological analyses.

We quantified the activity of the ETS as a proxy for metabolic rate (De Coen and Janssen, 2003). The measurement of ETS activity was based on the protocol of De Coen and Janssen (2003) adapted for damselflies (Janssens and Stoks, 2013). A 384-well microtitre plate was filled with 5 µl supernatant, 15 µl buffered substrate solution (0.13 mol l⁻¹ Tris HCl, pH 8.5, 15% polyvinyl pyrrolidone, 153 µmol l⁻¹ MgSO₄ and 0.2% Triton X-100) and 10 µl INT (8 mmol l⁻¹ p-iodonitrotetrazolim) to start the reaction. We monitored the increase in absorbance at 490 nm (TECAN infinite M200 spectrophotometer, Männedorf, Switzerland) and 20°C over a period of 5 min (measurements every 20 s at 20°C). We used the formula of Lambert–Beer to convert absorbance into the concentration of formazan (extinction coefficient 15.9 mol l⁻¹ cm⁻¹). Then we converted formazan concentration to cellular oxygen consumption based on the stoichiometric relationship that for each 2 µmol of formazan formed, 1 µmol of O₂ was consumed in the ETS system. The ETS activity was measured in quadruplicate and expressed as nmol O₂ min⁻¹.

Quantification of the fat content was based on a modified version of the protocol of Bligh and Dyer (1959). We mixed 8 µl of supernatant with 56 µl H₂O₄ (100%) in 2 ml glass tubes. The tubes were then heated for 20 min at 150°C and afterwards 64 µl Milli-Q water was added. We filled a 384-well microtitre plate with 30 µl of the sample and we measured the absorbance at 490 nm (at 25°C). Fat content was measured in triplicate and we converted the averaged absorbance per larva to its fat content using a standard calibration curve of glyceryl tripalmitate. Fat content was expressed as mg per individual.

PO activity was measured using a modified protocol of Stoks et al. (2006). PO catalyses the transformation of phenols into quinones, which polymerise non-enzymatically into melanin. We mixed 10 µl supernatant with 65 µl PBS in a 96-well microtitre plate. We then added 5 µl chymotrypsin (1 mg ml⁻¹) and incubated the plate for 5 min in room temperature. During this incubation period, all pro-enzyme (proPO) present was converted into PO. Subsequently, we added 120 µl l-DOPA and measured the absorbance at 490 nm over a period of 45 min (measurements every 30 s at 30°C). PO activity was quantified in duplicate as the slope of the linear part (400–1100 s) of the reaction curve. PO activity was expressed per mg protein.

We quantified the melanin content based on the protocol of Zhou et al. (2012). First, we mixed the homogenate again with the pellet and transferred 100 µl of this mixture to an Eppendorf tube. We added 25 µl of 5 mol l⁻¹ NaOH/50% DMSO and then incubated the tubes for 2 h at 80°C. After this incubation period, we centrifuged the samples for 10 min (7800 rpm). Afterwards, we filled a 384-well microtitre plate with 30 µl of the sample and we measured the absorbance at 480 nm (at 25°C). The melanin content was measured in triplicate. We converted the averaged absorbance per larva to melanin content using a standard calibration curve. The total melanin content was expressed as mg per individual.

We evaluated the effects of species and the heat wave treatment on the different response variables using separate linear mixed models. To correct for a potential effect of population, we added population as a random effect in all models. We added body mass as a covariate in all models, except for growth rate and PO activity. We further analysed significant interactions by comparing least-square means using Tukey post hoc tests. Survival was tested using a Fisher's exact test. Growth efficiency did not meet the assumption of normality, even after transformations. Based on large (>3) absolute values of the studentised residuals, we identified three data points as outliers. Therefore, we ran a non-parametric rank F-test test (Quinn and Keough, 2002, p.196) on the entire data set (including the outliers) by first ranking the growth efficiency data and then performing a linear mixed model on the ranked values. The ranking of the data gives less weight to the outliers. To evaluate two other potential functions in which PO is involved – cuticle hardening (Hopkins and Kramer, 1992; Sugumaran, 2002) and body darkness through the production of the pigment melanin (True, 2003) – we added growth rate and melanin content as covariates to the PO model. All statistical analyses were performed in the program R v3.2.2 (www.r-project.org/). We used the R package ‘lme4’ (https://CRAN.R-project.org/package=lme4) for running the linear mixed models and the ‘car’ package to compute Wald chi-square statistic and P-values for fixed effects (Fox and Weisberg, 2011).

Overall, I. elegans larvae had a higher growth rate compared with E. cyathigerum larvae (Table 1, Fig. 1A). The two species ingested the same amount of food, but I. elegans larvae had a higher growth efficiency (Table 1, Fig. 1B,C). Both ETS activity and fat content were higher in I. elegans than in E. cyathigerum (Table 1, Fig. 2). Ischnura elegans larvae also had a higher melanin content (Table 1, Fig. 3).

Heat wave exposure did not impose mortality in I. elegans (survival at 18°C: 90%, at 30°C: 91%, Fisher's exact test: P=1.00), but it did so in E. cyathigerum (survival at 18°C: 94%, at 30°C: 75%, P<0.001). Heat wave exposure resulted in an increased growth rate for both species compared with that at 18°C (Table 1, Fig. 1A). This was associated with a higher food intake under the heat wave, while growth efficiency was not affected (Table 1, Fig. 1B,C).

Exposure to the heat wave increased ETS activity and fat content in both species (Table 1, Fig. 2). For PO activity, there was a species×heat wave interaction (Table 1, Fig. 3A), indicating that the heat wave only resulted in increased PO activity for I. elegans larvae (Tukey post hoc test, heat wave effect for I. elegans: P=0035; for E. cyathigerum: P=0.49). Larval growth rate negatively covaried with PO activity (slope±s.e.m. −0.31±0.11, χ²₁=7.94, P=0.0048). Exposure to the heat wave increased the melanin content in both species and this increase was larger for E. cyathigerum larvae (species×heat wave interaction, Table 1, Fig. 3B). The melanin content at 30°C was, however, still higher in I. elegans than in E. cyathigerum. Melanin content showed no covariation with PO activity (χ²₁=1.21, P=0.27).

Exposure to heat wave temperatures can have negative effects on performance and can even cause mortality (Garrabou et al., 2009; Petter et al., 2014; Mislan and Wethey, 2015), as observed before in damselflies (Chang et al., 2007). We detected a striking species difference, with the heat wave only being lethal in E. cyathigerum and not in I. elegans. However, for both I. elegans and E. cyathigerum, larval performance measured as growth rate was higher under the simulated heat wave of 30°C compared with 18°C, the mean water temperature the studied larval instars would experience in natural populations in Belgium. Given that larval growth of both species was also higher at 30°C compared with the intermediate temperature of 24°C (M.V.D., unpublished results), our data indicate that the heat wave was beneficial in terms of growth rate (at least for the survivors) compared with the temperatures larvae would otherwise experience in natural populations. Beneficial effects of the heat wave were also detected for two other fitness-related traits: investment in energy storage (measured as fat content) and in immune components [PO activity (for I. elegans) and melanin content]. The beneficial effects of the heat wave on these traits were associated with an increase in food intake. We first discuss the general heat wave effects that were observed for both species and then focus on species differences.

Exposure to the extreme heat wave increased the metabolic rate (measured as ETS activity) of the damselfly larvae, probably causing higher cellular maintenance costs (Lemoine and Burkepile, 2012). Increases in metabolic rate with temperature may generate growth reductions when metabolic rate increases faster than food consumption under high temperature (Rall et al., 2010; Lemoine and Burkepile, 2012). Instead, we documented larger increases in food intake (39% for I. elegans and 97% for E. cyathigerum) than in metabolic rate (29% for I. elegans and 39% for E. cyathigerum) under the heat wave. This may explain why larvae of both damselfly species were able to increase their growth rate under the extreme heat wave. The greater increase in food intake than in metabolic rate may also explain the increased investment in energy storage. A higher fat content under a heat wave has been documented before in the study species I. elegans (Arambourou and Stoks, 2015) and in other insect taxa (butterflies: Karl et al., 2011; crickets: Adamo et al., 2012; but see Fischer et al., 2014; Dinh et al., 2016).

Similar to our results, food intake of the damselfly Enallagma vesperum was also higher at 30°C (Culler et al., 2014). Likewise, a study on I. elegans showed food intake to increase up to the highest temperature tested (27.5°C; Thompson, 1978). This pattern of increasing food intake up to extreme temperatures may be related to the tropical origin of Odonates (Pritchard et al., 1996). An increase in metabolic rate under a heat wave was also suggested in a study on a tropical butterfly (Karl et al., 2011). However, this contrasts with research showing a metabolic depression to overcome a short- to medium-term exposure to extreme temperatures (Pörtner and Farrell, 2008; Dinh et al., 2016); this is thought to occur to reduce energy depletion (Marshall and McQuaid, 2010). It is likely that, as the animals in the current study were fed ad libitum during the heat wave and increased their food intake, they accumulated enough energy to cope with the higher energy demand associated with the increasing metabolic rate at the high temperature (Rall et al., 2010; Lemoine and Burkepile, 2012; Culler et al., 2014).

Although we documented positive effects on growth and fat storage under the simulated heat wave, there could be hidden costs. These can be driven both directly by the heat wave (for example, a reduction in reproduction; Zhang et al., 2015) and indirectly as a result of the increases in growth rate and metabolic rate (for example, an increased production of reactive oxygen species causing oxidative damage; Mangel and Munch, 2005).

In line with previous studies comparing Ischnura and Enallagma damselfly larvae (McPeek, 1996, 1998; Stoks et al., 2005; Siepielski et al., 2011), I. elegans grew faster and had a higher metabolic rate than E. cyathigerum. As shown before when comparing the two genera (McPeek et al., 2001; McPeek, 2004), the higher growth rate in Ischnura was not caused by a higher food intake but by a higher efficiency of converting ingested food into biomass. This may also explain the observed higher fat and melanin content in I. elegans larvae. The higher melanin content could be associated with their presence in small, shallow water bodies (Dijkstra, 2006), where they are more exposed to UV. Damselfly larvae have been shown to increase melanin content in response to UV exposure (Debecker et al., 2015).

While the response to the extreme heat wave was for most non-lethal response traits similar in the two species, the heat wave only caused mortality in E. cyathigerum. This higher sensitivity to the heat wave was expected based on the species’ preference for deeper and cooler water bodies that have smaller temperature fluctuations. For most traits, E. cyathigerum reacted in the same way to the heat wave as I. elegans and there may be two non-exclusive reasons for this. First, for these other traits, the heat wave may still have been in the optimal range of temperatures; traits indeed may differ strongly in their thermal optimum (Sinclair et al., 2016). Second, survival selection may have occurred whereby the most sensitive E. cyathigerum larvae (that would have shown reduced performance values under the heat wave) were eliminated before they could be measured.

The two species differed in their response to the heat wave with regard to two traits related to investment in immune function. The heat wave resulted in the energetically costly (González-Santoyo and Córdoba-Aguilar, 2012; for damselflies: De Block and Stoks, 2008) higher PO activity only in I. elegans larvae. As for the investment in energy storage, the increased investment in immune function may reflect the higher food intake. In addition, the higher PO activity could be the result of a changed trade-off pattern with unmeasured traits. An increased PO activity in response to heat wave temperatures has been observed before in another insect order (crickets: Adamo and Lovett, 2011), and in the study species I. elegans (Arambourou and Stoks, 2015). Although PO is a key enzyme of the immune system in insects, it is also involved in other functions such as hardening of the cuticle and pigment synthesis (González-Santoyo and Córdoba-Aguilar, 2012). Yet, it is unlikely that in our study the other functions of PO caused the response to the heat wave, as we observed a negative correlation between growth rate and PO activity and no association between PO and melanin content. The former reflects the well-known trade-off between growth rate and PO activity (for damselflies: De Block and Stoks, 2008).

In addition, melanin content increased under the heat wave; while this increase was stronger in E. cyathigerum, the melanin content was still higher in I. elegans at 30°C. Melanin is another important component of an insect's immune function, being involved in the melanotic encapsulation response where it neutralises pathogens and parasites (Gillespie et al., 1997; Sugumaran, 2002). While melanin may also play a role in thermoregulation (Roulin, 2014), this cannot explain the increase under the heat wave temperature observed here, as we would then have expected lighter animals with less melanin.

Studies focusing on effects of heat waves on immune function have found mixed results with some studies showing an immunosuppression while other studies detected no or even a positive effect of heat waves on immune function (see Introduction). This is an important topic as a greater investment in immune function might be important during heat waves as pathogens may develop faster with increasing temperature (Karvonen et al., 2010). Our results showed increases of the two studied immune components under a long-term heat wave, suggesting an overall higher immune function. Yet, we cannot exclude that other unmeasured immune components (such as the number of haemocytes; Siva-Jothy et al., 2005) were suppressed by the heat wave. Indeed, trade-offs between different immune components have been reported, and some studies found opposing results of a heat wave on different immune components (Roth et al., 2010; Karl et al., 2011; Seppälä and Jokela, 2011). This may also explain the stronger heat wave-induced increase of melanin in E. cyathigerum, without an effect on PO activity.

Understanding how extreme temperatures influence species is important to determine which species can survive under global warming (Domisch et al., 2011; Rosset and Oertli, 2011). The emerging pattern of the impact of heat waves on species is that of strong opposing fitness effects on different species, with some species suffering (e.g. decreased immune function: Fischer et al., 2014; Dinh et al., 2016) and other species benefiting (e.g. increased fat content and/or immune function: Adamo and Lovett, 2011; Arambourou and Stoks, 2015). We added to this intriguing pattern by providing a mechanistic understanding of the beneficial effect of an extreme heat wave on growth rate, energy storage and two immune components. Exposure to the simulated heat wave (30°C) positively affected several performance traits in both I. elegans and E. cyathigerum larvae. This is especially striking as we exposed both species to a long and extreme heat wave while previous studies that reported no or a negative effect of a heat wave used a shorter exposure period (e.g. Fischer et al., 2014; Dinh et al., 2016). Furthermore, it has been shown that a longer exposure period can impose stronger effects (Leicht et al., 2013). In particular, the fact that I. elegans larvae could maintain high performance over a prolonged experimental heat wave suggests that this species was able to acclimatise to the increased temperature. The larvae did so by increasing their food intake and therefore were able to cope with the higher energy demands associated with the increased metabolic rate at the high temperature. In the more sensitive E. cyathigerum, which showed mortality under the heat wave, survival selection removing the larvae with the lowest performance at 30°C probably also a played role in causing positive effects of the heat wave. The dependence of the positive effects of the heat wave on the higher food intake supports the view that the impact of heat waves on fitness may critically depend on food availability (Adamo et al., 2012). Our results highlight the importance of studying not only fitness-related response variables but also the assumed underlying behavioural and physiological variables to unravel and predict the impact of extreme climate events on organisms.

We thank Jan Wouters (Natuurpunt Flanders) and Limburgs Landschap vzw for authorizing the collection of adult Ischnura elegans and Enallagma cyathigerum damselflies in the protected nature reserves of Torfbroek and Bergerven. We thank the whole ESEE group, Sarah Vanzeebroeck, Sarah Princen and Ria Van Houdt for assistance during the experiment and Rony Van Aerschot and Geert Neyens for technical support. Also, special thanks to Thomas De Preter and Karl Lauwers for their support during the experiment. Comments from two anonymous reviewers improved the manuscript. M.V.D. is a PhD fellow and L.J. is a postdoctoral fellow of the Fund for Scientific Research Flanders (FWO).