Introduction

Fever, the behavioural or physiological ability to maintain an elevated body temperature (T body) in response to pathogens or associated components (e.g. bacterial cell wall), exists in a range of animal taxa—from arthropods to mammals (Kluger 1990). Elevated T body can enhance survival after infection (Kluger 1990; Adamo 1998; but see Müller and Schmidt-Hempel 1993) and, therefore, can boost fitness. Yet, the presence and magnitude of a fever response varies within and between species (Adamo 1998; Moore 2002). We used empirical and meta-analytical approaches to address two hypotheses that may explain this variation in insects, including:

  1. 1.

    Fever is context dependent. Ectothermic animals alter their preferred body temperature (T pref) under different circumstances—e.g. by exhibiting a lower T pref during food limitation (Angilletta 2009) or a higher T pref during reproduction (Hedrick et al. 2002). If the increase in T pref in some insects is a modest 1–3 °C (the same magnitude as fever in mammals; Kluger et al. 1998), fever may be particularly difficult to detect if fever is context dependent.

  2. 2.

    Fever is rare in insects. Though well established in some taxa (e.g. mammals), fever may be the ‘exception’ rather than the ‘rule’ in insects.

Materials and methods

For the first study (described below), we used long-winged adult Gryllus texensis that were part of a long-term colony, which has been described previously (Adamo and Lovett 2011). Briefly, we supplied crickets with water ad libitum and housed crickets in a room maintained at 26 ± 1 °C and a 12:12 light/dark cycle except during T pref trials. All studies were approved by the Animal Care Committee of Dalhousie University (#I9-026) and are in accordance with the Canadian Council on Animal Care.

To measure T pref, we used a thermal gradient experimental design modified from one described previously (Adamo 1998) (see Electronic Supplementary Material (ESM): Preferred temperature trials). To ensure that G. texensis was capable of fever using our study system, we examined T pref after the administration of a pharmacological dose of prostaglandin E2 (PGE2) as a positive control (see ESM: Effects of prostaglandin E2 on T pref). Prostaglandins are eicosanoids associated with immune function that rapidly induce fever in vertebrates and invertebrates (Kluger 1990).

Effects of chronic food limitation, immune challenge, and adult stage on T pref

For the first study, we used a repeated 2 × 3 design to manipulate food availability and immune status before and after mating in a factorial fashion (n = 15–21 for each of six treatment groups). Over the span of 4 months, we isolated 22 cohorts of newly moulted (≤1 day post-adult moult) females and housed them individually in transparent 2,000-ml plastic containers. As described previously (Stahlschmidt et al. 2013), we manipulated body condition by manipulating each female’s access to food: ‘ad lib’ (ad libitum food access as above) or ‘intermittent’ (access to cat food for 3 h every 3 days, which results in G. texensis with similar levels of body fat as those found in the field; Adamo et al. 2012). We also manipulated the females’ immune status: ‘bacterial challenge’ treatment (injection with heat-killed bacteria (Serratia marcescens) every 3 days), ‘wounded’ treatment (pierced with sterile dissection pin every 3 days), or ‘control’ treatment (not handled). Elevated T body confers a survival advantage to G. texensis upon exposure to live S. marcescens (Adamo and Lovett 2011). For details, see ESM: Immune status manipulation.

On the evening of 11 days post-adult moult, we mixed females with sexually mature males at a ratio of 1 female per 1–2 males to elicit overnight mating, and we excluded from analyses females that did not mate (mating status was determined by the presence of spermatophore-filled spermatheca at 17 days post-adult moult). We performed T pref trials on each female at 8 and 15 days post-adult moult to determine the effect of adult stage on T pref while controlling for individual-level effects.

We performed analyses with SPSS (version 19, IBM Corp.). We determined two-tailed significance at α < 0.05 and display all values as mean±s.e.m. All data met the assumptions of parametric statistics or were transformed as necessary. We used linear mixed models to determine the main and interactive effects of treatments on T pref in the first study. We included cohort and individual as random effects. We included treatments (e.g. immune status) as fixed effects, and we included femur length (a proxy for body size) and the high temperature of each individual’s experimental lane as covariates.

Generality of fever in insects: a meta-analysis approach

For the second study, we performed a literature search on empirical studies of behavioural fever in insects. Due to a lack of statistical reporting in most studies, we used Fisher’s method of meta-analysis to determine the generality of behavioural fever in insects for studies that reported a specific P value (n = 31 host–pyrogen combinations).

Results

Effects of chronic food limitation, immune challenge, and adult stage on T pref

Immune status significantly influenced the mean T pref of female G. texensis. Yet, bacterially challenged females did not exhibit higher T pref than unhandled control females (Fig. 1, Table S1). However, PGE2 did induce fever in G. texensis because T pref was affected by a time × treatment effect (F 6,67 = 11, P < 0.001) with T pref peaking 30 min post-injection of PGE2 (Fig. 2).

Fig. 1
figure 1

Effects of treatment on the preferred body temperature of female Gryllus texensis that experienced varying access to food (ad libitum or intermittent) and were exposed to a bacterial challenge, a wound, or no immune challenge (unhandled control) before and after mating (to the left and the right side of the dashed vertical line, respectively). See text for details about dietary and immune-challenge treatments. Values are displayed as mean±s.e.m

Full size image
Fig. 2
figure 2

Preferred body temperature of female Gryllus texensis that experienced one of three treatments: ‘PGE2’ (black symbols: injection of 100 μg of PGE2 in vehicle, 2 μl of a 1:1 solution of ethanol and water), ‘vehicle control’ (grey symbols: injection of vehicle alone), and ‘unhandled control’ (white symbols: no injection). Values are displayed as mean±s.e.m

Full size image

We did not detect independent effects of time (F 3,67 = 2.2, P = 0.11) or treatment (F 2,24 = 0.12, P = 0.89) on T pref due to the administration of PGE2 (Fig. 2).

Generality of fever in insects: a meta-analysis approach

Studies used various pyrogens (e.g. fungi, live bacteria, and bacterial components) to induce fever in 18 species of insects (Table 1). Although it was dose dependent or not found in many studies (Table 1), fever was a general response in insects (X 2 = 83, P = 0.039). When fever was induced (n = 14 host–pyrogen combinations), it began 3.6 ± 0.9 days after pyrogen exposure, ended 1.5 ± 0.5 days thereafter, and resulted in a 3.4 ± 0.5 °C increase in T body when it was recorded.

Table 1 Previous research on fever in insects used in our meta-analysis, including information on insect taxon and associated pyrogen (fever-inducing agent), whether fever was detected, the magnitude of the fever response (change in T pref, preferred body temperature) when it was detected and recorded, the timing after treatment at which T pref was measured, and the corresponding reference
Full size table

Discussion

Although their immune function and survival to live S. marcescens increases with temperature (Adamo and Lovett 2011), G. texensis did not exhibit fever in response to heat-killed S. marcescens regardless of food availability or mating status (Fig. 1), which does not support our first hypothesis (fever is context dependent in insects). Yet, crickets were capable of a modest fever in response to a pharmacological dose of PGE2 (Fig. 2), suggesting that the ‘context’ that most influences fever involves the type of immune response activated.

Different pathogens activate different types of immune responses (Gillespie et al. 1997), and not all pathogens elicit behavioural fever. Typically macroparasites do not elicit fever in insects (Adamo 1998; Moore 2002). However, whether or not a pathogen induces behavioural fever is not entirely dependent on the pathogen type (e.g. bacterial or fungal species, see Table 1); therefore, activating behavioural fever likely relies on a more complex mechanism than simply which pathogens induce prostaglandin release. For example, the house cricket (Acheta domesticus) exhibits fever in response to an intracellular prokaryote parasite (Rickettsiella grylli), which confers a survival advantage to A. domesticus during R. grylli infection (Adamo 1998). However, A. domesticus does not exhibit fever in response to pathogens when fever does not confer a survival advantage, such as during infection by S. marcescens (bacterium) and Ormia ochracea (parasitoid fly) (Adamo 1998); thus, pyrogen specificity of insect fever may be adaptive. Yet, the mechanism(s) that would allow for this specificity remain a mystery.

Our meta-analysis provided weak support for our second hypothesis (fever is rare in insects). Most studies reported a negative result (absence of fever; Table 1) despite a general publication bias toward those with positive results over those with negative results across disciplines (Dirnagl and Lauritzen 2010). Nevertheless, behavioural fever occurs in many insect species, although significant variation existed with regard to the timing of T pref measurement, the specific pyrogens used, and the presence of fever across studies (Table 1). For example, one orthopteran insect (desert locust, Schistocerca gregaria) exhibits a 2–3 °C increase in T pref for 1–2 days after exposure to live S. marcescens (Bundey et al. 2003), but another orthopteran (A. domesticus) does not prefer warmer temperatures after exposure to the same pathogen (Adamo 1998). Further, the concentration of pyrogens to which animals were exposed varied across studies, and it influenced the fever response in studies that used varying dosages of pyrogen. For example, two tenebrionid beetle species exhibited fever in response to high (but not low) dosages of lipopolysaccharide (Table 1). Thus, the literature indicated that although fever is widespread in insects, its expression is sporadic.

We posit that the sporadicity of insect fever may be driven by cost–benefit trade-offs where fever in some insects entails more costs than benefits. Although the survival benefits of fever are clear in some cases (Adamo 1998; Blanford et al. 1998), there may be appreciable ecological or physiological costs associated with increased T pref (Elliot et al. 2005). Other crickets (Gryllus campestris) that spend more time basking outside their burrows tend to be predated more frequently (Otti et al. 2012). Further, fever is estimated to increase energetic expenditure by 75 % due to the additive effects of immune activation (estimated 25 % increase in metabolic rate; Ardia et al. 2012) and a 3.4 °C increase in T body, which is the average magnitude of insect fever (Table 1) and is expected to result in a 37 % increase in metabolic rate (Nespolo et al. 2003). Thus, research should continue to examine the benefits and costs of fever in insects.

In sum, we show that fever is widespread in insects, but sporadically expressed. Previous researchers have posited that fever in insects may be ubiquitous and rely upon the same fever-inducing mechanism (eicosanoids) as mammals (Bundey et al. 2003). Though we induced fever in G. texensis with a pharmacological dose of an eicosanoid (PGE2), our meta-analysis results suggest that eicosanoid release due to immune recognition and activation is not sufficient for fever induction in insects. Future work should examine the effects of specific immune pathways on a suite of proteins (e.g. eicosanoids) to better understand their role in behavioural fever. We also suggest a standardized experimental protocol that includes exposure to several concentrations of non-infective pathogens (e.g. 102, 104, and 108 cells of heat-killed bacteria, to preclude any parasitic manipulation of host thermoregulatory behaviour) and fixed time points of T pref measurement (e.g. 1 h prior to treatment followed by measurements 30 min, 60 min, 120 min, 24 h, and 48 h after treatment) or real-time T pref measurement (e.g. Hunt et al. 2011). Studying the expression of insect fever has informed pest management strategy (e.g. the limited effectiveness of fungal biocontrol measures to eradicate locusts; Blanford et al. 1998). It can also provide insight into how global climate change may alter their disease resistance (e.g. Adamo and Lovett 2011; Adamo et al. 2012).