Energetically costly behaviour and the evolution of resting metabolic rate in insects



    1. Department of Entomology, University of Kansas, Lawrence, KS, 66045, USA
    Search for more papers by this author
    • Present address: Institut für Evolutionsbiologie und Ökologie, Universität Bonn, An der Immenburg 1, D-53121 Bonn, Germany.


1. A general hypothesis is presented to explain interspecific differences in size-independent resting metabolic rate. This hypothesis is based on a presumed trade-off between a low resting metabolism and adaptations of metabolism during activity.

2. With such a trade-off, selection to reduce resting metabolism is less intense in active species than in species where resting metabolism constitutes a large proportion of the daily metabolic costs. Those animals that spend more energy on activity should therefore have a higher resting metabolic rate than animals that spend less energy on activity.

3. A literature review reveals that flying insects have higher resting metabolic rates than species that use energetically less demanding types of locomotion.

4. Insects producing acoustic advertisement signals can be shown to have higher mass-independent resting metabolic rates than closely related species without this energetically demanding behaviour.

5. Literature data on vertebrate resting metabolic rates are also consistent with the presented hypothesis: the more energy animals spend on activity, the higher the mass-independent resting metabolic rate.


Variation in resting metabolic rate (RMR) among species has attracted considerable attention beginning with Kleiber (1932). Most studies have focused on the allometric relationship between body size and RMR (e.g. Kleiber 1932; Kittel 1937; Hemmingsen 1960; Bartholomew & Casey 1978; Coelho & Moore 1989). Recently, there has been a growing interest in explaining interspecific differences in mass-independent RMR. These differences represent variation in RMR that remains after elimination of the influence of body size (Daan, Masman & Groenewold 1990; Konarzewski & Diamond 1995; Vinogradov 1995). Daan et al. (1990) proposed that such variation reflects differential masses of organs with high metabolic activity. Other studies have shown correlations between interspecies variation in mass-independent RMR and life history and feeding ecology (McNab 1986; McNab 1988; Koteja & Weiner 1993). In addition, animal species that have high metabolic rates in the field often have high RMRs (Daan et al. 1990; Koteja 1991; Ricklefs, Konarzewski & Daan 1996).

These correlations between RMR and other traits led to posthoc explanations concerning the evolutionary reason behind the observed variation in mass-independent RMR. Probably the best supported hypothesis to explain interspecific differences in RMR is the ‘aerobic capacity' hypothesis independently formulated by Regal (1978), Bennet & Ruben (1979), Benton (1979) and Pough (1980). The key element of this hypothesis is an assumed functional link between RMR and maximum metabolic rate for aerobic metabolism. When higher maximum metabolic rates are selected for, RMR will increase as a correlated response. Consistent with this prediction, species with high maximum metabolic rates usually have also higher RMRs (Taigen 1983; Hinds & Rice-Warner 1992; Bozinovic 1992; Walton 1993; but see Koteja 1987 and Sparti 1992).

Adaptations that increase the maximum sustainable metabolic rate are, however, not the only ones that might lead to an increased RMR. Adaptations that increase the efficiency of metabolism during activity or that increase the maximum speed at which metabolic rate can be increased or decreased might also lead to an increased RMR. In the following it is argued that selection in favour of these adaptations should be more intense in species that spend more energy on activity. Suppose there is a trade-off between a low resting metabolism and an efficient metabolism during activity. Such a trade off can occur in several ways:

1. Physiological mechanisms (e.g. protein activity, proton leak (Porter & Brand 1993) and oxygen availability) will probably be influenced by the physiological changes correlated with energy demanding activity. Under these circumstances, these mechanisms might be adapted to the physiological conditions either during activity or during rest.

2. An increased density of mitochondria and an increased mitochondria surface area might be adaptive for the metabolism during activity but will probably increase RMR.

3. The size of energetically active organs might vary according to the physiological requirements during either resting or activity (Daan et al. 1990; Hammond & Diamond 1997).

How should such a trade-off affect the evolution of RMR? Assume a mutation reduces RMR and at the same time increases the metabolic cost of daily activity by a comparable amount. Also assume that this mutation does not influence any other trait. In a species that spends more than half of its daily metabolic energy on resting metabolism, this mutation should spread. But in a species that spends less than half of its daily metabolism on resting metabolism, spread is unlikely. Similarly, a mutation that increases RMR but decreases the metabolic cost of activity will spread in active species where RMR constitutes a small amount of the daily energy requirements.

The differential fate of mutations outlined above should lead to active species maintaining higher RMRs than relatively inactive species. Here, this prediction is tested by comparing literature data of RMRs of insects that differ in the existence of energetically demanding behaviour. Two different approaches were used: first the RMR of flying insects was compared with the RMR of insects that use less costly modes of locomotion, and second the RMR of species with acoustic advertisement signalling was compared with the RMR of closely related species that show no such signalling. In accordance with the predictions of the proposed trade-off hypothesis, those species that show the costly behaviour have a higher RMR than closely related species that do not show this costly behaviour. In addition, previously published correlations between RMR and costly behaviour in vertebrates are generally consistent with the hypothesis that there is a trade-off between a low RMR and adaptations of metabolism during activity.

Materials and methods

Values for mass-specific resting metabolic rates (ml O2 consumption g–1 h–1) of insects were compiled from the available literature. RMR values were not included if authors stated that movements occurred during the metabolic measurements. The measured metabolic rates include the metabolic costs of these movements and thus would overestimate the actual RMR.

The RMR measurements available in the literature were conducted at different temperatures (range 15–28 °C) by different authors. To enable comparison, RMR values of all authors were adjusted to 25 °C assuming a Q10 of 2·0. Similar values for Q10 have been obtained in empirical studies (Gunn 1933; Richards 1963; Mispagel 1981; Morgan, Shelly & Kimsey 1985) and have been used in comparative studies (Coelho & Moore 1989). Even if the actual Q10 would deviate from 2 this should not introduce any systematic error into the analysis because the Q10 value and the presence of energetically costly behaviour should not be correlated.

Resting metabolic rates were compared between flying and non-flying insects as well as between insects with acoustic advertisement signalling and insects without this energetically demanding behaviour. For Table 1, mass-independent RMRs were calculated according to Kleiber (1932) as (ml g–0·75 h–1) to allow comparison of RMRs among species of different sizes.

Table 1.  . Comparison of the mass-independent resting metabolic rates (ml O2 g–0·75 h–1) of flying and non-flying insects from the literature Thumbnail image of


Insects with known RMRs were categorized as non-flying and flying species. This classification was done according to the descriptions in the source manuscript for the RMR values where possible. Values were not used where classification into flying and non-flying species could not be done decisively.


Values for the RMR of male insects with acoustic advertisement signalling were compiled from the literature. These values were compared with RMR values of closely related non-acoustic species. In order to account for mass differences, predicted values for RMR in non-acoustic insects were calculated from published correlations between logarithmic body mass and logarithmic RMR assuming the mass of the acoustic insects. The resulting values, labelled as ‘expected RMR’, were compared with the actual RMRs measured in the acoustic insects.



According to the hypothesis presented herein, those species that have a higher cost of activity should also have a higher RMR. Insects that fly, an energetically demanding behaviour, are thus expected to have a higher RMR than insects that do not fly.

This prediction is supported by the analysed data showing that flying insect species have higher RMRs than species that use less energy-expensive modes of locomotion (Fig. 1). There is almost no overlap between these two groups, and flying insects have resting metabolic rates that are about three times as large as those of non-flying insects (Fig. 1). The categorization as flyers vs non-flyers explains 64% of the variance in RMR that is not explained by body mass (ANCOVA, SS = 3·1; error: SS = 1·7, F = 81·1, P < 0·001). Some RMR values had been measured in closely related species and might be dependent because of common phylogeny. To reduce the confounding effect of common phylogeny, the average resting metabolic rate was calculated of flying and non-flying insects for each order. These average RMR values for insect orders also differ between flying and non-flying insects (Kolmogorov–Smirnov, n = 7,8; T = 0·86, P < 0·01).

Figure 1.

. Relationship between RMR and body mass in flying (○) and non-flying (○) insects on a double logarithmic scale.


Acoustic advertisement signalling is known to be an energetically demanding behaviour (Prestwich 1994), and insects with acoustic advertisement are therefore expected to have a higher RMR than their close relatives that do not produce acoustic advertisement signals.

To evaluate whether species with acoustic communication have an increased RMR, the RMR of species with acoustic advertisement signalling was compared with the RMR of closely related species without acoustic advertisement signalling. Compared with the RMR expected for cockroaches of the same mass, RMR is higher in all 10 orthopteran species that produce acoustic advertisement signalling (Table 2). Likewise, actual RMR is larger than expected in Achroia grisella, the Lesser Wax Moth, a moth species where males use ultrasonic signals to attract females (Jang & Greenfield 1997; Reinhold et al. 1998). Here, the expected RMR was calculated from a linear regression between logarithmic mass vs logarithmic RMR from 95 moth species (Bartholomew & Casey 1978). In addition, the RMR of the Tok-tok Beetle, Psammodes striatus, exceeds the RMR expected for tenebrionid beetles that do not produce advertisement signals by substrate vibration (Lighton & Fielden 1995). All species with acoustic advertisement signalling thus have higher resting metabolic rates than related species without this energetically demanding behaviour.

Table 2.  . Comparison between the actual RMR of insects with acoustic advertisement signalling and their expected RMR calculated from literature data on resting metabolic rates of related insects without acoustic advertisement signalling Thumbnail image of


The proposed hypothesis predicts that species that spend more energy on activity should have a higher RMR. The energetic cost of insect flight is large compared with the costs of resting metabolism. In moths, for example, the metabolic rate during flight increases, on average, by more than 100-fold when compared with RMR (Bartholomew & Casey 1978). Therefore, efficiency of metabolism during flight should be under intense selection. In addition, selection to reduce RMR should be relatively weak when these insects fly for an extended period of the day. In accordance with the expectation, flying insects have a higher RMR than non-flying insect species.

Additional evidence comes from a within species polymorphism in flight ability. In Gryllus firmus there is a short-winged flightless morph and a fully winged morph that can fly. Flight muscles of the fully winged morph have higher metabolic rates than flight muscles of short-winged individuals (Zera, Sall & Grudzinski 1997). There is also indirect evidence that fully winged individuals have higher RMRs than short-winged individuals (Mole & Zera 1993; Zera & Mole 1994). In ants, on the other hand, alates do not seem to have consistently higher RMRs than flightless workers (Lighton & Berrigan 1995). The low RMR of alates shows that flight ability does not necessarily lead to a high RMR and hints to the hypothesis that RMR may increase as a response to energetically costly behaviour. In ants, flight occurs only for a restricted period of time and selection to increase the efficiency of metabolism during flight might be much weaker than selection to reduce RMR. Especially where solitary queens found colonies, queens will depend on a low RMR during the initial stages of colony founding.

Within the flying insects the trade-off hypothesis predicts that those insects that fly for extended periods should have a higher RMR than those that fly only for shorter periods. As expected, insects that presumably ‘live on the wing’, e.g. honey-bees and larger flies, are those with the highest RMRs (Table 1). Further evidence comes from a comparison between dragonfly species that hunt on the wing and those that hunt from perches. Species that hunt on the wing have significantly higher RMRs than species that hunt from perches (May 1979).

Fine scaled interspecific comparisons can also be accomplished among the non-flying species. Some of those groups with a RMR below 200 μl O2 g–0·75 h–1, e.g. apterygote insects, termites and walking sticks, are known to be rather inactive. Even lower RMRs are found in antlion larvae (mean of two species: 0·015 ml O2 g–0·75 h–1, Lucas 1985) and ticks (mean of five species: 0·015 ml O2 g–0·75 h–1, Lighton & Fielden 1995); arthropods that probably spend most of their energy for RMR because they do not move around.

Insects with acoustic advertisement signalling can be expected to have a higher RMR because advertisement signalling is energetically demanding (Prestwich 1994). Also, selection for a low resting metabolism should be weaker in species that spend more energy for activity. In accordance with this prediction, acoustic insects have higher RMRs than related, non-acoustic species (Table 2). The largest ratio between observed and expected RMRs occurs in the two orthopteran species that signal with exceptionally high pulse rates (Prestwich 1994) and pay the largest metabolic cost of signalling among those species listed in Table 2. Since these two species also show endothermy and are able to fly, their relatively large resting metabolic rates might alternatively be caused by these costly activities.

Additional data to test the prediction that advertisement signalling leads to a higher RMR are available in vertebrates. Frogs invest heavily in advertisement signalling and the energetic cost of signalling can exceed the metabolic cost of forced locomotion (Gatten, Miller & Full 1992). In accordance with the trade-off hypothesis, frogs have, on average, 1·5 times the mass-independent RMR of salamanders, related amphibians that do not produce acoustic advertisement signals (Gatten et al. 1992). A similar difference can be found in two groups of birds: the basal metabolic rate of oscines is about 50% higher than RMR in non-passerines when species with similar mass are compared (Lasiewski & Dawson 1967). Although some non-passerine species also produce acoustic advertisement signals, on average, signalling is likely to be more frequent and the cost of signalling should be higher in oscines.

Whenever closely related species differ in the metabolic cost of activity, this allows additional testing of the trade-off hypothesis. In the following, some of the available data are discussed. Species from aquatic habitats with fast flow rates should have a higher RMR than species from habitats with a low flow rate because the energetic cost of staying in place, or swimming against the current might be high in habitats with high flow rates. In accordance with this expectation aquatic arthropods (Fox & Simmonds 1933; Fox, Simmonds & Washbourne 1935; Walshe 1948) from streams have higher resting metabolic rates than related species from ponds and lakes. Mammals that can afford to be slow moving, because they are protected by spines or plates such as armadillos, pangolins, echidnas, hedgehogs and tenrecs should have lower RMRs than other mammals of similar mass. In these species the RMR is likely to constitute an important part of the daily energy budget and the efficiency of the RMR should therefore be under intense selection. As predicted, RMR is comparatively low in these species (McNab 1986). Those species that spend more energy on locomotion because they move faster can be expected to have a higher RMR. In accordance with this prediction, RMR is correlated with maximum swimming speed in four species of cephalopds (O’Dor & Webber 1991). Maybe the most direct evidence for the trade-off hypothesis is the significant correlation found between mass-specific RMR and average field metabolic rate in birds and mammals (Daan et al. 1990; Koteja 1991; Ricklefs et al. 1996).

It is largely unknown whether the higher resting metabolic rates of more active animals are caused by larger sizes of those organs that have high metabolic activity or by higher metabolic rates in some or all organs. There is evidence for both mechanisms. Brain tissue of more active fish has a higher metabolic rate than brain tissue from less active fish (Vernberg & Gray 1953) and mice with increased RMR have higher masses of the most active organs (Konarzewski & Diamond 1995). In many mammals including humans increased metabolic costs for activity, caused, for example, by exercise, lead to a higher RMR as a physiological response (Hammond & Diamond 1997). Thus, some of the differences in RMR between insects with or without energetically costly behaviour may be caused by physiological responses to different activity levels. But such physiological responses can also be interpreted as outcomes of a changed optimal RMR in accordance with the proposed trade-off.

To conclude, a large part of the interspecific variation in mass-independent RMR can be explained by interspecific differences in the energetic costs of activity. Further possible tests of the hypothesis presented here might include a more direct test of the proposed trade-off between a low RMR and an efficient metabolism during activity. Artificial selection for low RMR should lead to higher net metabolic rates during activity, and artificial selection for a higher efficiency during activity should lead to a higher RMR.


I would like to thank Leif Engqvist, Mike Greenfield, Catherine Loudon and York Winter for comments on earlier versions of this manuscript. This study was supported by the DFG under grant number RE 1167/1–2.


  1. Present address: Institut für Evolutionsbiologie und Ökologie, Universität Bonn, An der Immenburg 1, D-53121 Bonn, Germany.