Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Thermodynamics Constrains Allometric Scaling of Optimal Development Time in Insects

  • Michael E. Dillon ,

    Affiliation Department of Zoology & Physiology and Program in Ecology, University of Wyoming, Laramie, Wyoming, United States of America

  • Melanie R. Frazier

    Address Current address: National Center for Ecological Analysis and Synthesis, Santa Barbara, California, United States of America

    Affiliation Pacific Coastal Ecology Branch, Western Ecology Division, United States Environmental Protection Agency, Newport, Oregon, United States of America

Thermodynamics Constrains Allometric Scaling of Optimal Development Time in Insects

  • Michael E. Dillon, 
  • Melanie R. Frazier


Development time is a critical life-history trait that has profound effects on organism fitness and on population growth rates. For ectotherms, development time is strongly influenced by temperature and is predicted to scale with body mass to the quarter power based on 1) the ontogenetic growth model of the metabolic theory of ecology which describes a bioenergetic balance between tissue maintenance and growth given the scaling relationship between metabolism and body size, and 2) numerous studies, primarily of vertebrate endotherms, that largely support this prediction. However, few studies have investigated the allometry of development time among invertebrates, including insects. Abundant data on development of diverse insects provides an ideal opportunity to better understand the scaling of development time in this ecologically and economically important group. Insects develop more quickly at warmer temperatures until reaching a minimum development time at some optimal temperature, after which development slows. We evaluated the allometry of insect development time by compiling estimates of minimum development time and optimal developmental temperature for 361 insect species from 16 orders with body mass varying over nearly 6 orders of magnitude. Allometric scaling exponents varied with the statistical approach: standardized major axis regression supported the predicted quarter-power scaling relationship, but ordinary and phylogenetic generalized least squares did not. Regardless of the statistical approach, body size alone explained less than 28% of the variation in development time. Models that also included optimal temperature explained over 50% of the variation in development time. Warm-adapted insects developed more quickly, regardless of body size, supporting the “hotter is better” hypothesis that posits that ectotherms have a limited ability to evolutionarily compensate for the depressing effects of low temperatures on rates of biological processes. The remaining unexplained variation in development time likely reflects additional ecological and evolutionary differences among insect species.


A rich literature investigates the relationship between body size and life history traits of diverse organisms [1][3]. In particular, development time has long fascinated ecologists because of its compounding effects on organism fitness and therefore maximum population growth rates (intrinsic rates of increase, r) [4], [5]. That development time should vary with body size is not controversial: it should take longer to build more tissue. However, controversy surrounds the precise nature of the relationship between body size and development time (the allometric scaling exponent and intercept) which has important implications at multiple hierarchical levels and has been measured for diverse organisms [2], [6]. The allometry of development time can reveal capacities and limits of the underlying processes of cell division and differentiation [7], as well as informing discussions of the life-history implications [2] and ecological consequences of body size [1]. For example, models addressing key questions about body size evolution such as why developing in colder temperatures typically results in larger body sizes for ectotherms (i.e., the temperature size rule; [8]), often hinge on the relative rates of growth (development time) and differentiation [9], [10]. Therefore, a better understanding of the scaling relationship between body size and development time can critically alter the conclusions of such models. At the ecological scale, differences in the scaling of development time with body size can dramatically alter predictions (based on body size) of organism population growth rates, space usage and resource demands [11], [12]. Finally, variation in development time not explained by body size begs explanation by other physiological and ecological hypotheses.

Documenting the empirical relationship between body size and life history traits is important in its own right, but there has also been a strong interest in mechanistic explanations for these patterns [11], [13], [14]. The metabolic theory of ecology (MTE) attempts to explain the scaling of biological variables with body mass using fundamental principles from physics and chemistry [11], [14]. According to the MTE, organisms supply their tissues via fractal-like branching networks which are space filling and optimized to minimize transportation costs [14]. Given these characteristics, metabolic rate is predicted to scale with mass0.75 [14]. Furthermore, the MTE predicts that, although metabolic rates may vary among taxonomic groups (i.e. different intercepts), the slope of the scaling relationship should be consistent across taxa [14], regardless of phylogenetic effects [15]. Ultimately the relationship between body mass and metabolic rate has cascading effects, resulting in quarter-power scaling predictions at the individual, population, and ecological levels [11]. Although controversial [16], [17], the MTE is appealing because it provides a potentially powerful approach to linking the physiology of individual organisms to ecological processes simply by knowing body size and temperature [11].

The ontogenetic growth model (OGM) builds on the MTE to mechanistically describe organism development in terms of the bioenergetics of growth [18]. The OGM posits that organism growth rates depend on the balance between how energy consumed by organisms is devoted to new growth relative to maintenance of existing biomass. Assuming that whole organism metabolic rate scales with mass0.75 and that the energy required to create and maintain tissues does not vary with body size or tissue type, the OGM predicts that the development time of diverse organisms will scale with mass0.25 [12], [18][20].

Many empirical studies find quarter power scaling of growth rates (and times) with body size, confirming the prediction of the OGM. These studies, largely limited to birds and mammals, suggest that embyronic and post-embyonic growth times, though variable, scale with body mass to the roughly 0.25 power [1], [2], [21], [22]. Beyond these groups, maturation times from viruses to mammals were found to scale with adult mass to the 0.26 power [6].

Despite the controversy and the large body of literature on vertebrate homeotherms, there is a dearth of synthetic studies on the scaling of development time with body size in ectotherms, and particularly in insects. There has been some interest in the effects of egg or neonate mass on embryonic development time [23][25], but, to our knowledge, there are no synthetic studies addressing the scaling of egg to adult development time with adult body size. Development times of diverse insects have been measured for over a century [26], [27] due to their abundance in virtually all ecosystems, and to their far-reaching and profound ecological and economic impacts. Given the abundance of independent studies of their development times, insects provide an ideal opportunity to test whether the scaling of development time with body size is consistent with empirical fits in other taxa and with theoretical expectations from the OGM.

One of the challenges of studying the scaling of development time in insects and other ectothermic organisms is the strong effect of temperature, independent of body size, on rates of development [20], [28], [29]. For a given species, development time decreases with increasing temperature to a minimum value (“minimum or optimal development time”) occurring at some optimal temperature (Topt). At temperatures higher than Topt, development takes longer, likely due to the physiological challenges of dealing with stressfully hot temperatures [30]. Among diverse insects, Topt for intrinsic rates of population growth, r, is correlated with environmental temperature [31], suggesting that minimum development time is an ecologically relevant metric of a species’ ability to mature in its native environment. Although insects have adapted to a wide range of thermal environments, they do not fully compensate for the depressing effects of low temperatures on rates of biological processes (the “warmer is better” hypothesis; [31], [32]). Consequently, species adapted to cold environments are predicted to have longer minimum development times than warm-adapted species, at their respective Topt.

Here we examine the scaling of minimum development time with body mass and Topt in a large and diverse insect data set to determine whether: (1) minimum development time scales with body mass according to theoretical predictions [18], [19], (2) body size alone explains most of the variation in minimum development time, and (3) whether adaptation of insects to different thermal environments drives differences in minimum development time (a test of the “warmer is better” hypothesis) [32].

Materials and Methods

We compiled literature measurements of egg to adult development time (days) for 361 insect species from 94 families and 16 orders (Table S1). For each species, development time was measured at 2–11 constant temperatures, with more than 80% of species measured at 4 or more temperatures (Table S1). In general, insect development time decreases with increasing temperature until it begins to level off or even increase at stressfully high temperatures (Fig. 1B, C) [12], [33], [34]. We determined the minimum development time (Tdev) and the temperature at which that minimum occurred (Topt) for each species (Fig. 1B, C filled points). When multiple independent estimates of minimum development time were available for a single species, we used the minimum available estimate.

Figure 1. Insect body lengths and representative reaction norms for development time.

(A) The length of each insect species included in the study (N = 361, Table S1) is represented by a vertical bar, arranged from smallest to largest, with vertical scale bars at 1, 5, 10, and 100 mm. The parasitic wasp (Trichogramma pretiosum ∼0.3 mm) is shown magnified 50× for comparison with the grasshopper (Taeniopoda eques ∼59 mm) which is shown to scale. Development time data for these two species are given in lower plots (B, C). For each species, development times were measured at a number of constant temperatures and the minimum development time and temperature at which that occurred (filled points) were used in subsequent analyses. (Wasp: P. Flinn, USDA, grasshopper: Wesley Sprinkle, CC).

For each species, we determined mean body length (the most commonly reported measure of insect size) from the literature or by personal communication with authors or specialists. When a range of body lengths were available, we used the midpoint. Body lengths spanned nearly 2.5 orders of magnitude, from 0.3 mm parasitic wasps to 80 mm walking sticks (Fig. 1A). We converted body length (mm) to dry body mass (mg) using equations derived from models based on large insect data sets (Table S2) [35][42]. Models were specific to insect order, thereby controlling for differences in body plan among taxonomic groups. For orders without models (Phasmatodea, Siphonaptera) we estimated body mass using general insect models. For most orders, multiple models were available (Table S2). Rather than attempting to identify the “best” model, we averaged dry mass estimates from all available models to obtain a consensus estimate. This ensemble approach should yield a lower mean error because the individual estimates come from models derived from independent data sets [43], [44].

Although based on the best available data for insect body sizes, indirectly estimating body mass from body length using these equations could add significant measurement error as suggested by the range of scaling exponents for the length-mass models (Table S2). We used a resampling approach to determine whether this source of error significantly altered the scaling relationships between body mass and development time. Independent length-mass models (Table S2) yielded 2–7 estimates of dry mass for each species, with 88% of species having 5 or more body mass estimates and 9% having only 2–3 estimates. The means and standard deviations of these dry mass estimates were used to generate random normal distributions of dry mass for each species. We then selected a single mass estimate for each species by random draw from their respective distributions and used these data in analyses of the scaling of development time with body mass (see below). This procedure was repeated 10,000 times to estimate the effects of mass measurement error on estimates of the scaling exponent of and variance explained by the scaling analysis.

We performed regression analyses using ordinary least squares (OLS), standardized major axis (SMA), and phylogenetic generalized least squares (PGLS) techniques. OLS analyses assume that the predictor variable (body mass) is measured without error so may underestimate allometric slopes if that is not the case [45], [46]. SMA analyses (model II or reduced major axis regression) assume equal measurement error in dependent and independent variables. SMA may therefore under- or overestimate slope values if the dependent or independent variables have more measurement error, respectively [46]. To account for the statistical non-independence of species data due to shared evolutionary history [47], we used PGLS which generalizes the phylogenetically independent contrasts (PICs) [47][49] method to deal with multichotomies and more complex models of evolution [50], [51]. For this analysis, we constructed a phylogeny for all taxa (Text S1) based on best available molecular and morphological data (Text S2), and assuming branch lengths of one given the lack of comparable branch length data across studies. From the phylogeny, a correlation structure was estimated for incorporation into a generalized least squares model [52]. We compared the performance of four evolutionary models that varied in the model of trait evolution: Ornstein-Uhlenbeck (OU) [53] vs. Brownian motion, and in branch lengths: Grafen’s transformation [48] vs. untransformed with all equal to 1. The OU model assumes that trait evolution is constrained, with the strength described by the parameter α [53], which was optimized by restricted maximum likelihood (REML) methods [52]. For each analysis, we chose from among these four models based on the lowest AIC value. We estimated R2 values for PGLS analyses following Paradis (2011) [52].

For comparison with other empirical studies, we first estimated the scaling of minimum development time (Tdev) with dry body mass (both variables ln-transformed) among all insect species, using all three regression methods. For the OLS and SMA regressions, we used robust methods to minimize the effect of potential outliers on slope estimates [54], but these methods were not available for the PGLS analysis. We then performed an OLS regression with minimum development time as the response variable and dry body mass, Topt, insect order, and all interactions as the predictor variables. For comparison with the OLS full-factorial model, we ran a similar PGLS model that included mass, Topt, and insect order, but excluded the Topt by insect order interaction because of singularities. In both cases, we removed non-significant interactions to obtain the final models. SMA methods were not available for models with covariate interactions. To visualize how optimal development time varied with Topt and insect order, we used the final OLS model to predict development time (mean and standard error) at Topt = 25 and 35°C for a 1 mg insect species in those orders with at least 10 species.

Analyses were performed in R [55] using the smatr package [56] for SMA and OLS analyses, and the caper [57] and nlme [58] packages for the PGLS analyses following the approach described by Paradis (2011) [52].


Among 361 insect species from 16 orders, spanning nearly 6 orders of magnitude in body mass (Table 1), development time increased significantly with body mass regardless of analytical approach. The SMA scaling exponent (0.259; Fig. 2, black line) was not significantly different from the predicted exponent of 0.25 (95% CI = 0.234–0.287; Fig. 2, black line). However, the OLS analysis yielded a slope of 0.123, which was significantly smaller than 0.25 (95% CI = 0.096–0.149; R2 = 0.26; Fig. 2, brown line). For the PGLS analysis, the best performing model assumed OU trait evolution with untransformed branch lengths (AIC = 461.3 vs. 468.1, 506.9, and 627.6 for the other three models). The allometric exponent from this PGLS analysis (0.097) was also significantly less than 0.25 (95% CI = 0.062–0.131, R2 = 0.28).

Figure 2. Scaling of minimum development time with body mass among insect species.

The standardized major axis (black line) scaling exponent of 0.259 was not significantly different from 0.25 (95% CI: 0.234–0.287, shown in gray). The slope estimated by ordinary least squares (brown line) was significantly lower than 0.25 (95% CI: 0.096–0.149; R2 = 0.26; shown in light brown). Shading indicates 95% confidence intervals.

Table 1. Ranges for dry mass, optimal developmental temperature (Topt), and minimum development time (Tdev) within the 16 insect orders and 361 insect species included in analyses.

Measurement error in body masses appears to have had little effect on model results. The scaling exponents and R2 values estimated for observed data (Fig. 2) closely matched 10,000 analyses of randomized data (Fig. S1) in which the body mass of each of the 361 species was randomly selected from a normal distribution with a mean and sd estimated from species-specific summarized mass body length equations (Table S2). The average (OLS) scaling exponent from the randomization approach was 0.119 (95% CI = 0.108–0.130) as compared to the observed (OLS) exponent of 0.123 (Fig. S1, A). R2 values from the randomization analysis averaged 0.22, with most of them (95% CI = 0.19–0.26) falling below the observed R2 of 0.26 (Fig. S1, B).

For the full-factorial OLS model including dry mass, Topt, and insect order, we removed the non-significant three-way interaction (mass: Topt:order; F9,307 = 0.94, P = 0.488), and the mass:order interaction (F14,316 = 1.36, P = 0.172) to generate the final model. Along with the main effects, the final, best-supported model (AIC = 479.8 vs 497.5 and 489.4 for models excluding the three-way or both the three-way and mass:order interaction, respectively) included Topt:order and Topt:mass interactions and explained 56% of the variation in Tdev (Table 2) Most of the variation in Tdev was explained by mass (26%) and insect order (16%), with Topt and the two interactions explaining 5% and 4% each, respectively (Table 2). To verify that these results (in particular the order effects) were not a byproduct of low numbers of species in some orders (Table 1), we ran these models again including only the seven orders with 10 or more species (Table 1). As before, the three-way and mass:order interactions were not significant (F6,303 = 1.13, P = 0.342; F6,309 = 1.18, P = 0.319, respectively) so were not included in the final model. All other effects were strongly significant (all P<0.002). The final model with only these seven orders explained 46% of the variation in Tdev, and as in the model including all orders, mass (25%) and order (9%) explained most of the variation in Tdev.

Table 2. Among insects, minimum development time (Tdev) varied with dry body mass, optimum developmental temperature (Topt), and insect order.

For the PGLS analysis, the best performing model assumed Brownian evolution along untransformed branch lengths (AIC = 370.0 vs. 382.9, 413.0, and 484.2 for other models). The final model included mass, Topt, and the mass:Topt interaction, and explained 54% of the variation in minimum development time (Table 2). In contrast to the OLS analyses, the PGLS analysis found no differences in Tdev among orders (Table 2), suggesting the phylogeny adequately captured variation in Tdev at the order level.

Both OLS and PGLS models suggested that insect species with higher Topt developed more quickly for their body size (OLS and PGLS Topt effect, both P<0.001; Table 2). Further, the relationship between Tdev and body mass depended strongly on Topt (OLS and PGLS mass: Topt interaction, both P<0.001; Table 2). Specifically, development time increased more steeply with mass for insect species with low Topt than for species with high Topt (Fig. 3). This interaction was not driven by covariance between Topt and dry mass (OLS: F1,344 = 0.23, P = 0.636; PGLS: t1,359 = 0.37, P = 0.5451). SMA analyses of raw development time (order-corrected analyses were not possible with SMA) showed similar patterns, with the slope estimate decreasing from 0.400 (95% CI: 0.333, 0.480) for species with Topt less than 28°C to 0.225 (95% CI: 0.190, 0.266) for species with Topt between 28 and 30°C, to 0.192 (95% CI: 0.153, 0.241) for species with Topt between 30 and 33°C. For species with Topt greater than 33°C, the SMA slope estimate was 0.233 (95% CI: 0.187, 0.289).

Figure 3. The scaling of minimum development time with body mass depends on optimal developmental temperature.

Development time as a function of body mass (both ln-transformed) for four ranges of Topt. Insect species with lower Topt showed steeper scaling of Tdev with body mass than did insect species with higher Topt (scaling exponents fall from 0.206 to 0.072 as Topt increased from less than 28 to over 33°C; Table 2, mass: Topt interaction for both OLS and PGLS, P<0.001).

Some insect orders developed more quickly than others after controlling for body size and Topt (i.e. variation in intercept; OLS order effect, P<0.001; Table 2). However, the relationship between mass and Tdev did not differ among orders (OLS mass:order interaction, P = 0.172).

The OLS analysis suggested that the relationship between Tdev and Topt differed among orders (OLS Topt:order interaction, P<0.01; Table 2). For most orders, after controlling for body size, species with higher Topt developed more quickly (Fig. 4). This effect was strong for the Hymenoptera, Diptera, and Coleoptera (Fig. 4), but less pronounced or not evident for the Orthoptera, Lepidoptera, and Hemiptera when Topt of 25 and 35°C were compared (Fig. 4). However, these results may be confounded by significant variation in Topt among orders (OLS: F15,344 = 4.33, P<0.001; PGLS: F1,15 = 3.126, P<0.001; Fig. S2).

Figure 4. The effect of optimal developmental temperature on the scaling of development time varies among insect orders.

Model-predicted (Table 2) development time (mean and 95% CI) for insects species corrected to 1 mg mass and Topt values of 25 and 35°C. For some orders (Hymenoptera, Diptera, Coleoptera), species with higher Topt developed much more quickly, whereas Topt had little to no effect on development time in other orders.


Across this large and diverse insect data set, body size was an unexpectedly weak predictor of development time, and we did not find strong support for quarter power scaling of development time with body mass as predicted by the ontogenetic growth model (OGM). When only body mass was included as a predictor, the scaling of minimum development time with body size varied with the analytical approach. The scaling exponents estimated by OLS and PGLS analyses were significantly smaller than 0.25 (0.123 and 0.097, respectively), but these approaches assume no error in the body mass estimates, so may underestimate the true scaling exponent [46]. The SMA analysis, which assumes equal error in estimates of body mass and minimum development time, yielded a scaling exponent that was indistinguishable from 0.25 (Fig. 2, Table 1), consistent with the OGM [12], [18], [19]. We don’t know the relative errors in estimates of Tdev and body mass, but errors in body mass estimates may be high because they include both measurement and equation error.

Length-mass relationships from diverse studies (Table S2) yielded large variation in body mass estimates for some species, creating a potentially significant source of error that might explain the relatively low scaling coefficients and poor predictive power of the regression models. However, the distribution of estimated exponents from a resampling simulation clearly overlapped the original estimate (Fig. S1), suggesting that error due to indirectly estimating body mass had little effect on slope estimates. This is likely because the error in body mass estimates was small relative to the range of body sizes included in the study. Although mass estimates clearly include error, the slope estimates from the regression analyses depend on relative error, which, given the large range in body masses, may be similar for the two axes. Consequently, the best estimate of the scaling exponent is likely between the OLS and SMA values.

The scaling of development time with body mass may differ from predictions because the assumptions of the OGM model are violated. One key assumption of the OGM is that whole organism metabolic rates scales with mass0.75. A number of studies suggest that metabolic scaling exponents can significantly deviate from the predicted ¾ power within and among species and across broader taxonomic groups [59][66]. However, two large scale studies in insects found that, after correcting for phylogenetic nonindependence, metabolic rate scaled with mass0.75 (391 insect species from 16 orders) [67], and with mass0.76 (419 species from 11 orders) [68]. Based on available data, it appears unlikely that the smaller exponents we found arise because of violation of this key assumption of the OGM. Limited evidence suggests that other assumptions of the OGM may also be violated, including invariance in energy metabolism among tissues [13] and constant cell size [69], but see [19] for further evaluation of some of these assumptions. Clearly, more studies testing the assumptions of the OGM are necessary to evaluate whether we expect its predictions of development time scaling to hold.

The scaling exponent may also differ from the predicted 0.25 power due to strong selection on development time for species with larger body sizes. The strong negative allometry of development time with body size described here (exponent between 0.12 and 0.25; Figs 2, 3) means that larger insects develop even more rapidly than expected from predictions based on bioenergetics alone [19], suggesting that there may be strong selective pressure for larger insects to develop more quickly.

Body size alone explained only 26–28% of the variation in development time, and there was approximately 10-fold variation in development time at any given body size (Fig. 2). Including the optimal temperature for development, Topt, in the PGLS model and Topt and order in the OLS model doubled the amount of variation explained (Table 2). Topt proved to be a strong predictor of minimum development time. Consistent with the “warmer is better” hypothesis [31], [32], we found that warm-adapted insects (i.e., those with higher Topt) had shorter development times at their optimal temperature than did cold-adapted species (i.e, those with lower Topt; Table 2, Fig. 4). This provides further evidence that although insects have adapted to a wide range of thermal environments, they do not fully compensate for the depressing effects of low temperatures on rates of biological processes [70][72], including development. The evolutionary adaptation of insects to their thermal environment therefore appears to be a particularly important factor driving minimum development times.

Beyond direct effects of optimal temperature on development time, the scaling relationship between body mass and development time varied with Topt (Table 2, Fig. 3). For insects with low Topt (i.e., those adapted to cold environments), development time increased relatively quickly with mass0.21; conversely, for insects with high Topt (i.e., warm-adapted), development time increased only slightly with mass0.07 (Fig. 3). This temperature dependence of the scaling of development time with body size is contrary to one of the fundamental predictions of the MTE and OGM–a single slope of 0.25 for all insects, regardless of other factors [14], [15]. These findings suggest that, rather than following universal scaling, insects adapted to cold environments pay a larger cost (in terms of time) to develop larger body sizes, with potentially broad ecological implications. For insects in warm tropical regions, one of the costs of evolving larger body sizes (longer development time) may be mitigated by the evolution of higher optimal temperatures. Relaxation of this selective pressure may more readily allow for evolution of larger body sizes, which is consistent with the prevalence of the largest insects in warm tropical environments.

The effect of Topt on development time was not consistent among insect orders (Table 2, OLS Topt by order interaction). With the exception of the Lepidoptera, the effect of Topt on development time was pronounced for holometabolous orders for which we had at least 10 species (Fig. 4; Coloeptera, Diptera, Hymenoptera). For hemimetabolous orders (Orthoptera, Hemiptera, and Thysanoptera), the effect of Topt was weaker, or not evident. However, significant variation in Topt among orders makes interpretation of these patterns problematic.

Even after accounting for body size and Topt (and insect order for OLS analyses), a substantial amount of variation in development time was unexplained. Ecological differences among taxonomic groups may drive differences in metabolic rates among invertebrates [67], [73] (e.g. active hunters with higher metabolic rates than detritivores) [68] and other organisms [74][76], potentially leading to differences in the scaling of development time with body size. More studies investigating potential ecological drivers of development time among diverse insects are necessary to better understand what drives the substantial variation in development time documented here. Finally, predictive relationships for development time as a function of body mass may be useful to both basic and applied ecologists, but we urge caution. High variability in development time at a given body mass cannot be ignored and could lead to large prediction errors.

Supporting Information

Figure S1.

Error in body mass estimates had little effect on allometric slope and R2 estimates. From 10,000 OLS regression analyses where body masses of each of the 361 species were drawn from random normal distributions based on means and standard deviations of multiple estimates from length-mass equations (Table S2), A) the distribution of estimated allometric slopes overlapped the original estimate (0.123, red line) with the 95% CI of the resampling distribution (0.108, 0.130) similar to the 95% CI of the original slope estimate (0.096, 0.149). B) The R2 value of the original slope estimate (0.26) was higher than the mean of the resampled estimates (0.22).



Figure S2.

Optimal developmental temperature varied among insect orders. For six of the seven orders with 10 or more species, Topt was centered around 30 °C. The Orthoptera tended to have higher Topt, with the mean near 35 °C.



Table S1.

Compiled species data.



Table S2.

Compiled equations for estimating dry mass.



Text S1.

Newick tree for all insect species.



Text S2.

Description of and references for phylogenetic tree construction.




Data collection for this project began while the authors were at the University of Washington, Seattle. We thank Christine Bell for help in gathering literature and in finding insect body lengths, and two anonymous reviewers for comments that greatly improved the manuscript.

Author Contributions

Conceived and designed the experiments: MRF MED. Performed the experiments: MED MRF. Analyzed the data: MRF MED. Contributed reagents/materials/analysis tools: MRF MED. Wrote the paper: MED MRF.


  1. 1. Peters RH (1986) The Ecological Implications of Body Size. CambridgeUK: Cambridge University Press. 329 p.
  2. 2. Calder WA (1996) Size, Function, and Life History. New York: Dover Publications. 431 p.
  3. 3. Spicer J, Gaston K (1999) Physiological Diversity and its Ecological Implications. OxfordUK: Blackwell Science. 241 p.
  4. 4. Fenchel T (1974) Intrinsic rate of natural increase: the relationship with body size. Oecologia 14: 317–326. doi: 10.1007/bf00384576
  5. 5. Gaston KJ (1988) The intrinsic rates of increase of insects of different sizes. Ecological Entomology 13: 399–409 doi:10.1111/j.1365-2311.1988.tb00372.x.
  6. 6. Blueweiss L, Fox H, Kudzma V, Nakashima D, Peters R, et al. (1978) Relationships between body size and some life history parameters. Oecologia 37: 257–272 doi:10.1007/BF00344996.
  7. 7. Bonner JT (1965) Size and Cycle: An Essay on the Structure of Biology. PrincetonNJ: Princeton University Press. 219 p.
  8. 8. Atkinson D (1994) Temperature and organism size-a biological law for ectotherms? Advances in Ecological Research 25: 1–58. doi: 10.1016/s0065-2504(08)60212-3
  9. 9. Walters RJ, Hassall M (2006) The temperature-size rule in ectotherms: may a general explanation exist after all? The American Naturalist 167: 510–523 doi:10.1086/501029.
  10. 10. Van der Have TM, de Jong G (1996) Adult size in ectotherms: temperature effects on growth and differentiation. Journal of Theoretical Biology 183: 329–340 doi:10.1006/jtbi.1996.0224.
  11. 11. Brown JH, Gillooly JF, Allen AP, Savage VM, West GB (2004) Toward a metabolic theory of ecology. Ecology 85: 1771–1789. doi: 10.1890/03-9000
  12. 12. Zuo W, Moses ME, West GB, Hou C, Brown JH (2012) A general model for effects of temperature on ectotherm ontogenetic growth and development. Proceedings of the Royal Society B 279: 1840–1846 doi:10.1098/rspb.2011.2000.
  13. 13. Von Bertalanffy L (1957) Quantitative laws in metabolism and growth. The Quarterly Review of Biology 32: 217–231. doi: 10.1086/401873
  14. 14. West GB, Brown JH, Enquist BJ (1997) A general model for the origin of allometric scaling laws in biology. Science 276: 122–126. doi: 10.1126/science.276.5309.122
  15. 15. Savage VM, Gillooly JF, Woodruff WH, West GB, Allen AP, et al. (2004) The predominance of quarter-power scaling in biology. Functional Ecology 18: 257–282. doi: 10.1111/j.0269-8463.2004.00856.x
  16. 16. Darveau CA, Suarez RK, Andrews RD, Hochachka PW (2002) Allometric cascade as a unifying principle of body mass effects on metabolism. Nature 417: 166–170. doi: 10.1038/417166a
  17. 17. Kozłowski J, Konarzewski M (2005) West, Brown and Enquist’s model of allometric scaling again: the same questions remain. Functional Ecology 19: 739–743. doi: 10.1111/j.1365-2435.2005.01021.x
  18. 18. West GB, Brown JH, Enquist BJ (2001) A general model for ontogenetic growth. Nature 413: 628–631. doi: 10.1038/35098076
  19. 19. Moses ME, Hou C, Woodruff WH, West GB, Nekola JC, et al. (2008) Revisiting a model of ontogenetic growth: estimating model parameters from theory and data. The American Naturalist 171: 632–645. doi: 10.1086/587073
  20. 20. Gillooly J, Charnov E, West G, Savage V, Brown J (2002) Effects of size and temperature on developmental time. Nature 417: 70–73. doi: 10.1038/417070a
  21. 21. Calder WA (1983) Ecological scaling: mammals and birds. Annual Review of Ecology and Systematics 14: 213–230. doi: 10.1146/
  22. 22. Reiss MJ (1991) The Allometry of Growth and Reproduction. New York: Cambridge University Press. 204 p.
  23. 23. Howe RW (1967) Temperature effects on embryonic development in insects. Annual Review of Entomology 12: 15–42. doi: 10.1146/annurev.en.12.010167.000311
  24. 24. Gillooly JF, Dodson SI (2000) The relationship of egg size and incubation temperature to embryonic development time in univoltine and multivoltine aquatic insects. Freshwater Biology 44: 595–604. doi: 10.1046/j.1365-2427.2000.00607.x
  25. 25. Gillooly JF, Dodson SI (2000) The relationship of neonate mass and incubation temperature to embryonic development time in a range of animal taxa. Journal of Zoology 251: 369–375. doi: 10.1111/j.1469-7998.2000.tb01087.x
  26. 26. Sanderson DE (1910) The relation of temperature to the growth of insects. Journal of Economic Entomology 3: 113–140.
  27. 27. Krogh A (1914) On the influence of the temperature on the rate of embryonic development. Zeitschrift für Allgemeine Physiologie 16: 163–177.
  28. 28. Wagner TL, Wu HI, Sharpe PJH (1984) Schoolfield RM, Coulson RN (1984) Modeling insect development rates: a literature review and application of a biophysical model. Annals of the Entomological Society of America 77: 208–225.
  29. 29. Charnov E, Gillooly JF (2003) Thermal time: body size, food quality and the 10°C rule. Evolutionary Ecology Research 5: 43–51.
  30. 30. Hochachka PW, Somero GN (2002) Biochemical Adaptation: Mechanism and Process in Physiological Evolution. New York: Oxford University Press. 466 p.
  31. 31. Frazier MR, Huey RB, Berrigan D (2006) Thermodynamics constrains the evolution of insect population growth rates: “warmer is better.”. The American Naturalist 168: 512–520 doi:10.1086/506977.
  32. 32. Angilleta MJ Jr, Huey RB, Frazier MR (2010) Thermodynamic effects on organismal performance: is hotter better? Physiological and Biochemical Zoology 83: 197–206 doi:10.1086/648567.
  33. 33. Davidson J (1944) On the relationship between temperature and rate of development of insects at constant temperatures. The Journal of Animal Ecology 13: 26–38. doi: 10.2307/1326
  34. 34. Kipyatkov VE, Lopatina EB (2010) Intraspecific variation of thermal reaction norms for development in insects: New approaches and prospects. Entomological Review 90: 163–184 doi:10.1134/S0013873810020041.
  35. 35. Rogers L, Hinds W, Buschbom R (1976) A general weight vs. length relationship for insects. Annals of the Entomological Society of America 69: 387–389.
  36. 36. Rogers L, Buschbom R, Watson C (1977) Length-weight relationships of shrub-steppe invertebrates. Annals of the Entomological Society of America 70: 51–53.
  37. 37. Schoener TW (1980) Length-weight regressions in tropical and temperate forest-understory insects. Annals of the Entomological Society of America 73: 106–109.
  38. 38. Sample BE, Cooper RJ, Greer RD, Whitmore RC (1993) Estimation of insect biomass by length and width. American Midland Naturalist 129: 234–240. doi: 10.2307/2426503
  39. 39. Hodar J (1996) The use of regression equations for estimation of arthropod biomass in ecological studies. Acta Oecologica 17: 421–433.
  40. 40. Ganihar SR (1997) Biomass estimates of terrestrial arthropods based on body length. Journal of Biosciences 22: 219–224. doi: 10.1007/bf02704734
  41. 41. Benke AC, Huryn AD, Smock LA, Wallace JB (1999) Length-mass relationships for freshwater macroinvertebrates in North America with particular reference to the southeastern United States. Journal of the North American Benthological Society 18: 308–343. doi: 10.2307/1468447
  42. 42. Smock LA (1980) Relationships between body size and biomass of aquatic insects. Freshwater Biology 10: 375–383. doi: 10.1111/j.1365-2427.1980.tb01211.x
  43. 43. Bates JM, Granger CWJ (1969) The combination of forecasts. Operational Research Quarterly 20: 451–468. doi: 10.1057/jors.1969.103
  44. 44. Araújo MB, New M (2007) Ensemble forecasting of species distributions. Trends in Ecology & Evolution 22: 42–47. doi: 10.1016/j.tree.2006.09.010
  45. 45. McArdle B (1988) The structural relationship: regression in biology. Canadian Journal of Zoology 66: 2329–2339. doi: 10.1139/z88-348
  46. 46. Warton DI, Wright IJ, Falster DS, Westoby M (2006) Bivariate line-fitting methods for allometry. Biological Reviews 81: 259–291. doi: 10.1017/s1464793106007007
  47. 47. Felsenstein J (1985) Phylogenies and the comparative method. The American Naturalist 125: 1–15. doi: 10.1086/284325
  48. 48. Grafen A (1989) The phylogenetic regression. Philosophical Transactions of the Royal Society of London Series B, Biological Sciences 326: 119–157. doi: 10.1098/rstb.1989.0106
  49. 49. Garland T, Harvey PH, Ives AR (1992) Procedures for the analysis of comparative data using phylogenetically independent contrasts. Systematic Biology 41: 18–32.
  50. 50. Garland T Jr, Ives AR (2000) Using the past to predict the present: confidence intervals for regression equations in phylogenetic comparative methods. The American Naturalist 155: 346–364. doi: 10.1086/303327
  51. 51. Blomberg SP, Lefevre JG, Wells JA, Waterhouse M (2012) Independent contrasts and PGLS regression estimators are equivalent. Systematic Biology 61: 382–391 doi:10.1093/sysbio/syr118.
  52. 52. Paradis E (2011) Analysis of Phylogenetics and Evolution with R. New York: Springer. 386 p.
  53. 53. Martins EP, Hansen TF (1997) Phylogenies and the comparative method: a general approach to incorporating phylogenetic information into the analysis of interspecific data. The American Naturalist 149: 646–667. doi: 10.1086/286013
  54. 54. Taskinen S, Warton DI (2011) Robust estimation and inference for bivariate line-fitting in allometry. Biometrical Journal 53: 652–672. doi: 10.1002/bimj.201000018
  55. 55. R Development Core Team (2013) R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. Available:
  56. 56. Warton DI, Duursma RA, Falster DS, Taskinen S (2012) smatr 3– an R package for estimation and inference about allometric lines. Methods in Ecology and Evolution 3: 257–259 doi:10.1111/j.2041-210X.2011.00153.x.
  57. 57. Orme D, Freckleton R, Thomas G, Petzoldt T, Fritz S, et al.. (2012) caper: Comparative Analysis of Phylogenetics and Evolution in R. Available:
  58. 58. Pinheiro J, Bates D, DebRoy S, Sarkar D, R Core Team (2013) nlme: Linear and Nonlinear Mixed Effects Models. Available:
  59. 59. White CR, Seymour RS (2003) Mammalian basal metabolic rate is proportional to body mass2/3. Proceedings of the National Academy of Sciences 100: 4046–4049. doi: 10.1073/pnas.0436428100
  60. 60. Bokma F (2004) Evidence against universal metabolic allometry. Functional Ecology 18: 184–187. doi: 10.1111/j.0269-8463.2004.00817.x
  61. 61. White CR, Seymour RS (2005) Allometric scaling of mammalian metabolism. Journal of Experimental Biology 208: 1611–1619. doi: 10.1242/jeb.01501
  62. 62. Glazier DS (2005) Beyond the “3/4-power law”: variation in the intra- and interspecific scaling of metabolic rate in animals. Biological Reviews 80: 611–662 doi:10.1017/S1464793105006834.
  63. 63. White CR, Phillips NF, Seymour RS (2006) The scaling and temperature dependence of vertebrate metabolism. Biology Letters 2: 125–127 doi:10.1098/rsbl.2005.0378.
  64. 64. Sieg AE, O’Connor MP, McNair JN, Grant BW, Agosta SJ, et al. (2009) Mammalian metabolic allometry: do intraspecific variation, phylogeny, and regression models matter? The American Naturalist 174: 720–733 doi:10.1086/606023.
  65. 65. Capellini I, Venditti C, Barton RA (2010) Phylogeny and metabolic scaling in mammals. Ecology 91: 2783–2793. doi: 10.1890/09-0817.1
  66. 66. DeLong JP, Okie JG, Moses ME, Sibly RM, Brown JH (2010) Shifts in metabolic scaling, production, and efficiency across major evolutionary transitions of life. Proceedings of the National Academy of Sciences 107: 12941–12945 doi:10.1073/pnas.1007783107.
  67. 67. Chown S, Marais E, Terblanche J, Klok C, Lighton J, et al. (2007) Scaling of insect metabolic rate is inconsistent with the nutrient supply network model. Functional Ecology 21: 282–290. doi: 10.1111/j.1365-2435.2007.01245.x
  68. 68. Ehnes RB, Rall BC, Brose U (2011) Phylogenetic grouping, curvature and metabolic scaling in terrestrial invertebrates. Ecology Letters 14: 993–1000. doi: 10.1111/j.1461-0248.2011.01660.x
  69. 69. Huey RB, Gilchrist GW, Carlson ML, Berrigan D, Serra L (2000) Rapid evolution of a geographic cline in size in an introduced fly. Science 287: 308–309. doi: 10.1126/science.287.5451.308
  70. 70. Angilletta MJ (2009) Thermal adaptation: a theoretical and empirical synthesis. Oxford University Press. 305 p.
  71. 71. Kingsolver JG, Huey RB (2008) Size, temperature, and fitness: three rules. Evolutionary Ecology Research 10: 251–268.
  72. 72. Hamilton WJ (1973) Life’s color code. McGraw-Hill. 258 p.
  73. 73. Glazier DS (2006) The 3/4-power law is not universal: evolution of isometric, ontogenetic metabolic scaling in pelagic animals. BioScience 56: 325–332. doi: 10.1641/0006-3568(2006)56[325:tplinu];2
  74. 74. Lovegrove BG (2000) The zoogeography of mammalian basal metabolic rate. The American Naturalist 156: 201–219. doi: 10.1086/303383
  75. 75. Symonds MRE, Elgar MA (2002) Phylogeny affects estimation of metabolic scaling in mammals. Evolution 56: 2330–2333. doi: 10.1111/j.0014-3820.2002.tb00156.x
  76. 76. Killen SS, Atkinson D, Glazier DS (2010) The intraspecific scaling of metabolic rate with body mass in fishes depends on lifestyle and temperature. Ecology Letters 13: 184–193 doi:10.1111/j.1461-0248.2009.01415.x.