Skip to main content
Search

Search

Quick Search in Journals
Quick Search anywhere
Advanced Search
Log In | Register
Skip main navigation
menuDrawerCloseTextmenuDrawerOpenText
Previous article
Next article
Free

Phylogenetic Analysis Supports the Aerobic-Capacity Model for the Evolution of Endothermy

ORCIDs: Nespolo, http://orcid.org/0000-0003-0825-9618; Solano-Iguaran, http://orcid.org/0000-0002-1319-8568.

Abstract

The evolution of endothermy is a controversial topic in evolutionary biology, although several hypotheses have been proposed to explain it. To a great extent, the debate has centered on the aerobic-capacity model (AC model), an adaptive hypothesis involving maximum and resting rates of metabolism (MMR and RMR, respectively; hereafter “metabolic traits”). The AC model posits that MMR, a proxy of aerobic capacity and sustained activity, is the target of directional selection and that RMR is also influenced as a correlated response. Associated with this reasoning are the assumptions that (1) factorial aerobic scope (FAS; MMR/RMR) and net aerobic scope (NAS; MMR − RMR), two commonly used indexes of aerobic capacity, show different evolutionary optima and (2) the functional link between MMR and RMR is a basic design feature of vertebrates. To test these assumptions, we performed a comparative phylogenetic analysis in 176 vertebrate species, ranging from fish and amphibians to birds and mammals. Using disparity-through-time analysis, we also explored trait diversification and fitted different evolutionary models to study the evolution of metabolic traits. As predicted, we found (1) a positive phylogenetic correlation between RMR and MMR, (2) diversification of metabolic traits exceeding that of random-walk expectations, (3) that a model assuming selection fits the data better than alternative models, and (4) that a single evolutionary optimum best fits FAS data, whereas a model involving two optima (one for ectotherms and another for endotherms) is the best explanatory model for NAS. These results support the AC model and give novel information concerning the mode and tempo of physiological evolution of vertebrates.

Online enhancements:   appendix.

Introduction

A fundamental problem in evolutionary biology is understanding the factors that promote or constrain adaptive evolution and assessing the role of natural selection in this process. From many perspectives, this has been a successful research agenda, especially regarding morphological and life-history evolution (e.g., Schluter 1993; Kingsolver et al. 2001; Abzhanov et al. 2006; Grant and Grant 2006; Lee et al. 2014). In vertebrates, an important adaptation is endothermy, or the capacity to elevate body temperature by producing and storing metabolic heat (Hayes 2010; Nespolo et al. 2011). Endothermy appeared independently in birds and mammals and possibly did so as well in other groups of vertebrates (Seymour et al. 2004; Bernard et al. 2010). Thus, birds and mammals are considered “endotherms,” and fish, amphibians, and (nonavian) reptiles are considered “ectotherms” (see, however, other definitions in Cossins and Bowler 1987; Angilletta 2009; Clarke and Pörtner 2010).

A number of mechanistic hypotheses have been proposed to explain the initial steps in the transition from ecto- to endothermy (see recent reviews in Clarke and Pörtner 2010; Nespolo et al. 2011; Lovegrove 2012; Little and Seebacher 2014). One of the most popular is the aerobic-capacity model (hereafter “AC model”; Bennett and Ruben 1979; Hayes and Garland 1995). This hypothesis is based on the fact that aerobic capacity, normally measured as maximum metabolic rate under forced exercise (MMR), is roughly a constant fraction of the resting rate of metabolism (RMR) across taxa (in both ecto- and endotherms); however, the absolute, or “net,” aerobic scope (the difference between MMR and RMR, hereafter “NAS”) is 1 order of magnitude larger in endotherms than in ectotherms (Bennett and Ruben 1979; Clarke and Pörtner 2010). Subsequently, the factorial aerobic scope (FAS, i.e., MMR/RMR) and NAS are two variables with different biological meanings (Killen et al. 2016). Whereas FAS is supposed to be an invariant feature of vertebrates, NAS represents how much energy is allocated for sustained activity and differs between ecto- and endotherms (Clarke and Pörtner 2010; Killen et al. 2014). From an evolutionary perspective, the underlying assumption of the AC model is simple: high RMR evolved as a by-product of selection acting on the high aerobic capacity needed for sustained work (i.e., MMR). In turn, increasing RMR raised the maintenance costs that are responsible for the characteristic high body temperatures of endothermic animals. Once endothermy was achieved, body temperature was regulated at high, stable levels; however, this stability did not occur until RMR had increased enough to be able to support higher levels of sustained activity. Hence, the essential premise of the AC model is that RMR and MMR should be correlated across a wide range of vertebrates (Else et al. 2004; Hochachka and Burelle 2004; Weibel et al. 2004; Killen et al. 2016).

Most of the efforts to test the AC model have been directed toward finding a correlation between RMR and MMR, yet the evidence supporting the existence of such a correlation is inconclusive. Whereas interspecific studies in specific groups of vertebrates (e.g., rodents, passerine birds, teleost fish) in general have found a positive correlation between these traits (Bozinovic 1992; Dutenhoffer and Swanson 1996; Killen et al. 2016), intraspecific studies have provided variable results (Nespolo et al. 2005; Sadowska et al. 2005; Gębczyński and Konarzewski 2009; Wone et al. 2015). These apparent contradictions might be due to differences in taxonomic sampling coverage between inter- and intraspecific studies.

Although searching for the correlation between resting and maximum metabolic rates is perhaps the most recurring test of the AC model, the hypothesis is considerably more general. It was initially delineated by Bennett and Ruben (1979), but later Hayes and Garland (1995) formalized the model in the context of multivariate selection theory by implicitly assuming different forms of selection, correlated responses, and the existence of evolutionary optima. Furthermore, Nespolo et al. (2011) discussed possible empirical tests of the model, which later were explored with numerical methods (Nespolo and Roff 2014). These extensions of the model are perhaps more explicit but are not contradictory to the originally proposed idea of Bennett and Ruben (1979). Here we apply, for the first time, a family of phylogenetic statistical methods that make use of information theory for testing explicitly the predictions of the aerobic model in an evolutionary context. These predictions are summarized in the following paragraphs, as arising from the hypothetical scenario (endothermy evolution based on the AC model) proposed by Bennett and Ruben (1979).

In an ancestral population of vertebrates, a biological innovation generated a fitness advantage in individuals with comparatively high capacities for sustained activity (i.e., high aerobic capacity; Bennett and Ruben 1979, pp. 650–651). Subsequently, positive directional selection began to act on MMR, a measure of aerobic capacity. A structural coupling between RMR and MMR—a design feature of vertebrates—led to a correlated increase in RMR (Hayes and Garland 1995, p. 839). The existence of this link in present-day vertebrates, in the form of correlated evolution between MMR and RMR, is the first prediction that is tested in this study.

A long period of directional selection targeting aerobic capacity would eventually result in extensive evolutionary change in MMR (and also in RMR, as a correlated response), until an evolutionary optimum was attained (e.g., Butler and King 2004). Then, the form of selection (most likely) changed gradually from directional to stabilizing. An alternative, unrealistic scenario is to assume that there was constant directional selection or that aerobic capacity increased indefinitely without costs (see Hulbert 1990; Koteja 2004). In this scenario of rapid phenotypic change as a response to selection, the theory predicts that trait diversification should exceed random expectations. These expectations are assumed to follow a Brownian motion (BM) model of trait evolution (see Felsenstein 1973), which can be demonstrated using a “disparity-through-time” analysis (see “Methods”). This is the second prediction that we test here (i.e., average trait diversification is greater than expected by BM).

After an evolutionary optimum was attained, stabilizing selection maintained trait values around an observable, statistically distinguishable optimum. As a consequence, the actual diversity of metabolic traits would have the signature expected for an evolutionary model assuming selection (the Ornstein-Uhlenbeck model for trait evolution, hereafter the “OU model”), as traits would have been evolving toward one evolutionary optimum for a long period of time. This is the third prediction tested here: a model assuming evolutionary optima in MMR and RMR fits the data better than alternative models, including a white-noise model (see Hansen 1997; Butler and King 2004; Beaulieu et al. 2012; also see “Methods”).

In principle, the above predictions also hold for NAS and FAS, as both are descriptors of aerobic capacity. However, the conceptual differences between NAS and FAS outlined above give rise to the fourth prediction that we test. If FAS is an invariant feature of vertebrates, it should remain constant across lineages. In this case, it is expected that a white-noise (nonphylogenetic) model fits the data better. If selection is an important factor shaping observed FAS values, then a model assuming one evolutionary optimum should best fit the data. NAS, on the other hand is supposed to be a proxy of aerobic capacity (the hallmark of the AC model; see Bennett and Ruben 1979; Clarke and Pörtner 2010; Killen et al. 2016). Hence, it should show two optima, one for ectotherms and another for endotherms. In the following sections we test these predictions, using data from 176 vertebrate species.

Methods

We used the time-calibrated Global Timetree of Life (Hedges et al. 2015), which includes data with standard errors and branch lengths for 50,632 species and was built with data from 2,274 molecular studies. We trimmed this phylogenetic tree to remove species for which no MMR or RMR data were available; therefore, a subtree was generated with 176 vertebrate species in which major vertebrate groups (bony fish, sharks, amphibians, nonavian reptiles, birds, and mammals) were represented (fig. 1). According to the original megatree, the branch lengths in the subtree are in millions of years. The compilation of data for MMR and RMR (109 studies; table A1) was generated by conducting several literature searches (Google, Google Scholar, Web of Science) using terms including “aerobic capacity,” “ V ˙ O 2 max ,” and “aerobic scope.” Studies were considered only if aerobic capacity was measured by indirect calorimetry combined with oxygen consumption ( V ˙ O 2 ; Lighton 2008). Furthermore, we prioritized studies where RMR was measured in postabsorptive individuals and (for ectotherms) at routine temperatures. Studies on humans, specialized species such as marine mammals, and domesticated species were excluded from the analysis. In a few cases the measured species was not available in the phylogeny. In these cases we assigned the MMR or RMR datum to other species of the same genus (if available; sensu Rezende et al. 2002).

Figure 1.
Figure 1.

Our working phylogenetic tree. This is a subtree of the Global Timetree of Life (Hedges et al. 2015), which included 50,632 species and was based on 2,274 molecular studies. Silhouettes were kindly drawn by Alejandra Tejada.

View Large Image Download PowerPoint

In general, MMR was recorded as the highest V ˙ O 2 obtained during forced, intense exercise using a wheel, a swim channel, or a treadmill (see details in table A1). Only one data set per species was used, and trait values were averaged. Data were transformed from V ˙ O 2 units (in mg or mL per hour) to watts, using the conversion 20.1 J/mL O2 (Walsberg and Wolf 1995). RMR and MMR were log transformed before any analysis. The general reliability of the sample was checked by assessing the dispersion of the bivariate MMR-RMR relationship as well as by generating linear regressions of each variable with body mass. We checked the residuals of these models to ensure that they were normally distributed. No significant correlation was found between factorial aerobic scope (FAS; i.e., MMR/RMR) and body mass, but NAS (MMR − RMR) was significantly correlated with body mass. We systematically removed outliers when Cook’s D was greater than 0.5.

In order to remove body mass effects from MMR, RMR, and NAS, all subsequent analyses used residuals that were calculated with three different methods (giving identical final results): ordinary least squares with body mass (nonphylogenetic), ordinary least squares with body mass and measurement temperature (also nonphylogenetic), and phylogenetic regression using generalized least squares (GLS), assuming a covariance structure where a constant (lambda) multiplies internal branch lengths in the variance-covariance phylogenetic matrix (corPagel option, for the “gls” command in nlme and ape; Martins and Hansen 1997). Given that all these options gave similar results, only the last approach is presented (phylogenetic residuals from linear regressions with body mass; see below). To check whether statistical removal of body mass effects was complete, we correlated the residuals of RMR and MMR (adjusted r2 [hereafter “adj-r2”] = −0.01, not significant in both cases). To determine whether there was a correlation between MMR and RMR taking into consideration phylogenetic relationships (i.e., to test the first prediction), we explored the bivariate distribution of independent contrasts (Felsenstein 1985). Furthermore, we computed the phylogenetic regression, again using GLS with the corPagel option (Martins and Hansen 1997). Since the models are nested, both models (i.e., including and excluding phylogenetic relationships) were compared using likelihood ratio tests (LRTs).

To explore trait diversification through time (i.e., to test the second prediction), we generated disparity-through-time plots and calculated the morphological disparity index (MDI; “dtt” function in the GEIGER package; Harmon et al. 2003; Swenson 2014). Disparity is a measurement of trait divergence at each node of a phylogeny; these values were scaled to the whole phylogeny and compared to a null distribution produced by simulating trait evolution according to Brownian motion (BM; Harmon et al. 2003). The MDIs were computed by comparing the observed disparities with the median of the expected disparities obtained from the BM simulations. Then, a line was drawn to connect the observed and expected disparities in geometric space, and the MDIs were calculated as the area of the resulting polygons (Harmon et al. 2003; Swenson 2014). Negative MDI values were interpreted as decelerated rates of diversification (i.e., on average, trait evolution progressed at a rate less than that expected via a random-walk model), whereas positive MDIs were taken as evidence of a constant or accelerating rate of trait diversification (i.e., on average, trait evolution exceeded random-walk expectations; see examples of evolutionary inferences based on MDIs in Harmon et al. 2003; Slater et al. 2010; Arbour and López-Fernández 2013; Colombo et al. 2015; Pincheira-Donoso et al. 2015).

For fitting and comparing a series of alternative evolutionary models of metabolic trait evolution for all of the traits measured in this study (i.e., to test the third prediction), we used the “fitContinuous” command in the GEIGER package. This is a likelihood-based approach, where models with different assumptions are fitted to the data and then compared on the basis of their goodness of fit (see details in Butler and King 2004). For all traits, we fitted the BM model, the “early-burst” (EB) model, the OU model, and the white-noise model to the data. The BM model assumes that traits evolved according to BM. Conversely, in the EB model, character change tends to be concentrated toward the base of the tree, while the OU model assumes a tendency toward a central value, such as occurs under constant stabilizing selection (see extensive explanations and assumptions of the models in Butler and King 2004; Ingram et al. 2012). This model allows one to estimate θ, the primary optimum, a hypothetical entity representing the trait value reached by an “infinite number of populations identical to the common ancestor” evolving independently (Hansen 1997, p. 1342; Price and Hopkins 2015).

Finally, the white-noise model is equivalent to drawing trait values from a single normal distribution (assuming no phylogenetic covariance structure; Harmon et al. 2008; Muschick et al. 2014). After discarding the white-noise model as a best description of the data, we addressed our fourth prediction (i.e., the existence of one global optimum for FAS and two evolutionary optima for NAS). To this aim, we used the OUwie package (Beaulieu et al. 2012). This is a procedure similar to the previously described approach for which we used the GEIGER package (i.e., model selection based on information theory), but with the OUwie package we specifically aimed to discriminate different evolutionary optima. We fitted a BM model, an OU1 model (an OU model assuming one optimum), and an OUM model (i.e., an OU model assuming different optima for ectotherms and endotherms).

The selection of the best models was performed with the Akaike information criterion (AICc [AIC corrected for small sample size] and AIC weights; Burnham and Anderson 2002). All statistical procedures were performed in R (R Development Core Team 2013). All data are provided in table A1.

Results

The linear regression between (the residuals of) MMR and RMR was significant ( F 1 , 174 = 765.5 , adj‐ r 2 = 0.82 , P < .0001 ; fig. A1). In addition, the regression of independent contrasts forced to the origin was significant ( F 1 , 174 = 497.4 , adj‐ r 2 = 0.74 , P < .0001 ; fig. 2). This suggests that there is a strong correlation between RMR and MMR even when phylogenetic effects are not considered. These findings support the first prediction of the AC model (i.e., “correlated evolution” between RMR and MMR). The model including phylogenetic effects (using generalized least squares and a covariance structure assuming a proportional correlational structure) had a similar fit ( log‐likelihood [LL] = 114.61 ), compared to the model without such effects (i.e., assuming a “star” phylogeny, LL = 114.05 ; LRT = 1.12 , P < .29 ).

Figure 2.
Figure 2.

Bivariate plot of the independent contrasts of the residuals of maximum metabolic rate (MMR) and resting metabolic rate (RMR). The adjusted regression coefficient is shown for the linear regression between the contrasts forced to the origin ( P < .001 ).

View Large Image Download PowerPoint

The disparity-through-time plots showed, in general, that diversification exceeded random-walk expectations most of the (relative) time (see fig. 3 for NAS and fig. A2 for the other variables). This observation was confirmed by the morphological disparity index (MDI), which was positive and significant for all variables, suggesting that net diversification exceeded Brownian motion expectations (figs. 3, A2). These results support the second prediction of the AC model (i.e., “accelerated evolution” in metabolic traits).

Figure 3.
Figure 3.

Disparity-through-time plot for the residuals of net aerobic scope (NAS; maximum metabolic rate [MMR] − resting metabolic rate [RMR]), showing trait diversification (disparity, solid line) at each node in the phylogeny. On the X-axis, 0 represents the most basal node in the phylogeny and 1 the actual species. The expected disparity, according to a null distribution of 999 randomizations assuming Brownian motion, is depicted by the dashed line. The shaded area represents the 95% confidence intervals. The morphological disparity index (MDI) represents the area between the observed and the expected disparities. Positive MDI values represent accelerated evolution and negative MDI values decelerated evolution (see text for details; P < .001 after a comparison with a null distribution). Similar plots for MMR, RMR, and factorial aerobic scope (MMR/RMR) are presented in figure A2.

View Large Image Download PowerPoint

Fitting different evolutionary models to the data showed that the OU model fits the data better than the alternative models (including the white-noise model) for all of the traits, according to AIC weights (table 1). This supports the third prediction of the model (i.e., the existence of “evolutionary optima” despite random fluctuations). In addition, the diversification pattern of both traits can be visualized in the phenograms, or the plots of the traits according to the phylogenetic relationships (fig. 4). The phenograms show different trait distributions for NAS and FAS for ecto- and endotherms (fig. 4). This pattern is confirmed by the OUwie output (table 2), which suggests that a model assuming a single optimum was better ranked for FAS (global optimum θ = 2.9 ± 0.038 ), whereas a model assuming two optima was better ranked for NAS (ectotherm θ = 0.08 ± 0.03 ; endotherm θ = 0.40 ± 0.06 ). This supports the fourth prediction of the model (i.e., two optima for NAS and a single optimum for FAS). Overlapping the actual trait distribution (i.e., without considering the phylogeny) with the estimated NAS evolutionary optima shows that the optimum approaches the mean value for ectotherms but departs considerably from the mean for endotherms (fig. 5).

Table 1.

AICc values according to different models of evolutionary diversification of vertebrate metabolic traits
Trait BM OU EB White
Maximum metabolic rate (MMR) 42.11 14.18 44.18 63.28
 AIC weight for MMR 0 1 0 0
Resting metabolic rate (RMR) −570.43 −577.35 −568.36 −474.42
 AIC weight for RMR .03 .96 .01 0
Factorial aerobic scope (FAS; MMR/RMR) 46.87 7.95 48.94 49.52
 AIC weight for FAS 0 1 0 0
Net aerobic scope (NAS; MMR − RMR) 48.64 13.69 50.72 48.55
 AIC weight for NAS 0 1 0 0

Note. The best model (underlined) is the one with the highest AIC (Akaike information criterion) weight (or lowest AICc [AIC value corrected for small sample size]). BM = Brownian motion model; OU = Ornstein-Uhlenbeck model with a single optimum; EB = early-burst model; White = white-noise model. We used the residuals of the phylogenetic regressions when traits were correlated with body mass (MMR, RMR, NAS; see “Methods” and “Results” for details). All variables were transformed to log ⁡ 10 ( x + 10 ) before the analysis.

View Table Image
Figure 4.
Figure 4.

Phenograms (combined plots of phylogenetic relationships with trait values; each line represents a lineage and the tips each present-day species) of factorial aerobic scope (a) and net (additive) aerobic scope (b), showing trait diversification over time, for endotherms (birds and mammals) and ectotherms (all the rest). A complete version including species names is provided in figure A3. Time calibration was obtained from the original megatree (Hedges et al. 2015).

View Large Image Download PowerPoint
Table 2.

Results of the OUwie analysis
Model θ (±SE) AICc AICwi
Factorial aerobic scope (FAS):      
 BM1 2.85 ± .04 46.9 0
 OU1 2.86 ± .04 8.0 .733
 OUM 2.83 ± .8 10.0 .267
Net aerobic scope (NAS):      
 BM1 .13 ± .04 48.6 0
 OU1 .81 ± .03 13.7 .001
 OUM .40 ± .06 −.43 .999

Note. The θ value represents the estimated optimum according to different Ornstein-Uhlenbeck (OU) models fitted to the data. The models assume either Brownian motion (BM), a single optimum (OU1), or two optima (OUM). The best model (underlined) has the highest Akaike information criterion weight (AICwi; or lowest AICc [AIC value corrected for small sample size]). All variables were transformed to log ⁡ 10 ( x + 10 ) before the analysis. For NAS, residuals of regressions with body mass were used (see “Methods” for details). See also figures 4 and 5.

View Table Image
Figure 5.
Figure 5.

Kernel density plots showing the actual (residuals, nonphylogenetic) distribution of trait values for the net aerobic scope (maximum metabolic rate − resting metabolic rate) of ectotherms and endotherms in our data set. The different evolutionary optima obtained by the OUwie procedure are indicated by the dotted lines (see mean and standard errors in table 2).

View Large Image Download PowerPoint

Discussion

The key finding of this study is the support it offers for the AC model and the novel information it provides for the mode and tempo of physiological evolution of vertebrates, particularly regarding the AC model. As indicated in the “Introduction,” a basic premise of the AC model is that RMR and MMR are correlated across a wide range of vertebrates. There are many examples where different aspects of this principle have practical consequences: in the training of humans (Saltin and Rowell 1980; Hochachka and Burelle 2004) and animals (Thompson and Withers 1997; Eme et al. 2009), when comparing lifestyles among populations of a single species (Sadowska et al. 2009), or when comparing active and sedentary species (e.g., Gomes et al. 2004; Careau et al. 2010; Killen et al. 2016). Our results provide support for the idea that this principle—the link between RMR and MMR—is very general and represents a “pervasive or nearly pervasive” feature of vertebrates (Wone et al. 2009, p. 3701). In addition, other aspects or consequences of the AC model were also addressed in this work (see below).

Accelerated versus Decelerated Evolution

According to the disparity-through-time analysis, we detected accelerated evolution in both traits (i.e., trait diversification exceeding random-walk expectations). The concept of disparity was originally developed by paleontologists, who were characterizing the variance of morphology in the fossil record (Foote 1997; Ciampaglio et al. 2001). Later, disparity-through-time analysis and the morphological disparity index (MDI) were proposed as a scale-independent metric (it is scaled to the size of the phylogeny) that permitted comparisons among taxonomic groups (Harmon et al. 2003). Although we are not aware of any MDI computation for physiological traits, this metric has proven to be very informative in morphological studies exploring the evolutionary dynamics of a lineage, under a given ecological context (e.g., Colombo et al. 2015; Jonsson et al. 2015; Meloro et al. 2015). For instance, the detection of positive MDIs has signaled that lizard body size is an example of accelerated evolution (Pincheira-Donoso et al. 2015), but decelerated evolution was inferred for the same trait in cetaceans because of the detection of negative MDIs (Slater et al. 2010). A more specific contrast between morphological evolution and ecological-niche evolution was performed by Burbrink et al. (2012), who reported negative MDIs for morphology (decelerated change) but positive MDIs for ecological niches (accelerated change). Hence, in this example the change in ecological opportunity was not explained by the observed change in morphological diversification (Burbrink et al. 2012).

Accelerated evolution is predicted when species experience homogenizing forces such as directional or stabilizing selection. With this logic, traits that have a direct impact on fitness should exhibit accelerated evolution (e.g., brain weight and body size in fish; see Gonzalez-Voyer et al. 2009). In fact, when different kinds of traits are compared in several species of yeasts (under common-garden conditions), traits that are associated with ancient diversification and low differentiation (e.g., fermentation capacity) show negative MDIs, whereas fitness-related traits show positive MDIs (Hagman et al. 2013; Paleo-López et al. 2016; J. J. Solano-Iguaran, R. Paleo-López, J. F. Quintero-Galvis, J. Figueroa, and R. F. Nespolo, unpublished results). In this sense, the results provided in our study are provocative, as we obtained large and positive MDIs for all metabolic traits (range: 0.22–0.4); this strongly suggests that an evolutionary force in addition to Brownian motion acted during the majority of the clade’s evolutionary history for metabolic traits (see figs. 3, A2).

Aerobic Scope: Two Proxies and Two Meanings

Our results confirmed a classic pattern of variation in two widespread indexes of aerobic capacity, the factorial aerobic scope (FAS) and the net aerobic scope (NAS; see reviews in Hinds et al. 1993; Clark and Pörtner 2010). Several authors (especially those working with ectotherms) use and interpret NAS (e.g., Nilsson et al. 2009; Norin and Malte 2012; Killen et al. 2014; Auer et al. 2015), whereas others (especially those working with endotherms; see Hinds et al. 1993; Careau 2013; Chappell et al. 2013; Dawson et al. 2013; Schippers et al. 2014) report FAS. Our results are, in this sense, interesting, as they provide a basis for the distinction between FAS and NAS. In theory, FAS is an invariant measurement of the extreme capacity for energy turnover in relation to resting expenditure (Hinds et al. 1993; Clark et al. 2013). NAS, on the other hand, represents the maximum capacity for simultaneous aerobic processes above maintenance levels (Clark et al. 2013; Killen et al. 2014, 2016). Since NAS represents an amplitude that increases disproportionately in animals with high aerobic capacity (this can be seen in fig. A1 after comparing endotherms with ectotherms), our analysis revealed that it differentiates endotherms from ectotherms, whereas FAS does not. In this sense, FAS and NAS represent biologically different variables, and our results of one optimum for FAS and two optima for NAS confirm this (our fourth prediction), at the evolutionary level.

Approaches in Comparative Biology

Evolutionary physiologists either analyze single species submitted to experimental treatments (including comparisons of different populations) or develop comparative analyses among species (as in this study). The two approaches have different assumptions. Whereas in single-species studies the datum is the individual and the conclusions are drawn to the population, in comparative analyses the datum is the species (the trait mean, as in this study) and the conclusions apply to this particular sample of species, represented by this particular phylogeny. Hence, regarding the AC model, it is not surprising to find that conclusions drawn from intraspecific studies do not necessarily coincide with conclusions drawn from interspecific studies, as has been discussed previously (see Sadowska et al. 2005, Nespolo and Roff 2014, and Wone et al. 2015 for discussions).

An important advantage of incorporating several species in the analysis is the possibility of controlling for phylogenetic relationships, for which the independent contrasts provided the first available method (Felsenstein 1985; Garland and Adolph 1994). Later, it became evident that a null model of trait evolution via Brownian motion (see Felsenstein 1973; Butler and King 2004) is not the best model that can fit comparative data, especially when selection is (hypothetically) involved (Butler and King 2004). In order to test the prediction that metabolic traits have undergone adaptive evolution, we followed the approach implemented by Harmon et al. (2008), which compares the goodness of fit of several models of trait evolution, including models that explicitly assume selection (the “OU models”). From this, our results showed that the OU model best explained the observed variation in the metabolic traits considered in this study; this provides support for the idea that selection acted on vertebrate metabolic traits (our third prediction).

It must be recognized that we are not the first to have performed comparative analyses aimed at testing the predictions of the AC model. For instance, multiple-regression models have been applied to test the AC model (and other questions in evolutionary physiology) to show that climate is a significant predictor of MMR in birds (Swanson and Garland 2009). Furthermore, Dlugosz et al. (2013) have analyzed MMR scaling in mammals, using phylogenetic generalized least squares, and Dutenhoffer and Swanson (1996) have tested the AC model in passerine birds, using independent contrasts (see also Brischoux et al. 2011; Killen et al. 2016). All of these authors applied regression models with branch-length transformations (some of these transformation are based on OU models) to infer correlated selection between physiology and morphology or among physiological traits (similar to our analysis of correlated evolution between RMR and MMR discussed above). While these are valuable contributions, we believe that they address questions different from those presented here. We are not aware of any studies that have applied OU models sensu stricto to physiological traits in any group of animals. In this context, our results are an interesting addition to the competing ideas about endothermy evolution that have been proposed so far (see below).

Conclusion

According to Koteja (2004, pp. 1046–1047), the currently available models for the evolution of endothermy “do not make a progressive chain of hypotheses that asymptotically approaches a ‘true’ reconstruction of the history of birds and mammals.” Instead, the models (and especially comparative data) provide predictions that can be compared and contrasted with observed physiological diversity. This is especially important today, as our knowledge of vertebrate physiology, genetics, and evolution has increased substantially (e.g., Hochachka and Burelle 2004; Seebacher et al. 2006 Little and Seebacher 2014; Nespolo and Roff 2014; and references therein). This new knowledge has inspired new and integrated views of endothermy origin and evolution (e.g., Clarke and Pörtner 2010; Lovegrove 2012). In order to rank some of these models above others, we encourage authors to produce specific predictions that can be confirmed (or not) with observations. For the case of the AC model, we believe that this study provides such a test. We elaborated four specific predictions representing two different aspects of the AC model: the link between resting and maximum metabolic rates (for which we found support) and the signature of adaptive evolution (for which we found evidence). We also provided theoretical justification for using the net aerobic scope over the factorial aerobic scope when comparing aerobic capacity among species. We hope that these conclusions will inspire new analyses to contrast our findings with alternative models for the evolution of endothermy.

This study was funded by Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT) grant 1130750 to R.F.N. and CAPES award FB002-2014, line 3, to F.B. and R.F.N. J.J.S.-I. thanks a Comisión Nacional de Investigación Científica y Tecnológica (CONICYT) fellowship. We also thank P. Saenz-Agudelo and two anonymous reviewers for their insightful suggestions. We thank R. Paleo for assistance with the descriptions of the methods of forced exercise and E. Giles for assisting us with English grammar. We are very grateful to L. Revell for his kind and pedagogical introduction to modern phylogenetic methods and the phytools package.

Appendix Supplementary Table and Figures

Table A1.

Complete data set used in this study
                References Method of stimulation
Class, species Order Family State Tb (ºC) Mb (g) RMR (W) MMR (W) RMR MMR
Actinopterigii:                    
 Albula vulpes Albuliformes Albulidae Ectotherm 22.5 348 .3099 .7042 Murchie et al. 2011 Murchie et al. 2011 Recirculating water tunnel
 Carassius auratus Cypriniformes Cyprinidae Ectotherm 20 87 .0119 .0135 Smit 1965 Smit 1965 Annular chamber
 Notemigonus crysoleucas Cypriniformes Cyprinidae Ectotherm 20.6 5.3 .0027 .0037 Beecham et al. 2009 Beecham et al. 2009 Swim tunnel (swimming)
 Micropterus salmoides Perciformes Centrarchidae Ectotherm 20 150 .0656 .4865 Beamish 1970 Beamish 1970 Recirculating water tunnel
 Centropomus undecimalis Perciformes Centropomidae Ectotherm 23 74.3 .0170 .1104 Tolley and Torres 2002 Tolley and Torres 2002 Swim tunnel (swimming)
 Oreochromis mossambicus Perciformes Cichlidae Ectotherm 28 63 .0219 .0633 Febry and Lutz 1987 Febry and Lutz 1987 Swim tunnel (swimming)
 Stizostedion vitreum Perciformes Percidae Ectotherm 8 10 .0015 .0395 Beamish 1990 Beamish 1990 Swim tunnel (swimming)
 Argyrosomus japonicus Perciformes Sciaenidae Ectotherm 22 390 .1589 .7947 Fitzgibbon et al. 2007 Fitzgibbon et al. 2007 Swim tunnel (swimming)
 Platichthys flesus Pleuronectiformes Pleuronectidae Ectotherm 15 395 .0567 .3318 Duthie 1982 Duthie 1982 Respiration chambers (swimming)
 Limanda limanda Pleuronectiformes Pleuronectidae Ectotherm 15 395.8 .0924 .3787 Duthie 1982 Duthie 1982 Swim tunnel (swimming)
 Oncorhynchus mykiss Salmoniformes Salmonidae Ectotherm 15 207.5 .0683 .4319 Dickson and Kramer 1971 Dickson and Kramer 1971 Annular chamber
 Oncorhynchus nerka Salmoniformes Salmonidae Ectotherm 15 55 .0153 .1924 Brett 1964 Brett 1964 Recirculating water tunnel
 Salvelinus namaycush Salmoniformes Salmonidae Ectotherm 12 20 .0054 .0495 Beamish 1990 Beamish 1990 Swim tunnel (swimming)
Amphibia:                    
 Discoglossus pictus Anura Alytidae Ectotherm 20 30.7 .0063 .0456 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Bombina orientalis Anura Bombinatoridae Ectotherm 20 2.6 .0008 .0069 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Rhinella marina Anura Bufonidae Ectotherm 22 252 .0619 .6893 Withers et al. 1988 Withers et al. 1988 Manual rotation of chamber
 Anaxyrus americanus Anura Bufonidae Ectotherm 25 40 .0078 .1864 Taigen and Pough 1981 Taigen and Pough 1981; Pough and Kamel 1984 Motorized rotation of chamber
 Anaxyrus boreas Anura Bufonidae Ectotherm 25 26.1 .0147 .1861 Hillman and Withers 1979 Hillman and Withers 1979 Manual rotation of chamber
 Epidalea calamita Anura Bufonidae Ectotherm 20 8.7 .0028 .0378 Taigen et al. 1982 Taigen et al. 1982 Manual rotation of chamber
 Anaxyrus cognatus Anura Bufonidae Ectotherm 25 39.59 .0250 .2586 Seymour 1973 Seymour 1973 Motorized rotation of chamber
 Anaxyrus woodhousii Anura Bufonidae Ectotherm 20 67.9 .0246 .5589 Fitzpatrick and Atebara 1974 Walsberg 1986 Manual rotation of chamber
 Colostethus inguinalis Anura Dendrobatidae Ectotherm 20 1.55 .0008 .0063 Taigen and Pough 1983 Taigen and Pough 1983 Motorized rotation of chamber
 Dendrobates auratus Anura Dendrobatidae Ectotherm 20 2.02 .0008 .0099 Taigen and Pough 1983 Taigen and Pough 1983 Motorized rotation of chamber
 Colostethus nubicola Anura Dendrobatidae Ectotherm 20 .27 .0002 .0008 Taigen and Pough 1983 Taigen and Pough 1983 Manual prodding in water
 Eleutherodactylus coqui Anura Eleutherodactylidae Ectotherm 20 4.1 .0010 .0067 Taigen et al. 1982 Taigen et al. 1982; Taigen and Pough 1983 Motorized rotation of chamber
 Pseudacris crucifer Anura Hylidae Ectotherm 19 1.3 .0008 .0080 Taigen et al. 1982; Taigen and Beuchat 1984 Taigen et al. 1982; Taigen and Beuchat 1984 Motorized rotation of chamber
 Pseudacris regilla Anura Hylidae Ectotherm 20 2.76 .0009 .0042 Bennett and Licht 1973 Bennett and Licht 1973 Electrical stimuli via electrodes
 Osteopilus septentrionalis Anura Hylidae Ectotherm 20 5 .0018 .0182 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Agalychnis callidryas Anura Hylidae Ectotherm 20 5.7 .0019 .0167 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Smilisca fodiens Anura Hylidae Ectotherm 20 15.1 .0027 .0335 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Hyla cinerea Anura Hylidae Ectotherm 27 5.1 .0038 .0290 Prestwich et al. 1989 Prestwich et al. 1989 Forced to hop on a wet floor
 Hyla arenicolor Anura Hylidae Ectotherm 20 3.4 .0018 .0158 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Hyla chrysoscelis Anura Hylidae Ectotherm 20 5.47 .0039 .0300 Kamel et al. 1985 Kamel et al. 1985 Motorized rotation of chamber
 Hyla gratiosa Anura Hylidae Ectotherm 29 13.85 .0074 .0963 Prestwich et al. 1989 Prestwich et al. 1989 Manual rotation of chamber
 Hyla squirella Anura Hylidae Ectotherm 28 2.2 .0020 .0220 Prestwich et al. 1989 Prestwich et al. 1989 Manual rotation of chamber
 Hyla versicolor Anura Hylidae Ectotherm 20 6.1 .0034 .0347 Kamel et al. 1985 Kamel et al. 1985 Motorized rotation of chamber
 Hyperolius viridiflavus Anura Hyperoliidae Ectotherm 25 .9 .0005 .0037 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Semnodactylus wealii Anura Hyperoliidae Ectotherm 20 6.3 .0019 .0230 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Kassina senegalensis Anura Hyperoliidae Ectotherm 20 3 .0013 .0139 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Engystomops pustulosus Anura Leptodactylidae Ectotherm 25 1.78 .0014 .0187 Bucher et al. 1982 Ryan et al. 1983 Manual prodding in water
 Odontophrynus americanus Anura Leptodactylidae Ectotherm 20 15.2 .0031 .0490 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Gastrophryne carolinensis Anura Microhylidae Ectotherm 20 1.9 .0006 .0086 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Kaloula pulchra Anura Microhylidae Ectotherm 20 30.7 .0045 .1193 Taigen et al. 1982 Taigen et al. 1982 Motorized rotation of chamber
 Xenopus laevis Anura Pipidae Ectotherm 18 22.1 .0100 .0891 Hillman and Withers 1981 Hillman and Withers 1981 Manual rotation of chamber in air
 Pyxicephalus adspersus Anura Pyxicephalidae Ectotherm 20 481.15 .0794 1.9203 Loveridge and Withers 1981 Loveridge and Withers 1981 Manual prodding in water
 Rana pipiens Anura Ranidae Ectotherm 10 38.4 .0090 .0315 Seymour 1973 Seymour 1973 Motorized rotation of chamber
 Lithobates catesbeiana Anura Ranidae Ectotherm 20 43.58 .0092 .0387 Seymour 1973 Seymour 1973 Motorized rotation of chamber
 Lithobates sylvaticus Anura Ranidae Ectotherm 20 12.7 .0060 .0528 Taigen et al. 1982 Taigen et al. 1982; Taigen and Beuchat 1984 Motorized rotation of chamber
 Spea hammondii Anura Scaphiopodidae Ectotherm 15 10.88 .0042 .0243 Seymour 1973 Seymour 1973 Motorized rotation of chamber
 Ambystoma gracile Caudata Ambystomatidae Ectotherm 15 26.43 .0035 .0113 Feder 1976a Feder 1977, 1978a Electrical stimuli via electrodes
 Ambystoma jeffersonianum Caudata Ambystomatidae Ectotherm 15 7.06 .0018 .0023 Feder 1976a Feder 1977, 1978a Electrical stimuli via electrodes
 Ambystoma macrodactylum Caudata Ambystomatidae Ectotherm 15 2.79 .0006 .0016 Feder 1976a Feder 1977, 1978a Electrical stimuli via electrodes
 Ambystoma tigrinum Caudata Ambystomatidae Ectotherm 15 36.25 .0086 .0157 Hutchison et al. 1977 Hutchison et al. 1977 Electrical stimuli via electrodes
 Amphiuma means Caudata Amphiumidae Ectotherm 18 104 .0272 .0823 Withers and Hillman 1981 Withers and Hillman 1981 Manual prodding in water
 Amphiuma tridactylum Caudata Amphiumidae Ectotherm 25 493 .0550 .1734 Preslar and Hutchison 1978 Preslar and Hutchison 1978 Electrical stimuli via electrodes in water
 Ensatina eschscholtzii Caudata Plethodontidae Ectotherm 14 3.47 .0007 .0022 Feder 1976b Feder 1977, 1978a Electrical stimuli via electrodes
 Plethodon glutinosus Caudata Plethodontidae Ectotherm 15 4.69 .0008 .0035 Feder 1976b Feder 1977, 1978a Electrical stimuli via electrodes
 Pseudoeurycea bellii Caudata Plethodontidae Ectotherm 15 13.13 .0011 .0043 Feder 1976b Feder 1977, 1978a Electrical stimuli via electrodes
 Aneides lugubris Caudata Plethodontidae Ectotherm 23 5.49 .0008 .0211 Feder 1976b Hillman et al. 1979 Manual rotation of chamber
 Batrachoseps attenuatus Caudata Plethodontidae Ectotherm 15 .78 .0002 .0008 Feder 1976b Feder 1977, 1978a Electrical stimuli via electrodes
 Bolitoglossa occidentalis Caudata Plethodontidae Ectotherm 25 .61 .0001 .0006 Feder 1978b Feder 1977, 1978a Electrical stimuli via electrodes
 Bolitoglossa subpalmata Caudata Plethodontidae Ectotherm 13 1.67 .0003 .0023 Feder 1987 Feder 1978a Motorized rotation of chamber
 Desmognathus ochrophaeus Caudata Plethodontidae Ectotherm 15 1.45 .0002 .0032 Feder 1985 Feder 1986 Sliding lead weight
 Desmognathus quadramaculatus Caudata Plethodontidae Ectotherm 15 20.05 .0024 .0076 Feder 1976b Feder 1977, 1978a Electrical stimuli via electrodes
 Eurycea longicauda Caudata Plethodontidae Ectotherm 15 1.41 .0003 .0016 Feder 1976b Feder 1977, 1978a Electrical stimuli via electrodes
 Gyrinophilus porphyriticus Caudata Plethodontidae Ectotherm 15 7.34 .0010 .0040 Feder 1976b Feder 1977, 1978a Electrical stimuli via electrodes
 Plethodon jordani Caudata Plethodontidae Ectotherm 15 1.83 .0003 .0015 Stefanski et al. 1989 Stefanski et al. 1989 Manual prodding
 Pseudoeurycea gadovii Caudata Plethodontidae Ectotherm 15 2.35 .0005 .0014 Feder 1976b Feder 1977, 1978a Electrical stimuli via electrodes
 Pseudoeurycea smithi Caudata Plethodontidae Ectotherm 15 4.55 .0006 .0025 Feder 1976b Feder 1977, 1978a Electrical stimuli via electrodes
 Pseudotriton ruber Caudata Plethodontidae Ectotherm 15 10.8 .0013 .0057 Feder 1976b Feder 1977, 1978a Electrical stimuli via electrodes
 Necturus maculosus Caudata Proteidae Ectotherm 15 101.9 .0068 .0176 Miller and Hutchison 1979 Miller and Hutchison 1979 Electrical stimuli via electrodes in water
 Notophthalmus viridescens Caudata Salamandridae Ectotherm 15 1.48 .0003 .0012 Feder 1976a Stefanski et al. 1989 Manual rotation of chamber
 Taricha torosa Caudata Salamandridae Ectotherm 15 10.41 .0016 .0047 Feder 1978b Feder 1977, 1978a Electrical stimuli via electrodes
 Geotrypetes seraphini Gymnophiona Dermophiidae Ectotherm 20 1.93 .0004 .0017 Bennett and Wake 1974 Bennett and Wake 1974 Electrical stimuli via electrodes
Birds:                    
 Anas castanea Anseriformes Anatidae Endotherm 39 956.55 4.0964 20.6024 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Anas rubripes Anseriformes Anatidae Endotherm 39 1,026 6.2635 79.3817 Berger et al. 1970 Berger et al. 1970 Mask (flight)
 Patagona gigas Apodiformes Trochilidae Endotherm 37 20 .3015 2.1738 Lasiewski et al. 1967 Rezende et al. 2002 Cold-induced MMR
 Sephanoides sephanoides Apodiformes Trochilidae Endotherm 37 6 .1062 .8879 López-Calleja and Bozinovic 1995 López-Calleja and Bozinovic 1995 Cold-induced MMR
 Columba livia Columbiformes Columbidae Endotherm 39 331.75 1.4537 9.4588 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Colinus virginianus Galliformes Odontophoridae Endotherm 38 218 1.1926 7.1436 Swanson and Weinacht 1997 Swanson and Weinacht 1997 Cold-induced MMR
 Coturnix coturnix Galliformes Phasianidae Endotherm 39 154 1.9773 4.3844 Warncke et al. 1988 Warncke et al. 1988 Treadmill locomotion (running)
 Cardinalis cardinalis Passeriformes Cardinalidae Endotherm 39 43 .5028 3.0437 Hinds and Calder 1973 Rezende et al. 2002 Cold-induced MMR
 Pheucticus ludovicianus Passeriformes Cardinalidae Endotherm 34.3 41 .5539 3.0098 Dutenhoffer and Swanson 1996 Dutenhoffer and Swanson 1996 Cold-induced MMR
 Phytotoma rara Passeriformes Cotingidae Endotherm 40.2 42 .5791 2.7408 Rezende et al. 2001 Rezende et al. 2001 Cold-induced MMR
 Junco hyemalis Passeriformes Emberizidae Endotherm 39.5 17 .3151 1.9692 Swanson 1990 Swanson 1990 Cold-induced MMR
 Spizella arborea Passeriformes Emberizidae Endotherm 33.9 19 .4391 2.5583 Dutenhoffer and Swanson 1996 Dutenhoffer and Swanson 1996 Cold-induced MMR
 Spizella passerina Passeriformes Emberizidae Endotherm 39 11 .2137 1.2005 Rezende et al. 2002 Rezende et al. 2002 Cold-induced MMR
 Spizella pusilla Passeriformes Emberizidae Endotherm 34.4 13 .2642 1.5443 Dutenhoffer and Swanson 1996 Dutenhoffer and Swanson 1996 Cold-induced MMR
 Zonotrichia capensis Passeriformes Emberizidae Endotherm 42 20 .3696 1.8199 Novoa et al. 1990 Novoa et al. 1990 Cold-induced MMR
 Taeniopygia guttata Passeriformes Estrildidae Endotherm 39 11.55 .2278 1.3498 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Carduelis flammea Passeriformes Fringillidae Endotherm 39 14 .2892 1.7037 Rosenmann and Morrison 1974 Rosenmann and Morrison 1974 Cold-induced MMR
 Carpodacus mexicanus Passeriformes Fringillidae Endotherm 39 22 .3918 2.1676 O’Connor 1995 O’Connor 1995 Cold-induced MMR
 Coccothraustes vespertinus Passeriformes Fringillidae Endotherm 39 59 .7972 11.2876 Berger et al. 1970 Berger et al. 1970 Short flights
 Icterus galbula Passeriformes Icteridae Endotherm 39 32 .5020 2.4259 Rezende et al. 2002 Rezende et al. 2002 Cold-induced MMR
 Dumetella carolinensis Passeriformes Mimidae Endotherm 34.5 34 .5201 2.7293 Dutenhoffer and Swanson 1996 Dutenhoffer and Swanson 1996 Cold-induced MMR
 Baeolophus griseus Passeriformes Paridae Endotherm 35 17 .3075 1.8676 Cooper 1997 Cooper 1997 Cold-induced MMR
 Poecile atricapillus Passeriformes Paridae Endotherm 39.8 13 .2758 1.9079 Cooper and Swanson 1994 Cooper and Swanson 1994 Cold-induced MMMR
 Poecile gambeli Passeriformes Paridae Endotherm 35 11 .2493 1.5640 Cooper 1997 Cooper 1997 Cold-induced MMR
 Dendroica coronata Passeriformes Parulidae Endotherm 38.5 12 .2499 1.3190 Swanson and Dean 1999 Swanson and Dean 1999 Cold-induced MMR
 Dendroica petechia Passeriformes Parulidae Endotherm 32.1 9 .1869 .9782 Dutenhoffer and Swanson 1996 Dutenhoffer and Swanson 1996 Cold-induced MMR
 Passer domesticus Passeriformes Passeridae Endotherm 40 26 .3033 2.2845 Daan et al. 1990 Koteja 1986 Cold-induced MMR
 Sitta carolinensis Passeriformes Sittidae Endotherm 39 20 .3696 2.2497 Liknes and Swanson 1996 Liknes and Swanson 1996 Cold-induced MMR
 Sturnus vulgaris Passeriformes Sturnidae Endotherm 39 73 1.0584 9.4119 Torre-Bueno 1978 Torre-Bueno 1978; Torre-Bueno and Larochelle 1978 Wind tunnel (flight)
 Diuca diuca Passeriformes Thraupidae Endotherm 41 34 .4120 3.0482 Sabat et al. 2010 Rezende et al. 2002 Cold-induced MMR
 Troglodytes aedon Passeriformes Troglodytidae Endotherm 33.7 10 .1898 1.3024 Dutenhoffer and Swanson 1996 Dutenhoffer and Swanson 1996 Cold-induced MMR
 Contopus virens Passeriformes Tyrannidae Endotherm 35.2 14 .2149 1.2950 Dutenhoffer and Swanson 1996 Dutenhoffer and Swanson 1996 Cold-induced MMR
 Tyrannus tyrannus Passeriformes Tyrannidae Endotherm 39 37 .4585 2.6996 Rezende et al. 2002 Rezende et al. 2002 Cold-induced MMR
 Vireo gilvus Passeriformes Vireonidae Endotherm 37.9 13 .2293 1.3760 Swanson 1995 Swanson 1995 Cold-induced MMR
 Zosterops lateralis Passeriformes Zosteropidae Endotherm 39.2 11 .1412 .9604 Maddocks and Geiser 1999 Maddocks and Geiser 1999 Cold-induced MMR
 Picoides pubescens Piciformes Picidae Endotherm 39 25 .4271 2.6586 Liknes and Swanson 1996 Liknes and Swanson 1996 Cold-induced MMR
 Melopsittacus undulatus Psittaciformes Psittaculidae Endotherm 39 35 .6650 3.6750 Tucker 1973 Tucker 1973 Wind tunnel (flight)
 Eudyptula minor Sphenisciformes Spheniscidae Endotherm 38.5 1,031.35 4.4581 19.2325 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Glaucidium nanum Strigiformes Strigidae Endotherm 39 98 .8042 3.6490 Rezende et al. 2002 Rezende et al. 2002 Cold-induced MMR
 Dromaius novaehollandiae Struthioniformes Dromaiidae Endotherm 39 37.5 .0525 .6054 Grubb et al. 1983 Grubb et al. 1983 Rubber mask–treadmill (running)
Chondrichthyes:                    
 Negaprion brevirostris Carcharhiniformes Carcharhinidae Ectotherm 22 1,050 .5545 1.3224 Bushnell et al. 1989 Bushnell et al. 1989 Swim tunnel (swimming)
 Scyliorhinus canicula Carcharhiniformes Scyliorhinidae Ectotherm 16 .19 .2473 .5231 Ferry-Graham and Gibb 2001 Ferry-Graham and Gibb 2001 Peak postfeeding oxygen consumption
 Dasyatis violacea Myliobatiformes Dasyatidae Ectotherm 20 10,700 1.2167 5.1931 Ezcurra 2001 Ezcurra 2001 RMR and MMR considered as minimum and maximum routine metabolic rates, respectively
 Squalus acanthias Squaliformes Squalidae Ectotherm 10 2,000 .1809 .4935 Brett and Blackburn 1978 Brett and Blackburn 1978 Swim tunnel (swimming)
Mammalia:                    
 Genetta tigrina Carnivora Viverridae Endotherm 39 1,595 4.1707 51.8493 Hennemann and Konecny 1980 Taylor et al. 1980 Forced exercise in enclosed running-wheel respirometer
 Sminthopsis crassicaudata Dasyuromorphia Dasyuridae Endotherm 35 15.8 .1072 1.1020 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Sminthopsis macroura Dasyuromorphia Dasyuridae Endotherm 32.7 8.59 .1172 .7972 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Petaurus breviceps Diprotodontia Petauridae Endotherm 34.9 122 .4689 2.7599 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Bettongia penicillata Diprotodontia Potoroidae Endotherm 37.2 961.55 3.1451 23.4695 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Potorous tridactylus Diprotodontia Potoroidae Endotherm 35.8 932.55 2.9408 40.9971 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Oryctolagus cuniculus Lagomorpha Leporidae Endotherm 38.3 1,236.9 .0813 .3549 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Dromiciops gliroides Microbiotheria Microbiotheriidae Endotherm 35.15 32.39 .1773 1.6427 Bozinovic et al. 2004 Dlugosz et al. 2013 Forced exercise in enclosed running-wheel respirometer
 Ornithorhynchus anatinus Monotremata Ornithorhynchidae Endotherm 32 1,214.6 3.6408 16.7003 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Tachyglossus aculeatus Monotremata Tachyglossidae Endotherm 32 3,202.2 2.5992 15.7591 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Cavia porcellus Rodentia Caviidae Endotherm 39 584 3.3253 10.9148 Turner et al. 1995 Turner et al. 1995 Forced exercise in enclosed running-wheel respirometer
 Meriones unguiculatus Rodentia Cricetidae Endotherm 39 67.7 .3130 3.7849 Chappell et al. 2007 Chappell et al. 2007 Forced exercise in enclosed running-wheel respirometer
 Notomys alexis Rodentia Cricetidae Endotherm 38 33 .2244 1.4068 White et al. 2006 White et al. 2006 Forced exercise in enclosed running-wheel respirometer
 Baiomys taylori Rodentia Cricetidae Endotherm 36 7.25 .0795 .6310 Hudson 1965 Seeherman et al. 1981 Forced exercise in enclosed running-wheel respirometer
 Peromyscus leucopus Rodentia Cricetidae Endotherm 27 24.35 .4297 1.6429 Segrem and Hart 1967 Segrem and Hart 1967 Forced exercise in enclosed running-wheel respirometer
 Mesocricetus auratus Rodentia Cricetidae Endotherm NA 114.5 .7032 3.7546 Pasquis et al. 1970 Pasquis et al. 1970 Forced exercise in enclosed running-wheel respirometer
 Myodes glareolus Rodentia Cricetidae Endotherm NA 23.64 .3200 1.7786 Gorecki 1968 Sadowska 2009 Forced exercise in enclosed running-wheel respirometer
 Lemmiscus curtatus Rodentia Cricetidae Endotherm 36.7 24.7 .1154 .9658 McNab 1992 Dlugosz et al. 2013 Forced exercise in enclosed running-wheel respirometer
 Liomys salvini Rodentia Heteromyidae Endotherm 33.2 40.15 .2795 1.7250 Hulbert et al. 1985 MacMillen and Hinds 1992 Forced exercise in enclosed running-wheel respirometer
 Heteromys desmarestianus Rodentia Heteromyidae Endotherm 33.8 79.4 .5527 2.8805 Hinds and MacMillen 1985 MacMillen and Hinds 1992 Forced exercise in enclosed running-wheel respirometer
 Chaetodipus fallax Rodentia Heteromyidae Endotherm 32.6 19.7 .1440 1.2192 Hulbert et al. 1985 MacMillen and Hinds 1992 Forced exercise in enclosed running-wheel respirometer
 Microdipodops megacephalus Rodentia Heteromyidae Endotherm 32.8 12.25 .1721 .9948 Hinds and MacMillen 1985 MacMillen and Hinds 1992 Forced exercise in enclosed running-wheel respirometer
 Dipodomys ordii Rodentia Heteromyidae Endotherm 34.6 43.25 .3687 1.7847 Hinds and MacMillen 1985 Dlugosz et al. 2013 Forced exercise in enclosed running-wheel respirometer
 Dipodomys merriami Rodentia Heteromyidae Endotherm 34.1 35.31 .2910 1.6028 Hinds and MacMillen 1985 Dlugosz et al. 2013 Forced exercise in enclosed running-wheel respirometer
 Dipodomys panamintinus Rodentia Heteromyidae Endotherm 33.9 68.42 .4344 2.9308 Hinds and MacMillen 1985 Dlugosz et al. 2013 Forced exercise in enclosed running-wheel respirometer
 Apodemus flavicollis Rodentia Muridae Endotherm 36.7 29.26 .2415 3.0130 Haim and Izhaki 1995 Dlugosz et al. 2013 Forced exercise in enclosed running-wheel respirometer
 Rattus colletti Rodentia Muridae Endotherm 36.2 165.65 .6866 3.8552 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Conilurus penicillatus Rodentia Muridae Endotherm 35.9 213.2 .9077 4.9739 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Rattus villosissimus Rodentia Muridae Endotherm 36 250.6 .8139 4.8500 Hinds et al. 1993 Hinds et al. 1993 Cold-induced MMR
 Octodon degus Rodentia Octodontidae Endotherm 37.6 173 1.0696 3.0681 Arends and McNab 2001 Dlugosz et al. 2013 Forced exercise in enclosed running-wheel respirometer
 Pedetes capensis Rodentia Pedetidae Endotherm 35.9 2,650 4.4296 97.4687 Müller et al. 1979 Seeherman et al. 1981 Forced exercise in enclosed running-wheel respirometer
 Tamias merriami Rodentia Sciuridae Endotherm 38.2 75 .4396 2.4828 Wunder 1970 Wunder 1970 Forced exercise in enclosed running-wheel respirometer
 Tamias minimus Rodentia Sciuridae Endotherm 38 40.1 .4562 2.3502 Jones and Wang 1976 Dlugosz et al. 2013 Forced exercise in enclosed running-wheel respirometer
 Ammospermophilus leucurus Rodentia Sciuridae Endotherm 40.5 100.51 .4165 4.8222 Hudson and Deavers 1976 Dlugosz et al. 2013 Forced exercise in enclosed running-wheel respirometer
Reptilia:                    
 Alligator mississippiensis Crocodylia Alligatoridae Ectotherm 35 300 .1005 .3349 Emshwiller and Gleeson 1997 Emshwiller and Gleeson 1997 Treadmill locomotion (running)
 Crocodylus porosus Crocodylia Crocodylidae Ectotherm 32 1,030 .1345 2.3666 Owerkowicz and Baudinette 2008 Owerkowicz and Baudinette 2008 Treadmill locomotion (running)
 Ophisaurus ventralis Squamata Anguidae Ectotherm 25 32.17 .0069 .2276 Kamel and Gatten 1983 Kamel and Gatten 1983 Burst of forced activity
 Anniella pulchra Squamata Anniellidae Ectotherm 25 4.94 .0018 .0134 Kamel and Gatten 1983 Kamel and Gatten 1983 Burst of forced activity
 Pituophis catenifer Squamata Colubridae Ectotherm 30 548 .1377 1.4256 Greenwald 1971 Greenwald 1971 Electric shocks; RMR and MMR estimated from graphic interpolation
 Thamnophis butleri Squamata Colubridae Ectotherm 25 19.04 .0063 .0495 Kamel and Gatten 1983 Kamel and Gatten 1983 Burst of forced activity
 Heloderma suspectum Squamata Helodermatidae Ectotherm 25 463.9 .1476 1.5383 John-Alder et al. 1983 John-Alder et al. 1983 Treadmill locomotion (walking)
 Dipsosaurus dorsalis Squamata Iguanidae Ectotherm 40 80.99 .0684 .6858 Donovan and Gleeson 2008 Donovan and Gleeson 2008 Treadmill locomotion (running)
 Sauromalus hispidus Squamata Iguanidae Ectotherm 37.5 574 .2820 1.7880 Bennett 1972 Bennett 1972 Electrical stimuli via electrodes
 Ctenosaura similis Squamata Iguanidae Ectotherm 35 690.16 .4025 3.5728 Donovan and Gleeson 2008 Donovan and Gleeson 2008 Treadmill locomotion (running)
 Iguana iguana Squamata Iguanidae Ectotherm 35 2149.83 1.0595 9.1848 Donovan and Gleeson 2008 Donovan and Gleeson 2008 Treadmill locomotion (running)
 Uta stansburiana Squamata Phrynosomatidae Ectotherm 35 3.76 .0106 .0717 Donovan and Gleeson 2008 Donovan and Gleeson 2008 Treadmill locomotion (running)
 Sceloporus occidentalis Squamata Phrynosomatidae Ectotherm 35 19.77 .0272 .2488 Donovan and Gleeson 2008 Donovan and Gleeson 2008 Treadmill locomotion (running)
 Anolis carolinensis Squamata Polychrotidae Ectotherm 20 5.1 .0018 .0088 Gatten 1985 Gatten 1985 Forced exercise by shaking and inverting the chamber
 Tupinambis nigropunctatus Squamata Teiidae Ectotherm 35 1,089 .8268 4.0852 Bennett and John-Alder 1984 Bennett and John-Alder 1984 Treadmill locomotion (walking)
 Trogonophis wiegmanni Squamata Trogonophidae Ectotherm 25 4.97 .0011 .0158 Kamel and Gatten 1983 Kamel and Gatten 1983 Burst of forced activity
 Varanus caudolineatus Squamata Varanidae Ectotherm 35 14.03 .0126 .5247 Thompson and Withers 1997 Thompson and Withers 1997 Treadmill locomotion (running)
 Varanus brevicauda Squamata Varanidae Ectotherm 35 17.44 .0151 .3148 Thompson and Withers 1997 Thompson and Withers 1997 Treadmill locomotion (running)
 Varanus eremius Squamata Varanidae Ectotherm 35 37.94 .0344 .5130 Thompson and Withers 1997 Thompson and Withers 1997 Treadmill locomotion (running)
 Varanus acanthurus Squamata Varanidae Ectotherm 35 63.51 .0348 .9490 Thompson and Withers 1997 Thompson and Withers 1997 Treadmill locomotion (running)
 Varanus rosenbergi Squamata Varanidae Ectotherm 35 1788.5 1.5854 10.3833 Thompson and Withers 1997 Thompson and Withers 1997 Treadmill locomotion (running)
 Terrapene ornata Testudines Emydidae Ectotherm 20 354 .0184 .4061 Gatten 1974 Gatten 1974 Electrical stimuli via electrodes
 Trachemys scripta Testudines Emydidae Ectotherm 20 305 .0186 .4706 Gatten 1974 Gatten 1974 Electrical stimuli via electrodes
 Chrysemys picta Testudines Emydidae Ectotherm 25 179 .0362 .1466 Stockard and Gatten 1983 Stockard and Gatten 1983 Spontaneous activity in diurnally active animals

Note. Tb = measuring temperature; Mb = body mass; RMR = resting metabolic rate; MMR = maximum metabolic rate; NA = not available.

View Table Image: 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9
Figure A1.
Figure A1.

Bivariate plots showing the scaling of (log) body masses and maximum metabolic rate (MMR; a), resting metabolic rate (RMR; b), factorial aerobic scope (FAS; c) and net aerobic scope (NAS; d).

View Large Image Download PowerPoint
Figure A2.
Figure A2.

Disparity-through-time plot for the residuals of resting metabolic rate (RMR; a), the residuals of maximum metabolic rate (MMR; b), and factorial aerobic scope (FAS, i.e., MMR/RMR; c), showing trait diversification (disparity, solid line) at each node in the phylogeny. On the X-axis, 0 represents the most basal node in the phylogeny and 1 the actual species. The expected disparity, according to a null distribution of 999 randomizations assuming Brownian motion, is depicted by the dashed line. The shaded area represents the 95% confidence interval. The morphological disparity index (MDI) represents the area between the observed and expected disparities. Positive MDI values represent accelerated evolution and negative MDI values decelerated evolution (see text for details; P < .001 after a comparison with a null distribution).

View Large Image Download PowerPoint
Figure A3.
Figure A3.

Phenograms (combined plots of phylogenetic relationships with trait values; a high-resolution image showing all species names is available on request) of factorial aerobic scope (a) and net (additive) aerobic scope (b), showing trait diversification over time, for endotherms (birds and mammals) and ectotherms (all the rest). Time calibration was obtained from the original megatree (Hedges et al. 2015).

View Large Image Download PowerPoint

Literature Cited

References Cited Only in the Online Appendixes

Previous article
Next article