Scaling of fibre area and fibre glycogen concentration in the hindlimb musculature of monitor lizards: implications for locomotor performance with increasing body size

Any factor that limits muscle force production is destined to contribute to limits on animal locomotor performance. The maximum force-generating capacity and maximum contraction speed of a muscle are important factors in determining how forcefully and quickly animals can move their limbs, ultimately limiting speed, manoeuvrability and locomotor performance (Biewener, 2005; Biewener and Patek, 2018). These limits become even more critical as animals increase in size because a muscle's force production capacity depends on cross-sectional area, which increases at a lower rate than body mass (∝M_b^2/3) (Dick and Clemente, 2017). Among geometrically similar animals, this biomechanical challenge suggests that larger animals will be proportionally weaker than smaller animals. As a result, larger animals compensate with allometric shifts in limb posture or musculoskeletal architecture and suffer reduced maximum sprint speed and increased duty factor (Cieri et al., 2021; Clemente et al., 2013; Dick and Clemente, 2017). For example, many mammals compensate for predicted decreases in proportional strength by adopting a more erect posture, thereby increasing the effective mechanical advantage of the extensor muscles and allowing animals to locomote effectively without substantial changes in muscle architecture (Biewener, 1989, 2005).

In varanid lizards, however, where posture does not vary across a substantial body size range (Clemente et al., 2011), allometric increases in muscle mass, physiological cross-sectional area (PCSA) and fascicle length occur in many hindlimb and forelimb muscles (Cieri et al., 2020; Dick and Clemente, 2016). Similar trends are displayed in felids, another tetrapod group that also maintain a crouched posture even at larger body sizes, however the allometry is less pronounced than varanids and is not significant when accounting for phylogeny (Cuff et al., 2016a,b). Sprawling locomotion was widespread in stem tetrapods, thus allometric increases in muscle size were likely important for the locomotion of larger animals even before the rise of animals with parasagittal gaits (Dick and Clemente, 2017).

Increased PCSA in the muscles of larger species can be accomplished either by larger animals simply having more muscle fibres in parallel (increasing fibre number, hyperplasia) or by increasing the cross-sectional area of individual muscle fibres (fCSA, hypertrophy). The fibres themselves can also vary in biochemistry. Larger muscle fibres, which typically are more glycolytic, have a higher tension cost (ATP required to generate tension) than smaller fibres, which tend to be oxidative (Bottinelli et al., 1994; Rome and Lindstedt, 1997). The small fCSA of slow oxidative fibres facilitates timely oxygen diffusion that is necessary for aerobic metabolism (Hardy et al., 2009; Kinsey et al., 2011; Van Wessel et al., 2010). By contrast, fibres with larger fCSA generally power muscle activity via glycolytic metabolism and are capable of generating larger forces because a greater proportion of their internal area is occupied by contractile myofilaments (reviewed in Schiaffino and Reggiani, 2011). Fast-twitch muscles, which use glycolytic metabolism, are often associated with anaerobic sprinting in lizards (Bonine et al., 2005; Jayne et al., 1990), suggesting that slow-twitch muscles are primarily involved in slower locomotion or body support. Although the relationship between muscle design and locomotion is complex, this correlation between fCSA and metabolic capacity is generally consistent within individual limb muscles and for different muscles within an organism in several species of lizards (Acevedo and Rivero, 2006; Gleeson and Harrison, 1988; Gleeson et al., 1980; Schiaffino and Reggiani, 2011) and in many, but not all, mammals (Rome and Lindstedt, 1997; but see Bottinelli and Reggiani, 2000).

Adding to this complexity, absolute fCSA differs substantially among individual muscles within an animal (Allen et al., 2014; Cieri et al., 2020; Cuff et al., 2016a; Dick and Clemente, 2016), fibre type distribution varies with muscle size (Eng et al., 2008) and other anatomical and physiological aspects of muscle function vary markedly between species. Thus, to investigate these trade-offs in muscle design, it would be ideal to compare multiple muscles across a large range of body sizes, in animals of similar shape and ecology, and to control for muscle fibre type.

Although few explicit comparisons have been made, the main driver of the difference in overall muscle size among species of different body sizes is presumably muscle fibre number (Rehfeldt et al., 1999). Despite fibre division being restricted to the early stages of development in humans, mosaic hyplerplasia allows fishes to continue recruiting new fibres throughout development (Johnston, 2006; Jorgenson et al., 2020). Data from black sea bass (Centropristus striata) suggest that differences in muscle size between relatively small individuals are due mainly to hypertrophy but that hyperplasia becomes more important at greater body sizes (Priester et al., 2011). The relationship between body size and fCSA both within a muscle and within individual fibre types, however, remains unclear. Among other factors, larger animals could theoretically increase muscle force-generating capacity by shifting to a higher proportion of fast-twitch fibres, yet when compared with smaller animals, larger animals generally appear to have more slow-twitch fibres (Schiaffino and Reggiani, 2011). The fact that slower muscle fibres have smaller fCSA than fast fibres (Rome and Lindstedt, 1997) suggests that average fCSA should decrease with body size, but support for this relationship is conflicting.

In mammals, individual muscle fCSA is often reported to either be invariant with body size (Rome and Lindstedt, 1997) or increase with body size generally (Eisenberg, 1983); however, the available morphological evidence suggests (albeit modestly) that fCSA increases with body size. For example, previous studies in mammals report slight positive scaling of individual muscle fCSA (∝M_b^0.05) (Hoppeler and Flück, 2002) and diameter (∝M_b^{0.02–0.0375}) (Seow and Ford, 1991). In reptiles, a variety of results for the scaling of muscle fCSA in locomotor muscles have been found (Bonine et al., 2001, 2005; Gleeson, 1983; Gleeson and Harrison, 1986, 1988; Young et al., 1990). In fishes and invertebrates, multiple studies have shown an increase in fibre diameter and reduction in fibre number with body size (Johnston et al., 2003, 2004, 2012).

Unlike results for fCSA, relationships between muscle fibre type and body mass have been more clearly identified in many taxa. In mammals, locomotor muscles are generally composed of fast fibre types (higher glycogen concentration) with higher shortening velocities (v) and maximum shortening velocities (v_max) in small animals, with the overall fibre composition becoming slower (lower glycogen concentrations) with increasing body size (reviewed in Schiaffino and Reggiani, 2011; Thomas et al., 1994). Lizards, however, do not show a similar transition from faster to slower muscle fibre phenotypes with increasing body size either within a species (Gleeson and Harrison, 1986) or among species (Bonine et al., 2001, 2005). In reptiles, three fibre types have been identified: (1) fast-twitch glycolytic (FG) which produce high force, power, v and v_max, but fatigue quickly; (2) slow oxidative (SO) fibres that produce lower force and power, v and v_max, but fatigue slowly; and (3) fast-twitch oxidative-glycolytic (FOG) fibres which produce intermediate force, v and v_max, and power, but have intermediate fatigue resistance (Gleeson and Johnston, 1987; Gleeson et al., 1980; Peter et al., 1972; Rome et al., 1990; Scales et al., 2009). Fast-twitch muscles are activated in varanids for high-speed burst locomotion (Jayne et al., 1990).

The metabolic properties of muscle cells related to fuelling locomotor activity also vary across species and body sizes. Enzymes responsible for glycolytic metabolism (e.g. lactate dehydrogenase) scale positively with body mass within several lizard species, including black spiny-tailed iguanas (Ctenosaura similis) (Garland, 1984) and central netted dragons (Ctenophorus nuchalis) (Garland and Else, 1987), as well as widely in fishes (Moyes and Genge, 2010). In adult male desert iguanas, however, total muscle enzymatic activity (glycolytic and oxidative combined) decreased with body size (Gleeson and Harrison, 1986). If positive scaling of glycolytic metabolic capacity is present or the percentage of fast-twitch fibres increases, it would enable the maintenance of high sprint speeds with increasing body mass by increasing anaerobic metabolic power for muscle contractions (Somero and Childress, 1980). This may be a compensatory mechanism, in combination with larger muscles, to partially overcome the challenges imposed by geometric scaling in the disparity of muscle force with body mass (Somero and Childress, 1980). Muscle performance characteristics have also been shown to vary with body size in the iliotibialis of lacertid lizards, maximal tetanus strength and muscle power increased, but fatigue resistance decreased with body size (James et al., 2015). Thus, there appears no clear consensus on how we might expect these variables to change with size across different taxonomic groups. Many studies – including several covering substantial body size ranges (Hardy et al., 2010; Jimenez et al., 2008; Kinsey et al., 2007) – report muscle fibre size measurements (e.g. Gleeson and Johnston, 1987; Johnston et al., 2004; Kielhorn et al., 2013; Velten et al., 2013; Gleeson and Harrison, 1986), but a comparative quantitative analysis of these data should provide greater insight into the general patterns regarding how muscle fibre type, muscle glycolytic capacity, or muscle enzyme capacity might be expected to change with size.

Here, we examined the scaling relationships of fCSA and muscle glycogen concentration in 11 hindlimb locomotor muscles from varanid lizards ranging in body mass from 105 g to 40,000 g (fCSA) and 265 g to 4025 g (glycogen concentration). Varanid lizards are an ideal clade for understanding scaling relationships of the musculoskeletal system in response to changes in body mass because they retain similar sprawling posture and locomotor kinematics across an enormous range in body size (Clemente et al., 2011; Dick and Clemente, 2016, 2017; Thompson et al., 1997). Our previous work has shown allometric increases in muscle mass and PCSA with body size in varanid forelimb and hindlimb muscles, however muscle force-generating capacity could also be increased by means of an increased proportion of faster fibre types. We predicted that, given the constraints imposed by the geometric scaling of muscle force: (1) both individual fCSA and fibre number would increase with body size in order to increase the force-generating capacity of locomotor muscles, and (2) muscle glycogen concentration (as an index of muscle fibre type) would increase with body size, reflecting a greater capacity for the higher peak forces with increased body size involved in varanid locomotion (Cieri et al., 2021). fCSA and glycogen content may also scale more clearly against muscle mass rather than body mass, given that muscle mass has been shown to scale with positive allometry in these animals (Cieri et al., 2020; Dick and Clemente, 2016).

Variation in specimen tissue preservation did not allow us to directly determine muscle fibre type using traditional methods. Instead, we measured muscle glycogen concentration, which represents the amount of fuel within a muscle cell for rapid anaerobic muscle activities that are associated with faster muscle fibre types (reviewed in Schaeffer and Lindstedt, 2013). While not a direct proxy for muscle fibre type, this trait does provide information regarding the glycolytic capacity of a muscle fibre, is generally correlated with fibre type (Burke et al., 1971; Schaart et al., 2004) and should therefore be relatively greater in more glycolytic (fast-twitch) muscle fibres. Finally, our understanding of how muscle architecture traits change with body size generally can best be examined by controlling for phylogenetic independence and comparing traits across multiple clades. We therefore also sought to contextualize our findings in a broader comparative trend by examining how fCSA varies with body size in animals using data available in published literature.

Eleven hindlimb muscles (Table 1) were dissected (Fig. 1B) from 15 varanid lizards, ranging in body size from 105 g to 40,000 g (Fig. 1A). Animals (except for Varanus komodoensis, which were cadavers from captive animals) were wild caught using a variety of pit trapping, funnel trapping, and hand foraging techniques. Animals that were obviously injured, sick or malnourished were excluded from the study. Lizards were collected under permits WISP11435612 (QLD), SF009075 (WA), 61540 (NT), 08-001092-5 (WA) and WA0001919 (QLD); and ethics SBS/195/12/ARC (QLD), ANA16104 (QLD) and RA/3/100/1188 (WA). The age of animals is not reported as they were largely caught from the wild, but the vast majority of the animals caught were males. No systematic differences in muscle architecture were detected between males and females previously (Dick and Clemente, 2016). The following animals were included for the study of fCSA and glycogen concentration: Varanus panoptes (2 individuals:= 997 g; 1060 g), Varanus tristis (1 individual: 265 g), Varanus varius (3 individuals: 810 g, 834 g, 4025 g).

Additional data were collected from these animals: Varanus gouldii (3 individuals: 429 g, 439 g, 459 g), V. komodoensis (2 individuals: 30,000 g, 40,000 g), V. panoptes (2 individuals: 1060 g, 2077 g), Varanus scalaris (1 individual: 158 g), V. tristis (1 individual: 265 g), V. varius (3 individuals: 810 g, 834 g, 4025 g), which may have been frozen at least once before analysis. Whole muscles were dissected and weighed. Fresh animals were fasted for 24 h prior to being euthanized.

During dissections, the whole muscle was excised and relevant muscle parameters were measured (Dick and Clemente, 2016). Next, the mid-belly of the muscle was determined to be half the distance between the origin and the insertion. The muscle was cut perpendicular to the fibre orientation at two points approximately 0.5 cm apart. This section was then fixed in 10% formalin and then dehydrated in ethanol. Each sample was mounted in a tissue cassette, oriented perpendicular to the long axis of the muscle and embedded in wax. Oblique sections of muscle fibres were identified where diameter in one axis was more than twice the diameter in the perpendicular axis and excluded from the analysis. Wax-mounted muscles were thin-sliced (1 µm) and mounted onto glass slides. Slides were rehydrated and stained for glycogen using the stains and protocol in a commercial periodic acid-Schiff kit (Sigma-Aldrich 395B-1KT). Briefly, the staining protocol was as follows: slides were incubated at 60°C for 30 min in an oven to melt the wax covering, and then deparaffinized and hydrated. Hydration consisted of the following steps with the slides vertically mounted in plastic slide staining racks and solutes in successive wells of a 12-well slide staining dish: submersion in xylene (twice for 5 min), submersion in 100% ethanol (twice for 5 min), and submersion in 80% ethanol (twice for 5 min). Still in the vertical slide staining racks and using the staining dish wells, slides were then immersed in periodic acid solution (1 g dl⁻¹ H₂O) for 5 min, rinsed in several changes of deionized water, immersed in Schiff's reagent for 15 min, and washed in running tap water for 5 min. Slides were then removed from the vertical staining racks, and hematoxylin solution (gill no. 3) was applied for 90 s with a plastic transfer pipette. Slides were then rinsed in running tap water, dehydrated back to xylene using a reverse of the hydration protocol in vertical staining racks (80% ethanol twice for 5 min, 100% ethanol twice for 5 min, and xylene twice for 5 min), and a slide cover was immediately applied with slide adhesive.

Photographs of slides were taken at 40× or 100× magnification using a Nikon Eclipse E200 light microscope and a Basler acA1920-150uc USB area scan camera (Basler AG, Arhrensburg, Germany). Muscle fibre areas were calculated in ImageJ (v.1.53a, National Institutes of Health) using the hand-traced polygon selection tool after the field of view was calibrated to a 1000 μm length scale at each magnification using a calibration slide. The areas of 10 fibres were calculated in four different, randomly selected regions on each muscle slide from 1–4 slides per muscle per individual. Fibre areas were calculated instead of fibre diameters because they are more directly related to changes in muscle force output, which relates to whole muscle physiological cross-sectional area (the cross-sectional area of muscles perpendicular to muscle fibre). In only slides where staining led to clear differentiation, fibre counts were performed using the multipoint tool in ImageJ. In four different regions of each muscle, the dark, medium and light fibres were counted, corresponding to high, medium and low glycogen concentrations, respectively. Other muscle parameters, including PCSA, muscle mass and fascicle length were measured previously (Cieri et al., 2020; Dick and Clemente, 2016). All fCSA values were multiplied by a conversion factor of 1.63212 to account for the reduction in fibre cross-sectional area that occurs as a result of 1 month of ethanol storage according to published decay equations (Leonard et al., 2021a,b). Storage in ethanol for 1 month after fixation in formalin causes a decrease in muscle mass of 41% compared with fresh tissue (Leonard et al., 2021a) and reductions in fCSA relate to reductions in muscle mass over longer periods by a reduced major axis regression [relative fCSA=(relative muscle mass×10,113.5)+5.9272; R²=0.9648] (Leonard et al., 2021b) such that fCSA can be expected to decrease by 61.27%. To control for differences in specimen preparation, fCSA was calculated among all samples, and also among a subset of the data where specimens were never frozen prior to analysis. Incorrect freezing of tissue can lead to the formation of larger ice crystals which dehydrate muscle fibres, leading to underestimates of fibre area (Leonard et al., 2021a). Thus, scaling exponents calculated from all samples may provide an underestimate, but cover a much wider range of body sizes, while the fresh samples may yield more accurate exponents but cover a narrower range of body sizes.

For the comparative analysis, fibre areas and body masses were taken from the literature for mammals (Hoppeler and Flück, 2002; Kielhorn et al., 2013; Kroeger et al., 2020; Marx et al., 2006; Seow and Ford, 1991; Velten et al., 2013), birds (Brown et al., 2019), phrynosomatid lizards (Bonine et al., 2005), as well as fish and invertebrates (Jimenez et al., 2013). Fibre diameters were converted into areas assuming a circular shape. For birds, data from the domestic chicken (Gallus domesticus) were excluded because the pectoralis muscle architecture of this species has likely been under intense artificial selection. In cases where fast and slow muscle fibre areas were provided for the same individual, these were averaged. In cases where the original articles provided details about muscle preservation, appropriate correction factors were calculated as above using regressions from Leonard et al. (2021a,b).

Statistical analyses were conducted using R 3.6.3 (https://www.r-project.org/). The relationships between anatomical parameters (body mass, muscle mass, muscle physiological cross-sectional area, fascicle length) and histological parameters (fCSA, proportion of low-glycogen fibres) were assessed using mixed effects models (LMMs) using the lmer.R function from the lme4 package (https://CRAN.R-project.org/package=lme4), including animal (individual) nested within species as a random factor (Bates et al., 2015). To investigate the overall scaling relationships of all muscle fibres together, muscle was also included as a random factor. The scaling of muscle fibre number was calculated by dividing the scaling of total muscle cross-sectional area by the scaling of fCSA. As a conservative measure, the relative proportion of low-glycogen content fibres only was analysed and was calculated as the ratio of light-stained category fibres over total counted stained fibres. All continuous variables were log-transformed to promote normality and homoscedasticity in data distribution. Ratio values are not log-transformed and scaling exponents for these variables are from semi-log regressions. Individual fixed effect regression slopes (Table 2) were extracted using sim_slopes.R and plotted using interact_plot.R from the interactions package (https://CRAN.R-project.org/package=interactions). Intercepts (Table 2) were calculated directly from the mixed effects model outputs. Effect sizes for LMMs are presented as η² from the t_to_eta2.R function from the effectsize package (Ben-Shachar et al., 2020). Other relationships were tested using Kruskal–Wallis tests, and post hoc comparisons were made with Dunn's tests with Bonferroni corrections using the dunn.test package (https://CRAN.R-project.org/package=dunn.test). For the comparative analysis, mixed effects models were used to investigate the relationship between body size and fCSA for each taxonomic group, holding species as a random effect.

Phylogenetically-corrected slopes for fCSA data were calculated using the gls.R function from the nlme package in R (https://CRAN.R-project.org/package=nlme), under the assumption of correlation=corBrownian(1,tree). Brownian motion was chosen over early burst, Ornstein–Uhlenbeck or Lambda models as the evolutionary model because it was best supported based on AIC. A phylogenetic tree containing all species (Fig. S1) was created from the Timetree database (Kumar et al., 2017). Where body masses are not reported for mammals, these were estimated from Pacifici et al. (2013). Post hoc comparisons between groups in the gls models were calculated using the posthoc.R function from the package postHoc (https://CRAN.R-project.org/package=postHoc).

Across all individuals, the smallest fibres were found in tibialis anterior (tibAnt; 3945±1939 µm²), and the largest fibres were found in femorotibialis (femTib; 10,169±5262 µm²). Raw and body size-normalized values are shown in Table S1. fCSA scaled positively with body mass taking all muscles together among all samples (∝M_b^0.197 LMM; d.f.=128.6, η²=0.42, P<0.0001; body size: 104.5–40,000 g) and among a smaller range of body sizes representing fresh samples only (∝M_b^0.278 LMM; d.f.=53.19, η²=0.34, P<0.0001; body size: 845–4025 g). Considering each muscle separately, fCSA scaled positively with body mass in all muscles (Fig. 2, Table 3). Among fresh samples, fCSA scaled positively with body mass in all muscles except for the caudofemoralis longus (cFem) and iliotibialis (ilTib). Among all samples, whole muscle cross-sectional area (PCSA) scaled ∝M_b^0.708 (LMM; d.f.=13.19, η²=0.97, P=0.001) so the number of muscle fibres therefore scaled ∝M_b^0.511 assuming that all increases in PCSA were due to fCSA or fibre number. Among the smaller range of body sizes representing fresh samples only, whole muscle cross-sectional area scaled ∝M_b^0.673 (LMM; d.f.=10.79, η²=0.97, P=0.001) so the number of muscle fibres therefore scaled ∝M_b^0.395 assuming that all increases in PCSA were due to changes in fCSA or fibre number. PCSA increased with fCSA among all samples (log–log slope: 0.377, LMM; d.f.=504.35, η²=0.03, P<0.0001; Fig. 3) and among a smaller range of body sizes representing fresh samples only (log–log slope: 0.454, LMM; d.f.=214.08, η²=0.05, P<0.005).

When muscle data from multiple taxonomic groups were aggregated and regressed in a phylogenetic framework, fCSA was found to scale ∝M_b^0.182 (GLS; P<0.0001). We further compared the scaling of fCSA between groups (GLS; P<0.001), finding that the scaling exponent was different between groups (Fig. 4, Table 3). fCSA scaled much more steeply with body mass in invertebrates (∝M_b^0.575) than in other groups, more steeply in fish (∝M_b^0.347) and non-varanid reptiles (∝M_b^0.308) than in varanids (∝M_b^0.267). fCSA scales still more steeply with body mass in varanids, than in birds (pectoralis muscle only) (∝M_b^0.134) and mammals (∝M_b^0.122). Significant differences in scaling exponents were detected using post hoc comparisons between invertebrates and mammals (P=0.014), and birds and invertebrates (P=0.022).

There was no significant relationship between the proportion of low-glycogen fibres with either body mass or muscle mass when all muscles were considered together in linear mixed effects models where animal and muscle were held as random effects. When muscle was included as a fixed effect, the proportion of low-glycogen fibres decreased significantly with body mass in the iliofibularis (ilFib; semi-log slope: −49.25; LMM; d.f.=175.83, η²=0.09, P=0.001) and femTib (semi-log slope: −36.40; LMM; d.f.=179.94, η²=0.06, P=0.011).

fCSA within each glycogen content category (low, medium, high) were also found to increase with body size: low-glycogen category fibres (∝M_b^0.330 LMM; d.f.=54.06, η²=0.09, P<0.05), medium-glycogen category fibres (∝M_b^0.317 LMM; d.f.=55.31, η²=0.09, P<0.05) and high-glycogen category fibres (∝M_b^0.269 LMM; d.f.=55.05, η²=0.07, P<0.05). When the fCSA for each glycogen concentration category was multiplied by the proportion of that category to form a weighted average, all fibres together scaled (∝M_b^0.337 LMM; d.f.=187.00, η²=0.11, P<0.001).

Muscle fCSA was significantly smaller in low-glycogen (1.10±0.11 µm²) than either medium (1.15±0.10 µm²) or high-glycogen (1.14±0.10 µm²) fibres (χ²=25.24, d.f.=2, P<0.001). Pairwise comparisons indicated significant differences in fCSA between all glycogen concentration fibre groups.

Here, we report the scaling of muscle fibre cross-sectional area (fCSA) in 11 hindlimb locomotor muscles of varanid lizards ranging in body mass from 105 g to 40,000 g and muscle glycogen concentration in 11 muscles of varanid lizards ranging in body mass from 265 g to 4025 g. The first main finding of this study is that muscle fCSA increased with body mass (∝M_b^{0.278–0.197}), corresponding to a scaling of fibre diameter ∝M_b^{0.140–0.099}. This finding supports our main hypothesis that increases in fCSA (analogous to hypertrophy) would contribute along with increases in fibre number (analogous to hyperplasia) to allometric increases in muscle PCSA with body size. The true scaling exponent is likely above the lower presented value because it is calculated from a dataset that includes once-frozen animals of higher body size, and the muscles from these animals may have shrunk because of dehydration before analysis (Leonard et al., 2021a). The true exponent may be closer to the higher presented value because this exponent was calculated using animals that were never frozen, but is based on data from animals covering a substantially smaller range of body sizes. Increases in fCSA made less of a contribution than increases in fibre number (analogous to hyperplasia). However, this increase is not due solely to changes in fibre type, as we also show that fCSA increases independently in fibres with high, low and medium levels of glycogen. These results are consistent with previous studies that also display increased fCSA in the iliofibularis muscle with body mass in lizards (Bonine et al., 2001, 2005; Gleeson and Harrison, 1986). On the other hand, Gleeson and Harrison (1988), found that fCSA increased with body size in Dipsosaurus in the fast glycolytic fibres of the gastrocnemius only, with no relationship in the iliofibularis or caudofemoralis. In our data, fCSA increased with body size in the iliofibularis and caudofemoralis, but scaling exponents for these muscles were among the lowest of all muscles.

Increases in fCSA partially compensate for the disproportionally higher force requirements of locomotion and body support at larger body sizes. As geometrically similar sprawling animals increase in size, force production should scale ∝M_b^0.67 as it depends on muscle cross-sectional area, while locomotor forces should scale ∝M_b¹ because they depend on animal mass. Thus, larger animals must become relatively stronger or change shape to locomote similarly to smaller animals. In varanid lizards, which do not change substantially in overall shape or posture (Clemente et al., 2011), peak ground reaction forces scale ∝M_b^0.89–0.99 (Cieri et al., 2021) and while PCSA and mass of the forelimb and hindlimb muscles scales with positive allometry, this is not sufficient to fully explain the scaling of ground reaction forces. An increased fCSA means that muscle fibres should generate more total force because of a greater number of sarcomeres in parallel (Bruce et al., 1997; Krivickas et al., 2011). Force per unit cross-sectional area in the iliofibularis of Sceloporus torquatus is greater in males, which are larger than females and have fibres with higher fCSA, even though the relative proportion of fibre types is not different between sexes (Quintana et al., 2014).

The implications of increased fCSA for sprint speed, however, are much less straightforward. Even though larger fCSA is associated with greater force production (Krivickas et al., 2011; Quintana et al., 2014), fCSA in the iliofibularis has been shown to be negatively correlated with sprint speed, despite being the best predictor of it, in adult male desert iguanas (Gleeson and Harrison, 1988). This could be due to a trade-off in muscle design, whereby larger fibres produce more force, but at the cost of decreased oxidative capacity (owing to limitations in the speed of oxygen diffusion) and, thus, decreased efficiency and endurance (Kinsey et al., 2007, 2011). If short-term fatigue due to muscle fibre diffusion limitations is present, the larger fCSA may partially explain the lower sprint speeds in desert iguanas (Gleeson and Harrison, 1988) and with body size in varanids (Clemente et al., 2009b) by imposing an upper limit on sustained locomotor activity through an increase in the fatigability of muscle fibres. A considerable body of work has shown, however, that most animal muscle fibres grow and are organized to avoid theoretical diffusion limitations, but are instead limited by reaction rates (Kinsey et al., 2007, 2011). If this is the case, fCSA may be responding more to changes in body mass in varanids, and sprint speed may instead be limited by safety factors imposed by the musculoskeletal system (Cieri et al., 2021; Clemente et al., 2011; Dick and Clemente, 2017).

Our second main finding, that fCSA increases generally with body size in different groups, but much more steeply in fishes and invertebrates, more steeply in non-varanid lizards and somewhat more steeply in varanids (Fig. 4), provides broad comparative insights into the locomotor consequences of increased fCSA. The optimal fibre size hypothesis suggests that larger fibres may be metabolically optimal because the lower surface area to volume ratio in larger fibres reduces the metabolic cost of maintaining the resting membrane potential (Jimenez et al., 2011, 2013; Johnston et al., 2003, 2004, 2006). In smaller muscle fibres, the cost per muscle contraction is lower but the considerable costs (Milligan and McBride, 1985; Zurlo et al., 1990) associated with maintaining the membrane potential are greater (Jimenez et al., 2013). This may be advantageous in endothermic mammals and birds, which maintain a higher basal metabolic rate and move more regularly (Bennett and Ruben, 1979; Farmer, 2003, 2020). Ectotherms, such as reptiles and invertebrates, however, which have lower basal metabolic rates and move less frequently, may benefit from larger muscle fibres that cost more to contract but have lower costs of membrane potential maintenance (Jimenez et al., 2013). Under this paradigm, the scaling exponent of varanids, which is higher than in birds and mammals but lower than in other reptiles may be explained by the elevated maximal metabolic rates of varanids compared with other reptiles (Clemente et al., 2009a; Thompson et al., 1997). Similarly, the steeper scaling of fCSA with body size in fishes and invertebrates (most species with published fCSA values are marine) may also reflect habitat differences, as the muscle activation necessary to support the body against gravity in terrestrial animals may select for smaller fibres with relatively lower contraction costs at the expense of higher maintenance costs.

The third main finding is that the proportion of fibres in the low-glycogen category decreases with body size in some varanid hindlimb muscles, lending conditional support to our hypothesis. Although glycogen staining is only a proxy of fibre type, these data suggest that these muscles may be composed of relatively more fast-twitch muscle fibres in larger varanids. Muscles composed of proportionally more fast-twitch fibres should produce more force, helping to close the gap between theoretical predictions of muscle force (∝M_b^0.67) and observed peak ground reaction forces (∝M_b^0.89–0.99). In other muscles in our study, however, glycogen concentration does not change with body size, suggesting that the value of disproportionate increases in force at higher body sizes differs between muscles. It may also suggest that slow-twitch fibres (likely the postural muscles) become more valuable in certain muscles at larger body sizes, perhaps because they are more resistant to fatigue and cost less energy to contract. Future work either directly measuring or simulating the roles of locomotor muscles across a range of body size may make more sense of these patterns.

Finally, a consideration of how animals grow may help to explain the pattern in our data. In mammals, muscle cell division seems to be restricted to the early stages of developmental growth (reviewed in Jorgenson et al., 2020), while cell division also occurs later in development in fishes, termed mosaic hyperplasia (Johnston, 2006, 2003). Thus, although exogenous factors substantially influence muscle growth (e.g. Xu et al., 2021), muscle design can nonetheless be understood as a compromise between the biomechanical needs of the adult and juvenile. Without substantial mosaic hyperplasia, increases in muscle force output to maintain functional similarity during growth must be accomplished in one of two ways. One option, a shift in the proportions of muscle fibre types in larger species towards glycolytic (fast-twitch) fibres would increase total muscle force output. Our glycogen content scaling data suggest that varanids use this pattern in some (but not all) muscles, so perhaps the need for mainly aerobic muscle fibres in the juveniles (which have higher relative maximal metabolic rates than adults) (Clemente et al., 2008; Thompson, 1996) of larger varanid species outweighs the advantage that the more forceful and powerful glycolytic fibres would provide the adults in other muscles. The second option for increasing total muscle force output is through proliferation of myofibrils within muscle cells, leading to hypertrophy and increased fCSA, which also occurs in varanids. In black sea bass, muscle fibre diameter increased substantially from 36 μm to 280 μm (corresponding roughly to fCSA of 1017–61,544 μm²) in individuals between 0.45 g and 2000 g, but increases in muscle size from individuals ranging from 2000 to 4840 g in body mass were accomplished through increased recruitment of smaller fibres (Priester et al., 2011). Fibres also increased in area along a similar range of body sizes in varanids, but the highest raw fCSAs in varanids were not as large as in sea bass, and there was no obvious body size in varanids where further increases in fCSA were solely due to changes in fibre number as in sea bass. This comparison suggests that larger fibres provide a greater advantage with size compared with small fibres in fish versus varanids. It may also indicate that the biomechanical advantages of larger fibres are similar in both groups but become more important at smaller body sizes in fish. A future ontogenetic analysis of muscle design and growth patterns could help to disentangle the effects of individual growth and interspecific size differences.

This study provides a more nuanced view of how the musculoskeletal system of varanid lizards responds to the biomechanical challenges of increasing body size. Previous work on scaling in varanids has shown that many hindlimb and forelimb muscles scale with positive allometry for PCSA, fascicle length, and muscle mass to deal with the size-related increases in stress imposed by geometric scaling (Cieri et al., 2020; Dick and Clemente, 2016). Taken together with this previous work, our results suggest that the increase in muscle mass is at least partially accomplished through an increase in muscle fCSA, rather than through an increase in muscle fibre number alone. Our results also suggest that relative increases in fast-twitch muscle fibres, or at least high-glycogen content fibres, may be another mechanism that larger animals use to deal with the outsized force requirements at larger body sizes. These data also suggest that, while fCSA generally increases with body size, the extent of this scaling has taken a different form in different clades, and may relate to broad differences in locomotor function, habitat and activity metabolism between endotherms and ectotherms.

The authors wish to thank P. Couper and A. Amey from the Queensland Museum for providing the V. komodoensis specimens as well as ample assistance, workspace, and helpful discussions. The authors also acknowledge data processing and staining assistance from J. Bettega and T Baker.