Moving to the city: testing the implications of morphological shifts on locomotor performance in introduced urban lizards

Abstract

Understanding how morphology affects performance in novel environments and how populations shift their morphology in response to environmental selective pressures is necessary to understand how invaders can be successful. We tested these relationships in a global colonizer, the common wall lizard (Podarcis muralis), translocated to Cincinnati, OH, USA 70 years ago. We investigated how morphology shifts in this population inhabiting a novel environment, how these morphological shifts influence locomotor performance and how performance changes in novel conditions. We compared the morphology of museum specimens and current lizards to see which aspects of morphology have shifted over time. Although overall body size did not change, most body dimensions reduced in size. We measured sprint speed with a full-factorial design of substrate type, incline angle and obstacles. We identified a pattern of negative correlation in sprint performance between conditions with and without obstacles. The locomotor advantage of larger body size was diminished when obstacles were present. Finally, there was no relationship between individual variation in contemporary morphology and sprint performance, providing no support to the hypothesis that these shifts are attributable to selective pressures on locomotor performance in the conditions tested.

incline locomotion, invasive species, lizard, morphology, obstacles, sprint performance, trade-offs

INTRODUCTION

The morphology of an organism, i.e. the way its body is shaped, directly determines how effectively it moves and carries out tasks in its environment (Arnold, 1983; Irschick, 2002). Subsequently, performance determines fitness, i.e. how well an individual can survive and successfully produce offspring (Arnold, 1983). This relationship provides insight into how the process of selection works in nature (Arnold, 1983). Environmental alterations will affect the performance ability of an organism, potentially leading to the selection of different traits, as described in the morphology–performance–fitness paradigm (Arnold, 1983; Battles et al., 2019; Irschick, 2002; Winchell et al., 2018). Although this has become the established paradigm over the past decades, numerous studies have demonstrated that these relationships are not so straightforward. For example, the performance measures must have a relationship with fitness and be measured in ecologically relevant conditions to allow inference (Irschick et al., 2008; Lambert et al., 2020), and organisms experience different environments where their morphology might not be as beneficial. Furthermore, organisms do not often perform an ecological task to their maximum capacity in laboratory environments (Irschick & Garland, 2001). To understand this evolutionary relationship, we first have to understand how certain traits affect the functional performance of an organism (Irschick et al., 2008). For example, this paradigm can provide insight into the mechanisms by which invasive species adapt to and inhabit novel environments and how they could spread in the future (Gherghel & Tedrow, 2019; Heym, 2013; Santana Marques et al., 2020). This aspect is of particular importance currently, because of increasing anthropogenic changes to the environment.

As humanity has increased in population, the rate of urbanization has increased dramatically (United Nations, 2018). Urban areas have large and widespread effects on ecological and evolutionary processes (Doherty et al., 2020; Johnson & Munshi-South, 2017; Putman & Tippie, 2020) and profoundly alter habitat structure (Szulkin et al., 2020). For instance, increased urbanization has been shown to facilitate the colonization potential of invasive species (Ferreira & Tidon, 2005; Johnson & Munshi-South, 2017; Numbere, 2018). The effects of urban environments specifically on lizard morphology are varied, resulting in altered limb dimensions (Winchell et al., 2018; Putman et al., 2019; Gómez-Benitez et al., 2020; Putman & Tippie, 2020), increased or decreased body length (Baxter-Gilbert et al., 2020; Putman & Tippie, 2020), increased parasite load and reduced body condition (Lazić, 2017), and taller, shorter and less curved claws (Falvey et al., 2020). Urbanization-related morphological changes such as these are likely to affect the performance capabilities of an individual. For instance, the morphological shifts in urban populations of Anolis cristatellus exhibit a positive correlation with sprint speed (Winchell et al., 2018). Brown et al. (1995a) found that introduced populations of the lizard Podarcis muralis (Laurenti, 1768) in North America had decreased home ranges, smaller territories and less distance between mates, which might relax selection for endurance traits and favour increased sprint speed. It is important to remember that urban adaptation is complicated, and the observed morphological shifts might be a result of plasticity or non-adaptive processes, such as founder effects, genetic drift and gene flow (Johnson & Munshi-South, 2017; Lambert et al., 2020).

In this study, we investigate how morphological traits change within a population over time, which morphological traits influence the performance of an organism, and how performance changes in various conditions. Understanding the mechanisms that drive variation in performance is crucial because it provides insight into how invasive species perform the way they do in novel environments. Importantly, the influence of morphological traits on performance, hence their potential to evolve, can be restricted by trade-offs (Arnold, 1983; Sathe & Husak, 2015; Wynn et al., 2015; Winchell et al., 2018; Tan et al., 2020). Morphology that is beneficial in one environment could be detrimental in another. It is necessary to understand this aspect, in order to explain why the best invaders are able to persist in environments that differ, sometimes drastically, from the one in which they have evolved (Matzek, 2012). There are numerous changes that accompany a shift to another environment, but differences in structural habitat can exert profound influences on locomotor performance (Van Damme et al., 1997; Gomes et al., 2016; Winchell et al., 2018; Battles et al., 2019; Druelle et al., 2019). Structural habitat used by lizards varies widely, ranging from tree trunks to sand to rock crevices (Pianka & Vitt, 2006). Changes in structural habitat, which happen when organisms switch to a new environment, can influence the evolution of morphology (Muñoz & Losos, 2018; Stuart et al., 2014; Putman et al., 2019; Winchell et al., 2018; Donihue et al., 2021).

Introduced populations of P. muralis in the novel environment of urban Cincinnati, OH, USA provide the chance to study how morphology changes in response to urbanization. Leveraging historical specimens collected in the 1980s and animals living in this environment today, our experiment tests both the relationship between contemporary morphology and performance and how morphology has changed over time. Specifically, we are able to test three primary hypotheses: (1) lizard morphology has shifted over time in this new environment; (2) Some aspects of this shift in morphology influence locomotor performance; and (3) individuals exhibit locomotor performance trade-offs in different environmental conditions. We predict that: (1) relative limb length will increase to enhance sprint performance; (2) these morphological traits will influence locomotor performance the most on artificial substrates similar to pavement; and (3) lizards will exhibit within-individual trade-offs such that those performing best in one condition will suffer performance decrements in other conditions. Taken together, our goal is to highlight important aspects of the morphology–performance relationship as organisms colonize a novel environment.

MATERIAL AND METHODS

Collection and captive housing

The common wall lizard, P. muralis, is a small lacertid lizard that is widely distributed throughout Europe (Speybroeck et al., 2016). It has been successful at establishing populations outside its native range, including England (Michaelides et al., 2015), Germany (Heym, 2013; Beninde et al., 2018) and multiple sites in North America, including Cincinnati, OH, USA and the surrounding areas (Hedeen, 1984; Brown et al., 1995a; Davis, 2011). Podarcis muralis was introduced to urban Cincinnati in the early 1950s, when several individuals were released into a yard following a vacation to Northern Italy (Hedeen, 1984; Brown et al., 1995a; Davis, 2011). Owing to the small number of individuals introduced, the P. muralis population in Cincinnati has experienced a population bottleneck and exhibits low levels of genetic diversity (Homan, 2013). Podarcis muralis is considered a climbing specialist and encounters a large variety of substrate conditions in nature (Braña, 2003; Davis, 2011; Druelle et al., 2019; P.L.V. and E.J.G. pers. obs.).

Historical specimens (N = 15) were collected between the years 1981 and 1993 (median, 1983) from the metropolitan area of Cincinnati (Fig. 1; Supporting Information, Table S1). Historical specimens were preserved in an aqueous solution of ethanol in the collection of the Cincinnati Museum Center (https://www.cincymuseum.org; Supporting Information, Table S1). The contemporary P. muralis used for this study (N = 26) were collected from five different sites around Cincinnati between 23 May and 2 June 2020 (Fig. 1; Supporting Information, Table S2). All sites sampled were close (< 50 m) to roads, and all individuals captured were on or near (< 1 m) anthropogenic structures.

Figure 1.

Map showing the locations of specimen collection for contemporary (blue) and historical (orange) common wall lizards (Podarcis muralis) in Cincinnati, OH, USA. Inset images show the habitat of contemporary collection locations. Historical locations are approximate, based on museum records (Supporting Information, Table S1).

Animals were collected using the lasso method (Blomberg & Shine, 2006; Fitzgerald, 2012). Immediately after capture, we measured the body temperature with a thermocouple probe inserted ~1 cm into the cloaca (HH806AU; Omega Engineering, Norwalk, CT, USA). Lizards were placed in cloth bags and transported to the laboratory on the day of capture. To eliminate the effects of gravidity and reproduction and to lower the risk of spreading an invasive species to a new location, females were excluded from this study. We identified male lizards from both historical and contemporary groups by relative head size, tail width and the presence of femoral pores.

Animals were housed in plastic tubs (60 cm × 42 cm × 34 cm) with a wood shaving substrate and a water dish. Each tub contained two structures that served as both basking platforms and hides for the animals. Ambient light was provided for 14 h/day, ultraviolet light for 12 h/day, and an incandescent heat bulb provided a temperature gradient within tubs of ~20–45 °C for 8 h each day in 2-h intervals. Each tub contained between three and five lizards. Lizards were fed mealworms (Tenebrio spp. larvae) dusted with calcium and vitamin powder three times a week. Cages were misted daily, and water was provided in a small dish ad libitum. Lizards were fasted for 48 h before each performance measure to ensure a postabsorptive state (Van Damme et al., 1991; Angilletta, 2001). We conducted performance measures once a day on individual animals on ≤ 3 days consecutively each week, after fasting. All research was conducted under the Ohio Division of Wildlife Wild Animal Permit (23-014), and all procedures were approved by Ohio Wesleyan University Institutional Animal Care and Use Committee (Spring_2020-21-04_Gangloff_B).

Performance measurements

We considered a lizard eligible for experimentation if he possessed a majority of his tail, because loss of the tail affects locomotor performance significantly (Brown et al., 1995b). Given that most museum specimens had partly regenerated or missing tails, tail length was not considered in the morphological analysis. Before experimentation, lizards were acclimated to captive conditions (range, 6–17 days). The amount of time in captivity before the start of trials did not affect lizard performance (including this effect in models yielded P = 0.422). At least 1 h before each measurement of performance, we placed the lizards in an incubator set to 36 °C, which is the optimal temperature for maximal aerobic scope (Gangloff EJ, Telemeco RS, unpubl. data) and close to the mean body temperature of male lizards collected at our study site (35.9 °C).

To test the effect of substrate on sprint speed, we constructed a racetrack (length, 2 m) with interchangeable substrates. We used a full-factorial design of substrate type, sprinting elevation (level/incline) and obstacles (presence/absence). We chose three ecologically relevant substrates for our performance measures. These were cork, sandpaper and artificial grass (turf). Cork was chosen because it allows for the claws of the lizard to interlock with the substrate, which provides better traction and increased performance (Van Damme et al., 1997; Tulli et al., 2012; Vanhooydonck et al., 2015). It is also a porous substrate that is representative of wood, which P. muralis are known to use in natural environments (Brown et al., 1995a; Davis, 2011). We used coarse-grain sandpaper (60 grit; grains ~400 µm in diameter) because it has characteristics of artificial terrain, such as pavement and walls, that P. muralis inhabit in anthropogenically altered environments (Brown et al., 1995a; Davis, 2011; Gomes et al., 2016). Artificial grass was chosen because it has similar properties to the real grass and foliage that surrounds the habitats of these lizards. In addition to sprint speed along a straight path, we also measured manoeuvrability and speed on an incline of 30°. This slope is within the range of inhabited landscapes in both native and introduced populations (P.L.V. and E.J.G., pers. obs.) and has been recommended as a standardized slope for measurement of locomotion on an incline, to enable comparison across studies (Birn-Jeffery & Higham, 2014). These are both ecologically relevant measures because P. muralis normally inhabit rocky outcrops, on which both incline locomotor speed and manoeuvring around obstacles are crucial to avoid predation and to capture prey. These factors can also influence speed and the potential for slips and other mistakes (Brown et al., 1995a; Wynn et al., 2015; Gomes et al., 2016; Adams & Gifford, 2020). To quantify manoeuvrability, we placed eight 10.2-cm-diameter polyvinyl chloride pipe obstacles at equal distances of 25 cm (running distance adjusted for curvature, 215 cm).

All performance measures were carried out during daylight active hours (between 10.00 and 15.00 h). Lizards were assigned randomly to one of three groups, which experienced the 12 running conditions in a random order. Furthermore, within each group the order in which each animal sprinted was randomized daily. Each individual ran three times in total, in quick succession, to ensure that maximum locomotor performance was achieved (total time after being removed from the incubator < 1 min). To record speed information, the racetrack was outfitted with photocells at 25 cm intervals. From those data, we extracted the fastest speed over an interval of 50 cm (Gangloff et al., 2019), which is within the natural range of P. muralis observed field movements (Braña, 2003; Monasterio et al., 2009). In the event of a mistrial (e.g. the lizard escaped the track or did not run), lizards were re-run on another day. All measures were completed within 59 days of lizards being brought into captivity.

Morphological measurements

We measured several morphological variables on the 41 lizards included in the experiment (Supporting Information, Fig. S1). We measured snout–vent length (SVL) from the tip of the snout to the posterior end of the anal scale. Head length (HL) was measured from the tip of the snout to the posterior end of the nuchals. Metatarsus length was measured from the proximal end of the metatarsus to the base of the fourth toe. The fourth toe was measured from the base of the toe to the toe tip. The zeugopodium was measured from the axilla to the tip of the zeugopodium. The stylopodium was measured from the tip of the stylopodium to the femorotibial joint. Scapular girdle width (SG) was measured immediately posterior to the forelimb insertion on the dorsal side. Pelvic girdle width (PG) was measured immediately anterior to the hindlimb insertion on the dorsal side (Lowie et al., 2019). We performed all measurements with a set of digital callipers (model CD-6, Mitutoyo, Japan; model 1130, Insize, USA), with precision to the nearest 0.01 mm. Each measure, excluding SVL, HL, PG and SG, was taken on both the left and right sides of the animal for each limb. Measurements were taken twice and re-measured if the coefficient of variation of the two measures was > 10%. We averaged the two measures for subsequent analyses. One historical specimen was missing the right rear toe, hence we used the left toe as the average toe length value for that individual.

We characterized the morphological phenotype with seven dimensions: three that were measured directly (SG, PG and HL) and four that were combined measurements [front limb length (FrontLimb), front foot length (FrontFoot), hindlimb length (HindLimb) and hindfoot length (HindFoot)]. FrontLimb and HindLimb were calculated by taking the average of the right anterior/posterior limb (stylopodium length + zeugopodium length) and the left anterior/posterior limb. FrontFoot and HindFoot were calculated by taking the average of the right anterior/posterior foot (metatarsus length + fourth toe length) and the left anterior/posterior foot. To scale these measures to body size, we created a log₁₀–log₁₀ regression of each measure on SVL and utilized the residual value for downstream analyses.

Statistical methods

Morphology

To test for changes in body size between historical and contemporary lizards, we performed a Student’s unpaired t-test to compare SVL between the two groups, in the programming language R (R Core Team, 2018). We described the multivariate morphological phenotype with the seven size-standardized measures (see previous subsection). We tested for differences between historical and contemporary lizards using a non-parametric multivariate analysis of variance (NP-MANOVA) with residual randomization in permutation procedures (RRPP; Collyer & Adams, 2018), implemented in R. This approach allowed us simultaneously to describe the major axes of variation between groups and test for differences between groups in multivariate space. We implemented 999 iterations of the residual randomization procedure to assess the significance of a single categorical factor, with two levels describing lizards that were either historical specimens or collected contemporarily. Following the protocol of Telemeco & Gangloff (2020), we then extracted the predicted values in principal component (PC) space for each individual and the least-squares means and 95% confidence intervals for each of the traits. The predicted PC values from the model are rotated such that PC1 separates the two groups (historical and contemporary specimens) maximally in multivariate space, and thus best serves as a factor in the model of sprint speed to test our hypothesis of interest (see Statistical Methods: Sprint speed). For visual comparison, we also conducted a principal components analysis with the prcomp function (Supporting Information, Fig. S2). The RRPP approach offers the advantage of providing a unified framework for constructing models, extracting least-squares means and graphically presenting the results, in addition to having the capacity to handle the inclusion of more morphological measurements relative to sample size (Collyer & Adams, 2018; Telemeco & Gangloff, 2020).

Sprint speed

We used linear mixed models to address simultaneously two of our motivating questions regarding the relationship between morphology and sprint performance and the effect of different running conditions on sprint performance. To do this, we created a model including categorical fixed effects of substrate type (cork/sandpaper/turf), the presence of obstacles (present/absent) and incline (flat/incline). The model also included the covariates of SVL and individual PC1 scores predicted from the multivariate model implemented in RRPP, which described the greatest axis of variation between historical and contemporary specimens (see Statistical Methods: Morphology). The initial model also included the three-way and all two-way interactions of running conditions and the two-way interactions of each running condition factor and both SVL and predicted PC1 score. We used a backward-selection procedure and removed unimportant interactions from the model sequentially (P > 0.10), retaining main effects and interactions that tested our specific hypotheses. We included a random intercept for individual to account for covariation of repeated measures made on the same individual. Sprint speed was square-root transformed before analysis to meet the assumption of normally distributed model residuals. The importance of fixed effects was assessed using type III sums of squares, correcting denominator degrees of freedom for F-tests (Kenward & Roger, 1997). We implemented models with the lme4 package (Bates et al., 2014) in R. Data figures were created with the ggplot2 package (Wickham, 2009).

To assess within-individual trade-offs in sprinting in different conditions, we estimated correlations and the significance of relationships in pairwise combinations of the sprinting conditions using the package corrplot (Wei & Simko, 2017) in R. We initially centred and standardized all sprint measures within each individual and then created a correlation matrix of all 66 pairwise combinations of the 12 sprinting conditions (3 substrate × 2 incline × 2 obstacle). In addition to estimating correlations between all pairwise conditions for sprint performance, we also tested whether correlations were stronger when comparing among substrates, between flat and incline and between conditions with and without obstacles. We did this by comparing correlation estimates for matched vs. different conditions within each aspect of sprint conditions. We then used a Student’s unpaired t-test to test whether these correlations differed in matched conditions (i.e. all pairwise combinations in which either both conditions included obstacles or both conditions were without obstacles) vs. unmatched conditions (i.e. comparing all pairwise combinations in which one condition was with obstacles and one without obstacles). We assessed significance of correlations with α = 0.05.

Data availability

All data used in the analyses are freely available in the Mendeley Data Repository (http://dx.doi.org/10.17632/fyk445ccrh.1).

RESULTS

Morphology

Contemporary and historical lizards did not differ in body size (SVL mean and range, in millimetres: historical, 60.3, 51.4–67.1; contemporary, 61.7, 47.4–70.3; t_37.2 = 0.774, P = 0.444; Supporting Information, Fig. S3). The NP-MANOVA with RRPP demonstrated that the multivariate morphological phenotype differed between groups (Fig. 2; F_1,39 = 11.1, P = 0.001). Historical and contemporary lizards were split almost perfectly along one axis of variation, which contrasted lizards with small shoulder girdles, front feet and hindlimb length vs. those with large pelvic girdles, head length, rear feet and front limbs (Table 1; Fig. 2). The 95% confidence intervals of least-squares means from the model were non-overlapping for five of the seven traits, indicating that contemporary lizards had larger shoulder girdles, whereas historical lizards had larger pelvic girdles and a longer head, hindfoot and front limb length (Fig. 3).

Table 1.

Variable loadings of predicted principal component (PC) values for first two axes of variation describing the morphological phenotype of historical and contemporary common wall lizards (Podarcis muralis)

Morphological measure	PC1	PC2
Pelvic girdle (PG)	0.569	−0.0198
Shoulder girdle (SG)	−0.357	−0.703
Head length (HL)	0.472	−0.006
Total front limb length (FrontLimb)	0.378	−0.005
Total front foot length (FrontFoot)	−0.096	0.001
Total hindlimb length (HindLimb)	−0.255	0.003
Total hindfoot length (HindFoot)	0.329	−0.711

Predicted values were generated using a non-parametric multivariate analysis of variance (NP-MANOVA) with randomized residuals in a permutation procedure (see main text for statistical details).

Figure 2.

Principal components (PC) analysis plot, generated using ‘RRPP’ tools in R, displaying the multivariate morphological phenotype of historical and contemporary common wall lizards (Podarcis muralis). Values are fitted values from a non-parametric multivariate analysis of variance (NP-MANOVA) with the RRPP model for each individual projected onto PC space. See main text for statistical details.

Figure 3.

Least-squares means and 95% confidence intervals generated from a non-parametric multivariate analysis of variance (NP-MANOVA) with the RRPP model for morphological traits of historical and contemporary common wall lizards (Podarcis muralis). Values shown are the predicted values from the model after accounting for covariation within the response matrix, displayed on a z-standardized scale. Abbreviations: HL, head length; PG, pelvic girdle; SG, shoulder girdle.

Sprint speed

Sprint speed in lizards was affected by both extrinsic (running conditions) and intrinsic (body size) factors. Overall, lizards ran faster on both cork and turf compared with sandpaper and, surprisingly, faster up an incline than on a flat surface (Table 2; Figs 4, 5B). The main effect of obstacle was not important for predicting sprint speed, but the presence of obstacles did affect sprint speed such that lizards were slowed more by obstacles on turf compared with the other substrates, and lizards were slowed more by obstacles on an incline compared with a flat surface (significant obstacle × substrate and obstacle × incline interactions). Size affected sprint speed, with larger lizards running faster overall, but the advantage of being larger decreased in the presence of obstacles (significant SVL × obstacle interaction). Individual scores on the first axis of morphological variation between historical and contemporary lizards, and its interactions with running conditions, did not affect sprint speed.

Table 2.

Results of linear mixed model analysis describing the effect of running conditions and morphology on sprint performance in adult male common wall lizards (Podarcis muralis)

Source of variation	Sprint speed
Incline
F (d.f.n, d.f.d)	19.0 (1, 276)
Pr > F	< 0.001
Substrate
F (d.f.n, d.f.d)	18.4 (2, 276)
Pr > F	< 0.001
Obstacle
F (d.f.n, d.f.d)	0.7 (1, 276)
Pr > F	0.416
SVL
F (d.f.n, d.f.d)	15.0 (1, 23)
Pr > F	< 0.001
Predicted morphology PC1
F (d.f.n, d.f.d)	0 (1, 23)
Pr > F	0.903
Incline × substrate
F (d.f.n, d.f.d)	0.7 (2, 276)
Pr > F	0.481
Incline × obstacle
F (d.f.n, d.f.d)	5.2 (1, 276)
Pr > F	0.02
Substrate × obstacle
F (d.f.n, d.f.d)	11.1 (2, 276)
Pr > F	< 0.01
Obstacle × SVL
F (d.f.n, d.f.d)	4.4 (1, 276)
Pr > F	0.04

The model also included a random intercept of individual to account for repeated measures. See main text for statistical details. d.f.n: numerator degrees of freedom for F-test; d.f.d: denominator degrees of freedom for F-test; Pr > F is the probability that the observed value is greater than that predicted by the F-distribution (P-value). Abbreviations: PC, principal component; SVL, snout–vent length.

Figure 4.

Boxplots and raw values of sprint speed across all conditions in adult male common wall lizards (Podarcis muralis). Abbreviated conditions are ordered such that the first letter is the incline (F = flat or I = 30° incline), the second is substrate type (T = turf, C = cork or S = sandpaper), and the last letter is the obstacle condition (N = no obstacles or Y = obstacles present). Tukey boxplots show the median, interquartile range and the limits of 1.5× the interquartile range of raw data values.

Figure 5.

Dependence of lizard sprint performance on body size and the presence of obstacles (A) and incline (B) in adult male common wall lizards (Podarcis muralis). A, linear regression lines ± SE of sprint speed as a function of snout–vent length in running conditions without and with obstacles, combining data across substrate type and incline. The results of linear mixed model analysis show that slopes differ between obstacle and no obstacle conditions (snout–vent length × obstacle interaction, P = 0.04; Table 2). B, back-transformed least-squares means ± SE for sprint speed in flat and incline (30°) conditions, combining data across substrate type and the presence/absence of obstacles. The results of linear mixed model analysis show that lizards sprint faster up an incline (P < 0.001; Table 2).

Correlation coefficients for sprinting in different conditions ranged from −0.49 to 0.34. Of the 66 correlations calculated, 35 were significantly negative and 19 significantly positive (Fig. 6). The five strongest negative correlations were between combinations of conditions including and excluding obstacles, whereas the four strongest positive correlations were between conditions with matched obstacle presence/absence. The clearest difference in correlation estimates from matched vs. unmatched conditions was with regard to the presence or absence of obstacles (mean correlation for matched conditions ± SD, −0.035 ± 0.208; mean correlation for unmatched conditions ± SD, −0.127 ± 0.190; t_59.5 = −1.87; P = 0.067). We found no support for differences in correlation estimates from matched vs. unmatched conditions with regard to incline (mean correlation for matched conditions ± SD, −0.100 ± 0.180; mean correlation for unmatched conditions ± SD, −0.072 ± 0.221; t_64.0 = 0.559; P = 0.578) or substrate (mean correlation for matched conditions ± SD, −0.064 ± 0.155; mean correlation for unmatched conditions ± SD, −0.102 ± 0.204; t_40.3 = −0.809; P = 0.423).

Figure 6.

Pairwise correlation matrix of the sprint performance of adult male common wall lizards (Podarcis muralis) between all combinations of conditions. Above the diagonal is a graphical representation of the strength and direction of correlations (correlations for P > 0.05 not shown). Below the diagonal is the correlation estimate. Abbreviated conditions are ordered such that the first letter is the incline level (F = flat or I = 30° incline), the second is substrate type (T = turf, C = cork, or S = sandpaper), and the last letter is the obstacle condition (N = no obstacles or Y = obstacles present).

DISCUSSION

Our experiment demonstrates that morphology has shifted dramatically over several decades in a population of invasive lizards that have successfully colonized an urban environment, but the aspects of morphology that have shifted do not drive variation in contemporary locomotor performance. We found a decrease in relative head length, pelvic girdle width, hind foot length and front limb length, but an increase in relative scapular girdle width, between lizards collected ~30 years after introduction and those collected ~70 years after introduction. Interestingly, despite the trend of reduced body part size, overall body size (SVL) did not change. We predicted that the morphological traits that shifted over time would influence locomotor performance, specifically to enhance speed on the sandpaper substrate, which mimics anthropogenic urban surfaces. This was not the case, because we found no evidence to suggest that the morphological traits that shifted over time influence locomotor performance in any of the variety of conditions tested. In the first 30 years after introduction, selection could already have purged alleles from the population, reducing the ability to detect morphology–performance relationships in contemporary lizards. We did, however, find strong evidence for within-individual performance trade-offs, suggesting that performing well in one condition will probably lead to a performance decrement in other conditions (Sathe & Husak, 2015; Tan et al., 2020). Our results show that the presence or absence of obstacles is a key component of within-individual trade-offs and a determinant of individual sprint performance in interaction with substrate and incline. This highlights the importance of accounting for complex interactions of extrinsic (environmental) and intrinsic factors in assessing the relationship between morphology and performance (Irschick, 2002).

Urban P. muralis populations show reduced body condition, increased parasite load and smaller home ranges (Brown et al., 1995a; Lazić, 2017). Our results display a reduction in four of seven morphological variables measured, but no discernible change in overall body size. These results match studies in Sceloporus (Sparkman et al., 2018; Putman et al., 2019), but not Anolis. In a review, Putman & Tippie (2020) found that the general trend for urban populations of lizards is to exhibit increased body size, limb length and digit length relative to populations in natural environments. In Anolis lizards, urban populations display longer limbs, wider pelvic and pectoral girdles, and taller, shorter and less curved claws (Winchell et al., 2018; Falvey et al., 2020). In contrast, urban populations of western fence lizards (Sceloporus occidentalis) exhibit reduced limb and toe lengths (Putman et al., 2019), in addition to diminished escape behaviour and lower sprint speed (Sparkman et al., 2018). Lizards occupy a diversity of habitat types and do not respond to urbanization in the same ways. This indicates that there are species-specific responses to urbanization that are related to how a species interacts with its environment and how that relationship changes in urban settings. This highlights the fact that researchers should avoid making generalizations based on the response of one clade.

We found drastic changes in morphology within the invasive P. muralis population over a period of several decades, but our data indicate that these changes do not benefit lizard performance in our measured conditions. Morphological evolution is multifaceted and complex, resulting from selective pressures in the environment and the genetic variation that underlies the morphological variation (McGlothlin et al., 2018). Our finding of major shifts in morphology unrelated to sprint performance suggests that other selective pressures might be shaping morphological traits over evolutionary time scales. For example, it is possible that selection is acting on other performance traits, such as squeezing into rock crevices (Žagar et al., 2017). Our results demonstrate that scapular girdle width is the only morphological measure that has increased significantly in relative size between historical and contemporary lizards. The scapular girdle is associated with muscles involved in limb, neck and body movements (Jenkins & Goslow, 1983; Tinius et al., 2020). Further work should be aimed at a more detailed investigation of the selective pressures shaping scapular girdle morphology in this species.

It is important to note that the urban P. muralis population in Cincinnati is also invasive, which is especially interesting genetically because the founding population was very small, with roughly ten individuals founding the population 70 years ago (Deichsel & Gist, 2001). Being founded from such a small number of individuals, the P. muralis population in Cincinnati was subject to a strong founder effect, which might influence the phenotypes of the present lizards (Michaelides et al., 2016). Evidence suggests that after experiencing a genetic bottleneck, populations of P. muralis can persist at low levels of genetic diversity in both their native (Sacchi et al., 2016) and invasive ranges (Homan, 2013). It is possible that shifts in morphology over time are attributable to low genetic variability and subsequent drift, which is one of the most clear evolutionary processes occurring in urban environments worldwide (Johnson & Munshi-South, 2017; Miles et al., 2019).

Our results show that P. muralis displays increased sprint performance on a 30° incline. This is a surprising and fairly novel result, because multiple studies have shown that most animals incur a loss of performance on inclined surfaces (Birn-Jeffery & Higham, 2014). Sceloporus undulatus (Pinch & Claussen, 2003) and Stellio stellio (Huey & Hertz, 1984) display decreased performance on inclines ranging from 15 to 60° in comparison to a flat surface. Conversely, Vanhooydonck & Van Damme (2001) found that several species of lizards, including P. muralis, did not experience any reduction in performance on an inclined surface of 70°. A 30° incline prompts Callisaurus draconoides to adopt a bipedal posture more often than when running on a flat surface (Irschick & Jayne, 1998). In P. muralis, incline could affect body posture, bringing the lizards to a more bipedal position and allowing their hindlimbs to increase the force applied to the substrate in a way that is not possible when running on flat surfaces (Aerts et al., 2003). This would be evolutionarily advantageous, because P. muralis is a climbing specialist (Braña, 2003; Druelle et al., 2019) that spends the majority of its time on vertical or heavily inclined surfaces, such as stone landscaping walls (Brown et al., 1995b; Davis, 2011). Further studies are required to investigate the kinematics of inclined and level locomotion to examine the reasons for this observed increase in performance on an incline.

Our results demonstrate that obstacles interact with substrate, incline and morphology to drive variation and trade-offs in lizard sprint performance, presenting some exciting avenues for future research. Previous work has demonstrated the effects of obstacles on running performance when animals are forced to turn, including work in lizards (Higham et al., 2001; Kohlsdorf & Biewener, 2006; Olberding et al., 2012; Tucker & McBrayer, 2012; Sathe & Husak, 2015; Parker & McBrayer, 2016; Adams & Gifford, 2020). A reduction in sprint speed when navigating obstacles can be attributable either to animals simply being unable to gain purchase on the substrate to anchor the body for rapid turning motions or to intentional slowing down to avoid potentially costly mistakes (Higham et al., 2001; Wynn et al., 2015; Grillner & El Manira, 2020). Our results suggest that the presence of obstacles alters the biomechanics of lizard running to limit performance, but only in specific circumstances. For example, we found that the reduction of sprint speed attributable to obstacles was greater when lizards were running up an incline. We speculate that greater loading on the rear legs when running up an incline reduces contact with the forelimbs, which are important for navigating turns (Birn-Jeffery & Higham, 2014). Furthermore, obstacles had little effect on flat surfaces of cork or sandpaper, but strongly reduced sprint performance on turf. This suggests that lizards are able to gain better traction on cork or sandpaper compared with turf, perhaps owing to the ability of the claws to interact with the substrate and facilitate navigation around turns (Van Damme et al., 1997; Vanhooydonck et al., 2015). We are currently designing experiments to test the importance of claw size and shape in locomotor performance on different substrates, including vertical climbing.

Obstacles were also at the centre of within-individual trade-offs in sprint performance. Although larger lizards ran faster, this advantage was reduced in the presence of obstacles (Fig. 5A). This is similar to results in other taxa, such as quolls, where larger size is advantageous in navigating wide turns, but not the sharpest turns (Wynn et al., 2015). Additionally, we found that individuals running faster around obstacles exhibited relatively slower speeds when running in a straight line, as demonstrated by the pattern of within-individual correlations (Fig. 6). Overwhelmingly, we found the strongest negative correlations in performance between conditions with and without obstacles, whereas the strongest positive correlations were in matched obstacle conditions. This implies that some suite of traits drives individual variation in performance on a continuum of specialization for running in straight or circuitous paths. We found no evidence that any of the morphological traits we measured are driving this variation and therefore conjecture that this might be attributable to variation in physiology, behaviour or other unmeasured morphological traits (e.g. claw shape). Given the prevalence of obstacles and uneven terrain we observed in habitats where lizards live, the traits underlying this variation in performance might be under strong selection and warrant further investigation (Tucker & McBrayer, 2012; Sathe & Husak, 2015; Adams & Gifford, 2020).

CONCLUSION

Our results provide a case study into the complexities shaping responses to novel environments in a highly successful urban colonizer. Urban environments are expanding, and more species, including lizards, are occupying these landscapes (Johnson & Munshi-South, 2017; Lambert et al., 2020; Putman & Tippie, 2020). Understanding the mechanisms by which successful urban colonizers inhabit these highly anthropogenic environments has become a pressing conservation issue and simultaneously offers many unique natural experiments to quantify ecological and evolutionary processes. Podarcis muralis transported from Italy to Cincinnati provide an excellent study system for such responses, especially given the known founding population size and date. Over a time period of ~30 years, wall lizards in Cincinnati dramatically shifted aspects of their morphology. These patterns, such as reduced relative limb length, are dissimilar to results found in Anolis (Winchell et al., 2018), but similar to Sceloporus (Sparkman et al., 2018; Putman et al., 2019), highlighting the need for species-specific studies. These taxa interact differently with the urban environment, and further research into urban adaptation should be based on trait–environment interactions specific to the taxa (Putman et al., 2020). Owing to the ecological differences between Sceloporus and Anolis, these results suggest that habitat selection and availability are key to understanding the potential fitness consequences of variation in locomotor performance. An important consideration in future studies of sprint performance in lizards is to recognize inherent trade-offs in performance in different conditions and to select ecologically relevant conditions for a given species or even population.

SUPPORTING INFORMATION

Additional Supporting Information may be found in the online version of this article at the publisher’s web-site:

Table S1. Historical specimen Podarcis muralis identification numbers, GPS coordinates, location name and year, as assigned by the Cincinnati Museum Center. The GPS coordinates were approximated from collection notes.

Table S2. Habitat and sample size information from sampled contemporary populations of Podarcis muralis in Cincinnati, OH, USA.

Figure S1. Diagram of measurements used to describe the morphology of the common wall lizard (Podarcis muralis): (1) snout–vent length (SVL); (2) head length (HL); (3) left posterior metatarsus; (4) fourth toe on left posterior foot; (5) left posterior zeugopodium; (6) left posterior stylopodium; (7) left anterior metatarsus; (8) fourth toe on left anterior foot; (9) scapular girdle (SG); and (10) pelvic girdle (PG). The illustration demonstrates the location of body positions measured, but does not indicate the specific landmarks used (see main text for details).

Figure S2. Principal components (PC) plot of multivariate morphological phenotype of historical and contemporary common wall lizards (Podarcis muralis). Axis labels identify the proportion of variance explained in the first two major axes of variation.

Figure S3. Boxplots of raw morphological data on historical and contemporary common wall lizards (Podarcis muralis) collected in Cincinnati, OH, USA.

ACKNOWLEDGEMENTS

We would like to acknowledge support from the Ohio Wesleyan University Summer Science Research Program, coordinated by L. Tuhela-Reuning. H. Farrington and the Cincinnati Museum Center graciously provided the museum specimens used in this experiment. J. Davis helped to locate populations of Podarcis muralis in Cincinnati. K. Koeller, A. Garner and K. Jenkins shared insight into the role of the scapular girdle, and two reviewers provided helpful comments on the manuscript. Richie’s and T. Beifong provided wonderful food and entertainment. The authors declare no conflict of interest. P.L.V. and E.J.G. designed the experiment, conducted the experiment and analysed the data. P.L.V. and W.M. designed the lizard racetrack and running conditions. W.M. contributed to field collection of lizards and experimental design. P.L.V. wrote the first draft of the manuscript. All authors contributed to revisions and the final manuscript draft.