Integration and Modularity in the Turtle Body Plan: Impacts on Disparity and Species Richness.

Integration and Modularity in the Turtle Body Plan: Impacts on Disparity and Species Richness. | Authorea try { document.documentElement.classList.add('js'); } catch (e) { } var _gaq = _gaq || []; _gaq.push(['_setAccount', 'G-8VDV14Y67G']); _gaq.push(['_trackPageview']); (function() { var ga = document.createElement('script'); ga.type = 'text/javascript'; ga.async = true; ga.src = ('https:' == document.location.protocol ? 'https://ssl' : 'http://www') + '.google-analytics.com/ga.js'; var s = document.getElementsByTagName('script')[0]; s.parentNode.insertBefore(ga, s); })(); Skip to main content Preprints Collections Wiley Open Research IET Open Research Ecological Society of Japan All Collections About About Authorea FAQs Contact Us Quick Search anywhere Search for preprint articles, keywords, etc. Search Search ADVANCED SEARCH SCROLL This is a preprint and has not been peer reviewed. Data may be preliminary. 10 June 2025 V1 Latest version Share on Integration and Modularity in the Turtle Body Plan: Impacts on Disparity and Species Richness. Authors : Taggert Butterfield 0000-0003-3500-6341 [email protected] , Mark Olson 0000-0003-3715-4567 , Jorge Contreras-Garduño 0000-0002-9231-0641 , and Macip Ríos Rodrigo Authors Info & Affiliations https://doi.org/10.22541/au.174954810.05814002/v1 259 views 195 downloads Contents Abstract Information & Authors Metrics & Citations View Options References Figures Tables Media Share Abstract Evolution is shaped by development, natural selection, and physiological limitations that bias the range of variation observed in organisms, influencing patterns of diversification. This study investigates how patterns of morphological integration and modularity impact disparity and species richness across freshwater and terrestrial turtles. Modularity refers to the idea that biological systems, like organisms, organs, or traits, are organized into relatively independent, semi-autonomous units, called modules. Integration refers to how strongly different traits are interconnected or correlated with each other. We first hypothesize that the most diverse turtle suborder, Cryptodira, exhibit weaker integration and higher modularity than Pleurodira, leading to greater morphological disparity and species diversity. Second, we hypothesize that at the family level weaker integration and higher modularity promotes morphological disparity and species richness. To test these hypotheses we take linear measurements of limb, shell, and head characteristics of 1652 turtle specimens belonging to 270 species (70% of species level diversity). Covariation matrices were used to test hypotheses in a phylogenetic framework. Results partially support our hypotheses: Cryptodira show lower integration and higher modularity but unexpectedly lower disparity than Pleurodira. At the family level, higher modularity and weaker integration correlate with higher species richness, while integration is positively correlated with increased disparity. The most diverse families that have evolved terrestrial and aquatic lifestyles, Emydidae and Geoemydidae, exhibit high modularity, weak integration, low disparity, and higher species richness, whereas Kinosternidae and Trionychidae which have strictly aquatic species, exhibit moderate levels of modularity, high integratation, and high disparity. These findings highlight how patterns of trait covariation can shape organismal diversity, and the depth of our sampling provides key insight on how patterns of covariation can influence the diversification in a major order of vertebrates. INTRODUCTION Development, biophysical limits, and physiological interactions shape the range of morphologies that organisms can produce (Gould & Lewontin, 1979; Alberch, 1989; Losos, 2011). These biases influence adaptation by limiting the pool of available variation, rather than allowing natural selection to act on an unrestricted continuum of traits (Olson, 2012). Understanding how selection and development mutually shape one another requires examining how different morphological components covary, as patterns of covariation can reveal developmental tendencies that either channel or facilitate evolutionary change. Organisms are coordinated systems, and their traits do not evolve independently. Instead, morphological features exhibit varying degrees of covariation, with some traits evolving in a tightly integrated manner, while others remain more modular (Pigliucci & Preston, 2004; Callebaut & Rasskin-Gutman, 2005). Strong integration, where traits covary tightly, can reinforce particular evolutionary trajectories, while modularity, where sets of traits covary more independently, may allow for greater morphological disparity and species diversity by enabling different regions of the body to be exposed to more potential variation (Wagner & Altenberg, 1996; Breuker et al. , 2006; Felice et al. , 2018). Studies across vertebrates have yielded mixed results on whether integration or modularity is more strongly associated with increased diversity (Goswami & Polly, 2010; Hu et al. , 2016; Felice & Goswami, 2018; Evans et al. , 2021). However, most studies have focused on localized trait complexes, such as cranial or limb structures, rather than examining large-scale patterns of covariation across major body regions (e.g. Larouche et al. , 2018). Examining covariation at the whole-body level seems certain to provide novel insights into how morphological integration and modularity influences evolutionary trajectories. Studies on integration and modularity often focus on covariation patterns within specific traits (e.g., comparing regions of the skull; Bardua et al. , 2020), but the integration of entire organismal body plans, such as the head, torso, and limbs, has received less attention. Larouche et al. , (2018) examined the integration and modularity between the head, trunk, and tail in ray-finned fish (Actinopterygii), revealing significant covariation between the trunk and tail across species. They demonstrated that the interaction between these regions and head characteristics helps explain the major axes of morphological variation. The coordination between the trunk and tail makes sense, as both regions serve distinct yet complementary functions, while the trunk supports the body’s core and internal systems, the tail plays a crucial role in locomotion. These interactions directly influence how the fish moves and uses resources in its environment (Wainwright, 1996). This relationship between the trunk, tail, and head underscores the importance of considering major body regions in studies of morphological diversity, as such insights would be overlooked in studies focusing only on individual structures like mandibles (Parsons et al. , 2012), fins (Du et al. , 2019), or craniofacial features (Navalón et al. , 2020). Investigating how major components of an organism’s body covary is essential to undestand the evolution of organisms as integrated wholes rather than disarticualted parts (Olson, 2019). Freshwater and terrestrial turtles offer a compelling system for investigating how patterns of covariation shape adaptive diversity. With only 356 species, it is feasible to collect data on most extant turtles (e.g., Stayton et al. , 2018), which exhibit a wide range of ecological adaptations. Freshwater and terrestrial turtles have successfully colonized nearly every major water system, from estuaries and marshes to swamps, ephemeral pools, and rivers of varying sizes (Ernst & Barbour, 1989). This diversity in habitat use underscores the ecological importance of their coordinated body plan. The head, shell, and limbs of turtles each serve distinct functions: the head is involved in feeding, defense, and sensory perception (Herrel et al. , 2002, 2018; Claude et al. , 2004; Stayton, 2011; Foth et al. , 2017); the shell provides hydrodynamic advantages, protection, and self-righting ability (Domokos & Várkonyi, 2008; Rivera & Claude, 2008; Berlant & Stayton, 2017; Stayton et al. , 2018); and the limbs facilitate locomotion in both aquatic and terrestrial environments (Pace et al. , 2001; Mayerl et al. , 2016, 2019; Young et al. , 2017). Because these structures are functionally interconnected, understanding their covariation patterns is essentioal for documenting the interaction between developmental possiblity and selection in turtle evolution. Turtle diversity exhibits two major levels of evolutionary divergence: (1) the deep split between the suborders Cryptodira and Pleurodira and (2) the subsequent diversification of major families (Lovich & Gibbons, 2021; Thomson et al. , 2021). Cryptodires, which retract their necks straight back into the shell, are the more species-rich and widespread suborder, inhabiting terrestrial, freshwater, and marine environments across the globe (Ernst & Barbour, 1989; Lovich & Gibbons, 2021). A key anatomical distinction between the two suborders is the attachment of the pelvic girdle to the shell. In cryptodires, the pelvic girdle is unattached, potentially allowing greater limb flexibility and a wider range of locomotor strategies, including walking, digging, and swimming (Mayerl et al. , 2016). In contrast, pleurodires, or side-necked turtles, retract their necks laterally and have a pelvic girdle fused to the shell, which limits hind limb mobility and may limit locomotor diversity (Mayerl et al. , 2016). This structural difference has been suggested as a possible factor contributing to the striking contrast in ecological diversity between the two groups, with the detachment of the pelvic girdle and neck retraction being potential traits that have facilitated the diversification of Cryptodira (Miller et al. , 2023). Notably, pleurodires have been present on the Australian continent for over 120 million years (Ferreira et al. , 2018), yet no fully terrestrial pleurodire species has evolved. Whether their restricted locomotor capability has played a role in preventing a terrestrial transition remains an open question, but it underscores the potential influence of limited developmental possibilities on turtle evolution (Galis et al. , 2018). Pleurodira are represented by three families: Chelidae, Pelomedusidae, and Podocnemididae, all of which are known for occurring in freshwater systems in the southern hemisphere (Turtle Taxonomy Working Group, 2021). Within Cryptodira, further diversification has given rise to six major families that have more than three species, each representing distinct evolutionary adaptations: Geoemydidae and Emydidae, which dominate freshwater and terrestrial systems; Trionychidae, known for being aquatic and for their unique softshell morphology; Testudinidae, which have adapted to fully terrestrial life with elephantine feet; Kinosternidae, a group of small-bodied freshwater turtles that have adapted to many different water regimes; and Chelydridae, snapping turtles known for their powerful jaws and bottom walking habits (Lovich & Gibbons, 2021). Understanding how these families have diversified provides key insights into the evolutionary processes and ecological factors shaping turtle biodiversity. In this study, our aim was to evaluate patterns of integration, modularity, disparity, and species diversity across the two major suborders and major turtle families to test two hypotheses. Hypothesis 1: Cryptodira, the most diverse turtle suborder that has invaded both terrestrial and aquatic habitats, exhibit weaker integration and higher modularity than Pleurodira, leading to greater morphological disparity and species diversity. This is driven by the increased flexibility in locomotion afforded by their unattached pelvic girdle. Hypothesis 2: At the family level that weaker integration and higher modularity promote disparity and species richness. We also compare integration, modularity, and disparity among families to investigate variation in these patterns across the diversity of turtles. To test these hypotheses, we measured limb, shell, and head characteristics of 270 freshwater and terrestrial turtle species to assess modularity, integration, and disparity within a phylogenetic framework. We applied a resampling method to generate distributions of modularity, integration, and disparity estimates for suborders and families. Specifically, we iteratively removed 10% of the species at the start of each run and repeated the analysis 100 times. An identical procedure was used to generate null distributions. We assess the presence of significant patterns of integration, modularity, and disparity across suborder and family by comparing the empirical data to their null distributions. We then test both hypothesis using a Markov chain Monte Carlo general linear mixed models, with the backbone phylogeny incorporated into the error structure. Specimens and morphological measurements We gathered data from 1652 turtle specimens liquid-preserved and wild caught turtle specimens, representing 270 species. This covers 77% of the species level diversity in extant freshwater and terrestrial turtles (Turtle Taxonomy Working Group, 2021). We aimed to measure only adult specimens, which were identified based on expected straight-line carapace sizes of adults from the literature (e.g. Ernst and Barbour 1989) and ossification of the shell (adults have a completely ossified shell). However, we made exceptions to include sub-adults of rare species (e.g. Yunnan box turtle, Cuora yunnanensis ), or sub-adults of species that are difficult to preserve in liquid (e.g. the large Indian soft-shell turtle Nilssonia gangetica ). When possible, three males and three females were measured for each species. Linear distances of limb, shell, and head characters were measured for each individual with a digital caliper to the nearest ± 0.5 mm. Limb characteristics measured included the manus (Hand; middle of the wrist crease to distal end of 3 rd digit where skin meets the nail), antebrachium (Ante; apex of elbow to wrist crease), pes (Foot; apex of heel to distal end of 3 rd digit where skin meets nail), and crus (Crus; apex of knee to apex of heel). Shell characteristics measured included straight-line carapace (SLC; nuchal scute to cleavage between supracaudal scutes), straight-line plastron (SLP; intergular scute to cleavage between anal scutes), bridge (Bridge; from axilla to inguinal pocket), plastral lobe (Lobe; maximal ventral width across femoral scutes were they fuse with abdominal scutes), plastron width (PW; length of seam that connects abdominal and pectoral plastral scutes), marginal width (MW; width between 5/6 marginal scutes), and shell height (SH; maximum vertical height from plastron to carapace). Head characteristics measured included head length (HL; premaxilla to posterior edge of supraoccipital), head width (HW; widest part of the skull), and head height (HH; highest part of the skull at posterior end of jaw). Phylogeny To incorporate phylogenetic relationships into our analyses, we relied on a published molecular phylogeny that includes 279 extant turtle species (Thomson et al. , 2021). This phylogeny was constructed using 15 nuclear markers and a Bayesian phylogenetic framework to estimate phylogenetic relationships (Thomson et al. , 2021). Of the 270 species in our dataset, 232 were represented in the Thomson et al., 2021 data, but 38 were not. To include all 270 species, we manually inserted taxa that were in our dataset, but not Thomson et al. 2021, as polytomies to their closest known relatives using the program Mesquite v3.5 (Fig. S1; Maddison & Maddison, 2023). We rely on the most up-to-date nomenclature using the Turtles of the World Checklist (Turtle Taxonomy Working Group, 2021). Modularity To minimize risk of Type II error with our morphological dataset containing 12 variables, we assessed modularity in suborders and families using an iterative resampling approach that accounts for body size, phylogeny, and sample size variation. We first filtered out families with at least three species and performed 100 iterations. In each iteration, we randomly removed 10% of the species from the data and reconstructed the phylogenetic tree for the remaining taxa using the keep.tip function in the ’ape’ package in R Statistical Environment 4.4.3 (Paradis & Schliep, 2019; R Core Team, 2022). The iteration then computed residuals of the log-transformed morphological variables while controlling for phylogenetic effects using phyl.resid function in the ’phytools’ package (Revell, 2012). Residuals were used to test a predefined modularity hypothesis in which traits were assigned to three modules: shell, limbs, and head, which represent matrices of the variables measured for those features. To quantify modularity, we used the phylo.modularity function in the ’geomorph’ package (Adams & Otárola‐Castillo, 2013; Baken et al. , 2021, which estimates the covariance ratio (CR), effect size (Z), and statistical significance ( P -value) for each iteration ). The results were compiled across all iterations to evaluate the patterns of modularity patterns in suborders and families. Smaller matrices are more susceptible to high sampling variance, which can lead to spurious correlations in families with fewer species, such as Podocnemidae (8 species in our data). To mitigate this issue, we compared each suborder and family to a null distribution of the same data. To generate the null distribution, we randomized species labels at the start of each iteration by shuffling the row names of the dataset and running 100 iterations, following the same procedure used for the empirical data. Integration We measured integration of suborder and family using the phylo.integration function from the ’geomorph’ package in R (Adams & Otárola‐Castillo, 2013; Baken et al. , 2021). This function quantifies the degree of morphological integration between two sets of variables by evaluating the strength of covariation while accounting for phylogeny using partial least squares (PLS) under a Brownian motion model of evolution. To assess integration patterns in our dataset, we evaluated the covariation between limb and shell variables, and the head variables. We apply a similar iterative approach to measuring integration as described for modularity, except when residuals of the limb, shell, and head matrices are extracted they are used to calculate the phylogenetic integration between the limbs-shell variables and the head variables using the phylo.integration function (Adams & Felice, 2014). This function estimates the partial least squares correlation coefficient (rPLS), effect size (Z), and statistical significance ( P -value) between two matrices. The results of our iterations were compiled to assess integration patterns across suborders and families, and we also create a null distribution by randomly shuffling the species labels at the start of each iteration to assess whether observed integration values exceed expectations under a null model. Disparity There are a wide range of methods to evaluate disparity in organisms (Guillerme et al. , 2020) . In this study, we use the morphol.disparity function in the ’ geomorph ’ package in R to evaluate disparity of turtle suborder and family (Adams & Otárola‐Castillo, 2013; Baken et al. , 2021). We used the same iterative approach as we used for integration and modularity, randomly removing 10% of the species in each iteration. Morphological variables were first log-transformed, and body size and phylogenetic independent residuals were extracted using phyl.resid in the ’phytools’ package under a Brownian motion model (Revell, 2012). Disparity was then calculated for the matrix containing all 12 morphological variables for each suborder and family using the morphol.disparity function in the ’ geomorph ’ package (Baken et al. , 2021), iterating 100 times to account for sampling effects. This approach allowed us to robustly assess morphological disparity while accounting for phylogenetic structure. A null model was also produced by randomizing species labels at the start of each iteration by shuffling the row names of the dataset and running 100 iterations. Statistical analysis and hypothesis testing To test hypothesis 1 for significant differences in integration, modularity, and disparity between suborders (Cryptodira and Pleurodira) we used a linear mixed model (lmm) or generalized linear mixed model (glmm) depending on the ideal distribution for model fit using the lmer and glmer functions in the ’lme4’ package in R (Bates et al. , 2015). To determine the best model we compared Akaike Information Criterion (AIC) values of models with a gaussian distribution with identity link (lmm), gaussian distribution with a log link (glmm), and a gamma distribution with a log link (glmm), selecting the model with the lowest AIC. Once the model with the lowest AIC was identified, the assumptions of the tests were assessed visually, to ensure that the data followed the corresponding distribution, that there weren’t any outliers, that the residuals were homooscedastic, and we assessed influential points by calculated Cook’s Distance with the influence function in the ’influence.ME’ package in R (Nieuwenhuis et al. , 2012). For models assuming normally distributed residuals, we also performed a Kolmogorov–Smirnov test using the ks.test function in the ’stats’ package (R Core Team, 2022) to test for significant departures from normality. We calculated P-values for independent variables using likelihood ratio tests (LRTs) compared against a χ² distribution, implemented with the drop1 function in the ´stats’ package. These tests addressed Hypothesis 1, under which we expected Cryptodira to exhibit significantly lower integration, higher modularity, and greater morphological disparity. To test Hypothesis 2, that modularity and integration drive species richness and morphological disparity, we used a Markov chain Monte Carlo generalized linear mixed models (MCMCglmm) with the MCMCglmm function from the ’MCMCglmm’ package in R (Hadfield, 2010). We use MCMCglmm models because it can handle multiple random effects including phylogeny and utilizes a robust Bayesian framework for uncertainty and estimates posterior distributions of parameters instead of relying on point estimates. The first model included disparity as the dependent variable, modularity and integration as dependent variables, family as a random effect, and inverse phylogenetic matrix as a random effect. The second model included integration as a dependent variable, species number as independent variable, family as a random effect, and inverse phylogenetic matrix as a random effect. The third model included modularity as a dependent variable, species number as independent variable, family as a random effect, and inverse phylogenetic matrix as a random effect. To assess model fit, we verified MCMC convergence using the Heidelberger-Welch test and the Geweke diagnostic with, respectively, the heidel.diag and geweke.diag functions from the ’coda’ package (Plummer et al. , 2006). Autocorrelation was assessed using the autocorr.diag function in the ’coda’ package. We visually assessed the diagnostic plots of the MCMCglmm models to ensure that data met the assumptions of model fitting and did not exhibit a skewed distribution. For each model, we used a weakly informative prior to ensure numerical stability and set the MCMC parameters to 100,000 iterations, a burn-in of 1,000, and a thinning interval of 10. P-values for the model and each family were interpreted using the summary function of the MCMCglmm object. To identify how integration, modularity, and disparity impact covariation patterns at the family level, we fit three MCMCglmm models with integration, modularity, and disparity as dependent variables, family as the independent variable, the null distribution as a random effect, and the inverse of the phylogenetic relationship matrix of turtle families as a random effect. RESULTS These results provide partial support for Hypothesis 1: while Cryptodira exhibited the expected patterns of weaker integration and higher modularity compared to Pleurodira, disparity was unexpectedly lower in Cryptodira (Fig. 1). After comparing models with different distributions for integration (rPLS scores) as a function of suborder, with null distribution as a random effect, we found that the lmm with a Gaussian distribution had the best fitting model (AIC = -1532.9). The Kolmogorov-Smirnov test revealed that residuals were significantly different from normal (D = 0.068, P = 0.047), and the q-q plot suggest this could be due to a slight right tail skew suggesting a gamma distribution, but the normal distribution had the better model fit based on AIC values, so we used the gaussian model. We did not detect substantial overdispersion or influence of outliers with Cook’s distance. The likelihood ratio test detected significant effect of integration on suborder ( X 2 = 568.81, P < 0.001), and model estimates demonstrated that Pleurodira had a positive effect on integration (intercept = 0.58 ± 0.9, estimate = 0.12 ±.0.003, random effects = 0.02 ± 0.14). Comparing models for modularity (CR) as a function of suborder, with null distributions as a random effect, we found that a glmm with a Gamma distribution had the best fitting model (AIC: -546.33). Data fit the gamma distribution, with a slightly skewed tail observable in the histogram plot, q-q plot, and fitted values with higher values exhibit higher distribution. We did not observe overdispersion or influence of outliers. The likelihood ratio test showed that modularity has a significant relationship with suborder ( X 2 = 37.95, P < 0.001), and that the Pleurodira body plan is less modular (higher CR value) than Cryptodira (intercept = -0.26 ± 0.07, estimate = 0.09 ±.0.01, random effects = 0.23 ± 0.15). Comparing models for disparity (Procrustes variance) against suborder with null distributions as a random effect revealed a better fit with the gamma distribution (AIC: -1332.79). This was further supported by visual diagnostics, including the histogram of residuals, the Q–Q plot, and the increasing variance in fitted values, all of which were consistent with a gamma distribution. No substantial overdispersion was observed, although the Cook’s Distance plot identified one potential outlier that may have contributed to the observed distribution. This test revealed significant effect of suborder on disparity (LRT = 182.53, P < 0.001), with Pleurodira having a higher disparity than Cryptodira (intercept = -0.52 ± 0.03, estimate = -0.12 ±.0.01, random effects = 0.007 ± 0.08). Figure 1. Jitter plots of integration (rPLS values), modularity (covariation ratio values), and disparity (procrustes variances) tests, representing the 100 iterations of the emperical data (darker) and null distributions (transparent), highlighting turtle suborders Cryptodira (red) and Pleurodira (blue). Hypothesis 2, weaker integration and higher modularity promote ecological diversification, resulting in higher number of species and greater disparity at the family level, was partially supported. Disparity was significantly and positively correlated to integration, but not modularity (Table 1). Species number was significantly and negatively correlated to modularity, with more modular body plans found in families with higher species number (Table 1). Species number was negatively correlated with integration, with lower levels of integration found in families with less number of species (Table 1). All of the models passed the Heidelberger-Welch test and Geweke diagnostic, no outstanding patterns were seen in diagnostic plots, no strong autocorrelation was observed, and effective sample sizes were all over 9000. Table 1. Results of multivariate generalized linear mixed models (MCMCglmm) used to test hypothesis 2, that modularity and integration drive disparity and species richness. Each model has family and phylogeny included as random effects. Model 1: Disparity ~ Modularity + Integration Fixed Effects Intercept 0.24 0.08 0.40 9147 0.008 Modularity (CR) 0.002 -0.01 0.01 9900 0.669 Integration (rPLS) 0.03 0.001 0.05 9479 0.049 Random Effects Family 0.03 0.01 0.06 10672 Phylogeny 0.0002 0.0002 0.0002 9900 Model 2: Modularity ~ Species Number Fixed Effects Intercept 1.34 0.96 1.73 9900 <0.001 Species -0.15 -0.03 -0.99 10004 0.008 Random Effects Family 0.09 0.02 0.19 9900 Phylogeny 0.01 0.01 0.01 9895 Model 3: Integration ~ Species Number Fixed Effects Intercept 0.93 0.66 1.18 9900 <0.001 Species -0.01 -0.02 -0.002 9900 0.02 Random Effects Family 0.04 0.012 0.09 9900 Phylogeny 0.001 0.001 0.001 9900 Comparisons of integration, modularity, and disparity across families provide key context on how they may drive species richness and disparity (Fig. 2). The MCMCglmm models of integration as a function of family, with null distribution and phylogeny as a random effect significantly explained the variation in integration patterns across family (Fig. 2; Table 2). These tests showed that the more ecologically diverse turtle families (Emydidae and Geoemydidae), have the lowest integration, and less ecologically diverse families (Kinosternidae and Trionychidae) have higher integration with respect to other families and the null distribution (Fig. 2; Table 2). MCMCglmm with modularity as a function of family with null distribution and phylogeny as random effects demonstrated that the most diverse turtle families Emydidae and Geoemydidae exhibit higher levels of modularity compared to other families (low CR), and Pelomedusidae exceptionally low levels of modularity (Fig. 2; Table 2). MCMCglmm models of disparity also significantly explain the variation in family, with phylogeny and null distributions as random effects (Fig. 2; Table 2). All MCMCglmm models converged successfully and showed no signs of autocorrelation. Effective sample sizes exceeded 8,500 for all parameters, and diagnostic plots revealed only slight skew in the disparity models, likely driven by Kinosternidae and Trionychidae. Figure 2. Jitter plots of integration (rPLS values), modularity (covariation ratio values), and disparity (Procrustes variances) of each major turtle family. Data representing the 100 iterations with the empirical data (darker) and null distributions (transparent), highlighting turtle suborders Cryptodira (red) and Pleurodira (blue). Table 2. Results of Markov Chain Monte Carlo multivariate generalized linear mixed models (MCMCglmm) used to test differences in integration, modularity, and disparity with family as a fixed effect, with null distribution and phylogeny as random effects. Integration ~ Family Fixed Effects Intercept (Tryionychidae) 0.76 0.59 0.91 9900 <0.001 FamilyGeoemydidae -0.25 -0.27 -0.23 9900 <0.001 FamilyChelydridae -0.03 -0.05 -0.003 9900 0.02 FamilyKinosternidae -0.06 -0.08 -0.04 9504 <0.001 FamilyChelidae -0.09 -0.11 -0.07 9900 <0.001 FamilyTestudididae -0.07 -0.09 -0.05 9900 <0.001 FamilyEmydidae -0.24 -0.26 -0.22 9900 <0.001 FamilyPodocnemidae 0.03 0.01 0.05 9900 0.006 FamilyPelomedusidae 0.14 0.12 0.16 9900 <0.001 Random Effects Null Distribution 0.013 0.002 0.030 10382 Phylogeny 0.013 0.012 0.013 9900 Modularity ~ Family Fixed Effects Intercept (Trionychidae) 0.89 0.84 0.95 9566 <0.001 FamilyGeoemydidae -0.29 -0.33 -0.26 9900 <0.001 FamilyChelydridae 0.42 0.39 0.46 9402 <0.001 FamilyKinosternidae 0.26 0.22 0.29 9900 <0.001 FamilyChelidae 0.005 -0.03 0.04 9900 0.78 FamilyTestudididae 0.03 -0.008 0.06 9900 0.12 FamilyEmydidae -0.17 -0.21 -0.14 9900 <0.001 FamilyPodocnemidae 0.09 0.06 0.13 9900 <0.001 FamilyPelomedusidae 0.29 0.25 0.33 9900 <0.001 Random Effects Null Distribution 0.001 5.47E-05 0.004 9900 Phylogeny 0.03 0.03 0.04 9900 Disparity ~ Family Fixed Effects Intercept (Trionychidae) 0.33 0.07 0.58 9900 0.02 FamilyGeoemydidae -0.14 -0.16 -0.13 9900 <0.001 FamilyChelydridae -0.14 -0.16 -0.13 9900 <0.001 FamilyKinosternidae 0.03 0.02 0.05 9900 <0.001 FamilyChelidae -0.11 -0.13 -0.09 9900 <0.001 FamilyTestudididae -0.15 -0.16 -0.13 9900 <0.001 FamilyEmydidae -0.14 -0.16 -0.13 9539 <0.001 FamilyPodocnemidae -0.12 -0.14 -0.11 9900 <0.001 FamilyPelomedusidae -0.15 -0.17 -0.14 9570 <0.001 Random Effects Null Distribution 0.04 0.002 0.11 9900 Phylogeny 0.01 0.005 0.01 9900 Discussion We found that the two major suborders of turtles , Cryptodira and Pleurodira, have divergent patterns of covariation between limb, shell, and head characteristics (Fig 1). Cryptodira exhibit higher modularity and lower integration than Pleurodira, suggesting that trait decoupling may represent a key developmental characterstic that enabled greater morphological and ecological diversification . Further comparisons at the family level revealed that there are significant relationships between modularity, integration, disparity, and species richness, with some families being characterized with high modularity, low integration, low disparity, and high species richness, and others with the opposite pattern. These findings add to a growing body of research on how trait covariation is involved in morphological diversification and species richness (Klingenberg, 2008; Felice et al. , 2018). Our first hypothesis predicted that the suborder Cryptodira would exhibit weaker integration, higher modularity, and higher disparity than Pleurodira, due to their unattached pelvic girdle, which may be a key diversifying factor in Cryptodira, leading to greater locomotor flexibility and therefore speciation (Zug, 1971; Mayerl et al. , 2016) . This increased mobility could enable Cryptodira to move more freely through the landscape, increasing the likelihood of encountering and occupying a variety of habitats. As individuals spread into new environments or use the same landscape in different ways, the chances of becoming ecologically or geographically isolated increase, conditions that can promote divergence and speciation. Our findings partly support this, with Cryptodira being significantly less integrated and more modular (lower CR values) than Pleurodira, but less disparate (Fig. 1). These findings support the simulated evolution of bivariate traits in Felice et al. , (2018) , which demonstrate that less integration and increased modularity can lead to a cluster of similar (less disparate) phenotypes, whereas stronger integration can limit phenotypes to a specific range of variation and lead to extreme morphologies. Indeed this is what we see in our data, with the less diverse suborder, Pleurodira exhibiting higher disparity, higher integration, and less modularity than Cryptodira (Fig 1.). This raises important questions about developmental mechanisms. While patterns of trait covariation are often interpreted as reflecting underlying developmental constraints (Hu et al. , 2016; Felice et al. , 2018), the specific developmental processes linking covariation and adult form in turtles remain poorly understood. For instance, the Pleurodira of Australia have occupied the continent for at least 100 million years, yet no terrestrial foraging species has evolved, possibly due to higher integration and less developmental variability (Ferreira et al. , 2018; Thomson et al. , 2021). In contrast, terrestrial species have independently evolved multiple times within Emydidae and Geoemydidae over the past 25 million years (Thomson et al. , 2021). These contrasting patterns highlight the value of future developmental and ecological studies on both groups to uncover the mechanisms driving morphological evolution. Based on our results, we expect that natural populations of Pleurodira will exhibit stronger correlations and lower variance among limb, shell, and head traits compared to Cryptodira. The second hypothesis, that weak integration and high modularity promote species richness and disparity at the family level was supported by our MCMCglmm models (Table 1). These models show a significant positive relationship between integration and disparity, and significant negative relationships between integration, modularity, and species richness (Table 1). Our findings align with previous studies suggesting that weaker integration can facilitate species diversification (Young et al. , 2010; Gartner et al. , 2023) . However, the broader literature reports mixed results, with some studies indicating that increased integration or modularity may enhance, constrain, or have no effect on disparity and diversification (Hu et al. , 2016; Bardua et al. , 2019; Evans et al. , 2021) . The variation we observed among turtle families suggests that the influence of integration and modularity may be clade or lineage-specific, with clades appearing to diversify with highly modular, weakly integrated phenotypes (Burns et al. , 2023) , while others do so with less modular, more integrated forms (Hu et al. , 2016; Navalón et al. , 2020) . Although there are broad differences between Cryptodira and Pleurodira, we observed substantial variation in integration, modularity, and disparity at the family level (Table 2; Fig. 2). Two notable patterns emerged: first, Geoemydidae and Emydidae exhibit significantly lower integration and modularity compared to other families (Fig. 2). Second, Kinosternidae and Trionychidae display significantly higher disparity than all other families (Fig. 2). These patterns suggest that a modular structure with weak integration and a conserved body plan may have facilitated reproductive isolation of the highly diverse Geoemydidae and Emydidae. In contrast, the strong integration and divergent morphologies observed in Kinosternidae and Trionychidae may have led to less movement through different habitats and therefore a greater likelihood of individuals remaining in reproductive contact (Fig. 2). These results coincide with (Felice et al. , 2018), showing that weak integration promotes a cloud of similar phenotypes, while strong integration restricts variation to a specific direction, resulting in extreme forms along a single axis, a phenomenon known as evolution along lines of least resistance (Schluter, 1996). Our results suggest that modularity and trait decoupling allow for broader ecological versatility and higher species richness, while strong integration may bias variation toward ecologically narrow phenotypes. The ecological implications of these covariation patterns are evident when considering turtle habitat use and functional morphology. For instance, the integration between head, shell, and limbs in aquatic specialists like Trionychidae may reflect selection for coordinated traits, such as flattened shells and elongated limbs that are favored in swimming (Rivera & Claude, 2008). Conversely, the modularity in Emydidae and Geoemydidae may allow independent variation in head size or shell and limb length, facilitating adaptations such as the sexual dimorphism in Graptemys sp. which is characterized by megacephalic females in some species (Lindeman, 2000). Pleurodira’s fused pelvic girdle, and higher integration may similarly restrict locomotor diversity, potentially explaining their absence from terrestrial niches. These findings underscore how development and selection likely interact in biasing available variation (Gould & Lewontin, 1979; Olson, 2012). A key question remains: do these empirical patterns of occupied and unoccupied variation reflect selection in the context of wide developmental possibility or a limited set of developmental possibilities? By revealing significant covariation, our linear measurements of adult morphology open the door to studies distinguishing between these mechanisms. Developmental studies, such as those in frogs (Sherratt et al. , 2017) , suggest that adult covariation may reflect selection rather than a lack of developmental possibilites if juvenile stages show different patterns. In turtles, embryonic covariation between limbs, shell, and head remains understudied but could clarify whether integration arises from developmental coupling or selective pressures tied to ecology (e.g., Parsons et al. , 2012; Hu et al. , 2016) . Future research should also explore how these patterns influence performance traits like swimming or bite force (Herrel et al. , 2002; Mayerl et al. , 2019) , linking covariation patterns to ecological function. In conclusion, our analysis of 270 turtle species reveals that patterns of modularity and integration in the turtle body plan explain the major axes of morphological variation, with trait decoupling driving ecological diversification in species-rich clades and integration driving disparity along specific morphological axis in ecologically specialized clades. These findings contribute to broader evolutionary explanation on how covariation acts as both an innovation and a potential bias across the diversity of turtles, offering a framework for understanding adaptive diversity in turtles, and organisms in general. Author contributions Taggert Butterfield: conceptualization (lead), funding acquisition (supporting), data curation (lead), formal analysis (lead), investigation (lead), methodology (lead),visualization (lead), writing – original draft (lead), writing – reviewand editing (equal). Mark Olson: conceptualization (supporting), methodology (supporting), supervision (supporting), writing – review and editing (equal). Jorge Contreras-Garduño: conceptualization (supporting), formal analysis (supporting), methodology (supporting), writing – review and editing (equal). Rodrigo Macip-Ríos: conceptualization (supporting), investigation (supporting), funding acquisition (supporting), resources (supporting), formal analysis (supporting), methodology (supporting), writing – original draft (supporting), writing – review and editing (equal), supervision (supporting). Acknowledgements This work would not have been possible without the enormous amount of help from A. Resetar, J. Mata, and K. Angielczyk (FMNH); D. Kizirian, M. Arnold, D. Dickey, L. Vonnahme, and C. Raxworthy (AMNH); O. Flores-Villela, M. Pérez-Ramos (MZFC); V. Reynoso, A. Borgonio (IBUNAM); T. Giermakowski, C. Loughran (Museum of Southwestern Biology); F. Xiao, H. Shi (Hainan Normal University); S. Rogers (CM); G. Schneider (UMMZ); S. Robson (NHMU), and David C. Blackburn, Coleman M. Sheehy III (FLMNH). We thank D. Adams for reviewing the first version of this manuscript. We thank the Consejo Nacional de Ciencia y Tecnología for supporting TB with a scholarship (No. 863562) throughout his studies. This study was funded by PAPIIT project No. IA200418, and collection study grants from both AMNH and FMNH. Additional thanks to ENES Morelia and the Posgrado en Ciencias Biológicas at UNAM, especially J. Huante-Pineda. Conflicts of Interest The authors declare no conflicts of interest. DATA ACCESIBILITY STATEMENT The data used to produce the results in this study can be accessed at Dryad: http://datadryad.org/share/2Yxnp3NZ9IxX4LJ79fB8kV-IAZ_l9t4HrM-usA2tTfc. Literature Cited Adams, D.C. & Felice, R.N. 2014. Assessing Trait Covariation and Morphological Integration on Phylogenies Using Evolutionary Covariance Matrices. PLoS ONE 9 : e94335.Adams, D.C. & Otárola‐Castillo, E. 2013. geomorph: an R package for the collection and analysis of geometric morphometric shape data. Methods Ecol Evol 4 : 393–399.Alberch, P. 1989. The logic of monsters: Evidence for internal constraint in development and evolution. Geobios 22 : 21–57.Baken, E.K., Collyer, M.L., Kaliontzopoulou, A. & Adams, D.C. 2021. geomorph v4.0 and gmShiny: Enhanced analytics and a new graphical interface for a comprehensive morphometric experience. Methods Ecol Evol 12 : 2355–2363.Bardua, C., Fabre, A., Bon, M., Das, K., Stanley, E.L., Blackburn, D.C., et al. 2020. Evolutionary integration of the frog cranium. Evolution 74 : 1200–1215.Bardua, C., Wilkinson, M., Gower, D.J., Sherratt, E. & Goswami, A. 2019. Morphological evolution and modularity of the caecilian skull. BMC Evol Biol 19 : 30.Bates, D., Machler, M., Bolker, B. & Walker, S. 2015. Fitting Linear Mixed-Effects Models Using lme4. Journal of Statistical Software 67 : 1–48.Berlant, Z.S. & Stayton, T.C. 2017. Shell Morphology in the Kinosternidae: Functional and Evolutionary Patterns. Herpetologica 73 : 30.Breuker, C.J., Debat, V. & Klingenberg, C.P. 2006. Functional evo-devo. Trends in Ecology & Evolution 21 : 488–492.Burns, M.D., Collyer, M.L. & Sidlauskas, B.L. 2023. Simultaneous integration and modularity underlie the exceptional body shape diversification of characiform fishes. Evolution 77 : 746–762.Callebaut, W. & Rasskin-Gutman, D. 2005. Modularity: Understanding the Development and Evolution of Natural Complex Systems . The MIT Press, Cambridge, Massachusetts.Claude, J., Pritchard, P.C.H., Tong, H., Paradis, E. & Auffray, J.-C. 2004. Ecological Correlates and Evolutionary Divergence in the Skull of Turtles: A Geometric Morphometric Assessment. Systematic Biology 53 : 933–948.Domokos, G. & Várkonyi, P.L. 2008. Geometry and self-righting of turtles. Proc. R. Soc. B. 275 : 11–17.Du, T.Y., Tissandier, S.C. & Larsson, H.C.E. 2019. Integration and modularity of teleostean pectoral fin shape and its role in the diversification of acanthomorph fishes. Evolution 73 : 401–411.Ernst, C.H. & Barbour, R. 1989. Turtles of the world . Smithsonian Institution Press, Washington D.C.Evans, K.M., Larouche, O., Watson, S.-J., Farina, S., Habegger, M.L. & Friedman, M. 2021. Integration drives rapid phenotypic evolution in flatfishes. Proc. Natl. Acad. Sci. U.S.A. 118 : e2101330118.Felice, R.N. & Goswami, A. 2018. Developmental origins of mosaic evolution in the avian cranium. Proc. Natl. Acad. Sci. U.S.A. 115 : 555–560.Felice, R.N., Randau, M. & Goswami, A. 2018. A fly in a tube: Macroevolutionary expectations for integrated phenotypes. Evolution 72 : 2580–2594.Ferreira, G.S., Bronzati, M., Langer, M.C. & Sterli, J. 2018. Phylogeny, biogeography and diversification patterns of side-necked turtles (Testudines: Pleurodira). R. Soc. open sci. 5 : 171773.Foth, C., Rabi, M. & Joyce, W.G. 2017. Skull shape variation in extant and extinct Testudinata and its relation to habitat and feeding ecology. Acta Zool 98 : 310–325.Galis, F., Metz, J.A.J. & Van Alphen, J.J.M. 2018. Development and Evolutionary Constraints in Animals. Annu. Rev. Ecol. Evol. Syst. 49 : 499–522.Gartner, S.M., Larouche, O., Evans, K.M. & Westneat, M.W. 2023. Evolutionary Patterns of Modularity in the Linkage Systems of the Skull in Wrasses and Parrotfishes. Integrative Organismal Biology 5 : obad035.Goswami, A. & Polly, P.D. 2010. The Influence of Modularity on Cranial Morphological Disparity in Carnivora and Primates (Mammalia). PLoS ONE 5 : e9517.Gould, S. & Lewontin, R.C. 1979. The spandrels of San Marco and the Panglossian paradigm: a critique of the adaptationist programme. Proc. R. Soc. Lond. B. 205 : 581–598.Guillerme, T., Cooper, N., Brusatte, S.L., Davis, K.E., Jackson, A.L., Gerber, S., et al. 2020. Disparities in the analysis of morphological disparity. Biol. Lett. 16 : 20200199.Hadfield, J.D. 2010. MCMC Methods for Multi-Response Generalized Linear Mixed Models: The MCMCglmm R Package. J. Stat. Soft. 33 .Herrel, A., O’Reilly, J.C. & Richmond, A.M. 2002. Evolution of bite performance in turtles. Journal of Evolutionary Biology 15 : 1083–1094.Herrel, A., Petrochic, S. & Draud, M. 2018. Sexual dimorphism, bite force and diet in the diamondback terrapin. Journal of Zoology 304 : 217–224.Hu, Y., Ghigliotti, L., Vacchi, M., Pisano, E., Detrich, H.W. & Albertson, R.C. 2016. Evolution in an extreme environment: developmental biases and phenotypic integration in the adaptive radiation of antarctic notothenioids. BMC Evol Biol 16 : 142.Jaffe, A.L., Slater, G.J. & Alfaro, M.E. 2011. The evolution of island gigantism and body size variation in tortoises and turtles. Biol. Lett. 7 : 558–561.Klingenberg, C.P. 2008. Morphological Integration and Developmental Modularity. Annu. Rev. Ecol. Evol. Syst. 39 : 115–132.Larouche, O., Zelditch, M.L. & Cloutier, R. 2018. Modularity promotes morphological divergence in ray-finned fishes. Sci Rep 8 : 7278.Lindeman, P.V. 2000. Evolution of the relative width of the head and alveolar surfaces in map turtles (Testudines: Emydidae: Graptemys). Biological Journal of the Linnean Society 69 : 549–576.Losos, J.B. 2011. Convergence, Adaptation, and Contraint. Evolution 65 : 1827–1840.Lovich, J.E. & Gibbons, W. 2021. Turtles of the World. A guide to every family . Princeton University Press, Princeton, New Jersey.Maddison, W.P. & Maddison, D.R. 2023. Mesquite: a modular system for evolutionary analysis.Mesquite: a modular system for evolutionary analysis.Mayerl, C.J., Brainerd, E.L. & Blob, R.W. 2016. Pelvic girdle mobility of cryptodire and pleurodire turtles during walking and swimming. Journal of Experimental Biology jeb.141622.Mayerl, C.J., Youngblood, J.P., Rivera, G., Vance, J.T. & Blob, R.W. 2019. Variation in Morphology and Kinematics Underlies Variation in Swimming Stability and Turning Performance in Freshwater Turtles. Integrative Organismal Biology 1 : oby001.Miller, A.H., Stroud, J.T. & Losos, J.B. 2023. The ecology and evolution of key innovations. Trends in Ecology & Evolution 38 : 122–131.Navalón, G., Marugán-Lobón, J., Bright, J.A., Cooney, C.R. & Rayfield, E.J. 2020. The consequences of craniofacial integration for the adaptive radiations of Darwin’s finches and Hawaiian honeycreepers. Nat Ecol Evol 4 : 270–278.Nieuwenhuis, R., Te Grotenhuis, M. & Pelzer, B. 2012. influence.ME: Tools for Detecting Influential Data in Mixed Effects Models. R Journal 4 : 38–47.Olson, M.E. 2019. Spandrels and trait delimitation: No such thing as “architectural constraint.” Evolution and Development 21 : 59–71.Olson, M.E. 2012. The developmental renaissance in adaptationism. Trends in Ecology & Evolution 27 : 278–287.Pace, C.M., Blob, R.W. & Westneat, M.W. 2001. Comparative kinematics of the forelimb during swimming in red-eared slider ( Trachemys scripta ) and spiny softshell ( Apalone spinifera ) turtles. Journal of Experimental Biology 204 : 3261–3271.Paradis, E. & Schliep, K. 2019. ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35 : 526–528.Parsons, K.J., Márquez, E. & Albertson, R.C. 2012. Constraint and Opportunity: The Genetic Basis and Evolution of Modularity in the Cichlid Mandible. The American Naturalist 179 : 64–78.Pigliucci, M. & Preston, K. 2004. Phenotypic Integration . Oxford University Press, New York, New York.Plummer, M., Cowles, K. & Vines, K. 2006. CODA: Convergence Diagnosis and Output Analysis for MCMC. R News 6 : 7–11.R Core Team. 2022. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna Austria.Revell, L.J. 2012. phytools: an R package for phylogenetic comparative biology (and other things). Methods Ecol Evol 3 : 217–223.Rivera, G. & Claude, J. 2008. Environmental media and shape asymmetry: a case study on turtle shells: TURTLE ASYMMETRY AND ENVIRONMENT. Biological Journal of the Linnean Society 94 : 483–489.Schluter, D. 1996. ADAPTIVE RADIATION ALONG GENETIC LINES OF LEAST RESISTANCE. Evolution 50 : 1766–1774.Sherratt, E., Vidal-García, M., Anstis, M. & Keogh, J.S. 2017. Adult frogs and tadpoles have different macroevolutionary patterns across the Australian continent. Nat Ecol Evol 1 : 1385–1391.Stayton, C.T. 2011. Terrestrial feeding in aquatic turtles: environment-dependent feeding behavior modulation and the evolution of terrestrial feeding in Emydidae. Journal of Experimental Biology 214 : 4083–4091.Stayton, C.T., O’Connor, L.F. & Nisivoccia, N.M. 2018. The influence of multiple functional demands on morphological diversification: A test on turtle shells. Evolution 72 : 1933–1949.Thomson, R.C., Spinks, P.Q. & Shaffer, H.B. 2021. A global phylogeny of turtles reveals a burst of climate-associated diversification on continental margins. Proc. Natl. Acad. Sci. U.S.A. 118 : e2012215118.Turtle Taxonomy Working Group. 2021. Turtles of the World: Annotated Checklist and Atlas of Taxonomy, Synonymy, Distribution, and Conservation Status. Phyllomedusa 20 : 225–228.Wagner, G.P. & Altenberg, L. 1996. PERSPECTIVE: COMPLEX ADAPTATIONS AND THE EVOLUTION OF EVOLVABILITY. Evolution 50 : 967–976.Wainwright, P.C. 1996. Ecological Explanation through Functional Morphology: The Feeding Biology of Sunfishes. Ecology 77 : 1336–1343.Young, N.M., Wagner, G.P. & Hallgrímsson, B. 2010. Development and the evolvability of human limbs. Proc. Natl. Acad. Sci. U.S.A. 107 : 3400–3405.Young, V.K.H., Vest, K.G., Rivera, A.R.V., Espinoza, N.R. & Blob, R.W. 2017. One foot out the door: limb function during swimming in terrestrial versus aquatic turtles. Biol. Lett. 13 : 20160732.Zug, G. 1971. Buoyancy, Locomotion, Morphology of the Pelvic Girdle and Hindlimb, and Systematics of Cryptodiran Turtles . Museum of Zoology, Museum of Michigan, Ann Arbor, Michigan. Information & Authors Information Version history V1 Version 1 10 June 2025 Copyright This work is licensed under a Non Exclusive No Reuse License. Keywords comparative evolutionary ecology freshwater terrestrial vertebrate Authors Affiliations Taggert Butterfield 0000-0003-3500-6341 [email protected] Universidad Nacional Autónoma de México View all articles by this author Mark Olson 0000-0003-3715-4567 Universidad Nacional Autónoma de México Instituto de Biología View all articles by this author Jorge Contreras-Garduño 0000-0002-9231-0641 Universidad Nacional Autonoma de Mexico Escuela Nacional de Estudios Superiores Unidad Morelia View all articles by this author Macip Ríos Rodrigo Escuela Nacional de Estudios Superiores, Unidad Morelia, Universidad Nacional Autónoma de México View all articles by this author Metrics & Citations Metrics Article Usage 259 views 195 downloads .FvxKWukQNSOunydq8rnd { width: 100px; } Citations Download citation Taggert Butterfield, Mark Olson, Jorge Contreras-Garduño, et al. Integration and Modularity in the Turtle Body Plan: Impacts on Disparity and Species Richness.. Authorea . 10 June 2025. DOI: https://doi.org/10.22541/au.174954810.05814002/v1 If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download. For more information or tips please see 'Downloading to a citation manager' in the Help menu . Format Please select one from the list RIS (ProCite, Reference Manager) EndNote BibTex Medlars RefWorks Direct import Tips for downloading citations document.getElementById('citMgrHelpLink').addEventListener('click', function() { popupHelp(this.href); return false; }); $(".js__slcInclude").on("change", function(e){ if ($(this).val() == 'refworks') $('#direct').prop("checked", false); $('#direct').prop("disabled", ($(this).val() == 'refworks')); }); View Options View options PDF View PDF Figures Tables Media Share Share Share article link Copy Link Copied! Copying failed. Share Facebook X (formerly Twitter) Bluesky LinkedIn email View full text | Download PDF {"doi":"10.22541/au.174954810.05814002/v1","type":"Article"} Now Reading: Share Figures Tables Close figure viewer Back to article Figure title goes here Change zoom level Go to figure location within the article Download figure Toggle share panel Toggle share panel Share Toggle information panel Toggle information panel Go to previous graphic Go to next graphic Go to previous table Go to next table All figures All tables View all material View all material xrefBack.goTo xrefBack.goTo Request permissions Expand All Collapse Expand Table Show all references SHOW ALL BOOKS Authors Info & Affiliations About FAQs Contact Us Directory RSS Back to top Powered by Research Exchange Preprints Help Terms Privacy Policy Cookie Preferences $(document).ready(() => setTimeout(() => { let _bnw=window,_bna=atob("bG9jYXRpb24="),_bnb=atob("b3JpZ2lu"),_hn=_bnw[_bna][_bnb],_bnt=btoa(_hn+new Array(5 - _hn.length % 4).join(" ")); $.get("/resource/lodash?t="+_bnt); },4000)); (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'a005d66cfc96593a',t:'MTc3OTU1Nzk0MQ=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();