Does Phenotypic Integration Promote Convergent Evolution?

Phenotypic integration is often perceived as being able to produce convergent evolution in the absence of selection, but specific mechanisms for this process are lacking and a connection has never been empirically demonstrated. A new model of the effect of integration on convergence provides such a mechanism, along with other predictions about the influence of integration on evolutionary patterns. I use simulations and data from three empirical systems-turtle shells, characiform fishes, and squirrel mandibles-to investigate the degree to which evolutionary integration is associated with high levels of convergent evolution. Levels of integration were varied in Brownian motion simulations and the resulting amounts of stochastic convergent evolution were quantified. Each empirical system was divided into modules, and the strength of integration, average amount of convergence, phenotypic disparity, and rate of evolution in each module was measured. Results from the simulations and from all three empirical systems converge on a common result: higher levels of phenotypic integration are indeed associated with higher levels of convergence. This is despite a lack of consistent association between the strength of phenotypic integration and evolutionary rate or disparity. The results here are only correlational. Further studies which more closely examine the influence of within-population drivers of evolutionary integration-for example, genetic or developmental integration-on convergence are required before it is possible to definitively establish when phenotypic integration can cause evolutionary convergence. Until then, however, the results of this study strongly suggest that phenotypic integration will often promote convergent evolution.

Challenges and advances in measuring phenotypic convergence.

Tests of phenotypic convergence can provide evidence of adaptive evolution, and the popularity of such studies has grown in recent years due to the development of novel, quantitative methods for identifying and measuring convergence. These methods include the commonly applied C1-C4 measures of Stayton (2015a), which measure morphological distances between lineages, and Ornstein-Uhlenbeck (OU) model-fitting analyses, which test whether lineages converged on shared adaptive peaks. We test the performance of C-measures and other convergence measures under various evolutionary scenarios and reveal a critical issue with C-measures: they often misidentify divergent lineages as convergent. We address this issue by developing novel convergence measures-Ct1-Ct4-measures-that calculate distances between lineages at specific points in time, minimizing the possibility of misidentifying divergent taxa as convergent. Ct-measures are most appropriate when focal lineages are of the same or similar geologic ages (e.g., extant taxa), meaning that the lineages' evolutionary histories include considerable overlap in time. Beyond C-measures, we find that all convergence measures are influenced by the position of focal taxa in phenotypic space, with morphological outliers often statistically more likely to be measured as strongly convergent. Further, we mimic scenarios in which researchers assess convergence using OU models with a priori regime assignments (e.g., classifying taxa by ecological traits) and find that multiple-regime OU models with phenotypically divergent lineages assigned to a shared selective regime often outperform simpler models. This highlights that model support for these multiple-regime OU models should not be assumed to always reflect convergence among focal lineages of a shared regime. Our new Ct1-Ct4-measures provide researchers with an improved comparative tool, but we emphasize that all available convergence measures are imperfect, and researchers should recognize the limitations of these methods and use multiple lines of evidence to test convergence hypotheses.

Are our phylomorphospace plots so terribly tangled? An investigation of disorder in data simulated under adaptive and nonadaptive models.

Contemporary methods for visualizing phenotypic evolution, such as phylomorphospaces, often reveal patterns which depart strongly from a naïve expectation of consistently divergent branching and expansion. Instead, branches regularly crisscross as convergence, reversals, or other forms of homoplasy occur, forming patterns described as "birds' nests", "flies in vials", or less elegantly, "a mess". In other words, the phenotypic tree of life often appears highly tangled. Various explanations are given for this, such as differential degrees of developmental constraint, adaptation, or lack of adaptation. However, null expectations for the magnitude of disorder or "tangling" have never been established, so it is unclear which or even whether various evolutionary factors are required to explain messy patterns of evolution. I simulated evolution along phylogenies under a number of varying parameters (number of taxa and number of traits) and models (Brownian motion, Ornstein-Uhlenbeck (OU)-based, early burst, and character displacement (CD)] and quantified disorder using 2 measures. All models produce substantial amounts of disorder. Disorder increases with tree size and the number of phenotypic traits. OU models produced the largest amounts of disorder-adaptive peaks influence lineages to evolve within restricted areas, with concomitant increases in crossing of branches and density of evolution. Large early changes in trait values can be important in minimizing disorder. CD consistently produced trees with low (but not absent) disorder. Overall, neither constraints nor a lack of adaptation is required to explain messy phylomorphospaces-both stochastic and deterministic processes can act to produce the tantalizingly tangled phenotypic tree of life.

Performance Surface Analysis Identifies Consistent Functional Patterns across 10 Morphologically Divergent Terrestrial Turtle Lineages.

Newly-developed methods for utilizing performance surfaces-multivariate representations of the relationship between phenotype and functional performance-allow researchers to test hypotheses about adaptive landscapes and evolutionary diversification with explicit attention to functional factors. Here, information from performance surfaces of three turtle shell functions-shell strength, hydrodynamics, and self-righting-is used to test the hypothesis that turtle lineages transitioning from aquatic to terrestrial habitats show patterns of shell shape evolution consistent with decreased importance of hydrodynamic performance. Turtle shells are excellent model systems for evolutionary functional analysis. The evolution of terrestriality is an interesting test case for the efficacy of these methods because terrestrial turtles do not show a straightforward pattern of morphological convergence in shell shape: many terrestrial lineages show increased shell height, typically assumed to decrease hydrodynamic performance, but there are also several lineages where the evolution of terrestriality was accompanied by shell flattening. Performance surface analyses allow exploration of these complex patterns and explicit quantitative analysis of the functional implications of changes in shell shape. Ten lineages were examined. Nearly all terrestrial lineages, including those which experienced decreased shell height, are associated with morphological changes consistent with a decrease in the importance of shell hydrodynamics. This implies a common selective pattern across lineages showing divergent morphological patterns. Performance studies such as these hold great potential for integrating adaptive and performance data in macroevolutionary studies.

Performance in three shell functions predicts the phenotypic distribution of hard-shelled turtles.

Adaptive landscapes have served as fruitful guides to evolutionary research for nearly a century. Current methods guided by landscape frameworks mostly utilize evolutionary modeling (e.g., fitting data to Ornstein-Uhlenbeck models) to make inferences about adaptive peaks. Recent alternative methods utilize known relationships between phenotypes and functional performance to derive information about adaptive landscapes; this information can then help explain the distribution of species in phenotypic space and help infer the relative importance of various functions for guiding diversification. Here, data on performance for three turtle shell functions-strength, hydrodynamic efficiency, and self-righting ability-are used to develop a set of predicted performance optima in shell shape space. The distribution of performance optima shows significant similarity to the distribution of existing turtle species and helps explain the absence of shells in otherwise anomalously empty regions of morphospace. The method outperforms a modeling-based approach in inferring the location of reasonable adaptive peaks and in explaining the shape of the phenotypic distributions of turtle shells. Performance surface-based methods allow researchers to more directly connect functional performance with macroevolutionary diversification, and to explain the distribution of species (including presences and absences) across phenotypic space.

The influence of multiple functional demands on morphological diversification: A test on turtle shells.

Organismal parts are often involved in the performance of more than one function. The role of trade-offs in influencing phenotypic evolution of such parts is well-studied; less well-understood is their role in influencing phenotypic diversity. Increases in the number of functions a part is involved in may inhibit subsequent diversification, as the number of trade-offs increases. Alternately, such an increase might promote phenotypic diversification, by increasing adaptive landscape complexity and promoting specialization for different roles. We compare these predictions by testing whether aquatic turtle shells, which resist loads, act as hydrodynamic elements, facilitate self-righting, and exchange heat with the environment, differ in phenotypic diversity from those of terrestrial species, which perform all the same functions except for hydrodynamics. We used 53 3D landmarks digitized on 2722 specimens of 274 hard-shelled turtle species to quantify shell shape variation, and a set of phylogenetic hypotheses to examine evolutionary patterns. Terrestrial turtles consistently had higher phenotypic diversity than aquatic species. Differences are not due to differences in the rates of evolution between the two groups, but rather differences in evolutionary mode. Thus this study supports the traditional view of the role of multiple functions in determining phenotypic diversity.

Warped finite element models predict whole shell failure in turtle shells.

Finite element (FE) models have become increasingly popular in comparative biomechanical studies, with researchers continually developing methods such as 'warping' preexisting models to facilitate analyses. However, few studies have investigated how well FE models can predict biologically crucial whole-structure performance or whether 'warped' models can provide useful information about the mechanical behavior of actual specimens. This study addresses both of these issues through a validation of warped FE models of turtle shells. FE models for 40 turtle specimens were built using 3D landmark coordinates and thin-plate spline interpolations to warp preexisting turtle shell models. Each actual turtle specimen was loaded to failure, and the load at failure and mode of fracture were then compared with the behavior predicted by the models. Overall, the models performed very well, despite the fact that many simplifying assumptions were made for analysis. Regressions of observed on predicted loads were significant for the dataset as a whole, as well as in separate analyses within two turtle species, and the direction of fracture was generally consistent with the patterns of stresses observed in the models. This was true even when size (an important factor in determining strength) was removed from analyses - the models were also able to predict which shells would be relatively stronger or weaker. Although some residual variation remains unexplained, this study supports the idea that warped FE models run with simplifying assumptions at least can provide useful information for comparative biomechanical studies.

What does convergent evolution mean? The interpretation of convergence and its implications in the search for limits to evolution.

Convergent evolution is central to the study of life's evolutionary history. Researchers have documented the ubiquity of convergence and have used this ubiquity to make inferences about the nature of limits on evolution. However, these inferences are compromised by unrecognized inconsistencies in the definitions, measures, significance tests and inferred causes of convergent evolution. I review these inconsistencies and provide recommendations for standardizing studies of convergence. A fundamental dichotomy exists between definitions that describe convergence as a pattern and those that describe it as a pattern caused by a particular process. When this distinction is not acknowledged it becomes easy to assume that a pattern of convergence indicates that a particular process has been active, leading researchers away from alternative explanations. Convergence is not necessarily caused by limits to evolution, either adaptation or constraint; even substantial amounts of convergent evolution can be generated by a purely stochastic process. In the absence of null models, long lists of examples of convergent events do not necessarily indicate that convergence or any evolutionary process is ubiquitous throughout the history of life. Pattern-based definitions of convergence, coupled with quantitative measures and null models, must be applied before drawing inferences regarding large-scale limits to evolution.

The definition, recognition, and interpretation of convergent evolution, and two new measures for quantifying and assessing the significance of convergence.

Convergent evolution is an important phenomenon in the history of life. Despite this, there is no common definition of convergence used by biologists. Instead, several conceptually different definitions are employed. The primary dichotomy is between pattern-based definitions, where independently evolved similarity is sufficient for convergence, and process-based definitions, where convergence requires a certain process to produce this similarity. The unacknowledged diversity of definitions can lead to problems in evolutionary research. Process-based definitions may bias researchers away from studying or recognizing other sources of independently evolved similarity, or lead researchers to interpret convergent patterns as necessarily caused by a given process. Thus, pattern-based definitions are recommended. Existing measures of convergence are reviewed, and two new measures are developed. Both are pattern based and conceptually minimal, quantifying nothing but independently evolved similarity. One quantifies the amount of phenotypic distance between two lineages that is closed by subsequent evolution; the other simply counts the number of lineages entering a region of phenotypic space. The behavior of these measures is explored in simulations; both show acceptable Type I and Type II error. The study of convergent evolution will be facilitated if researchers are explicit about working definitions of convergence and adopt a standard toolbox of convergence measures.

The colorful language of Australian flowers.

The enormous increase in phylogenetic information in recent years has allowed many old questions to be reexamined from a macroevolutionary perspective. We have recently considered evolutionary convergence in floral colors within pollination syndromes, using bird-pollinated species in Australia. We combined quantitative measurements of floral reflectance spectra, models of avian color vision, and a phylogenetic tree of 234 Australian species to show that bird-pollinated flowers as a group do not have colors that are significantly different from the colors of insect-pollinated flowers. However, about half the bird-pollinated flowers have convergently evolved a narrow range of colors with dominant long-wavelength reflection far more often than would be expected by chance. These convergent colors would be seen as distinctly different from other floral colors in our sample when viewed by honeyeaters (family Meliphagidae), birds with a phylogenetically ancestral type of color vision and the dominant avian pollinators in Australia. Our analysis shows how qualitative ideas in natural history, like the concept of pollination syndromes, can be given more precise definition and rigorous statistical testing that takes into account phylogenetic information.

Distinctive convergence in Australian floral colours seen through the eyes of Australian birds.

We used a colour-space model of avian vision to assess whether a distinctive bird pollination syndrome exists for floral colour among Australian angiosperms. We also used a novel phylogenetically based method to assess whether such a syndrome represents a significant degree of convergent evolution. About half of the 80 species in our sample that attract nectarivorous birds had floral colours in a small, isolated region of colour space characterized by an emphasis on long-wavelength reflection. The distinctiveness of this 'red arm' region was much greater when colours were modelled for violet-sensitive (VS) avian vision than for the ultraviolet-sensitive visual system. Honeyeaters (Meliphagidae) are the dominant avian nectarivores in Australia and have VS vision. Ancestral state reconstructions suggest that 31 lineages evolved into the red arm region, whereas simulations indicate that an average of five or six lineages and a maximum of 22 are likely to have entered in the absence of selection. Thus, significant evolutionary convergence on a distinctive floral colour syndrome for bird pollination has occurred in Australia, although only a subset of bird-pollinated taxa belongs to this syndrome. The visual system of honeyeaters has been the apparent driver of this convergence.

Effects of asymmetry on the strength of the chelonian shell: a comparison of three species.

A major focus of the field of organismal biology is to understand how morphology impacts performance. Although the functional implications of certain aspects of shape have been widely examined, the functional implications of a related parameter, symmetry, remain mostly unknown. We used finite-element models to examine the effects of turtle shell asymmetry on shell strength across three morphologically distinct emydid species. The goals of this study were to: 1) test the hypothesis that increased asymmetry (independent of differences in shape) is associated with increased stress levels for a given load, and thus with weaker shells, 2) ascertain how asymmetry and the position of load application interact to influence shell strength, and 3) determine how interspecific differences in shape influence the effect of asymmetry. We found that increased asymmetry does produce higher stresses for both midline and non-midline loads. Non-midline loads produce slightly larger and more variable stresses. Species-specific shell shape can mitigate the effects of asymmetry; stronger shapes are potentially more resistant to the negative effects of asymmetry. Our findings indicate that changes in asymmetry associated with relatively small changes in shape can have as much of an effect on stress incurred by the shell as the changes in shape themselves.

Biomechanics on the half shell: functional performance influences patterns of morphological variation in the emydid turtle carapace.

This study uses the carapace of emydid turtles to address hypothesized differences between terrestrial and aquatic species. Geometric morphometrics are used to quantify shell shape, and performance is estimated for two shell functions: shell strength and hydrodynamics. Aquatic turtle shells differ in shape from terrestrial turtle shells and are characterized by lower frontal areas and presumably lower drag. Terrestrial turtle shells are stronger than those of aquatic turtles; many-to-one mapping of morphology to function does not entirely mitigate a functional trade-off between mechanical strength and hydrodynamic performance. Furthermore, areas of morphospace characterized by exceptionally poor performance in either of the functions are not occupied by any emydid species. Though aquatic and terrestrial species show no significant differences in the rate of morphological evolution, aquatic species show a higher lineage density, indicative of a greater amount of convergence in their evolutionary history. The techniques employed in this study, including the modeling of theoretical shapes to assess performance in unoccupied areas of morphospace, suggest a framework for future studies of morphological variation.

Finite element modeling of shell shape in the freshwater turtle Pseudemys concinna reveals a trade-off between mechanical strength and hydrodynamic efficiency.

Aquatic species can experience different selective pressures on morphology in different flow regimes. Species inhabiting lotic regimes often adapt to these conditions by evolving low-drag (i.e., streamlined) morphologies that reduce the likelihood of dislodgment or displacement. However, hydrodynamic factors are not the only selective pressures influencing organismal morphology and shapes well suited to flow conditions may compromise performance in other roles. We investigated the possibility of morphological trade-offs in the turtle Pseudemys concinna. Individuals living in lotic environments have flatter, more streamlined shells than those living in lentic environments; however, this flatter shape may also make the shells less capable of resisting predator-induced loads. We tested the idea that "lotic" shell shapes are weaker than "lentic" shell shapes, concomitantly examining effects of sex. Geometric morphometric data were used to transform an existing finite element shell model into a series of models corresponding to the shapes of individual turtles. Models were assigned identical material properties and loaded under identical conditions, and the stresses produced by a series of eight loads were extracted to describe the strength of the shells. "Lotic" shell shapes produced significantly higher stresses than "lentic" shell shapes, indicating that the former is weaker than the latter. Females had significantly stronger shell shapes than males, although these differences were less consistent than differences between flow regimes. We conclude that, despite the potential for many-to-one mapping of shell shape onto strength, P. concinna experiences a trade-off in shell shape between hydrodynamic and mechanical performance. This trade-off may be evident in many other turtle species or any other aquatic species that also depend on a shell for defense. However, evolution of body size may provide an avenue of escape from this trade-off in some cases, as changes in size can drastically affect mechanical performance while having little effect on hydrodynamic performance.

Application of thin-plate spline transformations to finite element models, or, how to turn a bog turtle into a spotted turtle to analyze both.

Finite element (FE) models are popular tools that allow biologists to analyze the biomechanical behavior of complex anatomical structures. However, the expense and time required to create models from specimens has prevented comparative studies from involving large numbers of species. A new method is presented for transforming existing FE models using geometric morphometric methods. Homologous landmark coordinates are digitized on the FE model and on a target specimen into which the FE model is being transformed. These coordinates are used to create a thin-plate spline function and coefficients, which are then applied to every node in the FE model. This function smoothly interpolates the location of points between landmarks, transforming the geometry of the original model to match the target. This new FE model is then used as input in FE analyses. This procedure is demonstrated with turtle shells: a Glyptemys muhlenbergii model is transformed into Clemmys guttata and Actinemys marmorata models. Models are loaded and the resulting stresses are compared. The validity of the models is tested by crushing actual turtle shells in a materials testing machine and comparing those results to predictions from FE models. General guidelines, cautions, and possibilities for this procedure are also presented.

Is convergence surprising? An examination of the frequency of convergence in simulated datasets.

Although convergence is recognized as a central concept in evolutionary biology, very few tools are available for the quantitative study of this phenomenon. Moreover, although many evolutionary assertions assume that convergence should be rare in the absence of influences on organismal phenotypes such as natural selection or constraint, no studies have tested whether this is the case. I simulate random evolution (Brownian motion model) of quantitative characters along phylogenies with varying numbers of terminal taxa, numbers of traits, variance structure, and tree balance, and quantify the amount of convergence observed in these datasets using four metrics. The amount of convergence observed in a dataset increases with increasing number of taxa and decreasing number of traits, approaching the maximum possible amount of convergence under certain circumstances. Some convergence is expected in almost all datasets. Comparison of empirical datasets to those produced by random evolution provides a test of whether empirical datasets actually show elevated levels of convergence. Out of three test datasets, two show more convergence than expected. Given that high levels of convergence can be produced simply by random evolution, no explanation may be necessary for instances of convergence discovered in an evolutionary investigation.

Testing hypotheses of convergence with multivariate data: morphological and functional convergence among herbivorous lizards.

Despite its importance to evolutionary theory, convergence remains an understudied phenomenon and is usually investigated using qualitative data. This paper advances a new, multidimensional view of convergence. Three patterns indicative of convergence are discussed, and techniques to discover and test convergent patterns in a quantitative framework are developed. These concepts and methods are applied to a dataset of digitized coordinates on 1554 lizard skulls and 1292 lower jaws to test hypotheses of convergence among herbivorous lizards. Encompassing seven independent acquisitions of herbivory, this lizard sample provides an ideal natural experiment for exploring ideas of convergence among different systems (here, morphological and functional). Three related questions are addressed: (1) Do herbivorous lizards show evidence of convergence in skull and lower jaw morphology? (2) What, if any, is the morphospace pattern associated with this convergence? (3) Is it possible to predict the direction of convergence using functional models? Relative warp analysis and permutation tests reveal that the skulls and lower jaws of herbivorous lizards do show evidence of convergence. Herbivore skulls deviate from their carnivorous or omnivorous sister groups toward the same area of morphospace. Without a phylogenetic perspective, this pattern would not be recognizable. Lower jaws of herbivores are not convergent in morphology but are convergent in function: herbivores deviate away from their carnivorous sister groups toward higher values of mechanical advantage. These results illustrate the desirability of quantitative methods, informed by phylogenetic information, in the study of convergence.

Morphological evolution of the lizard skull: a geometric morphometrics survey.

Patterns of diversity among lizard skulls were studied from a morphological, phylogenetic, and functional perspective. A sample of 1,030 lizard skulls from 441 species in 17 families was used to create a lizard skull morphospace. This morphospace was combined with a phylogeny of lizard families to summarize general trends in the evolution of the lizard skull. A basal morphological split between the Iguania and Scleroglossa was observed. Iguanians are characterized by a short, high skull, with large areas of attachment for the external adductor musculature, relative to their sister group. The families of the Iguania appear to possess more intrafamilial morphological diversity than families of the Scleroglossa, but rarefaction of the data reveals this to be an artifact caused by the greater number of species represented in Iguanian families. Iguanian families also appear more dissimilar to one another than families of the Scleroglossa. Permutation tests indicate that this pattern is real and not due to the smaller number of families in the Iguanidae. Parallel and convergent evolution is observed among lizards with similar diets: ant and termite specialists, carnivores, and herbivores. However, these patterns are superimposed over the more general phylogenetic pattern of lizard skull diversity. This study has three central conclusions. Different clades of lizards show different patterns of disparity and divergence in patterns of morphospace occupation. Phylogeny imposes a primary signal upon which a secondary ecological signal is imprinted. Evolutionary patterns in skull metrics, taken with functional landmarks, allow testing of trends and the development of new hypotheses concerning both shape and biomechanics.