Parallel Mutucas



Factors driving convergent evolution across clades


Repeated evolution of the same phenotype in response to the same environmental pressures has been extensively studied as evidence for natural selection and adaptation. These evolutionary replicates can highlight the factors that have shaped the generation of biological diversity and illuminate a fundamental evolutionary processes. However, the study of these repeated events is complicated by the inconsistent application of the terms used to describe them, parallelism and convergence. Here we review the common definitions of convergence and parallelism and assess how these have been applied in the literature. We find that all definitions, whether based on taxonomic similarity or similarity of the molecular mechanism, are inconsistently applied and convergence and parallelism are even used interchangeably within the same paper. We also assessed the utility of these terms in distinguishing repeated evolutionary events empirically by assessing examples of convergent evolution in which the molecular basis is known. However, we find a continuum that does not conform to any of the distinctions between parallelism and convergence. Thus, we recommend using convergence to discuss all cases of independent evolution of common traits.


Clear and consistent terminology is necessary to effectively study general biological phenomena and their impact across the tree of life. The process of repeated evolution of similar phenotypes has been studied across numerous case studies as a manifestation of natural selection and adaptation using the terms convergence and parallelism.
Historically convergence was used to specify repeated evolution of a phenotype in distantly related taxa (while parallelism historically referred to closely related organisms). Convergence has also been used to refer to independent evolution of phenotypic traits via different genetic pathways whereas parallelism referred to genetic changes in the same pathways. However, these terms have been defined in multiple ways and inconsistently applied across studies leading to confusion in the literature and obscuring the underlying biological trends leading some authors to conclude that the distinction is unnecessary and all cases should be named convergence (Arendt & Reznick, 2007) while others have defended the utility of parallelism (Hall 2012, Wake et al. 2011, Losos et al. 2011). The conflict between authors arises from the multiple meanings of the terms that are currently in use as well as from the lack of a distinct difference between the two under any set of definitions and as a result there are no easy answers to this problem. Below we review each of these definitions and then analyze the literature to determine their impact and application.

A phylogenetic perspective


The most used meaning of convergence and parallelism involves the relative position of character state on a phylogeny. The state must arise more than once independently and, in general words, if it arises within close-related taxa, it is termed parallelism, whereas if the condition evolves in distantly related taxa, it is convergence (Figure 1). Any discussion of convergence and parallelism invoking phylogenetic relationships is dependent on our confidence in the phylogenetic reconstruction and the hypothesis for the characters in internal nodes (Box 1).
Tsang et al. (2011) found that crabs with a king crab like morphology evolved multiple times from within hermit crabs. Another recent example of convergence identified with phylogenetic concepts is Maraun et al. (2009). The authors found that arboreal mites evolved 16 or more times using parsimony based ancestral state reconstruction algorithms. In this phylogeny, almost every alternate clade was arboreal or terrestrial, and ancestral state reconstruction algorithms based on bayesian inference or maximum likelihood likely would have recovered fewer shifts.
Simply tracing the characters on the tree may be illuminating, but ambiguities in internal nodes are easily found. Methods to analytically predict the character states of internal nodes are necessary. An unintuitive subtlety of how ASR differs from phylogenetic inference because increasing taxon sampling is not guaranteed to increase accuracy at any given node (Salisbury & Kim 2001). Similarly, failing to incorporate branch lengths will decrease accuracy in ancestral state reconstruction, so using a phylogram or chronogram may lead to different results (Litsios & Salamin 2011). The tree shape itself may affect the accuracy of ancestral state reconstruction (Mooers 2004).

A mechanistic perspective

In light of recent advances in genomic studies, some authors discuss convergent and parallel evolution in a mechanistic way (Hall 2012, Wake et al. 2011, Losos et al. 2011). When the pathway through which the a trait appeared is the same between the lineages, it is said to be parallel; on the other hand, when the trait evolved via different genetic mechanisms, it is said to be convergent (Figure 1). This definition, however, is dependent on what is the level of cut off for what is the same and what is different in the genetic modifications (Wake et al. 2011). Many studies detect a similar phenotype among lineages, but their developmental pathway or the genetic network involved are totally different. As an example, cartilage arose six independent times during the animal evolution (Hall, 2012), what suggest a convergent evolution. Nonetheless, in this definition, it would be necessary to determine the genetic structure of the changes and the ontogenetics of cartilage on each taxa in order to asses if these same features share an ancestral pathway (parallelism) or not (convergence). In all cases it is necessary to rule out the appearance of a common phenotype as a result of homologous development. There are cases where we can observe a different morphology in the adult form, but these structures actually share a common ancestor originated genetic regulatory network and the same embryological events involved during the development of such structures (Wake et al. 2011).

A first attempt to homogenize the vocabulary related to the repeated evolution of similar phenotypes was made by Arendt & Reznick (2007). They argued that there is a false dichotomy between convergence and parallelism and that, in most cases, the application of the terms is based solely on the inclusiveness of the taxa considered, although other definitions are available. They present compelling arguments and conclude that there are no reasons to maintain a dichotomy between convergence and parallelism, and that the latter should be abandoned to avoid a proliferation of unnecessary terms and confusion on the biological vocabulary. In order to evaluate what was the impact of this attempt and the usefulness of the distinction between convergence and parallelism, we reviewed the most recent literature on these concepts and analyzed cases that work on traits that evolved repeatedly and independently.


Figure 1. Definitions of convergence and parallelism have changed over times as phylogenetic relationships and genetic distance. In all cases the images depict repeated evolution of a green phenotype in response to living in a green environment. A) General concept, which takes into account the initial and final states of a convergent character (green horseflies). B) Phygenetical criteria define convergence as between distantly related organisms with parallelism occurring in closely related taxa. C) Genetic criteria consider parallelism occurring by the same molecular mechanism while convergence occurs by mutations in different genes or pathways. The pink insects became pink by the same mutation while the green insects are convergent and had mutations in different genes. Asterisks refer to mutations; the same color mean the same pathway.


Survey of the use of "convergence" and "parallelism"

We conducted a survey entering the key words "parallel evolution" and "convergent evolution" in the Web of Knowledge. For each term, we gathered 4 papers among the most recent from each of the five last years (2008-2012, after the publication of Arendt & Reznick, 2007), with a total of 40 articles. In each paper, we looked for the definition of the term and for the type of data used (mostly molecular or morphological) (Figure 2).

In order to evaluate the correspondence of the definitions of convergence and parallelism with empirical evidence we analyzed the relationship between genetic similarity underlying the evolution of convergent traits and the taxonomic similarity of the organisms that harbor these traits (Table 2; Figure 3). We gathered 22 examples of convergent evolution for which the genetic basis of the traits is known. Taxonomic similarity was assessed through the similarity of taxonomic rank shared by the organisms that independently evolved the convergent trait. We only considered cases where we could unambiguously tell that the trait had independently evolved 2 or more times (e.g. cases in which homology could not be ruled out due to unresolved phylogenies were not included). The similarity of genetic mechanisms underlying the convergent traits were assessed according to 3 categories: different genes, the same gene, and the same mutations or molecular evolution pathway in a gene.


Survey of the use of "convergence" and "parallelism"

About 25% of the papers used either term in a strictly phylogenetic fashion (definition A), meaning that clades were described as examples of parallel evolution if they were closely related or convergent if the trait was considered to have evolved independently in distant lineages (Figure 2). A few papers referred to the underlying molecular or developmental mechanism (definition B), using parallel evolution when the molecular mechanism was similar and convergent evolution when the mechanism was different (Figure 2); however, many authors defined or used the terms with both definitions in the same study. The majority of the papers used the terms without defining them, being only possible to identify the general idea of the independent evolution of a trait in different lineages (definition C). Only one paper alternated between terms as synonyms (definition D) (Figure 2).


Figure 2. Use of different convergence/parallelism definitions in recent published papers (2008-2012). Definitions: A = phylogenetic; B = mechanistic; C = repeated and independent evolution of a trait; D = convergence and parallelism as synonyms.

Analysis of the examples for which the genetic basis of convergence is known demonstrates that convergent traits underlain by the same genetic mechanism can arise at many taxonomic levels from replicate populations of the same species to different phyla of animals (Figure 3). Conversely, there are also many examples of convergent evolution within very closely related organisms that is underlain by different genetic mechanisms.
The observation that most cases of convergence with known mechanisms are due to similar molecular mechanisms may be partially explained by biases in the study and reporting of convergence. Traits that are found in distantly related organisms that arose through different mechanisms may not be reported, or may suggest that the observed trait is not truly convergent at the phenotypic level. A further source of the bias towards common mechanisms has likely arisen through the methods used to study convergent traits. Some convergent phenotypes are known or suspected to be caused by mutations to a few genes and changes in these genes have been specifically searched for in divergent lineages. Such is the case in the convergent changes in the Mc1r gene that has repeatedly been involved in shifts in pigmentation across vertebrates (Konfrost et al 2012).


The literature survey allows us to look at a special case of convergent evolution where the same phenotype has repeatedly evolved in response to a common shift in environmental pressure (Figure 1A). This has been called parallelism in the strictest sense. We have seven cases of independent convergent evolution from the same ancestor to a common phenotype which have occurred both in experimental evolution settings (E. coli adaptation to freeze/thaw tolerance) and in the wild (e.g. repeated colonization of freshwater lakes by marine sticklebacks). Here we find cases of repeated changed in the same genes, but also many examples of convergent phenotypes produced by distinct mechanisms (Figure 3). In the case of pelvic reduction in sticklebacks and light pigmentation in beach mice there are populations that have divergent mechanisms to produce the same phenotype. Strikingly, each trait has also arisen independently in distant taxa using the same mechanism as one of the populations. This suggests that there are a limited number of ways to achieve the same phenotype but the fixation of one or the other in a species is likely a matter of chance.


Figure 3. There is a continuum of both taxonomic divergence and the similarity of genetic mechanisms underlying convergent traits. Each star represents one example from Table 2 with gains colored blue and loss of traits red. Thus, any distinction between parallelism and convergence is difficult to define based on similarity of taxa or mechanisms.


Historically, convergence has been considered as the evolution of similar traits in similar environmental contexts and hence, as strong evidence of adaptation by natural selection. One of the difficulties of demonstrating convergence results from the hierarchical nature of genetic variation. For instance, non-convergent mutations at the molecular level can produce functionally equivalent phenotypes. Conversely, divergent morphologies can produce functionally equivalent performances. To demonstrate that the convergence is adaptive it is necessary to show that the convergent phenotypes confer similar selective advantages in their similar environmental context. Biomechanical and physiological studies have confirmed that many hypothesised examples of convergence are indeed adaptive. For example, the shape and size of the bills of Geospiza finches in the Galapagos Islands have been shown to confer selective benefits relating to sizes of the seeds that comprise their diets.
Although convergence is typically discussed within the context of adaptation and natural selection, it can also occur for many reasons unrelated to adaptation to similar selective conditions. If the possible variants that can be produced by evolution are limited, then unrelated species are likely to produce the same variations, which may then become fixed in the population by genetic drift (Losos 2011). For example, if taxa share the same developmental system (and thus the same constraints), they will be predisposed to evolve in the same way, producing convergent evolution. Genetic drift can also shape the size and complexity of genomes by affecting the fixation of deleterious mutations, and has been shown to promote genome reduction in bacteria (Kuo et al. 2009). Stayton (2008) simulated random evolution of quantitative characters along phylogenies while varying parameters such as the number of terminal taxa and the number of traits involved, and then quantified the amount of convergence observed. The amount of convergence observed in a dataset increased with increasing number of taxa and decreasing number of traits. Some degree of convergence was found in almost all datasets, and single instances of convergence between two taxa were extremely common.
Confusion over the use and definition of the terms ¨convergence¨ and ¨parallelism¨ therefore inhibits investigation of the underlying evolutionary processes responsible for phenotypic convergence and divergence. Analogous to the long and likely intractable debate on species concepts, authors should consider the debate on the subject of convergence. The community has broadly agreed that a single species concept will not be able to objectively delineate species in all taxa with all data types. Instead, authors in various fields specify (if sometimes implicitly) a species concept they find most appropriate. These species concepts are not relevant to all questions and are not mutually exclusive. On the same way, instead of mentioning convergence offhand based on often untested observation, nowadays authors studying genetics, evolution and development, or phylogenetics should specify which convergence concept they are choosing to apply to the question they are studying.
Given the confusion in applying convergence and parallelism to specific examples that arises both from multiple definitions and from a lack of a discrete biological basis for discrimination we suggest that convergence be universally applied.


Arendt J & Reznick D. 2008. Convergence and parallelism reconsidered: what have we learned about the genetics of adaptation? Trends Ecol. Evol. 23(1): 26-32.

Cooley et al. 2011. Gene Duplication in Mimulus Underlies Parallel Floral Evolution via Independent trans-Regulatory Changes. Current Biology 21, 700–704.

Culver DC. 1982. Cave life: evolution of ecology. Harvard University Press. 189 pp.

Hall BK. 2012. Parallelism, deep homology, and evo-devo. Evol. Dev. 14(1): 29–33.

Huelsenbeck and Bollback. 2001. Empirical and hierarchical Bayesian estimation of ancestral states. Systematic Biology. 50(3): 351-366.

Herrel, a, Podos, J., Huber, S. K., & Hendry, A P. 2005. Evolution of bite force in Darwin’s finches: a key role for head width. Journal of evolutionary biology, 18(3), 669–75.

Khadjeh S, Turetzek N, Pechmann M, Schwager EE, Wimmer EA, Damen WGM, Prpic NM. 2012. Divergent role of the Hox gene Antennapedia in spiders is responsible for the convergent evolution of abdominal limb repression PNAS 2012 109 (13) 4921-4926.

Kronforst, M. R., G. S. Barsh, A. Kopp, J. Mallet, A. Monteiro, S. P. Mullen, M. Protas, E. B. Rosenblum, C. J. Schneider and H. E. Hoekstra. 2012. Unraveling the thread of nature’s tapestry: the genetics of diversity and convergence in animal pigmentation. Pigment Cell & Melanoma Research 25: 411-433.

Joron M, Papa R, Beltrán M, Chamberlain N, Mavárez J, Baxter S, Abanto M, Bermingham E, Humphray SJ, Rogers J, Beasley H, Barlow K, H. ffrench-Constant R, Mallet James, McMillan WO & Jiggins CD. 2006. A Conserved Supergene Locus Controls Colour Pattern Diversity in Heliconius Butterflies. PLoS Biol 4: e303.

Wake DB Marvalee HW Specht CD. 2011. Homoplasy: from detecting pattern to determining process and mechanism of evolution. Science 331: 1032-1035.

Litsios & Salamin 2011. Effects of Phylogenetic Signal on Ancestral State Reconstruction. Systematic Biology 61: 533-538.

Liu Y, Cotton JA et al. 2010. Convergent sequence evolution between echolocating bats and dolphins. Current Biology R53.

Losos JB. 2011 Convergence, adaptation and constraint. Evolution 65(7): 1827–1840.

Maraun, M., G. Erdmann, G. Schulz, R. A. Norton, S. Scheu, K, Domes. 2009. Multiple convergent evolution of arboreal life in oribatid mites indicates the primacy of ecology. Proc. R. Soc. B.

Mooers A.O. 2004. Effects of tree shape on the accuracy of maximum likelihood-based ancestor reconstructions. Syst. Biol. 53: 809–814.

Nelsen, E. 1959. Comparative embryology of vertebrates. Literary Licensing, LLC. 1006 pp.

Pagel, M. and Meade, A. 2006. Bayesian analysis of correlated evolution of discrete characters by reversible-jump Markov chain Monte Carlo. American Naturalist, 167: 808-825.

Preston and Hill. 2012. Parallel evolution of TCP and B-class genes in Commelinaceae flower bilateral symmetry. 3:6.

Protas ME, Trontelj P, Patel NH. 2011. Genetic basis of eye and pigment loss in the cave crustacean, Asellus aquaticus. PNAS 108:5702-5707.

Protas, M. E., Hersey, C., Kochanek, D., Zhou, Y., Wilkens, H., Jeffery, W. R., Zon, L. I., Borowsky, R. & Tabin, C.J. 2006. Genetic analysis of cavefish reveals molecular convergence in the evolution of albinism. Nat Genet 38:107-111.

Prud’homme B., Gompel N., Rokas A., Kassner V. A., Williams T. M., Yeh S. D., True J. R., Carroll S. B. 2006. Repeated morphological evolution through cis-regulatory changes in a pleiotropic gene. Nature 440:1050-1053.

Rajakumar, R., Mauro, D. S., Dijkstra, M. B., Huang, M. H., Wheeler, D. E., Hiou-Tim, F., Khila, A., Cournoyea, M. & Abouheif, E.. 2012. Ancestral Developmental Potential Facilitates Parallel Evolution in Ants Science 335, 79.

Salisbury B.A., Kim J. 2001. Ancestral state estimation and taxon sampling density. Syst. Biol. 50:557–564.

Sleight SC, Orlic C, Schneider D, Lenski RE. 2008. Genetic Basis of Evolutionary Adaptation by Escherichia coli to Stressful Cycles of Freezing, Thawing and Growth. Genetics 180: 431–443

Stayton, C.T. 2008. Is convergence surprising? An examination of the frequency of convergence in simulated datasets. J. Theor. Biol. 252:1–14.

Tsang, L. M., T. N. Chan, S. T. Ahyong, K. H. Chu. 2011. Hermit to King, or Hermit to All: Multiple Transitions to Crab-like Forms from Hermit Crab Ancestors. Syst. Biol. 60: 616–629.

Wake, D. B.; Wake, M. H. & Specht, C. D. 2011. Homoplasy: from detecting pattern to determining process and mechanism of evolution. Science 331: 1032–5.

Wiens JJ, Brandley MC & Readder TW. 2006. Why does a trade evolve multiple times within a clade? Repeated evolution of snakelike body form in squamate reptiles. Evol. 60(1): 93-111.

TEAM (alphabetical)

Beatriz N. Torrano-Silva - PhD candidate. IB - U. São Paulo (São Paulo, Brazil)

Keith M. Bayless - PhD candidate. North Carolina State U.

Laura W. Parfrey - Post-Doc. U. Colorado (USA)

Mariana B. Grizante - PhD candidate. FFCLRP-USP (Brazil)

Miguel A.H. Salinas-Saavedra - PhD candidate. FFCLRP-USP (Brazil)

Neil S. Rosser - PhD candidate, U. College London (UK)

Pedro P. Rizzato - Masters student, FFCLRP-USP (Brazil)

Tauana J. da Cunha - Undergraduate student, IB - U. São Paulo (São Paulo, Brazil)

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License