Evolution: Convergent and Parallel Evolution Caro-Beth Stewart, State University of New York at Albany, Albany, New York, USA

Advanced article Article Contents . Introduction . Similarity between Sequences . Categories of Convergent and Parallel Evolution

Convergence and parallelism are two forms of independent evolution of similar or identical traits; they are often taken as evidence for adaptive evolution, but can also occur due to design limitations or by chance.

. Sequence Homoplasy – Adaptation or Chance? . Homoplasy in Protein Sequences . Functional Convergence in Proteins

doi: 10.1002/9780470015902.a0005115.pub2

Introduction Observed similarities between biological entities or taxa (such as genes, genomes or species) are most often due to retention of traits that were found in their common ancestors. Such retained similarity is considered to be evidence of homology, or evolutionary relationship, between the entities. Similarity can also result from the independent evolution of a given trait in two or more lineages through convergence, parallel evolution or reversals; these phenomena are often collectively called homoplasy. Importantly, independent evolution of similar or identical traits in different taxa that have been subjected to similar environmental pressures has long been considered to be strong evidence of Darwinian adaptation, a keystone concept in evolutionary biology. Yet, certain levels of homoplasy can arise merely by chance, or can be due to design constraints of biological systems. Thus, care must be taken before invoking adaptive explanations for homoplastic similarities. For over a century, numerous attempts have been made by various authors to create strict definitions for terms that describe the various types of biological similarity (Doolittle, 1994; Fitch, 2000; Gould, 2002; Haas and Simpson, 1946; Lankester, 1870; Sneath and Sokal, 1973). Few authors have fully agreed with one another, and some are outright contradictory. For example, current working definitions of the term homoplasy range from the ‘dichotomous opposite of homology’ (Gould, 2002) to merely ‘parallelisms and reversals’ in homologous molecules (Fitch, 2000). Such confusion in terminology has resulted, at least in part, from the fact that the terms used to describe these evolutionary processes predate modern evolutionary genetics, and have been recycled and redefined over the years, rather than being replaced by more precise terms. See also: Homologous, Orthologous and Paralogous Genes For these reasons, this article will be more descriptive than prescriptive. The major terms involved in describing various types of independent evolution will be explained, in the manner in which they are often used by various fields (developmental biologists, phylogeneticists, molecular evolutionists and structural biochemists). When possible, these concepts will be illustrated using the independent evolution of foregut fermentation in several groups of distantly related animals – the ruminant artiodactyls, the colobine Old World monkeys and a leaf-eating bird, the hoatzin. In all three taxa, the enzyme lysozyme was

independently recruited to digest bacterial cell walls in the animal’s true stomach, and the lysozyme proteins appear to have adapted to the new environment through homoplastic mechanisms. Thus, this case provides examples of homoplasy at both the organismal and molecular levels. Most emphasis will be placed upon molecular examples and definitions, as this article is aimed at elucidating the subject of independent evolution as it relates to genetics and genomics.

Similarity between Sequences The major direction of evolution is divergent (Figure 1a). That is, after a branching event (such as speciation or a gene duplication), differences steadily accumulate in the separate lineages (here, the separate species or gene duplicates) until the descendants are both different from each other and different from their common ancestor. Thus, some genetic and phenotypic similarity is lost over time, but some also remains. Evolution of related entities can sometimes proceed, at least in part, through nondivergent processes such as convergence (Figure 1b) or parallelism (Figure 1c). In addition, unrelated entities can sometimes evolve similar characteristics, a process that is also referred to as convergent evolution (Figure 2). Thus, similarities between biological entities can be due to a variety of reasons. Two such reasons are the retention of ancestral traits – called shared ancestral traits or symplesiomorphies – and the gain of traits in the common ancestor of a given lineage – called shared derived traits or synapomorphies (Figure 3a). These two types of similarity are the immediate products of common ancestry, and thus are clear-cut examples of homology. These types of homologous similarity are sometimes called patristic similarity (Sneath and Sokal, 1973). Ancestral and derived similarities within a protein sequence are illustrated in Figure 3a. Derived similarities provide the strongest ‘signal’ for phylogenetic tree inference, and are the types of changes upon which the branches of phylogenetic trees are ideally constructed. Homologous sequences can also be similar due to independent acquisition of identical characters in two or more lineages. This type of similarity is called homoplastic

ENCYCLOPEDIA OF LIFE SCIENCES & 2007, John Wiley & Sons, Ltd. www.els.net

1

Evolution: Convergent and Parallel Evolution

% Sequence difference

100

0

(a)

0

100

% Divergence

4

3

2 Time

0

1

0

Figure 2 Geometric representation of convergent evolution of unrelated protein sequences. Two completely unrelated proteins might be expected to have between about 5% and 20% sequence identity, depending on amino acid composition, as represented at time 4. If two proteins, or other biological entities, gained significant similarity over time (as indicated by the solid lines), then they would be considered to have ‘converged’ in sequence and/or function.

(b)

0

R R R

Shared ancestral

R K

(c) 4

3

2

1

0

K

(a)

K

Time

R

K

R

K

V

K

S

K

R

R

K

R

R Figure 1 Geometric representations of three directions that the evolution of homologous entities can follow. The x-axis represents time and the y-axis represents percentage divergence from the common ancestor, with 0% represented by the faint dotted lines. (a) The usual direction of evolution is divergent, with the descendant lineages becoming progressively more different from their common ancestor (indicated by the filled circle at the node) and from each other (indicated by the solid lines) over time. (b) Convergent evolution occurs when lineages first diverge and then subsequently regain similarity over time, as shown here between time 1 and 0. (c) Parallel evolution occurs when the descendant lineages neither lose nor gain similarity over time, but follow parallel evolutionary paths, as show here from time 2 through time 0. Both convergence and parallel evolution will slow the normal divergence of lineages, as indicated by the dashed lines in (a).

similarity, and can arise through parallelisms, convergences and reversals (Sneath and Sokal, 1973). These processes are illustrated in Figure 3b, and are discussed in greater detail below. Homoplasy is the major ‘noise’ problem in phylogenetic inference, and is especially troublesome for parsimony. Computer-assisted analysis of traits upon phylogenetic trees is the most objective way to partition observed similarity into its various patristic and homoplastic components.

Categories of Convergent and Parallel Evolution Convergent and parallel evolution are concepts over which considerable confusion and debate have occurred. Much of this problem has to do with the fact that these terms are 2

R

Parallel events

Convergent events

Reversal

R (b)

Shared derived

Figure 3 Evolutionary mechanisms that can produce the same amino acid at a given position in the sequences of homologous proteins. (a) In the case of patristic similarities, the descendant sequences are identical due to shared common ancestry. (b) In the case of homoplastic similarities, the two descendants are identical due to independent gain of the same amino acid at the same position in the protein sequences. See text for further explanation of the different subtypes within two categories. Similar mechanistic principles apply to nucleotide sequences and organismal traits.

used ambiguously to mean different things at different times by different authors. Fortunately, most definitions of convergent and parallel evolution share the following unifying theme: parallelism is generally considered to be the production of apparently identical traits by the same generative systems, whereas convergence is generally considered to be the production of similar traits by different generative systems. This distinction can apply whether one is discussing complex developmental pathways (Gould, 2002) or simple nucleotide substitutions. See also: Developmental Evolution

ENCYCLOPEDIA OF LIFE SCIENCES & 2007, John Wiley & Sons, Ltd. www.els.net

Evolution: Convergent and Parallel Evolution

Myriad definitions of convergence and parallelism can be boiled down to three categories, which can be called phylogenetic, geometric and mechanistic. Haas and Simpson (1946) eloquently elaborated upon the phylogenetic and geometric definitions, and opted in favor of the geometric ones. Sneath and Sokal (1973) also favored geometric definitions, and further illustrated molecular evolutionary processes in mechanistic terms. These three categories of definitions often overlap and impinge on each other, as discussed below. See also: Developmental Evolution

Phylogenetic The basic idea behind the phylogenetic definitions is that parallelism occurs between closely related forms and convergence between distantly related ones. However, there is no universal standard for how distantly related two organisms (or homologous sequences) need be before the homoplasy between them should be termed convergence rather than parallelism. The phylogenetic definitions are not operational definitions, and do not have the analytical powers of the mechanistic and geometric ones. Although many authors have pointed out these pitfalls of the phylogenetic definitions, they have continued to permeate the evolutionary literature and widely persist to this day. This persistence is probably due to the fact that the phylogenetic definitions seem intuitive, and are somewhat useful in describing certain broad evolutionary phenomena. For example, they distinguish between the superficial similarity of whales and fishes (convergence) and the repeated gain of similar coloration in closely related species of insects (parallelism). By the phylogenetic definitions, one might consider the independent evolution of foregut fermentation in the artiodactyls and the colobines to be a case of convergence, due to the fact that these two mammalian groups are not close relatives. The phylogenetic argument for convergence is even stronger for the hoatzin in relation to the mammals. The phylogenetic definitions are sometimes taken to embody the concept of similarity arising from either different (convergent) or similar (parallel) genetic backgrounds, on the apparent assumption that closely related species are more likely to use similar generative methods than are more distantly related ones. By this criterion, whether the independent evolution of foregut fermentation should be considered parallelism or convergence would depend on whether the same or different genetic programs gave rise to the expanded foregut in the different taxonomic groups. At present, this is not known. Another problem with the phylogenetic definition of convergence when discussing organisms is that all living species are related; thus, it makes no sense to talk about similar traits arising in ‘unrelated’ species, as many authors do. Genes and proteins, however, are another matter. There appear to be hundreds of unrelated protein families, which offer numerous opportunities for convergent evolution of the type illustrated in Figure 2.

Geometric The geometric definitions of parallel and convergent evolution can best be envisioned by considering the geometric concepts implied by the words divergent, parallel and convergent (Haas and Simpson, 1946). These concepts are illustrated in Figure 1 and Figure 2. Divergence means that homologous sequences become increasingly different from each other over evolutionary time (Figure 1a), whereas convergence means that sequences become more similar to each other. Convergent evolution could theoretically occur between either related (homologous; Figure 1b) or unrelated (analogous; Figure 2) sequences. If the sequences are homologous, then they must first diverge before they can later converge. Parallel means that the sequences retain the same degree of similarity through time, even though each lineage may be sustaining substitutions. Thus, proteins that evolved in parallel have neither diverged nor become more similar in overall sequence through time, but have followed parallel paths of evolution (Figure 1c). Parallel evolution only makes sense if speaking about homologous entities. When considering homologous sequences, parallelism and convergence are not radically different processes. Both will slow the divergence of the molecules in question, as indicated by the dashed lines in Figure 1a. As discussed below, significant sequence convergence in unrelated proteins (Figure 2) has yet to be demonstrated (Doolittle, 1994; Fitch, 2000).

Mechanistic For the purpose of sequence analysis, relatively clear-cut mechanistic definitions exist for the various types of homoplasy (Sneath and Sokal, 1973). These mechanisms are illustrated in Figure 3b using amino acid replacements for the examples. Parallel replacements are those that occurred identically and independently along two or more separate lineages. In the example shown here, an arginine (R) was replaced by a lysine (K) in two divergent molecules. Parallel events can occur at the same or different times along different lineages (Haas and Simpson, 1946; Sneath and Sokal, 1973). Convergent replacements are those that result in identical amino acids in the descendants, but which arose from different intermediates (Sneath and Sokal, 1973). This definition is in the spirit of requiring convergent events to arise from different genetic backgrounds. In the example provided in Figure 3b, the ancestral R was first replaced by a valine (V) and then by a K along one lineage, whereas along the other lineage the R was first replaced by a serine (S) and then by a K. Thus, the two descendant molecules independently acquired K at the same position in their amino acid sequences, but did so through different pathways. Convergent amino acid replacements in homologous proteins usually occur in positions that are relatively rapidly evolving, as they require multiple substitutions in the history of a given position.

ENCYCLOPEDIA OF LIFE SCIENCES & 2007, John Wiley & Sons, Ltd. www.els.net

3

Evolution: Convergent and Parallel Evolution

An evolutionary reversal occurs when, for example, an amino acid changes from a derived state back to a previous ancestral state. In the example here, one lineage replaced the ancestral R with a K, and then subsequently replaced the K with an R. Reversals can occur along more than one lineage, and can involve either divergent or parallel replacements at the intermediate steps. Divergent, convergent and parallel evolution of proteins – in the geometric sense of the terms – result from amino acid replacements that occur by the mechanisms discussed above. Thus, the mechanistic and geometric definitions are reliant upon each other. The mechanistic definition is applied to individual characters or sites within aligned sequences, whereas the geometric definition can be used to evaluate the evolution of individual characters or whole molecules.

Sequence Homoplasy – Adaptation or Chance? Inherent in many authors’ definition of homoplasy, and especially of convergence, is that it results from adaptation to similar environments (Gould, 2002). However, some authors contend that unless the homoplasy can be related to a change in function it is merely due to chance events. Chance homoplasy is particularly apparent in nucleic acids sequences, as the presence of only four bases severely limits the mutational options; thus, even high levels of homoplasy at the DNA sequence level may not be evidence of adaptive evolution. With 20 amino acids to work with, significant homoplasy at the protein sequence level is more likely to suggest adaptation. Homoplasy in protein sequences can be demonstrated objectively using the mechanistic and/or geometric definitions. How, then, may one decide if the homoplasy is ‘significant’? If given homoplastic replacements can be linked to specific functional changes, then this suggests that they may be biologically significant. Another way to demonstrate significance is if the number of homoplastic replacements seen between two sequences is statistically significantly greater than can be explained by chance, as has been found for the stomach lysozymes from artiodactyl ruminants and colobine monkeys (Stewart et al., 1987). Under such circumstances, the observed homoplasy can be deemed evolutionarily significant, even if adaptive explanations are not readily available for each of the homoplastic replacements.

Homoplasy in Protein Sequences Convergence, parallelism and reversals could all produce important changes in different proteins, and thus make them more similar in structure and/or function to one another than they are to the other proteins within their family. Therefore, from a functional standpoint, it does not matter through which mechanism the similarity arose. The 4

distinction between convergent and parallel evolution is, however, of theoretical interest because of what the two processes imply about the structure/function relationship of proteins. Convergent evolution implies that there may be only one way to solve a given problem (through the gain of a specific amino acid), regardless of the original sequence. Parallel evolution need not imply this, as the selection pressure might have been to lose a given amino acid, rather than to gain one. This distinction is especially pertinent if one considers the possibility of sequence convergence in evolutionarily unrelated molecules (Figure 2). Significant levels of sequence convergence in unrelated proteins would imply that there are limited strings of amino acid sequence that can satisfy a given structural or functional requirement. Many biochemists have searched for indisputable cases of significant sequence convergence in unrelated proteins, but none has been found (Doolittle, 1994). This is consistent with the observation that many primary structures, even highly divergent ones, can support the same or similar functions (Orenga et al., 1993). However, if significant levels of sequence convergence did occur between unrelated molecules, it would be difficult – but perhaps not impossible (Fitch, 2000) – to distinguish from true homology. Interestingly, several examples exist of seemingly unrelated proteins that nonetheless have very similar folds; this suggests either that these protein motifs are actually very distantly related, or that they convergently evolved the same tertiary structures using different sequences (Orengo et al., 1993). See also: Protein Families: Evolution; Protein Structure Often when biochemists and biologists speak of ‘convergence’, they are using the word loosely to refer to all forms of homoplasy. Indeed, when most cases of homoplasy between homologous sequences have been examined using the mechanistic or geometric criteria, they have shown to be cases of parallel evolution, and not convergence. In contrast, the stomach lysozymes from foregutfermenting cows and langur monkeys share a statistically significant number of homoplastic amino acid replacements, of both the parallel and convergent varieties (Stewart et al., 1987). Furthermore, the lysozyme from langur monkeys gained a small amount of overall sequence similarity (about 4%) with cow stomach lysozyme after the langur lineage diverged from the baboon lineage, whereas baboon lysozyme continued on a divergent path. Thus, langur and cow stomach lysozymes meet the geometric definition of convergent evolution, as outlined in Figure 1b. It should be noted, moreover, that these two lysozymes are clearly homologous proteins, and thus do not meet the definition of convergence that is outlined in Figure 2 (Stewart et al., 1987).

Functional Convergence in Proteins Convergence to a common function based upon different primary and tertiary structures does appear to have occurred during the evolutionary process. (This is yet

ENCYCLOPEDIA OF LIFE SCIENCES & 2007, John Wiley & Sons, Ltd. www.els.net

Evolution: Convergent and Parallel Evolution

another definition of convergent evolution that is commonly used in the literature. It is in keeping with the spirit of convergence being based on different underlying structures.) A textbook example of this type of convergent evolution involves the catalytic sites of the trypsin- and subtilisin-like serine proteases. The catalytic mechanisms of these two unrelated proteases use a charge relay system involving aspartate, histidine and serine side chains, which are in the same spatial arrangement in the two enzymes. However, these two protein families display no sequence similarity: the active site residues are at different positions along the peptide chains and the three-dimensional structures of the enzymes are completely different. Thus, they are unrelated proteins that have converged to a similar enzymatic mechanism using different primary and tertiary structures. Many other examples of functional or mechanistic convergence exist (Doolittle, 1994). Although functional convergence in unrelated proteins may be somewhat common, many early claims of this phenomenon were not verified by tertiary structural analysis. A pertinent example of this comes from the lysozyme literature. Chicken- and goose-type lysozymes were thought to be a case of convergence of unrelated proteins to a common enzymatic function because these proteins have no significant sequence similarity. However, comparative studies of the tertiary structures of the chicken-, goose- and phage-type lysozymes showed that they share large regions of structural correspondence and have spatial similarities in their active sites. This has been taken to suggest that all three families of lysozymes diverged from a common evolutionary precursor, but that obvious sequence similarity has been lost over time (Weaver et al., 1985). We have already seen that there are numerous ways in which proteins can evolve and adapt. As the emerging fields of comparative genomics and proteomics continue to expand our understanding of protein structure, function and evolution, we can expect that new types of ‘convergent evolution’ will be described, such as that of protein surfaces (de Rinaldis et al., 1998). Thus, it will become ever more important for authors to define precisely how a given term is being used in a given context when writing about independent evolution of structural or functional characteristics of proteins and other biological entities. See also: Evolution: Views of; Phylogenetics

References Doolittle RF (1994) Convergent evolution: the need to be explicit. Trends in Biochemical Sciences 19: 15–18. Fitch WM (2000) Homology: a personal view on some of the problems. Trends in Genetics 16: 227–231. Gould SJ (2002) The Structure of Evolutionary Theory. Cambridge, MA: Belknap Press of Harvard University Press. Haas O and Simpson GG (1946) Analysis of some phylogenetic terms, with attempts at redefinition. Proceedings of the American Philosophical Society 90: 319–349.

Lankester ER (1870) On the use of the term homology in modern zoology, and the distinction between homogenetic and homoplastic agreements. Annals and Magazine of Natural History 6: 34–43. Orengo CA, Flores TP, Jones DT, Taylor WR and Thornton JM (1993) Recurring structural motifs in proteins with different functions. Current Biology 3: 131–139. de Rinaldis M, Ausiello G, Cesareni G and Helmer-Citterich M (1998) Three-dimensional profiles: a new tool to identify protein surface similarities. Journal of Molecular Biology 284: 1211–1221. Sneath PHA and Sokal RR (1973) Numerical Taxonomy: The Principles and Practice of Numerical Classification. San Francisco, CA: WH Freeman. Stewart C-B, Schilling JW and Wilson AC (1987) Adaptive evolution in the stomach lysozymes of foregut fermenters. Nature 330: 401–404. Weaver LH, Gru¨tter MG, Remington SJ et al. (1985) Comparison of goose-type, chicken-type, and phage-type lysozymes illustrates the changes that occur in both amino acid sequence and three-dimensional structure during evolution. Journal of Molecular Evolution 21: 97–111.

Further Reading Butler AB and Saidel WM (2000) Defining sameness: historical, biological, and generative homology. BioEssays 22: 846–853. Fernald RD (2006) Casting a genetic light on the evolution of eyes. Science 313: 1914–1918. Hildebrand M (1962) The terminology of structural and functional similarities. Systematic Zoology 11: 186–187. Irwin DM (1996) Molecular evolution of ruminant lysozymes. EXS 75: 347–361. Messier W and Stewart CB (1997) Episodic adaptive evolution of primate lysozymes. Nature 385: 151–154. Patterson C (1988) Homology in classical and molecular biology. Molecular Biology and Evolution 5: 603–625. Peacock D and Boulter D (1975) Use of amino acid sequence data in phylogeny and evaluation of methods using computer simulation. Journal of Molecular Biology 95: 513–527. Prager EM (1996) Adaptive evolution of lysozyme: changes in amino acid sequence, regulation of expression and gene number. EXS 75: 323–345. Sanderson MJ and Hufford L (eds) (1996) Homoplasy: The Recurrence of Similarity in Evolution. San Diego, CA: Academic Press. Wake DB (1991) Homoplasy: the result of natural selection, or evidence of design limitations? American Naturalist 138: 543–567. Wood TE, Burke JM and Rieseberg LH (2005) Parallel genotypic adaptation: when evolution repeats itself. Genetica 123: 157– 170. Zuckerkandl E and Pauling L (1965) Evolutionary divergence and convergence in proteins. In: Bryson V and Vogel HJ (eds) Evolving Genes and Proteins, pp. 97–166. New York: Academic Press.

ENCYCLOPEDIA OF LIFE SCIENCES & 2007, John Wiley & Sons, Ltd. www.els.net

5