Convergence and parallelism: is a new life ahead of old concepts?

Laure Desutter-Grandcolas*, Frederic Legendre, Philippe Grandcolas, Tony Robillardand Jerome Murienne

Museum National d’Histoire Naturelle, Departement Systematique et Evolution, UMR 5202 CNRS, CP n�50 (Entomologie), 45 rue Buffon, F-75231

Paris Cedex 05, France

Accepted 6 October 2004

Abstract

In comparative biology, character observations initially separate similar and dissimilar characters. Only similar characters areconsidered for phylogeny reconstruction; their homology is attested in a two-step process, firstly a priori of phylogeny reconstructionby accurate similarity statements, and secondly a posteriori of phylogeny analysis by congruence with other characters. Any patternof non-homology is then a homoplasy, commonly, but vaguely, associated with ‘‘convergence’’. In this logical scheme, there is noway to analyze characters which look similar, but cannot meet usual criteria for homology statements, i.e., false similarity detecteda priori of phylogenetic analysis, even though such characters may represent evolutionarily significant patterns of charactertransformations. Because phylogenies are not only patterns of taxa relationships but also references for evolutionary studies, wepropose to redefine the traditional concepts of parallelism and convergence to associate patterns of non-homology with explicittheoretical contexts: homoplasy is restricted to non-similarity detected a posteriori of phylogeny analysis and related to parallelism;non-similarity detected a priori of phylogenetic analysis and necessarily described by different characters would then correspond to aconvergence event s. str. We propose to characterize these characters as heterologous (heterology). Heterology and homoplasycorrespond to different non-similarity patterns and processes; they are also associated with different patterns of taxa relationships:homoplasy can occur only in non-sister group taxa; no such limit exists for heterology. The usefulness of these terms and concepts isillustrated with patterns of acoustic evolution in ensiferan insects.� The Willi Hennig Society 2005.

�Ce que l’on concoit bien s’enonce clairement, et les mots pour

le dire arrivent aisement.� (Boileau, 1674, L’art Poetique)

The study of similarity is a critical and recurrent themein phylogenetic analysis and evolutionary studies (Hen-nig, 1966). It has developed and been discussed in severalmajor contexts, such as the definition of character andcharacter states (Hawkins et al., 1997; Scotland andPennington, 2000), and the appraisal of similarity due tocommon ancestry (Hall, 1994). In this way, the study ofsimilarity has brought about the definitions and discus-sions of many fundamental issues in the science ofevolution (Rieppel and Kearney, 2002; Kluge, 2003).

In the present paper we will examine more partic-ularly the patterns of similarity associated with the

processes of parallelism and convergence. These termsare old ones and one could think that all has been saidabout them or that they are outdated. However, even abrief survey of the literature shows that this is not thecase. With the development of the phylogenetic methodand the general use of a phylogenetic reference toexamine problems in evolutionary biology, more andmore case studies document characters in which evolu-tion is incongruent with taxa relationships. Unfortu-nately, contradictory definitions are currently given forparallelism and convergence and no explicit series ofcharacter transformation are clearly associated witheither of these processes, while in the mean time a priorihypotheses of evolutionary processes are commonlyassociated with patterns of character similarity. Asemphasized by Wiens et al. (2003) among others,many authors equate convergence and parallelism, and*Corresponding author.

E-mail address: desutter@mnhn.fr

� The Willi Hennig Society 2005

Cladistics

www.blackwell-synergy.com

Cladistics 21 (2005) 51–61

associate it with a case of homoplasy. This results inmuch confusion, which is necessarily detrimental tostudies in evolutionary biology.

The problems are then first to decide whether suchdifferent things as parallel and convergent evolutionsactually exist, and second, if they do, to characterizethem by explicit patterns of character transformation: inthe present state of phylogenetic methodology, this isthe only way to test evolutionary hypotheses of charac-ter evolution (Eldredge and Cracraft, 1980; Coddington,1988, 1990; Carpenter, 1989; Grandcolas et al., 1994,1997; Desutter-Grandcolas, 1997; Grandcolas andD’Haese, 2003). These are the aims of the presentpaper. We will successively make a brief review of thedefinitions given by different authors for parallelism andconvergence, and propose a rationale to separate bothprocesses, based on explicit patterns of character evo-lution. In a second step, we will show their usefulness inre-analyzing briefly the evolution of tegminal stridula-tion in ensiferan insects (Desutter-Grandcolas, 2003)and scrutinizing the recent literature.

Parallelism and convergence: a brief review

Parallelism and convergence have been amply dis-cussed in the phylogenetic and evolutionary literature,and it is not possible to analyze all these papers here. Wehave consequently chosen a few authors and checked fortheir agreement concerning the problems listed above.We have thus reviewed, in alphabetical order, Brooks(1996), Brooks and McLennan (2002), Eldredge andCracraft (1980), Fitch (2000), Hennig (1966), Kitchinget al. (1998), Patterson (1982, 1988), Simpson (1961)and Wiley (1981). We will not discuss here the problemof character delimitation in phylogenetic analysis, northe criteria for determining putative homologies, asthese have been discussed at length elsewhere (e.g.,Rieppel and Kearney, 2002).

Parallelism and convergence, as well as divergence,have their origin in geometry (Haas and Simpson, 1946),and were used to describe the general evolutionarytendencies of different lineages. This conception mayhave been adequate in the narrative context of earlyevolutionary biology, but it is now ineffective in the lightof modern phylogenetic methodology (contra Fitch,2000). In more recent times, there have been two verydifferent ways of considering parallelism and conver-gence. In the first, parallelism and convergence aredefined according to the degree of relationships of thetaxa involved, and the pattern of character transforma-tion documented on the tree, especially between apicaltaxa and their immediate ancestors. This approachcombines hypotheses about similarity and about phylo-geny: we will call it the relatedness approach. Among theauthors listed above, this approach is supported by

Brooks (1996), Brooks and McLennan (2002), Eldredgeand Cracraft (1980), Simpson (1961), Wiley (1981), andin some measure Fitch (2000).

The second approach to parallelism and convergencegreatly differs from the one described above: it relies onPatterson’s (1982, 1988) analysis of similarity (see alsoKitching et al., 1998) and considers hypotheses of taxarelationships only in the final step of the procedure. Wewill refer to it as the character similarity approach.

Hennig (1966) must be considered separately here. Heactually recognizes the different processes correspond-ing to convergence and parallelism, adding the conceptof homoiology to designate ‘‘a form of homology inwhich the particular character has been acquiredindependently by close relatives’’ (Plate, 1928, in Hen-nig, 1966, p. 117). His definitions rely on the transfor-mation series of the characters and character states, andon the relationships of the taxa: ‘‘In convergence, twoforms with similarities in directly adaptive structures (aswolf and marsupial ‘‘wolf’’) have come from radicallydifferent ancestors with basically different patterns oforganization; in true parallelism, both habitus andheritage are similar; the ancestral types were closelyrelated, and evolutionary progress, stage by stage, hasbeen closely comparable in the two or more lineagesconcerned’’ (Romer, 1949, in Hennig, 1966, p. 117). Inother words, ‘‘One speaks of parallelism in the narrowtrue sense if different transformation series that arosefrom a common primary condition at first diverge, butlater progress in parallel or finally even convergently.’’(Hennig, 1966, p. 119). Nevertheless, according toHennig (1966), convergence and parallelism are butloosely defined and of no great significance for phylo-genetic systematics.

The relatedness approach

Parallelism (Fig. 1) is here defined as the occurrenceof a similar character in closely related species or inparaphyletic groups. According to Eldredge andCracraft (1980) for example (Fig. 1A), parallel evolu-tion corresponds to the independent occurrence of thesame character state (here a¢) in closely related species(A and B), whose immediate ancestor (X) has retainedthe ancestral character state (a). In a very similar way,Brooks (1996) defined parallelism as the occurrence ofthe same character in paraphyletic taxa, here E and F(Fig. 1B). Wiley (1981, p. 10) insisted on the ancestralstate of the character and defined ‘‘parallel develop-ment … as the independent development of similarcharacters from the same plesiomorphic character.’’(see also Fitch, 2000). Wiley distinguished two simplesituations in which parallelism may occur: a firstsituation (Fig. 1C) corresponding to Brooks’s (1996)definition; and a second situation (Fig. 1D), in which

52 L. Desutter-Grandcolas et al. / Cladistics 21 (2005) 51–61

the derived state occurs independently in two sub-clades.

As defined here, parallelism is associated with acertain amount of common ancestry, because commonancestry is assumed to imply common developmentalcharacteristics or common reaction to environmentalpressures. This is clearly stated by Eldredge andCracraft (1980, p. 72): ‘‘In parallelism, because closelyrelated species have similar genetic backgrounds andtherefore similar developmental potentials, they arelikely to respond to similar selection forces in a similarmanner; hence parallelism is to be expected in closerelatives living under similar environmental conditions.’’Similarly, Simpson (1961, p. 78) considered that ‘‘Par-allelism is the development of similar characters sepa-rately in two or more lineages of common ancestry andon the basis of, or channelled by, characteristics of thatancestry.’’

By contrast, but in a complementary way, a conver-gence (Fig. 2) is most often defined as the occurrence ofone character state in distantly related species or inpolyphyletic groups. Eldredge and Cracraft (1980, p. 70)stated for example ‘‘… The traditionally accepteddefinition of convergence refers to similarities independ-ently acquired and not derived from a common ances-tor.’’: state (a¢) appeared independently twice in taxa Aand D, whose ancestors retained the ancestral characterstate (a) (Fig. 2A). For Brooks (1996), a convergence isthe occurrence of one derived state in two distantlyrelated taxa, belonging to subclades of a polyphyleticgroup (Fig. 2B). As for parallelism, Wiley (1981) insis-ted on the ancestral states of characters and defined aconvergence (Fig. 2C) as ‘‘the development of similarcharacters from different preexisting characters.’’ (ibid.,

p. 10). Fitch (2000) uses the same definition, as will becommented below. Finally, Simpson (1961, p. 78)adopted a multifactor definition, which took intoaccount the adaptive nature of the involved characters:‘‘Convergence is the development of similar charactersseparately in two or more lineages without a commonancestry pertinent to the similarity but involving adap-tation to similar ecological status.’’

What are the consequences of defining convergenceand parallelism on taxa relationship and patterns ofcharacter transformation on existing topologies?

First, in this view parallelism and convergence can berecognized only a posteriori of phylogenetic reconstruc-tion, and not before tree building. Second, all the authorsconsider that parallelism and convergence are difficult toseparate and some authors even advise not to distinguishthem and ignore one term: ‘‘In conclusion, we recom-mend that the concept of parallelism be omitted fromsystematic studies. We suggest that the term convergencebe applied to all cases of non-homologous charactersimilarities, identified through character conflicts arisingfrom normal analysis of the patterns of similarity.’’(Eldredge and Cracraft, 1980, p. 74).

In the context described above, parallelism andconvergence are actually badly defined, because thetwo criteria used to define both processes are unclear.First, the distribution of character states used to defineparallelism and convergence are ambiguous, and they donot correspond to parsimonious optimization proce-dures. Obviously it is usually not possible to knowcharacter states at ancestral nodes, only to parsimoni-ously hypothesize them by optimization. Thus the figureproposed by Eldredge and Cracraft (1980) to defineparallelism could be modified to fit a hypothesis of

A CB D E F G

1(0)

1(1)1(1)

Aa' CaBa' Da

Xa

Ya

ZaA B

C D

Fig. 1. Definitions of parallelism according to: (A) Eldredge and Cracraft (1980); (B) Brooks (1996); (C, D) Wiley (1981). For explanations, seetext.

53L. Desutter-Grandcolas et al. / Cladistics 21 (2005) 51–61

Aa' CaBa Da'

Xa

Ya

Za

A CB D E F G

2(0)

2(1) 2(1)

A B

C

Fig. 2. Definitions of convergence according to: (A) Eldredge and Cracraft (1980); (B) Brooks (1996); (C) Wiley (1981). For explanations, seetext.

A

B

C

Fig. 3. Alternative scenarios for previous definitions of parallelism and convergence. (A) Eldredge and Cracraft (1980) parallelism; (B)Brooks’s (1996) parallelism; (C) Wiley’s (1981) convergence. For explanations, see text.

54 L. Desutter-Grandcolas et al. / Cladistics 21 (2005) 51–61

common ancestry (Fig. 3A), as acknowledged byEldredge and Cracraft themselves (op. cit., p. 72).Brooks’s definition fits a hypothesis of common ancestrywith a subsequent reversal (Fig. 3B). In much the sameway Wiley’s figure for a convergence may be modified toget a more parsimonious character transformationseries, with one synapomorphic change followed by anautapomorphic change (Fig. 3C). Second, criteria todecide if species are closely related or not, or thedefinition of paraphyly and polyphyly may depend onthe size and composition of the ingroup studied. Onecould remark here that the distribution of characterstates acknowledged as parallelism by Wiley (Fig. 1D)fits Brooks’ s definition of a convergence (Fig. 2B). Asacknowledged by Simpson (1961, p. 103), always in thelight of the two criteria mentioned, ‘‘Parallelism may bedifficult or practically impossible to distinguish fromhomology on one hand and convergence on the other.’’

In this general frame, how is similarity betweencharacters and character states categorized? All authorsagree to refer similarity due to common ancestry ashomology. Among many authors, Wiley (1981, p. 121)stated for example: ‘‘A character of two or more taxa ishomologous if this character is found in the mostcommon ancestor of these taxa, or, two characters …are homologous if one directly … derived from theother(s).’’ In much the same way, Simpson (1961, p. 78)considered that: ‘‘Homology is resemblance due toinheritance from a common ancestry.’’ On the reverse,any similarity not due to common ancestry is called ahomoplasy, with two alternative options: either conver-gence and parallelism were considered similar (Wiley,1981), or they were separated (Simpson, 1961).

This first general approach of parallelism and con-vergence is summarized here in Fig. 4: the phylogeny isreconstructed from the observations of characters, andaccording to the patterns of character state distributionon the resultant topology, and on the degree of taxarelationships, hypotheses of homology (i.e., common

ancestry) or homoplasy (i.e., parallelism ⁄convergence),are proposed.

Clearly, this approach finds its logic in the character-ization of natural groups. As stated by Wiley (1981,p. 121): ‘‘The important point is not whether the terms[convergence, parallelism] can be readily distinguishedby some objective criterion. Rather, the importance liesin recognizing that both types of similarities are non-homologies and thus both are irrelevant to justifyingnatural taxa.’’ As a consequence, any similarity notrelated to common ancestry is rejected as non-inform-ative, or even a source of error. Today, this idea isactually very common (Coddington, 1994; Kluge, 2001);it could even have been reinforced by, and havecontributed to, the equally common practice of associ-ating non-homology with adaptation (Pagel, 1994),which is nothing more than basic adaptationism. Infact, there are two problems here. First, restrictingcommon ancestry to homology, and adaptation tohomoplasy could be viewed as an oversimplification:phylogenetic information exists also in homoplasticcharacters (Kallersjo et al., 1999; Wenzel and Siddall,1999), and adaptation is not the only way to explainsimilarity of non-homologous characters (Gould andLewontin, 1979; Desutter-Grandcolas et al., 2003;Grandcolas and D’Haese, 2003). Second, does thisgeneralization apply to any ‘‘non-homologous’’character?

The character similarity approach

In fact, Patterson (1982, 1988) insisted on the obser-vation of similarity, which he considered the onlydirectly available information in phylogenetic studies,and which can be submitted to explicit criteria such asthose proposed by Remane (1952). In this context,Patterson defined his three well-known tests of similar-ity, conjunction and congruence (Table 1), which we will

Fig. 4. Diagram summarizing the tree similarity approach. For explanations, see text.

55L. Desutter-Grandcolas et al. / Cladistics 21 (2005) 51–61

not detail here. Patterson’s approach can be summarizedas shown in Fig. 5. Character observations are firstsubmitted to similarity and conjunction tests. If theyfail, they correspond to a convergence. If they pass thetests, they are used to define hypotheses of homology,which are then submitted to the congruence test andseparated into homologies and parallelism. Convergenceand parallelism are grouped together as non-homology.In this scheme, a homology always corresponds to asimilarity due to common ancestry, as for the authorspreviously referred to; a homoplasy corresponds to anon-homology, with two different situations associatedwith parallelism and convergence.

Whatever the difficulties that may exist in applyingPatterson’s tests, especially the similarity and conjunc-tion tests (Hawkins, 2000), Patterson’s approach isremarkable in the present context for several reasons.First, it gives character analysis the priority in thelogical steps associated with phylogenetic analysis. Thispoint of view could be considered as evidence, but thispriority clearly tends to be forgotten today, especiallywith the preponderance of phylogenies reconstructedwith only molecular data. Second, Patterson recognizesand clearly defines the logical situations corresponding

to parallelism and convergence, using explicit criteria:‘‘In my understanding, the distinction between homol-ogy, parallelism and convergence is simple. As discussedin the section on testing homologies, there are threeways of rejecting a potential homology: it fails the test ofsimilarity, it fails the test of congruence with otherhomologies, or it fails the test of conjunction, when twosupposed homologs are found in the same organism.Considering only the similarity and congruence tests,convergences are those similarities which fail both tests,and parallelisms are those which pass the similarity testbut fail the congruence test…’’ (Patterson, 1982, p. 46).Said in another way, ‘‘… parallelisms are rejected ashomologies solely because they fail to characterizemonophyletic groups, whereas convergences fail tocharacterize groups, and are not closely similar, or�not really the same�.’’ Third, Patterson clearly separatesthe homology hypotheses proposed before phylogenybuilding according to character similarity, from thehomology hypotheses which are supported by phylo-geny reconstruction. This has been acknowledged andmore formally analyzed by subsequent workers (dePinna, 1991; Brower and Schawaroch, 1996). Whateverthe terminology adopted, it is the separation of what isachieved before and after phylogeny reconstruction, thatremains significant and powerful.

Two problems still remain unsolved.First, even if parallelism and convergence are theor-

etically clearly separated by the tests of similarity, thepatterns of character change which are associated withthem remain uncertain. Patterson did not give precisedefinitions of parallelism and convergence in terms ofcharacter transformation series. In these conditions,hypotheses of parallelism and convergence cannot betested.

Table 1Patterson’s definitions of homology, parallelism and convergence,according to the tests of similarity, conjunction and congruence (afterPatterson, 1982)

RelationCongruencetest

Similaritytest

Conjunctiontest

Homology pass pass passParallelism fail pass passConvergence fail fail fail

Fig. 5. Diagram summarizing the character similarity approach. For explanations, see text.

56 L. Desutter-Grandcolas et al. / Cladistics 21 (2005) 51–61

Second, Patterson always defined a homoplasy as anon-homology, even though he postulated that ahomoplasy could potentially correspond to two verydifferent evolutionary contexts from the point of view ofthe characters themselves, which are, paradoxically,Patterson’s priority. Anyway, it is clear that the logic ofcharacter analysis and similarity tests implies that thesimilarity associated with parallelism is not identicalfrom the similarity related to convergence. This finallymeans that the concept of non-homology, that is thesimilarity not associated with a pattern of commonancestry, cannot be reduced to homoplasy alone.

Parallelism and convergence: new definitions for a new life

ahead

Under these conditions, we propose to modifyPatterson’s approach to distinguish both types of non-homology and define explicit patterns of characterchange (Fig. 6): as in Patterson’s approach, observa-tions are the basic step of the procedure. If they pass thesimilarity tests, they are used to define hypotheses ofhomology, which are submitted to the congruence test.The congruence test separates homology due to com-mon ancestry, and homoplasy s. str. related to parallel-ism. Contrary to Patterson’s approach, however, wepropose to clearly identify those cases where observa-tions fail the similarity tests. This situation clearlymeans, from a cladistic point of view, that the observedstructures correspond to extremely similar features, thatmust be described by different characters. We propose torestrict the definition of convergence to this precise

pattern of similarity: characters that have evolved to beso similar that they could not be recognized as non-identical by immediate observation, but cannot meetusual criteria of homology statements, i.e., charactersthat are similar but not the same. In the past,such characters would have been designated as analog-ous (analogy), but this concept intrinsically involveshypotheses about character function (e.g., Haas andSimpson, 1946) or, more recently, character selectivevalue (Simpson, 1961; Pagel, 1994), so that its use in anovel prospect looks difficult. These characters could beunambiguously identified as heterologous (heterology):this term was first used by Cope (1868) to designategenera (or higher level taxa) that had evolved similarlyin different lineages; in molecular biology it alsodesignates a molecular probe used to study an homol-ogous sequence in another species (Moritz and Hillis,1996). In an evolutionary context, this term is the exactopposite to homology, allowing a clear distinctionbetween the two situations; in much the same way, itcannot be confounded with homoplasy.

The distinction homology ⁄heterology proposed herecan apply to any character type, not only structural ones(Lauder, 1994), as soon as their definitions are based onprimary homology statements. Under these conditions,parallelism and convergence are distinct and they can betested in a historical context. Moreover parallelism, justas common ancestry, is documented only after phylo-geny reconstruction, while convergence is recognizedbefore phylogeny building.

One objection could be immediately raised to ourscheme: if the ‘‘very similar, but anyway different’’structures are described by different characters, why

Fig. 6. Diagram summarizing the approach recommended here to analyze character similarity and derive hypotheses of homology, homoplasy andheterology. For explanations, see text.

57L. Desutter-Grandcolas et al. / Cladistics 21 (2005) 51–61

bother about them anymore, why getting into trouble?This argument could hold in a strict perspective ofphylogeny making, the aim of which is to document taxarelationships and define monophyletic groups. But, phy-logeny is also the basic reference in evolutionary biology: asa statement of the chronicle of evolutionary events(O’Hara, 1988), a phylogenetic hypothesis documentscharacter transformation series, and these patterns can beused in turn to test hypotheses of evolutionary processes(Eldredge and Cracraft, 1980; Carpenter, 1989; Brooksand McLennan, 1991, 2002; Grandcolas et al., 1994,1997; Desutter-Grandcolas, 1997; Grandcolas andD’Haese, 2003). The similarity of non-homologous char-acters is potentially very interesting from an evolutionarypoint of view, because it could result from developmentalpathways, including heterochrony (Abouheif et al., 1997;Brooks andMcLennan, 2002), gene expression (Hu et al.,2003; Wittkopp et al., 2003), biophysical constraints(Eisthen and Nishikawa, 2002), reaction to environmen-tal constraints including adaptation (Grandcolas andD’Haese, 2003), recurring occurrence of intraspecific orinterspecific interplay (sexual selection, competition,mutualism, mimicry: Ricklefs and Miller, 1999), and soon, with diverse interactions between different levels ofbiological organization (Abouheif et al., 1997; Albertsonet al., 2003). This implies that all patterns of characterchange are explicitly identified and unambiguouslynamed, in order to be associated with, and allow the testsof, hypotheses of evolutionary processes.

Fitch (2000) acknowledged the necessity of definingdifferent patterns for non-homologous character similar-ity and actually distinguished the processes of parallelismand convergence, associating them with patterns ofhomology and analogy, respectively. This greatly remindsour own definitions. However, Fitch’s definitions arebased on patterns of character changes, much as Wiley(1981) did, and on the ‘‘common sense meanings’’ of thewords. Such transformational relations can be explicitlydocumented only in reference to a topology and dependon optimization procedures (see Fig. 3). The alternative isto use ad hoc hypotheses of character change prior to treebuilding, which may bias the phylogenetic analysis anddecrease the logical independence of phylogenetic tests ofevolutionary scenarios (Farris, 1983; Deleporte, 1993;Grandcolas et al., 1994). Comparing Fitch’s approachwith ours thus makes clear, if still necessary, that anunambiguous phylogenetic terminology taking intoaccount all patterns of character change and allowingthe test of all evolutionary hypotheses related to charactersimilarity is deeply needed.

Examples

To demonstrate the usefulness of this scheme forstudies in evolutionary biology, let us apply it to the

evolution of the tegminal stridulum in ensiferan insects,a particularly significant example in terms of evolutionof structures. Within Ensifera, true crickets, molecrickets and tettigonids sing with a stridulum locatedon their forewings, by rubbing the pegs of a stridulatoryfile located on one forewing with a thick and hardenedpart of the other forewing. Several hypotheses have beenproposed to explain this distribution, with or without aphylogenetic reference: they imply either a separateoccurrence of stridulation in tettigonids in one hand,and crickets and mole crickets on the other, or a uniqueoccurrence of stridulation in the common ancestorof the three groups (Ander, 1939; Alexander, 1962;Gwynne, 1995). A phylogenetic analysis of ensiferanvenation, using vein homologies assessed after articula-tory sclerites, reveals that the stridulatory file is locatedon the first anal vein in both true crickets and tettigo-nids, while it occurs on a branch of the posterior cubitusin mole crickets (Desutter-Grandcolas, 2003 and Fig. 7).Thus although true crickets and mole crickets are sistergroups, their stridulums, which are quite similar, are not

A

Grylloidea (true crickets)

Gryllotalpoidea

Rhaphidophoridae

Schizodactylidae

Gryllacrididae

Stenopelmatidae

Anostostomatidae

Tettigonidae

B

Fig. 7. Acoustic evolution in ensiferan insects: (A) phylogeny ofEnsifera (after Desutter-Grandcolas, 2003); (B) venation schemes of:(a) a cricket, and (b) a mole cricket. Symbols: black dotted line, clavalfold; gray, subcosta; orange, radius; yellow, anterior media; green,posterior media; red, anterior cubitus; blue, posterior cubitus; purple,anals; reduced costa not represented; stridulatory file located onthickened vein and ‘‘harp’’ colored.

58 L. Desutter-Grandcolas et al. / Cladistics 21 (2005) 51–61

the same. Within Ensifera, there are consequently twovery different situations as far as the stridulatory file isconcerned: the homoplastic occurrence of one and thesame character in two distantly related taxa (parallel-ism), and the occurrence of two different, yet verysimilar, characters, in sister taxa (convergence).

Parallelism and convergence patterns are actuallyamply documented in recent phylogenetic literature(Armbruster, 1993; Sinha and Kellogg, 1996; Swensen,1996; Bruneau, 1997; Henry et al., 1999; Parra-Olea andWake, 2001; Teeling et al., 2002; Hu et al., 2003;Wittkopp et al., 2003; Guilbert, 2004, etc…). Unfortu-nately, the methodological options of the authorsregarding character coding often impede further analysisof the series of character transformation and to under-stand how the observed similarity occurred.

In evolutionary biology, the traits of interest oftenconsist in complex structures or behavioral syndromes.It has been clearly shown that the evolution of thesetraits can be analyzed only if they are discretized intoindividual, homology-based characters because this isnecessary to optimization procedures; moreover, thediscretization of the traits of interest is necessary even ifthe resultant characters do not participate in phylogenybuilding (Luckow and Bruneau, 1997; Grandcolas et al.,2001; Desutter-Grandcolas and Robillard, 2003). Thispoint is particularly important for convergence studies,which are directly concerned with hypotheses about theancestral conditions and diverse evolutionary pathwaysthat have resulted in similar, yet not the same charac-ters. The reverse procedure, i.e., mapping a complextrait onto a phylogeny to reveal a convergence event, ismethodologically very limited, because it impedes fur-ther analytical approaches. This procedure is perhaps ameans of quickly characterizing the evolution of thestudied trait, but its limits in term of explanatory powershould clearly be denounced. Studying the evolution ofheterospory, Bateman (1996, p. 114) concluded: ‘‘As haslong been recognized, morphological parallelisms (inclu-ding parallel reversals) must be regarded as errors ofa priori homology assessment (e.g., Patterson, 1988).Thus, there may be 10 origins of heterospory, butfurther scrutiny will identify differences between modesof heterospory in different lineages.’’ In fact, the trait ofinterest (heterospory) should have been discretized intoas many characters as necessary to describe all itscharacteristics and their variation, and all these charac-ters should have been optimized separately. The errorhere resides more probably in the vague definition of thestudied character than in the obtained pattern.

Conclusions

It may seem odd, at the beginning of the 21st century,to consider so old evolutionary concepts as those

examined here, especially if one considers the progressin the phylogenetic methodology and in analysis possi-bilities achieved in the last 20 years. However, thisprogress has been accompanied by a change in focusamong phylogeneticists. In the past, phylogenies werebuilt mostly on morphological data sets and problemsrelated to similarity were essential in the phylogeneticlandscape. Now that molecular data are relatively easyto obtain and no longer costly, this landscape ischanging: homology is not analyzed in the same wayas it was for morphological or behavioral features.However, most features significant from an evolutionarypoint of view, i.e., features that so many authors try toanalyze in a historical context, relate to morphology,physiology or behavioral ecology (Wheeler, 2004). Eventhough these features are often not put in the datamatrix for phylogeny building, they must be correctlydescribed according to homology assessment (Grandco-las et al., 1997, 2001). So a clear understanding of thedifferent types of character similarity are necessary, evenmore now.

This problem is not only a terminological one,because confusion in the words most often showsconfusion in the ideas (Boileau, 1674). The tests ofevolutionary hypotheses require that each hypothesis isassociated with an explicit pattern of character change(Eldredge and Cracraft, 1980; Carpenter, 1989; Brooksand McLennan, 1991, 2002; Grandcolas et al., 1997),and that no ad hoc explanation about evolutionaryprocesses is associated with them. This in turn makes itnecessary to describe theoretical patterns of characterchange in the first place. In addition, these definitionsshould not depend on the centers of interest of theauthors, for the sake of transparency.

Finally, similarity not due to common ancestry hasbeen up to now thrown away as ‘‘someone not to invite totea’’, to paraphrase JohnWenzel (1992, in Nelson, 1994).This appears however, more and more as an oversimplis-tic way of thinking (Desutter-Grandcolas et al., 2003),and defining natural groups based on synapomorphiesshould not also preclude the consideration of the charac-ters which have been shown to be non-homologous as aresult of the phylogenetic analysis, in an evolutionarycontext. Talking about the development of the so-calledcomparative biology, Nelson (1994, p. 116) stated: ‘‘Theidea here is that biological phenomena are explicableeither as the effects of phylogeny (homologies, synapo-morphies) or as the effects of adaptation (homoplasies).’’That evolution is more complex than that, is today anevidence that the polarization homology ⁄non-homologydoes not allow us to grasp correctly. We are nowcompelled to modify our theoretical framework to takeinto account all types of non-homology and considerproperly their meaning in term of evolutionary modali-ties. This alone will give phylogeny and phylogeneticstudies their real value in the biology of evolution.

59L. Desutter-Grandcolas et al. / Cladistics 21 (2005) 51–61

Acknowledgments

This work was presented at the XXIIIth meeting ofthe Willi Hennig Society (Paris, 18–23 July 2004). Wethank J. Carpenter for the opportunity to publish thiscontribution.

References

Abouheif, E., Akam, M., Dickinson, W.J., Holland, P.W.H., Meyer,A., Patel, N.H., Raff, R.A., Roth, V.L., Wray, G.A., 1997.Homology and developmental genes. Trends Genet. 13, 432–433.

Albertson, C.R., Streelman, J.T., Kocher, 2003. Directional selectionhas shaped the oral jaws of Lake Malawi cichlid fishes. Proc. NatlAcad. Sci. USA, 100, 5252–5257.

Alexander, R.D., 1962. Evolutionary change in cricket acousticalcommunication. Evolution, 16, 443–467.

Ander, K., 1939. Vergleichend anatomische und phylogenetischeStudien uber die Ensifera (Saltatoria). Berlingska Boktryckeriet,Lund.

Armbruster, W.S., 1993. Evolution of plant pollination systems:hypotheses and tests with the neotropical vine Dalechampsia.Evolution, 47, 1480–1505.

Bateman, R.M., 1996. Nonfloral homoplasy and evolutionary scenar-ios in living and fossil land plants. In: Sanderson, M.J., Hufford, L.(Eds.), Homoplasy: the Reccurrence of Similarity in Evolution.Academic Press, San Diego, pp. 91–130.

Boileau, N., 1674. Oeuvres completes. Societe des Belles Lettres(Bude). 1966, Paris.

Brooks, D.R., 1996. Explanations of homoplasy at different levels ofbiological organization. In: Sanderson, M.J., Hufford, L. (Eds.),Homoplasy: the Reccurrence of Similarity in Evolution. AcademicPress, San Diego, pp. 3–36.

Brooks, D.R., McLennan, D.A., 1991. Phylogeny, Ecology andBehavior. University of Chicago Press, Chicago.

Brooks, D.R., McLennan, D.A., 2002. The Nature of Biodiversity. AnEvolutionary Voyage of Discovery. University of Chicago Press,Chicago.

Brower, A.V.Z., Schawaroch, V., 1996. Three statements of homologyassessment. Cladistics, 12, 265–272.

Bruneau, A., 1997. Evolution and homology of bird pollin-ation syndromes in Erythrina (Leguminosae). Am. J. Bot. 84, 54–71.

Carpenter, J.M., 1989. Testing scenarios: wasp social behavior.Cladistics, 5, 131–144.

Coddington, J.A., 1994. The roles of homology and convergence instudies of adaptation. In: Eggleton, P., Vane-Wright, R. (Eds.),Phylogenetics and Ecology. The Linnean Society of London,London, pp. 53–78.

Coddington, J.A., 1988. Cladistic tests of adaptational hypotheses.Cladistics, 4, 3–22.

Coddington, J.A., 1990. Bridges between evolutionary pattern andprocess. Cladistics, 6, 379–386.

Cope, E.D., 1868. On the origin of genera. Proc. Acad. Nat. Sci.Philad. 20, 242–300.

Deleporte, P., 1993. Characters, attributes and tests of evolutionaryscenarios. Cladistics, 9, 427–432.

Desutter-Grandcolas, L., 1997. Studies in cave life evolution: arationale for future theoretical developments using phylogeneticinference. J. Zool. Syst. Evol. Res. 35, 23–31.

Desutter-Grandcolas, L., 2003. Phylogeny and the evolution ofacoustic communication in extant Ensifera (Insecta, Orthoptera).Zool. Scr. 32, 525–561.

Desutter-Grandcolas, L., D’Haese, C., Robillard, T., 2003. Theproblem of characters susceptible to parallel evolution in phylo-genetic analysis: a reply to Marques and Gnaspini (2001) withemphasis on cave life phenotypic evolution. Cladistics, 19, 131–137.

Desutter-Grandcolas, L., Robillard, T., 2003. Phylogeny and theevolution of calling songs in Gryllus (Insecta, Orthoptera, Grylli-dae). Zool. Scr. 32, 173–183.

Eisthen, H.L., Nishikawa, K.C., 2002. Convergence: obstacle oropportunity? Brain Behav. Evol. 59, 235–239.

Eldredge, N., Cracraft, J., 1980. Phylogenetic Patterns and theEvolutionary Process, Method and Theory in Comparative Bio-logy. Columbia University Press, New York.

Farris, J.S., 1983. The logical basis of the phylogenetic system. In:Platnick, N.I., Funk, V.A. (Eds.), Advances in Cladistics. Colum-bia University Press, New York, pp. 7–36.

Fitch, W.M., 2000. Homology. A personal view on some of theproblems. Trends Genet. 16, 227–231.

Gould, S.J., Lewontin, R.C., 1979. The spandrels of san Maroc andthe panglossian paradigm: a critique of the adaptationist pro-gramme. Proc. R. Soc. Lond. B, 205, 581–598.

Grandcolas, P., D’Haese, C., 2003. Testing adaptation with phylogeny:how to account for phylogenetic pattern and selective valuetogether? Zool. Scr. 32, 483–490.

Grandcolas, P., Deleporte, P., Desutter-Grandcolas, L., 1994. Why touse phylogeny in evolutionary ecology ? Acta Oecol. 15, 661–673.

Grandcolas, P., Deleporte, P., Desutter-Grandcolas, L., 1997. Testingevolutionary processes with phylogenetic patterns: test power andtest limitations. In: Grandcolas, P. (Ed.), The Origin of Biodiver-sity in Insects: Phylogenetic Tests of Evolutionary Scenarios.Memoires du Museum national d’Histoire naturelle, Paris, 173, pp.53–71.

Grandcolas, P., Deleporte, P., Desutter-Grandcolas, L., Daugeron, C.,2001. Phylogenetics and ecology: as many characters as possibleshould be included in the cladistic analysis. Cladistics, 17, 104–110.

Guilbert, E., 2004. Do larvae evolve the same way as adults inTingidae (Insecta: Heteroptera)? Cladistics, 20, 139–150.

Gwynne, D.T., 1995. Phylogeny of the Ensifera (Orthoptera): ahypothesis supporting multiple origins of acoustical signalling,complex spermatophores and maternal care in crickets, katydidsand wetas. J. Orthopt. Res. 4, 203–218.

Haas, O., Simpson, G.G., 1946. Analysis of some phylogenetic terms,with attempts at redefinition. Proc. Am. Phil. Soc. 90, 319–347.

Hall, B.K., 1994. Homology. The Hierarchical Basis of ComparativeBiology. Academic Press, San Diego.

Hawkins, J.A., 2000. A survey of primary homology assessment:different botanists perceive and define characters in different ways.In: Scotland, R., Pennington, R.T. (Eds.), Homology and System-atics. Taylor & Francis, London and New York, pp. 22–53.

Hawkins, J.A., Hughes, C.E., Scotland, R.W., 1997. Primary homologyassessment, characters and character states. Cladistics, 13, 275–283.

Hennig, W., 1966. Phylogenetic Systematics. University of IllinoisPress, Urbana.

Henry, C.S., Martinez Wells, M.L., Simon, C.M., 1999. Convergentevolution of courtship songs among cryptic species of the carneagroup of the green lacewings (Neuroptera: Chrysopidae: Chryso-perla). Evolution, 53, 1165–1179.

Hu, F.Y., Tao, D.Y., Sacks, E., Fu, B.Y., Li, J., Yang, Y., McNally,K., Khush, G.S., Paterson, A.H., Li, Z.-K., 2003. Convergentevolution of perenniality in rice and sorghum. Proc. Natl Acad. Sci.USA, 100, 4050–4054.

Kallersjo, M., Albert, V.A., Farris, J.S., 1999. Homoplasy increasesphylogenetic structure. Cladistics, 15, 91–93.

Kitching, I.J., Forey, P.L., Humphries, C.J., Williams, D.M., 1998.Cladistics. The Theory and Practice of Parsimony Analysis. OxfordUniversity Press, Oxford.

Kluge, A.G., 2001. Parsimony with and without scientific justification.Cladistics, 17, 199–210.

60 L. Desutter-Grandcolas et al. / Cladistics 21 (2005) 51–61

Kluge, A.G., 2003. The repugnant and mature in phylogeneticinference: atemporal similarity and historical identity. Cladistics,19, 356–368.

Larson, A., Losos, J.B., 1996. Phylogenetic systematics of adaptation.In: Rose, M.R., Lauder, G.V. (Eds.), Adaptation. Academic Press,San Diego, pp. 187–219.

Lauder, G.V., 1994. Homology, form, and function. In: Hall, B.K.(Ed.), Homology. The Hierarchical Basis of Comparative Biology.Academic Press, San Diego, pp. 151–196.

Luckow, M., Bruneau, A., 1997. Circularity and independence inphylogenetic tests of ecological hypotheses. Cladistics, 13, 145–151.

Moritz, C., Hillis, D.M., 1996. Molecular systematics: context andcontroversies. In: Hillis, D.M., Moritz, C. (Eds.), MolecularSystematics, 2nd edn. Sinauer Associates, Sunderland, MA, pp.1–13.

Nelson, G., 1994. Homology and systematics. In: Hall, B.K. (Ed.),Homology. The Hierarchical Basis of Comparative Biology.Academic Press, San Diego, pp. 102–151.

O’Hara, R.J., 1988. Homage to Clio, or, toward an historicalphylosophy for evolutionary biology. Syst. Zool. 37, 142–155.

Pagel, M.D., 1994. The adaptationist wager. In: Eggleton, P., Vane-Wright, R.I. (Eds.), Phylogenetics and Ecology. The LinneanSociety of London, London, pp. 29–51.

Parra-Olea, G., Wake, D.B., 2001. Extreme morphological andecological homoplasy in tropical salamanders. Proc. Natl Acad.Sci. USA, 98, 7888–7891.

Patterson, C., 1982. Morphological characters and homology. In:Joysey, K.A., Friday, A.E. (Eds.), Problems of PhylogeneticReconstruction. Academic Press, London and New York,pp. 21–74.

Patterson, C., 1988. Homology in classical and molecular biology.Mol. Biol. Evol. 5, 603–625.

de Pinna, M.C.C., 1991. Concepts and tests of homology in thecladistic paradigm. Cladistics, 7, 367–394.

Remane, A., 1952. Die Grundlagen des Naturlichen Systems derVergleichenden Anatomie und der Phylogenetik. Geest und PortigK.G., Leipzig.

Ricklefs, R.E., Miller, G.L., 1999. Ecology, 4th edn. W.H. Freeman,New York.

Rieppel, O., Kearney,M., 2002. Similarity. Biol. J. Linn. Soc. 75, 59–82.Scotland, R., Pennington, R.T., 2000. Homology and Systematics.

Taylor & Francis, London and New York.Simpson, G.G., 1961. Principles of Animal Taxonomy. Columbia

University Press, New York.Sinha, N.R., Kellogg, E.A., 1996. Parallelism and diversity in multiple

origins of C4 photosynthesis in the grass family. Am. J. Bot. 83,1458–1470.

Swensen, S.M., 1996. The evolution of actinorhizal symbioses:evidence for multiple origins of the symbiotic associations. Am.J. Bot. 83, 1503–1512.

Teeling, E.C., Madsen, O., van den Bussche, R., de Jong, W.W.,Stanhope, M.J., Springer, M.S., 2002. Microbat paraphyly and theconvergent evolution of a key innovation in Old World rhinophoidmicrobats. Proc. Natl Acad. Sci. USA, 99, 1431–1436.

Wenzel, J.W., Siddall, M.E., 1999. Noise. Cladistics, 15, 51–64.Wheeler, Q.D., 2004. Taxonomic triage and the poverty of phylogeny.

Phil. Trans. R. Soc. Lond. B, 359, 571–583.Wiens, J.J., Chippindale, P.T., Hillis, D.M., 2003. When are phylo-

genetic analyses misled by convergence? A case study in Texas cavesalamanders. Syst. Biol. 52, 501–514.

Wiley, E.O., 1981. Phylogenetics. The Theory and Practice ofPhylogenetic Systematics. John Wiley & Sons, New York.

Wittkopp, P.J., Williams, B.L., Selegue, J.E., Carroll, S.B., 2003.Drosophila pigmentation evolution: divergent genotypes underlyingconvergent phenotypes. Proc. Natl Acad. Sci. USA, 100, 1808–1813.

61L. Desutter-Grandcolas et al. / Cladistics 21 (2005) 51–61