Portal de Periódicos da CAPES

Homology and historiography

2012; Wiley; Volume: 28; Issue: 6 Linguagem: Inglês

10.1111/j.1096-0031.2012.00415.x

1096-0031

James S. Farris,

Evolution and Paleontology Studies

Sir, Nelson (1994, p. 127) criticized cladists for including symplesiomorphy in homology: Characters seen as part of the same transformation series were claimed by early cladists to be homologous, and it is now evident that the claim is defective, for it construes the transformation series as “fins”-“arms”-wings, that is as including symplesiomorphy. He did not name anyone specifically, but it is not hard to discover whose view Nelson was portraying as defective (Hennig, 1966, p. 93): Different characters that are to be regarded as transformation stages of the same original character are generally called homologous. ‘‘Transformation’’ naturally refers to real historical processes of evolution. Neither, strangely, did Nelson (1994) identify the origin of the supposition that symplesiomorphy should be excluded from homology, and discovering that will turn out to be of some interest in connection with more recent discussions of homology. Wägele (2004, p. 110) traced the conception to Patterson: Some confusion exists because certain authors equate homology with synapomorphy (Patterson, 1982), a usage that certainly (and rightly so) would have been rejected by Hennig. Patterson (1982, p. 28) did indeed hold that view: Homologous features are those which characterize monophyletic groups... The definition implies either that paraphyletic groups are not characterized by homologies, or that shared primitive characters are non-homologous. According to Patterson (1982, p. 29), however, the idea was not new in 1982, but was already widely accepted: Although the equivalence of homology and Hennig’s concept of synapomorphy is implicit in Hennig’s work (e.g. 1966, p. 95), I did not become fully aware of the equation until early in 1976, through a manuscript by Gareth Nelson. That manuscript is just published (Nelson and Platnick, 1981), but appreciation of the equivalence of homology and synapomorphy is fairly widespread (Wiley, 1975, 1976; Bonde, 1977, p. 779; Bock, 1977, p. 888; Szalay, 1977, p. 16; Platnick and Cameron, 1977; Nelson, 1978; Cracraft, 1978; Patterson [1978]; Platnick [1978]; Gaffney [1979]). Yet most of those attributions seem to reflect wishful thinking on Patterson’s part. Not only did Hennig’s (1966, p. 93; quoted above) concept of homology include symplesiomorphy (which of course was precisely Nelson’s complaint), equating homology with synapomorphy was exactly what Hennig (1966, p. 95) did not say: In general we speak only of the homology of organs, but a “character” may also be the absence of an organ... The absence of wings in the Anoplura [sucking lice] and Mallophaga [biting lice] is a synapomorphous character, whereas in the Collembola, Protura, etc. it is a symplesiomorphous character. Wiley (1975, p. 235) used Hennig’s concept: Within this system two (or more) characters are said to be homologous if they are transformation stages of the same original character present in the ancestor of the taxa which display the characters (modified from Hennig, 1966). Furthermore (Wiley, 1975, p. 242): The terms apomorphous and plesiomorphous (and their derivatives) convey precise concepts which are logical derivatives of the phylogenetic definition of homology. As such, they should be substituted for the word homologous in systematic studies. E. O. Wiley (pers. commun.), moreover, emphatically rejects the contention that he (Wiley, 1975, 1976) would have excluded symplesiomorphy from homology. Szalay’s (1977) page 316 was taken up by a tree diagram. Bock (1977, p. 888) said nothing about ruling out symplesiomorphies as homologies: It should be noted that the determination of homologues does not automatically provide synapomorphs or symplesiomorphs for that matter. Neither did Platnick and Cameron (1977, p. 383): The discovery of homologous features in two languages demonstrates that they are related, since the homologies are both shared primitive features of the two languages themselves and shared derived features of a common ancestor of the two. Patterson’s (1978) own earlier paper said nothing about whether symplesiomorphies would be homologies—it instead concerned verifiability. Platnick (1978) never mentioned homology at all (and so did not say that homology would exclude symplesiomorphy), and Cracraft’s (1978, p. 215) only mention of homology was to identify convergence with non-homologous similarity. Neither Nelson (1978) nor Gaffney (1979) ever mentioned symplesiomorphy (and so did not say that symplesiomorphy would not be homology), and Gaffney’s (1979, p. 96) odd comment would seem to imply, if anything, that homology is to be equated with symplesiomorphy: A homology at one level is a synapomorphy at some higher level. Bonde (1977, p. 779) did not fit Patterson’s portrayal either, as he took note of Hennig’s comment quoted above: Hennig’s reason (1966: 95) for coining the term synapomorphy is that a feature may be the loss of a structure, and it seems artificial to talk about the homology of a non-existent structure. Bonde (1977, p. 779), however, was familiar with the idea in question—which he associated with Patterson: Therefore the concepts of synapomorphy and (phylogenetic) homology are almost identical (see Patterson, this vol.). Bonde was actually referring to Patterson’s oral presentation in the same symposium, which was held in the summer of 1976. Patterson (1982) reported having gotten the idea from Nelson’s manuscript early in 1976, and in fact the published version of that manuscript did suggest that symplesiomorphy would not be homology (Nelson and Platnick, 1981, p. 138): What might seem to be a homologous resemblance [presence of heavy hair] between the chimpanzee and orang-utan is symplesiomorphic, i.e., retention of a primitive character of mammals—a mammalian synapomorphy. The idea of excluding symplesiomorphy from homology thus came from Nelson and was propagated with Patterson’s help, beginning in 1976. But why, then, make up other sources? Of course proponents of fringe views often pretend to enjoy wide support, but the manufactured appearance of acceptance was particularly important for Patterson because his position was in fact based entirely on a switch of meaning. Patterson (1982, p. 30) knew that symplesiomorphies fit the conventional definition of homology:1 The absence of wings in all these organisms is homologous by the conventional evolutionary definition (similarity due to common ancestry). He arrived at his conclusion by drastically changing the definition (Patterson, 1982, p. 28): Homologous features are those which characterize monophyletic groups... The definition implies either that paraphyletic groups are not characterized by homologies, or that shared primitive characters are non-homologous, which is certainly counter-intuitive, and which is discussed further in the section on homology as synapomorphy. That was even stranger than it sounded. Patterson meant “characterize” to be taken literally, as indicating that a feature would be both necessary and sufficient (NAS) for membership in a group. In fact, Patterson’s way of reaching what he himself considered a counterintuitive conclusion consisted of a fantasy-scenario (discussed in detail by Farris, 2012a) in which an imaginary systematist placed imaginary insects on the basis of the NAS-trait principle, for example (Patterson, 1982, p. 32): The systematist must insist that the only character of Phthiriaptera [lice] is the absence of wings: no louse can have wings and no pterygote may lack them. To see the effect of the redefinition, consider that under Lankester’s concept, the wings of (say) bees and cockroaches are considered homologous because the shared trait is attributable to inheritance from the common ancestor of those taxa. Even though shared possession of wings is attributable to inheritance, however, under Patterson’s definition, the same shared possession of wings is not to be called “homology”, just because there is no monophyletic group for which possession of wings is NAS (wings having been secondarily lost in some pterygotes). But why switch “homology” to such a concept? Patterson (1982, p. 21) did so, he said, because: Defining homology in terms of common ancestry (cladistic, evolutionary) causes problems which are avoided by defining it in terns of monophyly: homology is the relation which characterizes monophyletic groups. That was obviously a contrived pretext, as monophyly is itself defined in terms of common ancestry, and in any case Patterson had simply inserted the “characterize” gratuitously. Evidently Patterson was determined to find an excuse to make “homology” refer to NAS traits. As Farris (2012a) has pointed out, Patterson’s actual reason for adopting NAS-“homology” was that his own method, ppa (pattern analysis of Patterson, 1980, 1988), worked by counting NAS traits. That connection with ppa turned out to be particularly unfortunate. Whereas Hennig (1966) emphasized the importance of distinguishing apomorphy from plesiomorphy and using only synapomorphy as grounds for grouping, Patterson (1988, p. 76) contended: It is not necessary to polarize characters, or to sort them into primitive and derived, to find a cladogram that is unambiguously rooted. Patterson did not mean to advocate the procedure, often used with Wagner trees (cf. Farris, 1970), in which apomorphies are implicitly distinguished from plesiomorphies by rooting an undirected tree—in fact Patterson strongly disapproved of that method. Instead, ppa did not even try to make the distinction, so that it simply grouped by all states—plesiomorphies as well as apomorphies. In that respect, ppa differed from Hennig’s approach and resembled phenetic clustering, which resemblance Patterson (1988, p. 83) indeed pointed out himself: The impact of neutral theory and the clock on systematics is to make phenetic and cladistic [sic—he meant ppa] methods equivalent at the level of DNA. Nonetheless, Patterson (1980, 1988) insisted on mislabelling ppa as a “cladistic” method, and to expose that deceptive usage I called ppa counterfeit cladistics (Farris, 2012a). The “cladistic” label seems especially unsuitable because ppa can readily form paraphyletic groups, as is illustrated by an example from Farris (2011), using Wheeler et al.’s (1993) data on arthropods and relatives. For those data, parsimony jackknifing (Farris et al., 1996) yields the tree (Fig. 1) one would reasonably expect, all the presently recognized major groups appearing as monophyletic. For ppa (Fig. 2), in contrast (Farris, 2011, p. 211): Parsimony jackknife tree for Wheeler et al.’s (1993) data with Trilobita omitted, redrawn after fig. 4 of Farris (2011). Molluscs (Loligo, Lepidochiton) are the outgroup. Strict consensus of optimal ppa trees for Wheeler et al.’s (1993) data with Trilobita omitted. Redrawn after fig. 5 of Farris (2011). Nephila is not an outgroup, ppa trees instead being rooted according to similarity in all states. ppa displaces highly divergent whip-spider Nephila to the base of the tree, in the process breaking up Araneae, Chelicerata, and Arthropoda. It creates paraphyletic groups ((Mollusca Annelida) Onychophora) based only on symplesiomorphy, and it manages to insert those taxa inside Arthropoda, so making Arthropoda seem paraphyletic. Finally, it simply ignores the synapomorphies of Myriapoda as well as those of Myriapoda+Insecta. As a final note, ppa’s syncretistic behaviour has an amusing connection with 3tans, advocates of Nelson and Platnick’s (1991) three-taxon analysis (3ta). 3tans revere Patterson as a pattern-taxonomic icon, and they also infamously and insistently concoct excuses to denigrate everything un3ta as “phenetic”—except for ppa (Farris, 2012a; p. 228):2 Williams and Ebach ([2012]) denounce Wagner parsimony... as “phenetic cladistics.’’ Well, ppa actually is phenetic, but Williams and Ebach have never been known to call Patterson a pheneticist... [thus] Patterson is still valuable as an indicator of the depth of 3tan sincerity. Alternative history. It is not only 3tans who carefully overlook the embarrassing parts of Patterson’s views. Brower and de Pinna (2012, hereinafter BdP) do so as well (BdP, p. 2): It is surprising to see cladists still arguing about methodology in terms of whether one ought to be “grouping by synapomorphy’’ or by ‘‘all states’’ (cf. Farris, 2012[a]; Nixon and Carpenter, 2012). Such terminology no doubt inspires (albeit misplaced) accusations of pheneticism. Patterson (1988, p. 83; quoted above) himself called ppa equivalent to phenetics, but BdP have never heard of that—or so one would think. In fact Farris (2012a) also quoted Patterson’s remark, as did Farris (2011). Evidently BdP are simply unwilling to admit what Patterson (1988) said. They seem more open to some of Patterson’s (1982) misattributions, however (BdP, p. 3): The cladistic equation of homology with synapomorphy brought precise and simple clarity to a fundamental yet formerly nebulous idea. The insight was so appealing that several thinkers came to it independently (e.g. Wiley, 1975, 1976; Bonde, 1977; Platnick and Cameron, 1977; Cracraft, 1978; Nelson, 1978; Patterson, 1978; Gaffney, 1979; Platnick, 1979; Nelson and Platnick, 1981), prior to the widely cited paper by Patterson (1982). Although these publications have been cited many times in the mainstream literature, we have little doubt that they will be ignored many times more and that Patterson will continue to be credited/scapegoated as creator of the idea. That acts as a smokescreen. What BdP avoid saying is that Patterson got “homology equals synapomorphy” (HES) from Nelson’s manuscript in 1976—and that by HES, Nelson and Platnick (1981) and Patterson (1982) meant to exclude symplesiomorphy from homology. Having thus concealed where HES came from and what it meant to its original advocates, BdP are free to introduce their own version (BdP, p. 1):3 Homology as a notion relevant to the recognition of clades remains equivalent to synapomorphy. Some symplesiomorphies are ‘‘homologies’’ inasmuch as they represent synapomorphies of more inclusive taxa; others are complementary character states that do not imply any shared evolutionary history among the taxa that exhibit the state. BdP’s concept is not like Nelson’s. In Nelson and Platnick’s (1981, p. 138; quoted above) example of hairy apes, being a synapomorphy of a more inclusive group would make the symplesiomorphy homology by BdP’s reasoning (as of course it is by Lankester’s definition as well), and this contradicts Nelson and Platnick’s (or Patterson’s) assertion that the symplesiomorphy is not homology. The same would naturally apply to almost any symplesiomorphy. The other symplesiomorphies—purely plesiomorphic shared absences—would seem to be homologies too, at least primary homologies (BdP, p. 4): As noted by Nelson (1978, p. 340), ‘‘absence of a character is not a character,’’ but it can still be a symplesiomorphy (Sereno, 2007). We view complementary absences as primary homologues (epistemologically identical character states) but not as secondary homologues. Of course “absence of a character is not a character”, but as Hennig (1966, p. 95; quoted above) said, the absence of an organ may be a character! Nelson often resorts to word games (cf. Farris, 2012b), but this one seems unusually shameless. If it were taken seriously, however, Nelson’s (1978)“argument” would also dispose of shared apomorphic (secondary) absences as characters. Secondary absences are of particular interest because 3ta notoriously fails to recognize secondary absences as synapomorphies, so denying (for example) Hennig’s (1966, p. 95; quoted above) observation that the winglessness of lice is a synapomorphous character.4 BdP’s idea that homologies should reflect shared evolutionary history is useful in this case, as secondary absences can obviously meet that requirement. Thus BdP’s shared history criterion seems incompatible with Nelson’s position. More generally, BdP’s homology seems to be effectively like the conventional concept (Lankester’s), so that rather than bringing “precise and simple clarity”, their version of HES seems primarily to add a layer of terminology. But in that case, why bother with HES? According to de Pinna (1991, p. 270): The various papers cited above [most of Patterson’s list, quoted earlier], especially Patterson (1982), provide clear and consistent reasons to equate homology and synapomorphy. Evidently de Pinna has a keen sense of humour. Again, Patterson (1982) meant to exclude symplesiomorphy from homology, whereas de Pinna (1991, p. 272) obviously did not: Symplesiomorphic similarities are obviously homologous, but every symplesiomorphy is a synapomorphy at a higher level. Patterson’s “clear and consistent reasons”, moreover, involved NAS-”homology” and fantasy-scenarios—subjects that de Pinna (1991) carefully avoided mentioning, as do BdP now. BdP’s and de Pinna’s (1991) HES thus seems meant to paper over another embarrassing part of Patterson’s views. BdP aside, the HES slogan by itself can encourage confusion, as is illustrated by a second kind of counterfeit cladistics, what Williams et al. (2010, p. 187) comically named “the cladistic redux” (cf. Farris, 2012d). Williams et al. (2010, p. 187) proposed to group by homology: Classification is meant to make sense of relationships by looking for sameness, which is observable, rather than an event, which is only partially observable. In this sense, relative relationships, in the sense of sameness or homology, are better ways to classify and summarize overall taxic relationships than inferring genealogies or phylogenies. In reality there is no reason why observed samenesses could not be symplesiomorphies, in which case “the cladistic redux” could lead to paraphyletic groups. Pattern taxonomists being as they are, however, Williams et al. will surely insist that their homologies cannot be symplesiomorphies—because “homology is equivalent to synapomorphy”! It would be better for systematics if we could all be spared such nonsense. Dr J.M. Carpenter provided extensive and invaluable aid in obtaining information from the older literature, and of course it was T.N. Nayenizgani’s (1990) entertaining exposé that first unmasked Patterson’s “cladistic” syncretism.