Syst. Biol. 46(3):441^63, 1997

A MODEL SYSTEM OF STRUCTURAL DUPLICATION: HOMOLOGIES OF ADDUCTOR MANDIBULAE MUSCLES IN TETRAODONTIFORM FISHES

JOHN P. FRIEL AND PETER C. WAINWRIGHT Department of Biological Science, Florida State University, Tallahassee, Florida 32306-4370, USA;

E-mail: [email protected] (J.P.F.), [email protected] (P.C.W.) Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020

Abstract.—We critically reviewed the homologies of the jaw muscles in tetraodontiform fishes (Triacanthoidea, Balistoidea, Tetraodontoidea), as first described in Winterbottom's phylogenetic monograph (1974, Smithson. Contrib. Zool. 155:1-201), as a case study in structural duplication. Within this order of teleost fishes, the two main adductor mandibulae muscles, Al and A2, are duplicated one or more times in some subclades. The number of descendant Al and A2 muscles ranges from as few as the original two muscles in triplespines to as many as eight muscles in some filefishes. As first pointed out by Winterbottom, the homologies of some muscles are unclear, particularly in comparisons between the superfamilies Balistoidea (boxfishes, triggerfishes, file- fishes) and Tetraodontoidea (pursefishes, molas, puffers, porcupinefishes). We reassessed the ho- mologies (orthologs and paralogs) of these Al and A2 muscles based on their origins, insertions, and relative masses in representative taxa and their congruence with a phylogeny for these taxa. New names that reflect the homologies of these muscles are presented. Ten muscle duplications by subdivision and three phylogenetic losses of muscles have occurred in this system. No rela- tionship was found between the number of separate muscles and the relative masses of the Al or A2 muscles, suggesting that muscle duplication events essentially repackage existing muscle tissue. However, both Al and A2 muscle masses are correlated with each other and with the feeding ecology of these fishes. Durophagous taxa have relatively larger Al and A2 muscles, whereas planktivores and benthic grazers have relatively smaller A2 muscles. [Homology; mor- phological duplications; muscle evolution; orthology; paralogy; Tetraodontiformes.]

The concept of structural duplication evolution of the protein structural diversi- has played a central role in the search for ty (Chothia, 1994), and the evolution of general patterns and repeating themes in some developmental mechanisms in ver- die evolution of organismal design (Ver- tebrates (Holland et al., 1994; Holland and meij, 1973a, 1973b; Liem, 1980b; Lauder, Garcia-Fernandez, 1996). 1981, 1990; Liem and Wake, 1985; Schaefer Although numerous examples exist of and Lauder, 1986; Emerson, 1988; Lauder repeated elements in animals and plants and Liem, 1989). The central idea is that (e.g., appendages, blood vessels, body seg- phylogenetic increases in the number of ments, hairs, leaves, muscles, scales), rela- structural units with the same primitive tively few cases have been developed to function promote functional diversification the point that specific consequences of by several routes. Repeated elements in- structural duplication can be tested within crease the overall complexity of design, a rigorous phylogenetic framework (Laud- which may underlie an increase in mor- er, 1990). Here, we describe a system of phological and functional diversity within structural duplication in the jaw muscula- a clade (Vermeij, 1973a; Liem, 1980b; ture of tetraodontiform fishes. Within this Schaefer and Lauder, 1986, 1996). The clade, the Al and A2 adductor mandibulae functional redundancy created by dupli- muscles have been duplicated repeatedly cation of morphology or molecules may in different subclades and extant taxa may also release constraints on one of the prim- possess from two to eight separate Al and itive elements, permitting it to evolve a A2 muscles. Our primary purpose here is new form or function (Lauder, 1981). For to review the homology and phylogenetic example, redundancy is thought to be a history of these tetraodontiform jaw mus- primary mechanism in the evolution of cles and present a nomenclature that re- gene function (Ohno, 1970; Ohta, 1989), the flects our hypotheses of homologies. We

441 442 SYSTEMATIC BIOLOGY VOL. 46

Triacanthoids Balistoids Tetraodontoids

11 genera 4 genera 31 genera 11 genera 14 genera 1 genus 3 genera 20 genera 6 genera 20 species 7 species 95 spedes 40 spedes 33 spedes 1 spedes 3 spedes 147 spedes 19 spedes

Triacanthodidae Triacanthidae Monacanthidae Balistidae Ostraciidae Triodontidae MoNdae Tetraodontidae Diodontidae Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020

FIGURE 1. Phylogeny of extant tetraodontiform families based on the work of Winterbottom (1974b), Mat- suura (1979), Tyler (1981), Lauder and Liem (1983), and Winterbottom and Tyler (1983). Images represent the general body form of fishes in these families. The number of genera and spedes are from Nelson (1994). also present initial data on the functional miliar tetraodontiform fishes placed in the consequences of muscle duplication other two superfamilies. The superfamily through an analysis of muscle sizes. Balistoidea contains the Ostraciidae (box- fishes, cowfishes), Balistidae (triggerfish- Adductor Mandibulae Muscles of es), and Monacanthidae (filefishes). Their Tetraodontiform Fishes sister group, the superfamily Tetraodon- The teleost order Tetraodontiformes is a toidea, contains the Triodontidae (purse- diverse group of primarily marine fishes fishes), Molidae (ocean sunfishes), Tetra- broadly distributed throughout the tropi- odontidae (puffers), and Diodontidae cal and temperate regions of the world. (porcupinefishes). This order is represented today by nine One of the most distinctive features of families containing approximately 101 most tetraodontiforms are their robust genera with 365 species (Fig. 1; Nelson, teeth and powerful oral jaws. Unlike most 1994). The tetraodontiform phylogeny other bony fishes, tetraodontiform fishes used in this study was proposed by Win- use their oral jaws not only to capture but terbottom (1974b), was corroborated by also to process prey (Turingan and Wain- Lauder and Liem (1983), and is supported wright, 1993; Wainwright and Turingan, by additional data provided by Matsuura 1993; Turingan, 1994). As in other fishes, (1979), Tyler (1980), and Winterbottom the muscles responsible for closing and and Tyler (1983). Tetraodontiforms are di- generating the biting forces of the oral jaws vided into three large clades. The relative- are those of the adductor mandibulae com- ly basal superfamily Triacanthoidea con- plex (Lauder, 1985). In most teleost fishes, tains the Triacanthodidae (spikefishes) and this muscle complex consists of at least Triacanthidae (triplespines). These triacan- four muscles, Al, A2, A3, and Aw, which thoids are the sister taxon to the more fa- are distinguished based on their relative 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 443 origins and insertions on the jaws (Winter- mations necessary to explain apparent bottom, 1974a). All of these muscles orig- morphological differences between "ho- inate on the palatal arch and are innervat- mologous" muscles in both taxa. Thus, we ed by branches of the fifth cranial nerve. began to question these homologies, par- Winterbottom (1974a) hypothesized that ticularly in comparisons between balis- all adductor mandibulae muscles of teleost toids and tetraodontoids. fishes plesiomorphically inserted solely on In addition to reviewing the morpholo- the lower jaw and Al has arisen phyloge- gy described by Winterbottom (1974b), we Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 netically by encroachment of muscle fibers examined additional taxa and here provide upon the maxilla-mandibular ligament, data on the relative masses of the Al and which connects the upper and lower jaws. A2 muscles in representatives of six of the Thus, all adductor mandibulae muscles ex- nine tetraodontiform families. Mass is gen- cept Al retain this plesiomorphic insertion erally reflective of the force-producing ca- on the lower jaw, and Al may retain a pacity of skeletal muscles (Powell et al., close connection with A2 in basal tetra- 1984), and we analyzed the general pat- odontiforms such as triacanthoids. In most terns of adductor mass in light of this teleost fishes including all tetraodonti- functional interpretation. We asked two forms, Al and A2 are the most easily ob- specific questions. First, as the adductor served jaw muscles given their large size muscles are duplicated, does the overall and relatively superficial positions. A3 and mass of the adductor muscle complex in- Aw lie deep to both Al and A2 and always crease or are new muscles essentially insert on the lower jaw. carved out of existing tissue? Second, are Within tetraodontiforms, the number of patterns of diversity in adductor muscle Al and A2 muscles has increased in some size reflective of differences among species clades, whereas the A3 and Aw muscles in feeding habits? when present are always singular (Winter- bottom, 1974b). The number of Al muscles Homobgies of Duplicated Muscles ranges from a single undivided Al in all Our review and new hypotheses of mus- triacanthoids and triodontids to as many cle homologies are based on the assump- as six separate Al muscles in some mon- tion that all the duplicated Al and A2 acanthids. Similarly, the number of A2 muscles in this clade of fishes have been muscles varies from a single A2 in triacan- produced by physical subdivision of pre- thids to as many as three separate A2 mus- existing muscles. Specifically, all Al and cles in balistids and tetraodontids. This A2 muscles have arisen in a phylogenetic duplication of adductor mandibulae mus- sense by repeated subdivision of singular cles, along with variation in their origins Al and A2 muscles present in the common and insertions, has made identifying ho- ancestor of all tetraodontiforms and in an mologous muscles between families and ontogenetic sense by repeated subdivision particularly superfamilies difficult. Win- of singular Al and A2 muscles present terbottom's (1974b) classic monograph on during the development of an individual the myology of tetraodontiform fishes doc- fish. We favor this model of muscle evolu- uments the diversity of adductor mandib- tion over a de novo model in which a new ulae muscles in this clade and provided jaw muscle could arise from a novel con- the first nomenclature for these muscles. densation of myogenic tissue. Although we During initial research on a representative do not regard the de novo origin of mus- balistoid (the gray triggerfish, Batistes ca- cles as an impossibility, a subdivision priscus) and a tetraodontoid (the southern model is consistent with the repeated ob- puffer, Sphoeroides nephalus), we identified servations of incompletely divided muscles muscles following Winterbottom (1974b). in some taxa (e.g., Al muscles of ostraciids However, like Winterbottom, we had dif- and A2 muscles of monacanthids). Such ficulty identifying some muscles in Sphoe- intermediate morphologies would not be roides and in comprehending the transfor- produced by a de novo model but might 444 SYSTEMATIC BIOLOGY VOL. 46 be expected with a subdivision model. Al- though no ontogenetic data on tetraodon- tiforms is available, study of the adductor mandibulae complex in other teleost fishes does support a subdivision model (Ad- riaens and Verraes, 1996). Another important aspect of the subdi- vision model is that it produces a hierar- Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 chical pattern of ancestral and duplicated descendant muscles. Such a phylogenetic framework for muscles, like a phylogeny for taxa, allows for explicit predictions about the descendant muscles based on A1ab' A1ab" A1ab"' |Aipb'm Aipb"m pleisiomorphic features (i.e., functions, mo- tor patterns, physiology, etc.) inherited ^^4 from the ancestral muscle. Muscles that ap- pear de novo, by definition, have no clear homology to any preexisting muscle and thus no predictable pleisiomorphic fea- tures when they first appear. A2g"b| |A2p1 We concur with Lauder (1990) that du- plication of morphological structures such (C) A2a A2p as jaw muscles of tetraodontiforms is anal- ogous to the more familiar phenomenon of VA2 gene duplication. Granted, this analogy is FIGURE 2. Muscle duplication by subdivision, (a) not exact because duplication of genes is Muscle duplication events mapped onto a phylogeny explained only by phylogeny and dupli- of three hypothetical taxa. An ancestral muscle, Al, is cation of morphological structures is ex- subdivided into two descendant muscles, Ala and Alp, in the common ancestor of taxa X, Y, and Z. Aip plained by both phylogeny and ontogeny is later subdivided into two additional descendant (Patterson, 1987). Nevertheless, duplication muscles, Aip' and Aip", in the ancestor of taxon Z. of structures (i.e., genes or morphology) Different copies of a duplicated muscle in the same should require the recognition of both par- individual or in different taxa are paralogous muscles alogs and orthologs if possible (Fitch, 1970; (e.g., Ala and Aip in taxa X and Y). The same copy of a duplicated muscle in different taxa are ortholo- Patterson, 1987, 1988). Simply put, para- gous muscles (e.g., Ala in taxa X, Y, and Z; Aip in logs are different copies of a structure, taxa X and Y). (b). Subdivision pattern of Al muscles derived from a single ancestral structure, in tetraodontiform fishes. Muscle names reflect the that are present in the same or different historical relationships between duplicated muscles. individuals (e.g., a and p hemoglobin Boxes at the same level in the figure and suffixes b (balistoids or balistids), m (monacanthids), o (ostra- genes, Ala and Alp muscles; Fig. 2a), ciids), and t (tetraodontoids or tetraodontids) denote whereas orthologs are the same particular muscles produced by parallel subdivision events in copy in different individuals (e.g., a he- different clades. (c) Subdivision pattern of A2 muscles moglobin genes, Ala muscles; Fig. 2a). in tetraodontiform fishes. Properly identifying orthologs in different taxa is critically important in any compar- ative or phylogenetic study involving du- dividuals (i.e., paralogy) have been point- plicated structures (Patterson, 1987; Doyle, ed out by Bateson (1892, 1894), Danforth 1992; Hillis, 1994). (1930), Van Valen (1982), Roth (1984), Wag- In practice, the identification of paralogs ner (1989a, 1989b), Mabee (1993), and oth- and orthologs of morphological structures ers. For example, it may be difficult if not is often not possible. Problems in homolo- impossible to identify orthologs when one gizing duplicated structures both between particular copy of a duplicated structure is individuals (i.e., orthology) and within in- not morphologically distinct relative to 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 445 others in different individuals (e.g., a scale 143 mm), C. macrocerus (CM, 1, 325 mm), on the body of a fish vs. another scale on and Monacanthus hispidus (MH, 2, 93-215 a different fish) or when the total number mm); eight tetraodontids, Arothron mani- of duplicated structures and thus the rel- lensis (AM, 1,185 mm), Canthigaster rostrata ative position or identity of any one copy (CR, 1, 61 mm), Chelonodon patoca (CPA, 1, differs between individuals (e.gv vertebra 115 mm), Torquigener hicksi (TH, 1, 83 mm), 15 in a fish with 50 total vertebrae vs. ver- Lagocephalus lagocephalus (LL, 1, 506 mm), tebra 15 in another fish with 55 total ver- Sphoeroides maculatus (SM, 2, 130-172 mm), Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 tebrae). Furthermore, in most cases of se- S. nephalus (SN, 3, 205-235 mm), and S. tes- rial homology (e.g., body segments, tudineus (ST, 1, 164 mm); and four diodon- vertebrae, fin rays, teeth), homonomy (e.g., tids, Chilomycterus antennatus (CA, 1, 130 hairs, scales, blood cells), and redundant mm), C. schoepfi (CS, 1, 96 mm), Diodon hol- functional linkages (e.g., jaw opening lig- ocanthus (DHQ 1, 107 mm), and D. hystrix aments in teleost fishes; Lauder and Liem, (DHY, 1, 235 mm). Fishes were initially 1989), the duplicated structures cannot be fixed in 10% formalin and then transferred homologized with a single structure in a to 70% ethanol prior to dissection. All hypothetical ancestor or outgroup taxon specimens except Parahollardia lineata (McKitrick, 1994). (FMNH 46686) are from the laboratory col- In contrast, the duplicated jaw muscles lection of P. C. Wainwright at Florida State described in this study can be homolo- University. One or two individuals of each gized with single ancestral muscles just species were dissected along with an ad- like duplicated genes can be homologized ditional size series of 10 striped burrfish, with single ancestral genes. The general Chilomycterus schoepfi (41-166 mm). These concepts of homology between muscles additional burrfish were included to ob- within an individual and the evolution of tain preliminary data on the allometry of muscles via subdivision have previously the Al and A2 muscles in tetraodontiform been suggested by McKitrick (1994). Here, fishes. Fishes were weighed to the nearest we build upon these concepts and present gram prior to dissection, and all Al and the first attempt to explicitly distinguish A2 muscles were carefully dissected and orthologs and paralogs for duplicated removed from one side of the head from morphological structures. all specimens except Parahollardia lineata. Individual muscles were gently patted dry MATERIALS AND METHODS to remove excess fluid and weighed to the This review is based in part on mor- nearest 0.01 g. Most of the anatomical il- phology described by Winterbottom lustrations that supplement our descrip- (1974b). In addition, we examined the jaw tions were produced from camera lucida musculature of 24 species from seven of drawings of representative specimens dis- the nine families. Taxa examined, taxon ab- sected in this study. breviations used in figures, number of in- All mass data were log transformed be- dividuals examined, and standard length fore statistical analyses were performed. are as follows: one triacanthodid, Parahol- The effect of body size was removed by lardia lineata (1, 120 mm); two triacanthids, regressing muscle mass on body mass and Triacanthus biaculeatus (TB, 2, 174-225 mm)calculating residuals. Any apparent corre- and Trixiphichthys zveberi (TW, 2, 124-143lations between the raw data points or re- mm); two ostraciids, Acanthostracion quad-siduals could be due to the confounding ricornis (AQ, 2, 51-141 mm) and A. poly- effects of phylogeny because species are gonius (AP, 1, 128 mm); three balistids, not statistically independent data points Balistes capriscus (BC, 2, 124-256 mm), B. (Felsenstein, 1985; Garland et al., 1992). To vetula (BV, 2, 155-187 mm), and Xanthich- determine whether such correlated char- thys ringens (XR, 2, 147-168 mm); four acter evolution was real, we also calculated monacanthids, Aluterus schoepfi (AS, 2, phylogenetically independent contrasts 127-169 mm), Cantherhines pullus (CPU, 1, with the computer program CAIC (Purvis 446 SYSTEMATIC BIOLOGY VOL. 46 and Rambaut, 1995). Using an incomplete- dorsalis (ERD), hypertrophied branchios- ly resolved phylogeny and an assumption tegal ray (HBR), levator arcus palatini of equal branch lengths, 19 standardized (LAP), levator operculi (LO), maxilla (MX), contrasts were generated. All regressions palatine (PAL), premaxilla (PMX), protrac- involving contrasts were forced through tor hyoidei (PHY), ramus mandibularis the origin as required (Garland et al., (RMD), and retractor arcus palatini (RAP). 1992). All osteological terms follow Tyler (1980). To reflect our hypotheses of muscle ho- Because the ultimate test of any hypoth- Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 mologies, we produced a new nomencla- esis of homology is that of congruence ture for the Al and A2 muscles of tetra- with the homologies of other characters odontiform fishes. We tried to maintain (Patterson, 1982,1988), we used a variation current names when possible. Orthologous of this test to argue for our new set of ho- muscles have the same unique name in all mologies. We compared transformation se- taxa possessing them, and cases of parallel ries for the Al and A2 muscles, one based muscle evolution are always clearly iden- on current homologies and the other based tified (Figs. 2b, 2c). To preserve the histor- on our new homologies, for the same phy- ical relationships among adductor man- logeny of tetraodontiform families (Fig. 1). dibulae muscles, each subdivision event Although this phylogeny was originally requires new names for all descendant based in part on Winterbottom's homolo- muscles. There has been a tendency to re- gies of Al and A2 muscles, it is supported tain the ancestral muscle name for the one by other characters presented by Winter- descendant muscle that diverges least from bottom (1974b), Matsuura (1979), Tyler the ancestral muscle morphology. Al- (1980), and Winterbottom and Tyler though this practice reduces the number of (1983). The same phylogeny was produced new names needed, it obscures the ho- by parsimony analysis of all available data mologies between the descendant muscles minus any Al and A2 characters, so there and their ancestral muscle. was no circularity in optimizing these In our new nomenclature, muscle names muscle homologies on this phylogeny. begin with either Al or A2 followed by suffixes to reflect the transformational his- Transformation series were optimized tories of these muscles. The descendant using assumptions of both accelerated muscles of a primary subdivision event are (ACCTRAN) and delayed (DELTRAN) designated by a and (3, those of a subse- transformation in MacClade (Maddison quent secondary subdivision event by sin- and Maddison, 1992). These options do not gle, double, or triple primes if necessary change the number of character steps but (as in monacanthids), and those of a sub- can reveal equally parsimonious transfor- sequent tertiary subdivision event by a mation series. The ACCTRAN option max- and p again. In addition, the suffixes b, m, imizes reversals by favoring early gains of o, and t, for balistoids or balistids, mona- muscles near the root of the phylogeny canthids, ostraciids, and tetraodontoids or and later losses of muscles. In contrast, the tetraodontids, respectively, are used to dis- DELTRAN option maximizes parallelisms tinguish descendant muscles produced by by favoring independent gains of muscles parallel (i.e., nonhomologous) subdivision near the tips of the phylogeny. This pro- events in different clades. This nomencla- cedure allowed us to determine the mini- ture and hierarchy of muscles is summa- mum number of evolutionary steps neces- rized in Figures 2b and 2c. The correspon- sary to explain the observed morphology dence between old and new muscle names in the terminal taxa given a particular set is summarized in the Appendix. Abbrevi- of muscle homologies and to compare ations are used in figures for nonadductor competing hypotheses of homologies. The mandibulae muscles and other elements: preferred set of homologies was the one adductor arcus palatini (AAP), dentary that required the fewest steps and least (DEN), dilatator operculi (DO), erectores amount of homoplasy. 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 447

RESULTS Alp' (=Alpb"o) (Fig. 3d). The separation This review is divided into four sections: between these two ventral muscles is slight homologies of Al muscles within balis- in some taxa (including the Acanthostracion toids, homologies of Al muscles between examined here) but is well developed in balistoids and tetraodontoids, homologies other ostraciids examined by Winterbot- of A2 muscles between balistoids and tet- tom (1974b). All three Al sections insert via a common tendon on the distal portion raodontoids, and comparisons of the rela- Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 tive masses of Al and A2 muscles in rep- of the maxilla. resentative taxa. Because the first three In balistids, most of Al is superficially sections deal with competing hypotheses covered by A2 muscles and Al is subdi- of homologies, it is necessary to use the vided once into a superficial muscle, Ala two different nomenclatures. By conven- (=Alab) and a deeper muscle, Alp tion, in the morphological descriptions the (=Alpb) (Figs. 3e, 3f). Ala (=Alab) orig- more familiar names of Winterbottom inates in front of the eye along the ethmoid (1974b) are listed first and are immediately region and narrowly inserts on the distal followed by the new names in parentheses, end of the maxilla. Alp (=Aipb), in con- e.g., Ala (=Alab'). In all other sections trast, originates relatively ventrally, on the and in all figures, muscle names follow the metapterygoid, and inserts narrowly on new nomenclature, which is summarized the distal end of the maxilla. Both Al mus- in Figures 2b and 2c and the Appendix. cles insert on the upper jaw via a common tendon. This balistid pattern is modified in mon- Homologies of Al Muscles in Balistoids acanthids by further duplication of Al Most triacanthoids retain the plesiomor- muscles. Most monacanthids have at least phic Al condition for all tetraodontiforms. two superficial and three deep Al muscles There is a single superficial Al, dorsal to (Figs. 3g, 3h). The superficial Al muscles A2, which originates beneath the eye and include Ala (=Alab;), which inserts on inserts on both the upper and lower jaws the maxilla, and Ala' (=Alab"), which in- (Fig. 3a). The only exceptions to this pat- serts on both the maxilla and upper lip. tern are the long-snouted triacanthodid The deep Al muscles include Alp genera Macrorhamphosodes and Halimochi- (=Aipb'm) and Alp' (=Aipb"m), which rurgus, in which Al shares a tendon with both insert on the maxilla, and AI7 A2 and inserts solely on the lower jaw (=Alab'"), which uniquely inserts on the (Winterbottom, 1974b). In all triacan- palatine. Variations of this monacanthid thoids, Al is separated anteriorly from A2 pattern occur in a few genera. Ala' by the path of the ramus mandibularis of (=Alab") is absent in Aluterus, and Alp' the trigeminal nerve (Fig. 3a). However in (=Aipb"m) is absent in Anacanthus. Ala" triacanthids, Al is not completely separat- (=Alab'P) is uniquely present in Oxymon- ed from A2 posteriorly (Fig. 3a) as it is in acanthus dorsal to Ala' (=Alab'a) and in- triacanthodids. serts on the maxilla. Alp" (=Aipb"mp) is In contrast to the condition in triacan- uniquely present in Paraluteres dorsomedi- thoids, Al is duplicated one or more times al to Alp' (=Aipb"ma) and inserts on the in all balistoids and all tetraodontoids ex- maxilla. cept triodontids. Some of these Al muscles Based on the pattern of Al muscles in in both balistoids and tetraodontoids were balistoids, Winterbottom (1974b) hypothe- considered orthologs by Winterbottom sized that Al was subdivided into a su- (1974b). perficial Ala (=Alab) and a deep Alp In ostraciids, Al is superficially covered (=Aipb) in the common ancestor of all by A2 muscles and is incompletely subdi- balistoids. In addition, his recognition of vided into three sections: a large dorsal di- Alp' in both ostraciids (=Aipb"o) and vision, A1B" (=Alab), and two smaller monacanthids (=Aipb"m) could be con- ventral divisions, Alp (=Aipb'o) and strued as indicating that these muscles are 448 SYSTEMATIC BIOLOGY VOL. 46

A1pb"o DO A1ocb Alpb'o

A1 Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020

1 cm (f) (g) 1 cm (h) FIGURE 3. Jaw musculature of representative balistoids. See text for muscle and other element abbreviations, (a) Triacanthus biacukatus, superficial view; Al does not overlie A2, and both muscles are incompletely subdi- vided posteriorly, (b) T. biacukatus, deep view of jaw musculature with Al and A2 removed, (c) Acanthostracion quadricornis, superficial view; A2 is subdivided into A2a and A2|3, and these two muscles lie superficial to the Al muscles, (d) A. quadricornis, deep view with A2a and A2|J removed; Al is subdivided into Alab, Al($b"o, and Aipb'o. (e) Batistes capriscus, superficial view; A2 is subdivided into A2a, A2(3'b, and A2|3"b, and these three muscles lie superficial to the Al muscles, (f) B. capriscus, deep view with A2a and A2p'b removed; Al is subdivided into Alab and AlfJb. (g) Monacanthus hispidus, superficial view; A2 is subdivided into A2a and A2p, and these two muscles lie superficial to all Al muscles except Alab' and Alab". (h) M. hispidus, deep view with Alab', Alab", A2a, and A2(3 removed; note the three deep Al muscles, Alab'", Alfib'm, and Aipb"m. orthologs and that AlfJ' was present in the and within this clade, Ala' (=Alab") and common ancestor of all balistoids. This Al Ala" (=Alab'P) have subdivided from pattern for the common ancestor of balis- Ala, Alp" (=Al(Jb"m|S) has subdivided toids requires the following transforma- from Alp, and Al-y (=Alab'") has arisen tions to explain the morphology observed with uncertain affinities to any other Al in extant taxa. Subsequently in the lineage muscles. This transformation series for Al leading to ostraciids, Ala (=Alab) was muscles in balistoids, using Winterbot- lost and Alp' (=Al|Sb"o) was subdivided tom's homologies and nomenclature, is to form Alp" (=Alab). Apparently, Alp' mapped on a phylogeny of tetraodonti- was lost in the lineage leading to balistids. forms in Figure 4a. In the lineage leading to monacanthids Our reexamination of adductor mandib- 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 449

Triacanthoids Balistoids Tetraodontoids

1 1 A1P 1 1 1 •IT A1a A1a' A1P' A1P AIO A1p Aip/i A1y AIP' A1a A1a A1 A1 Al /^^v AIP" A1P Alfl J^ "s • -to ij 1 ^—-0^ V Triacanthodidae Triacanthidae Monacanthidae Balistidae Ostraciidae Triodontidae Molidae Tetraodontidae Diodontidae Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 \>- A1a' ^#A1|3' lost^ AIY ^AIP" ^^^^1a lost Reversal ^^^^^ X (loss or fusion) ^^^t^ A1 a shifts origin ^^Aip shifts origin A1P'

FA1a&A1p (a)

Triacanthoids Balistoids Tetraodontoids 1 \ A1ab' Aipb'm A1ab A1cct A1ab" Aipb"m A1ab'" A1at A1at Aipt

TriacanthodidaeTriacanthidae Monacanthidae Balistidae Ostraciidae Triodontidae Molidae Tetraodontidae Diodontidae Aipb'm&Aipb"m 1ab',A1ab"&A1ctb' A1ot&A1pt

(b) FIGURE 4. Competing hypotheses of the evolution of Al muscles in tetraodontiform fishes, (a) Transfor- mation series (ACCTRAN) based on muscle names and homologies proposed by Winterbottom (1974b). This series requires a minimum of 10 steps (five gains, two shifts in the relative position of muscles, three reversals by loss or fusion), (b) Alternative transformation series (ACCTRAN and DELTAN) based on new muscle names and homologies described here. This series requires a minimum of five steps (all muscle duplications by sub- division). Note the parallel evolution of Ala and Aip in balistoids and tetraodontoids and Aip' and Alp" in ostraciids and monacanthids. ulae muscles suggests an alternative sce- with other Al muscles. In all balistoids, nario of evolution of Al muscles within Alab is the dorsalmost Al muscle and al- balistoids. We hypothesize that the ances- ways lies anterodorsal to the path of the tral Al condition for balistoids was not ramus mandibularis branch of the trigem- like that of balistids but was instead a par- inal nerve. In addition, new data on the rel- tially subdivided Al like that of ostraciids. ative masses of adductor mandibulae mus- We consider Winterbottom's (1974b) A1(T cles reveal that Alab is consistently the of ostraciids and Ala of balistids and largest Al muscle in all balistoids. monacanthids to be orthologs, which we In our scenario of Al evolution within rename Alab. This hypothesis of homol- balistoids, we hypothesize that Alab com- ogy is suggested by the relative position pletely separated from the ventral portion and mass of this Al muscle as compared of Al, AlfJb, and shifted more anterodor- 450 SYSTEMATIC BIOLOGY VOL. 46 sally to a superficial position in the com- positions of the Al muscles is reversed as mon ancestor of balistids and monacan- compared with balistoids. Ala (=Aipt) thids. We do not consider Winterbottom's originates superficially over the A2 mus- (1974b) Alp or Alp' of ostraciids and cles and inserts broadly on the distal end monacanthids to be orthologs. Instead, we of the maxilla. Alp (=Alat) originates in hypothesize these muscles have arisen in- front of the eye and inserts broadly on the dependently through separate subdivision proximal portion of the maxilla. events in both clades and thus have re- Winterbottom (1974b) did not explicitly Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 named them Aip'o and Aip"o in ostra- state whether the undivided Al of triodon- ciids and Aip'm and Aip"m in monacan- tids is a plesiomorphic condition as in tria- thids, respectively. canthoids or a derived condition (i.e., a re- Several additional duplication events versal by either loss of one muscle or and losses of muscles have produced a fusion of both). He did however explicitly complex pattern of Al muscles in mona- homologize Ala (=Alab) in balistoids canthids. It is unclear whether Alab', (Figs. 3e, 3f) and Ala (=Alpt) in tetra- Alab", and Alab'" have been produced by odontoids (Figs. 3b, 3d) (Winterbottom, a single subdivision event or by two se- 1974b:74). This interpretation of Al ho- quential subdivision events because all mologies implies that Al was subdivided three muscles appear simultaneously on into Ala (=Alab and Alpt) and Alp the phylogeny. Although other subdivision (=Aipb and Alat) in the common ances- events in tetraodontiforms have produced tor of balistoids and tetraodontoids. How- only two descendant muscles, it is more ever, Winterbottom (1974b:74) went on to parsimonious to assume that a single sub- state that Alp (=Aipb) of balistoids and division event of Alab produced these Alp (=Alat) of tetraodontoids were prob- three muscles. The absence of Alab' in ably not homologous. These statements Aluterus and of Aipb"m in Anacanthus is create a paradox because any subdivision interpreted as reversal by loss or fusion. event produces at least two pairs of or- Both genera are relatively derived mona- thologous muscles in descendant taxa. canthids and these muscles are present in There are only two alternatives; either both their sister taxa (Matsuura, 1979). Alab"a Ala and Alp are orthologs in balistoids and Alab"p in Oxymonacanthus and and tetraodontoids or neither of them are. Aipb"ma and Aipb"mp in Paraluteres have If one were to accept that Ala (=Alab) clearly arisen through two additional du- in balistoids and Ala (=Aipt) in tetra- plication events within these genera. This odontoids are orthologs, then Al must transformation series for Al muscles for have been subdivided in the common an- balistoids, using the new homologies and cestor of balistoids and tetraodontoids. nomenclature, is mapped on a phylogeny This hypothesis then requires the loss of a of tetraodontiforms in Figure 4b. muscle or fusion of both muscles to ex- plain the undivided Al in triodontids and Homologies of Al Muscles between Balistoids apparent absence of Ala (=Alab) in os- and Tetraodontoids traciids. Such a scenario for Al evolution, In triodontids, the sister group to all using Winterbottom's (1974b) names and other tetraodontoids, Al is an undivided homologies, is optimized (ACCTRAN) on superficial muscle that originates beneath a phylogeny of tetraodontiforms in Figure the eye and inserts broadly on the proxi- 4a. Based on this set of homologies, at least mal portion of the maxilla (Fig. 5a). In con- 10 steps are required (five gains of mus- trast, Al in molids, tetraodontids, and di- cles, two shifts of muscle origins, and three odontids is subdivided into a superficial reversals via loss or fusion of muscles) to ventral muscle, Ala (=Aipt), and a super- explain the origin of the Al muscles in the ficial dorsal muscle, Alp (=Alat) (Figs. common ancestors of each tetraodontiform 5b-e). These muscles are not covered by family. The only difference under DEL- A2 muscles, and the relative dorsoventral TRAN is that Alp' is gained independent- 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 451

A2oc A1 PAL MX PMX Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020

RMD

1 cm (e) FIGURE 5. Jaw musculature of representative tetraodontoids. See text for muscle and other element abbre- viations, (a) Triodon macropterus, superficial jaw musculature (adapted from original; Winterbottom, 1974b); Al is undivided and A2 is subdivided into A2a and A2|3. A23 passes medial to the path of the ramus mandibular- is. (b) Sphoeroides nephalus, superficial view; Al is subdivided into Alat and Aipt, AlfJt lies superficial to some A2 muscles, and A2 is subdivided into A2a, A2(3't, and A20"t. A23"t passes lateral to the path of the ramus mandibularis. (c) 5. nephalus, deep view with AlJ3t and A20't removed; A2a inserts on the maxilla, (d) Chilo- mycterus schoepfi, superficial view; Al is subdivided into a small Alat and a large Aipt. A13t lies superficial to some A2 muscles, (e) C. schoepfi, deep view with Alpt and part of the maxilla removed. The ramus man- dibularis passes entirely within A2f$ from the neurocranium to the lower jaw. 452 SYSTEMATIC BIOLOGY VOL. 46 ly in both monacanthids and ostraciids dia), and we followed Winterbottom's pro- rather than being gained in the common posal. Exceptions to this triacanthodid ancestor of balistoids and later lost in bal- pattern are the long-snouted genera, Mac- istids. rorhamphosodes and Halimochirurgus, which Alternatively, we hypothesize that the have an undivided A2 (Winterbottom, undivided Al of triodontids is the plesio- 1974b). The path of the RMD in these two morphic condition for tetraodontids be- genera is unknown. As in these specialized cause Al was undivided in the common triacanthodids, all triacanthids have an un- Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 ancestor of both balistoids and tetraodon- divided A2 muscle, which lies medial to toids. Therefore no duplicated Al muscles the path of the RMD (Fig. 3a). of tetraodontoids are orthologs of any du- In balistoids, A2 is always represented plicated Al muscles of baslistoids. This hy- by at least two muscles. In ostraciids, both pothesis is completely consistent with our a dorsal A2a and a ventral A2p are pres- scenario for Al evolution within balistids. ent, and both muscles pass lateral to the Subsequently, Al was subdivided in the path of the RMD (Fig. 3c). Plesiomor- common ancestor of molids, tetraodontids, phically, these muscles are completely sep- and diodontids to form Alat and Al(3t. arated in ostraciids but are fused posteri- Alat has remained in the relatively dorsal orly in one ostraciid subclade, the tribe position of Al in triodontids while Aipt Aracanini (Winterbottom and Tyler, 1983). has expanded ventrally over the A2 mus- In balistids, there are three A2 muscles in cles. A transformation series for Al evo- a dorsoventral sequence, A2a, A2(3 lution, using our nomenclature and ho- (=A2p'b), and A2T (=A2(3"b), respectively mologies, is optimized on a phylogeny of (Fig. 3e). In most balistids, A2a passes lat- tetraodontiforms in Figure 4b. Our scenar- eral to the path of the RMD, whereas both io requires only five steps (all gains by A2(3 (=A2p'b) and A2^ (=A2(Tb) are me- muscle duplication by subdivision) to ex- dial to the path of this nerve. Variation is plain the origin of these Al muscles in the found in the balistid Balistipus undulatus, common ancestors of each tetraodontiform where the RMD passes through A2a be- family. This transformation series is the fore passing lateral to A2£ (=A2(3'b) (Win- same under ACCTRAN or DELTRAN. terbottom, 1974b). In monacanthids, only two A2 muscles, A2a and A2(3, are present Homologies of Al Muscles in Balistoids and (Fig. 3g). In some genera (e.g., Chaetoderma, Tetraodontoids Monacanthus, Paramonacanthus, Paraluteres, As in other teleost fishes, A2 is the main Pervagor, Stephanolepis), the ventral portion section of the adductor mandibulae mus- of A2(3 may be slightly differentiated (Fig. culature, which plesiomorphically inserts 3g) but is never completely separated as is on the lower jaw. In tetraodontiforms, this A27 (=A2p"b) in balistids (Figs. 3e, 3f). In muscle may be undivided or may consist all monacanthids, A2a passes lateral and of up to three separate muscles. To identify A2p passes medial to the path of the RMD orthologous muscles in taxa with more (Fig- 3g). than a single A2 muscle, Winterbottom Like balistoids, all tetraodontoids also (1974b) proposed using their positions rel- have at least two A2 muscles. In triodon- ative to the path of the ramus mandibular- tids, A2 is represented by a dorsal A2a is of the trigeminal nerve to the lower jaw. and a ventral A2(3, and the RMD passes In most triacanthodids, A2 is represent- medial to A2a and lateral to A2fi much ed by two muscles, a posterodorsal A2a like it does in most balistids and all mon- and an anteroventral A2(3, which lies me- acanthids (Winterbottom, 1974b) (Fig. 5a). dial to the RMD (Winterbottom, 1974b). This triodontid A2 pattern differs however Gosline (1986) suggested that A2a in tria- from that of all other tetraodontids. In canthodids may in fact be a portion of Al. molids, tetraodontids, and diodontids, However, this is not the case in the tria- A2a (=A2p or A2p't and A2^"t) and A2(3 canthid examined in this study (Parahollar- (=A2a) are present but the relative origins 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 453 of these muscles have apparently shifted if terbottom's names and homologies, is op- A2p is identified as the A2 muscle medial timized (ACCTRAN) on a phylogeny of to the path of the RMD (Figs. 5b-e). In tetraodontiforms in Figure 6a. This scenar- these three families, A2a (=A20 or A2p't io requires a minimum of six steps (two and A2p"t) now lies ventral to A2p (=A2a) gains, one shift in the path of the RMD, and is in the same relative position as A2fi two shifts of muscle origins, and one re- in triodontids. versal by loss or fusion) to explain the or- In tetraodontids, A2a (=A2(3) has two igin of A2 muscles in the common ances- Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 distinct heads (=A2(3't and A2J3"t) (Win- tors of each tetraodontiform family. The terbottom, 1974b). The ventral head only difference under DELTRAN is that (=A2P"t) is present in all tetraodontid gen- subdivision of A2 into A2a and A2B has era examined here, but its separation from occurred independently in triacanthids the dorsal head (=A2£'t) may be partially and in the common ancestor of balistoids obscured in superficial view by Ala and tetraodontoids. (=Aipt). Furthermore, the insertion of We also consider A2 to have been sub- A2S (=A2a) has shifted from the lower divided in the common ancestor of all tet- jaw to the maxilla in the three species of raodontiforms and the undivided A2 in Sphoeroides examined. some triacanthodids and all triacanthids to The path of the RMD varies in tetra- be independent reversals. However, this odontids and diodontids and differs from hypothesis is tentative because under DEL- the clear pattern in balistids, monacan- TRAN it is equally parsimonious to as- thids, and triodontids. The RMD may start sume that the undivided A2 of triacan- medial to A2a (=A2p't and A2(Tt) but thids is plesiomorphic and the divided A2 passes through the anteroventral portion of most triacanthodids is independently (A2f$"t) just before it reaches the lower jaw derived from that of balistoids and tetra- (e.g., Arothron, Canthigaster, Chelonodon, odontoids. In either case, the A2 homolo- Sphoeroides) or it may travel entirely within gies we propose for balistoids and tetrao- A2a (=A2S) before reaching the lower jaw dontoids are unaffected. (e.g., Chibmycterus, Diodon). Our scenario of A2 evolution differs Based on the pattern of A2 muscles, from Winterbottom's (1974b) mainly be- Winterbottom (1974b) hypothesized that cause we hypothesize that A2a and A2B A2 was subdivided in the common ances- have retained their relative positions in all tor of all tetraodontiforms and the undi- tetraodontiforms while the path of the vided A2 of some triacanthoids had arisen RMD has shifted at least twice. Based on through two separate character reversals our observations, variation in the path of by loss or fusion, once in triacanthids and the RMD between species is greater than once in the common ancestor of Macro- originally suggested by Winterbottom. rhamphosodes and Halimochirurgus. Winter-Thus, we consider the path of the RMD an bottom used the path of the RMD to dis- unreliable landmark for identifying de- tinguish A2a and A2(3 in most balistoids scendant A2 muscles. Use of this criterion and all tetraodontoids. This hypothesis re- leads to the misidentification of the orthol- quires at least one shift of the RMD in os- ogous A2 muscles in molids, tetraodon- traciids and a relative shift of the origins tids, and diodontids. of both A2 muscles in the ancestor of mo- A transformation series for the evolution lids, triodontids, and diodontids to main- of A2 muscles based on our names and ho- tain these homologies. A separate A2-y mologies is optimized (ACCTRAN) on a (=A2B"b) has arisen in the lineage leading phylogeny of tetraodontiforms in Figure to balistids by subdivision of A2B. Winter- 6b. As with Winterbottom's (1974b) ho- bottom however did not provide a separate mologies, the only difference under DEL- name for an analogous subdivision of A2fJ, TRAN is that the subdivision of A2 into which occurs in tetraodontids. A transfor- A2a and A2(3 has occurred independently mation series for A2 muscles, using Win- in triacanthids and the common ancestor 454 SYSTEMATIC BIOLOGY VOL. 46

Triacanthoids Balistoids Tetraodontoids

A2p A2P A2a A2a

TriacanthodidaeTriacanthidae Monacanthidae Balistidae Ostraciidae Triodontidae MdkJae Tetraodonttdao Diodontidae Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 Reversal (loss or fusion) A2a shifts origin A2p shifts origin

(a) A2a&A2P

Triacanthoids Balistoids Tetraodontoids

TriacanthodidaeTriacanthidae Monacanthidae Balistidae Ostraciidae Triodontidae Molidae Tetraodontidae Diodontidae Reversal ^^^^P ** sion) ^S^^" mandibularis shifts ^ ^ Path of ramus mandibularis shifts

(b) A2O&A2P FIGURE 6. Competing hypotheses of the evolution of A2 muscles in tetraodontiform fishes, (a) Transfor- mation series (ACCTRAN) based on muscle names and homologies proposed by Winterbottom (1974b). This series requires a minimum of six steps (two gains, one shift of the ramus mandibularis, two shifts in the relative positions of muscles, one reversal by loss or fusion), (b) Alternative transformation series (ACCTRAN) based on new muscle names and homologies described here. This series requires a minimum of six steps (three duplications by muscle subdivision, two shifts in the path of the ramus mandibularis, one reversal by loss or fusion). This series is more parsimonious than that based on the homologies proposed by Winterbottom (1974b) because it explains the origin of additional muscles not previously recognized in tetraodontids.

of balistoids and tetraodontoids. We hy- path of the RMD shifted to pass medial to pothesize that the ancestor of balistoids both A2a and A2p. Triodontids retain the and tetraodontoids had a dorsal A2a and relatively plesiomorphic A2 pattern for tet- ventral A2(3 with the RMD passing lateral raodontiforms. This pattern was modified to A2a. In the lineage leading to balistids in the common ancestor of molids, tetra- and monacanthids, A2|3 had a slightly dif- odontids, and diodontids when the path ferentiated ventral section. This ventral of the RMD again shifted to pass medial section completely subdivided to form to A2p. In the tetraodontid lineage, A2p A2p'b and A2p"b in the common ancestor independently subdivided to form A2p't of balistids. In the ostraciid lineage, the and A2(3"t. In total, our scenario requires 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 455 six steps (three gains, two shifts in the 1.00 path of the RMD, and one reversal by loss or fusion) to explain the origin of A2 mus- cles in the common ancestors of each tetra- odontiform family and includes a step for the origin of A20't and A2p"t in tetraodon- tids, which were not recognized as sepa- rate muscles by Winterbottom. Thus, the Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 scenario based on our new homologies is more parsimonious than that based on Winterbottom's homologies. Relative Masses of Al and Al Muscles Based on an ontogenetic series of 10 -1.75 burrfish, Chilomycterus schoepfi (41-166 mm 0.75 1.25 1.75 2.25 2.75 3.25 standard length), most Al and A2 muscles log(body mass) showed slight positive allometry (slopes of least squares regressions of log muscle 0.80 mass on log body mass: Alat = 0.936; Al(3t = 1.126; A2a = 1.169; A2{* = 1.126). Only Alat differs from other Al and A2 muscles by having slight negative allome- try. This muscle is extremely atrophied in all diodontids examined (e.g., Figs. 5d, 5e). Interspecifically, the Al and A2 muscles of 22 other tetraodontiform species also showed slight positive allometry. An ex- ample of this relationship is shown in Fig- O ure 7a, where the log of the mass of all Al muscles is plotted against the log of body mass for 10 C. schoepfi along with single 0.00 representatives of 22 other species. A least -0.10 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 squares regression fitted to the log-trans- Contrast in log(body mass) formed data of all species has a slope of FIGURE 7. Slight positive allometry of Al muscle 2 mass for most tetraodontiform fishes, (a) Plot of 1.04, a y-intercept of -2.51, and an r of log(Al mass) and log(body mass) for 10 individuals 0.80. The only clear outliers to this trend of Chilomycterus schoepfi (x) plus single individuals of are the ostraciids, Acanthostracion polygoni-22 tetraodontiform species (•). See text for species ab- us and A. quadricornis, and one triacanthid, breviations, (b) Plot of contrasts in log(Al mass) and Trixiphichthys xveberi. Standardized inde- contrasts in log(body mass) for 19 independent con- trasts. A line fitted to these contrasts and forced pendent contrasts of Al mass and body through the origin has a slope of 1.06 and an r2 of mass also show a strong positive correla- 0.89. tion (slope = 1.06; r2 = 0.89) (Fig. 7b). The percentage of adductor muscle mass comprised by the Al muscle varies consid- greatest intrafamilial variation in tetra- erably among tetraodontiforms, ranging odontids and balistids. from 23% in one tetraodontid, Sphoeroides A pattern does emerge when the resid- nephalus, to 61% in one balistid, Xanthich- uals of log Al mass and log A2 mass re- thys ringens. Thus, there is considerable gressed against log body mass were plot- variation in the amount of muscle tissue ted against each other (Fig. 8a). There was devoted to moving the upper or lower a strong tendency for species with relative- jaws. There is also considerable overlap in ly large Al muscles to also have relatively variation between most families, with the large A2 muscles and for species with rel- 456 SYSTEMATIC BIOLOGY VOL. 46

0.50 - mass (slope = 1.12; r2 = 0.55) (Fig. 8b). 0.40 - (a) #CR Those taxa that have relatively large ad- 0.30 - CPA XR. AM BV pHY.ST SN ductor muscles (most tetraodontids, di- 0.20 - odontids, and Balistes vetula) are duropha- 0.10 - TB* * TH CA* - BC 0.00 - gous and feed on prey such as molluscs, AS* -0.10 - # # echinoderms, and crabs (Randall, 1967). CPU CM -0.20 - *DHO Species with relatively small Al and A2 Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 -0.30 - LL muscles (ostraciids, several monacanthids, -0.40 - one triacanthid) tend to be grazers that • -0.50 - TW feed on algae and relatively soft-bodied in- -0.60 - vertebrates (Randall, 1967). Morphological -0.70 - outliers tend to be trophic outliers as well. -0.80 - •AP #AQ For example, the tetraodontid Lagocephalus -0.90 - is a pelagic puffer that feeds largely on soft -1.00 - -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 prey such as squid (Wheeler, 1969). The Residual of log(A2 mass) triggerfish Xanthichthys is a zooplanktivore 0.40 and has a relatively small A2 muscle (Tur- (b) ingan, 1994). • • No relationship was seen between the 0.30 number of adductor mandibulae muscles possessed by a species and the relative size O 0.20 of the adductor mandibulae complex (Fig. 9a). Many taxa with only four or five Al g 0.10 and A2 muscles, such as balistids, tetra- o odontids, or diodontids, have much larger u K % • residuals than some monacanthids, with "5 o.oo as many as seven Al and A2 muscles. This • lack of correlation is also supported when -0.10 - residuals (standardized contrasts of log[Al and A2 mass] regressed against standard- -0.20 ized contrasts in log[body mass]) are plot- -0.15 -0.10 -0.05 0.00 0.05 0.10 0.15 0.20 ted against standardized contrast in the Residual of contrast in log(A2 mass) number of Al and A2 muscles (slope = 2 FIGURE 8. Positive correlation between Al mass 0.03; r = 0.01) (Fig. 9b). and A2 mass in tetraodontiform fishes, (a) Plot of re- The relative contributions of individual siduals of log(total Al mass) and the residuals of Al descendant muscles to total Al mass log(total A2 mass) both regressed against log(body mass) for single individuals of 23 tetraodontiform spe- for eight representative species are shown cies. An individual located at the intersection of both in Figure 10a. In all balistoids examined axes would represent a fish with average Al and A2 (ostraciids, balistids, monacanthids), Alab masses for all tetraodontiforms of that body size. See is the largest Al descendant muscle, and text for species abbreviations, (b) Plot of residuals of this relationship is maintained if its de- contrasts in log(Al mass) and log(A2 mass) both re- gressed against contrast in log(body mass) for 19 in- scendant muscles, e.g., Alab', Alab", and dependent contrasts. A line fitted to these contrasts Alab'" in monacanthids such as Monacan- and forced through the origin has a slope of 1.12 and thus, are pooled together. This observation an r2 of 0.55. that Alab is the largest Al muscle in bal- istoids is also consistent with our hypoth- esis that Winterbottom's (1974b) Alp" of atively small Al muscles to have relatively ostraciids (e.g., Acanthostracion) is the or- small A2 muscles. This positive correlation tholog to Alab of balistids and monacan- is supported by residuals of standardized thids. contrasts for Al and A2 mass regressed Unlike balistoids, the relative masses of against standardized contrasts of body Al descendant muscles in tetraodontoids 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 457

0.50- toids. At one extreme is the tetraodontid (a) SN« Lagocephalus with a large Alat and a small 0.30- CR»ST Aipt, and at the other extreme are diodon- DHY»CS «CPA tids such as Chilomycterus, which have a 0.10- minute Alat and an enormous AlfJt. Oth- AM MH . XR! # er tetraodontids examined in this study •TB DHO BC •CM have relatively equal-size Al descendant -0.10 - Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 muscles such as those of Sphoeroides (Fig. •AS •CPU •LL 10a). -0.30- The relative contributions of individual •TW A2 descendant muscles to total A2 mass -0.50- •AQ for the same eight species are shown in •AP Figure 10b. In most balistoids and tetra- -0.70- odontoids, A2p is larger than A2a. This 2 3 4 5 6 7 pattern is reversed in all monacanthids Number of A1 & A2 muscles such as Monacanthus and in a single balis- tid, Xanthichthys, where A2a is larger than ^ 0.25 A2(3. This reversal is more likely due to a s reduction in A2(3 rather than to an increase in A2a because these same taxa have neg- ative A2 residuals as compared with other tetraodontiforms (Fig. 8). One other nota- ble difference is the relatively small A2a and large A2p't and A2f*"t of Sphoeroides as compared with other tetraodontids. This pattern may be associated with the fact that A2a uniquely inserts on the upper jaw in this genus.

DISCUSSION

-0.15 The adductor mandibulae system of tet- -0.80 -0.40 0.00 0.40 0.80 1.20 raodontiform fishes provides a clear ex- Contrast in number of A1 & A2 muscles ample of phylogenetic repetition of mor- FIGURE 9. No correlation between muscle mass phological elements. This system is well and number of jaw muscles in tetraodontiform fishes, (a) Plot of residuals of log(Al mass + A2 mass) re- suited to an analysis of the functional con- gressed against log(body mass) and total number of sequences of structural duplication for sev- Al and A2 muscles for single individuals of 23 tetra- eral key reasons. First, there is a well-cor- odontiform species. Data points for Diodon hystrix roborated phylogeny of the major lineages (DHY) and Chilomycterus schoepfi (CS) overlap on thisof tetraodontiform fishes. Second, the new plot. See text for other species abbreviations, (b) Plot of residuals of contrasts in log(Al mass + A2 mass) muscle nomenclature presented here will regressed against contrasts in log(body mass) and to- permit specific comparisons to be made tal number of Al and A2 muscles for 19 independent between the functional attributes of an an- contrasts. A line fitted to these contrasts and forced 2 cestral muscle and the descendants that re- through the origin has a slope of 0.025 and an r of sult from a splitting event. Third, including 0.01. muscle evolution within tetraodontiform families, we recognize a total of 10 cases examined (tetraodontids, diodontids) vary of muscle subdivision and 4 cases of re- considerably between taxa, which is con- versals by either loss or fusion of muscle sistent with our hypothesis that the Al de- subdivisions in this system. This large scendant muscles of tetraodontoids have number of independent changes in muscle arisen independently in this clade and are number makes it possible to repeatedly not orthologs of any Al muscles of balis- test specific hypotheses about the conse- 458 SYSTEMATIC BIOLOGY VOL. 46

Triacanthus biaculeatus Xanthichthys ringens Acanthostracion quadhcomis Sphoeroides nephalus Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 Triacanthidae Balistidae Ostraciidae Tetraodontidae

Batistes caphscus Monacanthus hispidus Lagocephalus lagocephalus Chilomycterus schoepfi Balistidae Monacanthidae Tetraodontidae Diodontide

(b)

Triacanthus biaculeatus Xanthichthys ringens Acanthostracion quadricomis Sphoeroides nephalus Triacanthidae Balistidae Ostraciidae Tetraodontidae

Batistes capriscus Monacanthus hispidus Lagocephalus lagocephalus Chilomycterus schoepfi Balistidae Monacanthidae Tetraodontidae Diodontidae FIGURE 10. Pie charts of muscle masses for individuals of eight species from six tetraodontiform families, (a) Al muscles. All descendant Ala muscles in balistoids are shaded the same. The Ala and AlfJ muscles have arisen independently in both balistoids and tetraodontoids. (b) A2 muscles. The A2(3' and A2fJ" muscles have arisen independently in both balistids and tetraodontids. quences of subdivision. Because muscle flect new hypotheses of homology. Trans- subdivision is not a phylogenetically or formation series based on these new ontogenetically unique event, it is possible homologies are more parsimonious for to generate statistically useful sample both Al and A2 muscles than are those sizes, thus addressing a problem that has based on homologies proposed by Winter- plagued comparative studies (Garland et bottom (1974b). al., 1992). We hypothesize that a minimum of 10 separate duplication events have produced Muscle Homologies the diversity of Al and A2 muscles in This review of the Al and A2 muscles these fishes. This review revealed two of tetraodontiforms revealed that several trends in the evolution of jaw muscles in muscles currently identified by the same tetraodontiforms. First, the origins of Al name in different subclades are not or- and A2 muscles and the relative positions thologs and thus require new names to re- of descendant muscles to each other are 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 459 conservative features in this clade. For idae) (Bellwood and Choat, 1990; Bell- these fishes, position is a more reliable cri- wood, 1994). Loricarioids and scarids terion for identifying muscles than is the could potentially serve as other model path of the RMD, which has been used in cases of duplication because phylogenies other studies (Winterbottom, 1974b; Gos- are available for both groups (Howes, 1983; line, 1986). The path of this nerve is a re- Schaefer, 1987; Bellwood, 1994). Complex liable landmark for Al muscles in balis- patterns of duplicated jaw muscles occur toids but not for A2 muscles in in clades of fishes that have highly derived Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 tetraodontoids. oral jaw morphologies and novel feeding Second, duplication of muscles by sub- modes. Most bony fishes are ram or suc- division is more common whereas second- tion feeders and use their oral jaws for ary loss of duplicated muscles is less com- prey capture but not processing (Liem, mon than previous studies have suggested. 1980a; Lauder, 1985). In contrast, tetra- A similar trend in phylogenetic increases odontiform fishes and parrotfishes use in muscle number was reported by Gosline their robust oral jaws to both capture and (1986), who surveyed higher groups of te- process food items, which are often quite leost fishes and found mat muscles pro- hard (e.g., molluscs, corals) (Bellwood and duced by primary subdivision of the ad- Choat, 1990; Turingan and Wainwright, ductor mandibulae were seldom 1993; Wainwright and Turingan, 1993; secondarily lost. Some of the differences in Turingan, 1994). In loricarioid catfishes, the muscle gain versus loss ratio for the the oral jaws are highly modified to scrape transformations series examined in this food items such as algae off objects or to study are expected because homoplasy rasp plant material (Schaefer and Lauder, that was originally due to secondary losses 1986, 1996). of "homologous" muscles is now ex- Our recognition of homologies different plained instead by parallel gains of "non- from those of Winterbottom (1974b) is due homologous" muscles through additional to two main factors. First, our homologies subdivision events. We recognize at least are based on an explicit model of muscle three such cases of parallel muscle evolu- evolution by subdivision. This model pro- tion: Alab and Al|3b of balistoids versus duces paralogs, which we distinguished Alat and Alpt of tetraodontoids, AlfJb'o from their common ancestral muscle and and Al(3b"o of ostraciids versus Aipb'm each other. Although Winterbottom also and AlfJb"m of monacanthids, and A2|3'b suggested that new muscles are produced and A2|3"b of balistids versus A2|3't and by subdivision, he often retained an ances- A2B"t of tetraodontids. tral muscle name for the one descendant Duplication of jaw muscles by subdivi- muscle that resembled the ancestral mus- sion has undoubtedly occurred several cle in outgroup taxa. This practice ulti- times within teleost fishes. According to mately obscured the homologies between Gosline (1989) "Al" and "A2" muscles these muscles. have arisen once in ostariophysan fishes Second, Winterbottom's (1974b) analysis (Cypriniformes, Characiformes, Silurifor- predates the widespread availability of mes, Gymnotiformes) and again indepe- computer programs for analyzing data and nently in higher teleost fishes (Beryciformes, investigating alternative optimizations of Zeiformes, Perriformes, Scorpaeniformes, characters. As Winterbottom stated (1974b: Tetraodontiformes). Other more complex 73), his analysis methods precluded the adductor mandibulae subdivisions similar possibility of any explanations based on to those described here for tetraodonti- parallel or convergent evolution. Although forms have occurred in loricarioid catfishes the topology of his "hand generated" phy- (Trichomycteridae, Nematogenyidae, Cal- logeny is identical to one produced using lichthyidae, Scoloplacidae, Astroblepidae, a computer algorithm and the same data, Loricariidae) (Howes, 1983; Schaefer and it lacked explicit character transformations Lauder, 1986, 1996) and parrotfishes (Scar- such as those investigated in this study. A 460 SYSTEMATIC BIOLOGY VOL. 46 posteriori analysis of the transformations mass that clearly varies with trophic habits series for homoplastic characters can some- (Fig. 8a). Constructional constraints will times reveal alternative and possibly more undoubtedly place some upper boundary parsimoniously interpreted homologies. on the size of the adductor mandibulae Although these new homologies may not musculature, but the flexibility that is in- change the phylogeny itself, they may rad- dicated by ecomorphological patterns sug- ically affect any hypotheses based on the gests that increases in the number of ad- refuted homologies. ductor muscles could theoretically Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 influence the amount of muscular tissue. Masses of Al and Al Muscles Although there have been numerous A correlation between adductor muscle muscle duplications in this system, most mass and diet has been previously report- descendant muscles retain their primitive ed for other fishes. Bellwood and Choat attachment to a particular jaw element (i.e., (1990) found that parrotfishes that bite and Al muscles insert on the maxilla, and A2 excavate pieces of hard coral skeletons muscles insert on the mandible). Only two while feeding had relatively larger jaw examples were found where muscle inser- bones and greater total adductor mandibu- tions had changed to different bony ele- lae mass than did parrotfishes that only ments. In monacanthids, the insertion of scrape the algae-encrusted surfaces of old Alab'" has shifted from the maxilla to the coral skeletons while feeding. Similarly, palatine bone (Fig. 3g) (Winterbottom, Turingan (1994) found that durophagous 1974b). In tetraodontids of the genus tetraodontiforms such as balistids, tetra- Sphoeroides, A2a insertion has shifted from odontids, and diodontids had relatively the mandible to the maxilla (Fig. 5c). In massive A2 muscles, jaws, and suspension monacanthids, this shift is associated with elements as compared with a single graz- a subdivision event, whereas in Sphoeroides, ing monacanthid examined. However, he the shift clearly occurred after the subdi- did not find any significant differences in vision event. Al mass between taxa, as we have found Currently, we are further exploring the here with the larger number of nondu- analogy drawn in this study between du- rophagous taxa (i.e., triacanthids, ostra- plications of genes and duplications of ciids, and several monacanthids) exam- morphological elements. A major para- ined. Although some monacanthids had digm in molecular evolution is that gene Al residuals similar to those of balistids, duplication leads to divergence in nucleo- tetraodontids, and diodontids, most did tide sequence and ultimately divergence in not. the function of paralogous genes (Ohno, Our sample of 23 tetraodontiform spe- 1970; Ohta, 1989; Chothia, 1994; Holland cies from six families shows no support for et al., 1994; Holland and Garcia-Fernandez, the idea that overall adductor mass will in- 1996). Thus, is duplication of muscles also crease as new muscles evolve (Figs. 9a, 9b). associated with divergence in function, During phylogenetic increases in the num- and if so, are there general patterns in ber of muscles, the existing muscle tissue muscle evolution? To address these ques- apparently is subdivided not only in a tions, we are using electromyography to qualitative sense, as our model states, but quantify one aspect of muscle function, the also in a quantitative sense. One interpre- individual activity patterns of muscles tation of this pattern could be that con- during feeding events. Because paralogous structional constraints (Barel, 1983) on the muscles are derived from the same ances- space available to adductor muscle tissue tral muscle, they should have the same ple- limit increases in total adductor muscle siomorphic function (i.e., activity pattern) mass. However, this factor is probably not unless functional divergence has occurred. limiting the ability of the adductor muscle If one or more of the duplicated muscles to change in overall size, as indicated by examined in this study have functionally the existing diversity in overall adductor diverged, we should find differences in the 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 461 relative timing (i.e., onset and duration) GOSLINE, W. A. 1989. Two patterns of differentiation and/or intensity of muscular activity be- in the jaw musculature of teleostean fishes. J. Zool. Lond. 218:649-661. tween paralogous muscles. Such differ- HILLIS, D. M. 1994. Homology in molecular biology. ences in activity patterns should have Pages 339-368 in Homology: The hierarchical basis functional consequences for the feeding of comparative biology (B. K. Hall, ed.). Academic mechanisms in these fishes. Press, San Diego. HOLLAND, P. W. H., AND J. GARCIA-FERNANDEZ. 1996.

ACKNOWLEDGMENTS Hox genes and chordate evolution. Dev. Biol. 173: Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 382-395. We thank J. Grubich, K. Ralston, and T. Thys for HOLLAND, P. W. H., N. A. WILLIAMS, AND A. SIDOW. thoughtful comments on a draft of this manuscript 1994. Gene duplication and the origins of vertebrate and R. Winterbottom for his feedback on several as- development. Development (Suppl.) 1994:125-133. pects of this research. We also thank R. Turingan for HOWES, G. J. 1983. The cranial muscles of loricarioid providing some of the specimens used in this study catfishes, their homologies and value as taxonomic and M. Westneat for the loan of specimens from the characters (Teleostei: Siluroidei). Bull. Br. Mus. Nat. Field Museum of Natural History. Support for this re- Hist. 45:309-345. search was provided by NSF grant IBN 9306672 to P. LAUDER, G. V. 1981. Form and function: Structural C. Wainwright. analysis in evolutionary morphology. Paleobiology 7:430-^142. REFERENCES LAUDER, G. V. 1985. Aquatic feeding in lower verte- ADRIAENS, D., AND W. VERRAES. 1996. Ontogeny of brates. Pages 210-229 in Functional vertebrate mor- cranial musculature in Clarias gariepinus (Siluroidei: phology (M. Hildebrand, D. M. Bramble, K. F. Liem, Clariidae): The adductor mandibulae complex. J. and D. B. Wake, eds.). Harvard Univ. Press, Cam- Morphol. 229:255-269. bridge, Massachusetts. BAREL, C. D. N. 1983. Towards a constructional mor- LAUDER, G. V. 1990. Functional morphology and sys- phology of the cichlid fishes (Teleostei, Perciformes). tematics: Studying functional morphology in a his- Neth. J. Zool. 33:357-^24. torical context. Annu. Rev. Ecol. Syst. 21:317-340. BATESON, W. 1892. On the numerical variation in LAUDER, G. V, AND K. F. LIEM. 1983. The evolution teeth, with a discussion of the concept of homology. and interrelationships of the actinopterygian fishes. Proc. Zool. Soc. Lond. 1892:102-115. Bull. Mus. Comp. Zool. 150:95-197. BATESON, W. 1894. Materials for the study of variation LAUDER, G. V., AND K. F. LIEM. 1989. The role of his- treated with especial regard to discontinuity in the torical factors in the evolution of complex organis- origin of species. Macmillan, New York. mal functions. Pages 63-78 in Complex organismal BELLWOOD, D. R. 1994. A phylogenetic study of the functions: Integration and evolution of vertebrates parrotfishes family Scaridae (Pisces: Labroidei), (D. B. Wake and G. Roth, eds.). Wiley and Sons, with a revision of genera. Rec. Aust. Mus. Suppl. New York. 20:1-86. LIEM, K. F. 1980a. Acquisition of energy by teleosts: BELLWOOD, D. R., AND R. CHOAT. 1990. A functional Adaptive mechanisms and evolutionary patterns. analysis of grazing in parrotfishes (family Scaridae): Pages 299-334 in Environmental physiology of fish- The ecological implications. Environ. Biol. Fishes 28: es (M. A. Ali, ed.). Plenum, New York. 189-214. LIEM, K. F. 1980b. Adaptive significance of intra- and CHOTHIA, C. 1994. Protein families in the metazoan interspecific differences in the feeding repertoires of genome. Development (Suppl.) 1994:27-33. cichlid fishes. Am. Zool. 20:295-314. DANFORTH, C. H. 1930. Numerical variation and ho- LIEM, K. F., AND D. B. WAKE. 1985. Morphology: Cur- mologies in vertebrae. Am. J. Phys. Anthropol. 14: rent approaches and concepts. Pages 366-377 in 463^81. Functional vertebrate morphology (M. Hildebrand, DOYLE, J. J. 1992. Gene trees and species trees: Mo- D. M. Bramble, K. F. Liem, D. B. Wake, eds.). Har- lecular systematics as one-character taxonomy. Syst. vard Univ. Press, Cambridge, Massachusetts. Bot. 17:144-163. EMERSON, S. 1988. Testing for historical patterns of MABEE, P. M. 1993. Phylogenetic interpretations of change: A case study with frog pectoral girdles. Pa- ontogenetic change: Sorting out the actual and ar- leobiology 14:174-186. tefactual in an empirical case study of centrarchid FELSENSTEIN, J. 1985. Phytogenies and the compara- fishes. Zool. J. Linn. Soc. 107:175-291. tive method. Am. Nat. 125:1-15. MADDISON, W. P., AND D. R. MADDISON. 1992. FITCH, W. M. 1970. Distinguishing homologous from MacClade: Analysis of phylogeny and character analogous proteins. Syst. Zool. 19:99-113. evolution, version 3.05. Sinauer, Sunderland, Mas- GARLAND, T., JR., P. H. HARVEY, AND A. R. IVES. 1992. sachusetts. Procedures for the analysis of comparative data us- MATSUURA, K. 1979. Phylogeny of the superfamily ing phylogenetically independent contrasts. Syst. Balistoidea (Pisces: Tetraodontiformes). Mem. Fac. Biol. 41:18-32. Fish. Hokkaido Univ. 26:49-169. GOSLINE, W. A. 1986. Jaw muscle configuration in MCKITRICK, M. C. 1994. On homology and the on- some higher teleostean fishes. Copeia 1986:705-713. tological relationship of parts. Syst. Biol. 43:1-10. 462 SYSTEMATIC BIOLOGY VOL. 46

NELSON, J. S. 1994. Fishes of the world, 3rd edition. ships among Caribbean tetraodontiform fishes. J. Wiley and Sons, New York. Zool. Lond. 233:493-521. OHNO, S. 1970. Evolution by gene duplication. TURINGAN, R. G., AND P. C. WAINWRIGHT. 1993. Mor- Springer-Verlag, Heidelberg, Germany. phological and functional bases of durophagy in the OHTA, T. 1989. Role of gene duplication in evolution. queen triggerfish, Batistes vetula (Pisces, Tetraodon- Genome 31:304-310. tiformes). J. Morphol. 215:101-118. PATTERSON, C. 1982. Morphological characters and TYLER, J. C. 1980. Osteology, phylogeny, and higher homology. Pages 21-74 in Problems of phylogenetic classification of the fishes of the order Plectognathi reconstruction (K. A. Joysey and A. E. Friday, eds.). (Tetraodontiformes). NOAA Tech. Rep., NMFS Circ. Academic Press, New York. 434:1-422. Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 PATTERSON, C. 1987. Introduction. Pages 1-22 in Mor- VAN VALEN, L. M. 1982. Homology and causes. J. phology and molecules in evolution: Conflict or Morphol. 173:305-312. compromise (C. Patterson, ed.). Cambridge Univ. VERMEIJ, G. 1973a. Adaptation, versatility and evolu- Press, New York. tion. Syst. Zool. 22:466-477. PATTERSON, C. 1988. Homology in classical and mo- VERMEIJ, G. 1973b. Biological versatility and earth lecular biology. Mol. Biol. Evol. 5:603-625. history. Proc. Natl. Acad. Sci. USA 70:1936-1938. POWELL, P., R. R. ROY, P. KANIM, M. A. BELLO, AND V. WAGNER, G. P. 1989a. The biological homology con- EDGERTON. 1984. Predictability of skeletal muscle cept. Annu. Rev. Ecol. Syst. 20:51-69. tension from architectural determinations in guinea WAGNER, G. P. 1989b. The origin of morphological pig hind limbs. J. Appl. Physiol. 57:1715-1721. characters and the biological basis of homology. PURVIS, A., AND A. RAMBAUT. 1995. Comparative Evolution 43:1157-1171. analysis by independent contrasts (CAIC): An Ap- WAINWRIGHT, P. C, AND R. G. TURINGAN. 1993. Cou- ple Macintosh application for analysing compara- pled versus uncoupled functional systems: Motor tive data. Comp. Appl. Biosci. 11:247-251. plasticity in the queen triggerfish Balistes vetula. J. RANDALL, J. E. 1967. Food habits of reef fishes of the Exp. Biol. 180:209-227. West Indes. Stud. Trop. Oceanogr. 5:655-847. WHEELER, A. 1969. The fishes of the British Isles & ROTH, V. L. 1984. On homology. Biol. J. Linn. Soc. 22: N. W. Europe. Michigan State Univ. Press, East Lan- 13-29. sing. SCHAEFER, S. A. 1987. Osteology of Hypostomus ple- WINTERBOTTOM, R. 1974a. A descriptive synonymy of costomus (Linnaeus), with a phylogenetic analysis of the striated muscles of the Teleostei. Proc. Acad. the loricariid subfamilies (Pisces: Siluroidei). Con- Nat. Sci. Phila. 125:225-317. trib. Sci. (Los Angel.) 394:1-31. WINTERBOTTOM, R. 1974b. The familial phylogeny of SCHAEFER, S. A., AND G. V. LAUDER. 1986. Historical the Tetraodontiformes (Acanthopterygii: Pisces) as transformation of functional design: Evolutionary evidenced by their comparative myology. Smithson. morphology of the feeding mechanisms of loricari- Contrib. Zool. 155:1-201. oid catfishes. Syst. Zool. 35:489-508. WINTERBOTTOM, R., AND J. C. TYLER. 1983. Phyloge- SCHAEFER, S. A., AND G. V. LAUDER. 1996. Testing his- netic relationships of aracanin genera of boxfishes torical hypotheses of morphological change: Bio- (Ostraciidae: Tetraodontiformes). Copeia 1983:902- mechanical decoupling in loricarioid catfishes. Evo- 917. lution 50:1661-1675. Received 16 July 1996; accepted 20 February 1997 TURINGAN, R. G. 1994. Ecomorphological relation- Associate Editor: Paula Mabee 1997 FRIEL AND WAINWRIGHT—HOMOLOGIES OF TETRAODONTIFORM JAW MUSCLES 463

APPENDIX. Correspondence between the jaw mus- cle names used by Winterbottom (1974b) and the new names used in this study of tetraodontiform fishes.

Winterbot- Family tom (1974b) This study Triacanthodidae Al Al A2a A2a

A2p A2P Downloaded from https://academic.oup.com/sysbio/article-abstract/46/3/441/1651360 by guest on 01 June 2020 Triacanthidae Al Al A2 A2 Ostraciidae Alp Alpb'o Aip' Aipb"o Aip" Alab A2a A2a A2p A2p Balistidae Ala Alab Aip Alpb A2a A2a A2p A2p'b A27 A2p"b Monacanthidae Ala Alab' Ala' Alab" or Alab"aa Ala" Alab"p Al-y Alab'" Alp Alpb'm Alp' Aipb"m or Aipb"ma< Aip" Aipb"mp A2a A2a A2p A2p Triodontidae Al Al A2a A2a A2P A2P Molidae Ala Alpt Aip Alat A2a A2p A2p A2a Tetraodontidae Ala Alpt Alp Alat A2a A2p't and A2p"t A2p A2a Diodontidae Ala Alpt Aip Alat A2a A2p A2p A2a * Muscle duplications unique to some genera within this family.