130 



R.D. MOOI AND AC. GILL 



III 



rPM .Dl 



Fig. 8 An unusual Type 1 epaxial morphology in Normanichthys crockeri (USNM 176507, 63.4 mm SL). I - portion of the epaxial that 

 inserts on to the anterior pterygiophores largely separate from the main body of the epaxial, with only a few fibres shared from each 

 myoseptal section. The exceptions are the insertions on the two anteriormost pterygiophores which have many of their fibres originating 

 from the main epaxial muscle body. II - portion inserting on to pterygiophores that is not separate from the main epaxial body. Ill - 

 portion inserting on to the ptergygiophores bearing segmented rays, is mostly separate until just beyond the last ray where it merges with 

 the rest of the epaxial musculature. RPT, rayless pterygiophore; other abbreviations and methods of presentation as in Figs 1, 3. Scale bar 

 = 5 mm. 



al., 1989: 482) have regarded Hypopterus as a valid, mono- 

 typic genus. We provisionally retain the Centropomidae 

 (Centropomus only) until its relationships are better under- 

 stood. 



The Trachinoidei as defined by Pietsch & Zabetian (1990) 

 exhibit a variety of epaxial morphologies (Table 1). 

 Ammodytids and chiasmodontids have Type 0, champsodon- 

 tids and cheimarrichthyids have Type 1 , and Type 2 is found 

 in the creediids, percophidids, pinguipedids and ura- 

 noscopids. Considering the discussion by Johnson (1993: 

 13-15), this epaxial character distribution casts further doubt 

 on the integrity of this suborder as currently constituted. 

 Although it seems likely that the epaxial morphologies as 

 defined here have evolved more than once among acantho- 

 morphs, it is difficult to reconcile their distribution with the 

 phylogeny provided by Pietsch & Zabetian (1990). One of 

 their phylogenetic hypotheses is a sister group relationship 

 between the Chiasmodontidae and the Champsodontidae. 

 The Chiasmodontidae do not exhibit any muscle insertions on 

 the dorsal-fin pterygiophores, whereas the Champsodontidae 

 have a Type 1 condition very similar to that of scorpaenoids 

 and serranids. Ammodytids, considered a derived trachinoid 

 group, exhibit the primitive Type condition, while a puta- 

 tive basal taxon, Cheimarrichthys , has Type 1, usually a 

 derived morphology. Reversals are possible and structural 

 homologies are uncertain (as discussed below), but the incon- 

 sistencies among these taxa suggest a more thorough investi- 

 gation of the composition of the Trachinoidei sensu Pietsch & 

 Zabetian (1990) is warranted. 



There are differences even among those trachinoids that 

 share a Type 2 morphology. Parapercis has a separate muscle 

 that runs the entire length of the dorsal fin, with only 

 intermittent epaxial fibres contributing to the muscle body. 

 The posterior end of this separate muscle has some fibre and 

 fascia connection with the supracarinalis posterior and only 

 very weak attachments to the dorsal-fin pterygiophores that 

 bear segmented rays. These pterygiophore insertions become 

 strong anteriorly on spine-bearing pterygiophores, and the 

 muscle is continuous with the supracarinalis anterior. This 

 morphology is reminiscent of that of the Mullidae, described 

 above, but shows an even closer association with the supra- 

 carinalis muscles, suggesting a supracarinalis derivation, 

 rather than an epaxial one, for these pterygiophore inser- 

 tions. This is completely different from the condition in 



percophidids (Bembrops), which have a more typical Type 2 

 morphology with epaxial insertions on to the five pterygio- 

 phores of the anterior dorsal fin and to the first pterygiophore 

 of the second, and with the anterior and posterior supracari- 

 nalis muscles entirely separate from the epaxial musculature. 

 Of course, such differences can be interpreted as autapomor- 

 phies for families and genera among the trachinoids, but can 

 also be considered suggestive of non-relationship. 



Epaxial/pterygiophore associations can also strengthen 

 hypotheses about monophyly of currently recognized groups. 

 Although not unique among perciforms, the occurrence of 

 the Type 1 attachment in Niphon spinosus (Fig. 13) and its 

 proposed relatives, the serranids, lends support to Johnson's 

 (1983) placement of Niphon within this family based on other 

 characters. Niphon had previously been aligned with the 

 Percichthyidae, a family that exhibits Type epaxial mor- 

 phology. 



Among blennioids (sensu Springer, 1993), the Type 1 

 epaxial morphology has been found in all examined taxa, but 

 there is some variation in details. Tripterygiids, dacty- 

 loscopids, clinids, chaenopsids and blenniids have a separate, 

 more-or-less fan-shaped, anterior slip of the epaxial muscle 

 bundle that inserts on to the distal portions of the anterior 

 dorsal-fin pterygiophores and extends forward to the skull 

 (Fig. 14a-c). We have not found this anterior slip elsewhere 

 among acanthomorphs with epaxial attachments to dorsal-fin 

 pterygiophores, and interpret it as a synapomorphy of the 

 Blennioidei. This corroborates Springer's (1993) hypoth- 

 esized monophyly of the suborder. However, labrisomids are 

 an exception among blennioids in exhibiting a more typical 

 Type 1 morphology, without an anterior slip to the skull (Fig. 

 14d). On the basis of molecular work, Stepien et al. (1993) 

 hypothesized that the Labrisomidae are nested within the 

 Blennioidei. Morphological characters provided by Springer 

 (1993) also suggest that the Labrisomidae are not a basal 

 blennioid family; for example, labrisomids, clinids, blenniids, 

 and chaenopsids are more derived than tripterygiids and 

 dactyloscopids in having the dorsalmost pectoral-fin ray 

 articulating only with the dorsalmost proximal radial (vs with 

 the scapula). Therefore, the absence of an anterior extension 

 of the dorsal epaxial slip to the skull is most parsimoniously 

 interpreted as a reversal, and a synapomorphy of the Labriso- 

 midae. 



It is also possible that the discovery of epaxial/ 



