PH. GREENWOOD 



cirrhitoids, do any of these rays clearly extend beyond the fin 

 membrane, and none is markedly thickened. In having the 

 lowermost 5-9 rays unbranched, these fishes are, however, 

 typically cirrhitoid. This latter condition can be considered 

 one of the synapomorphies uniting cirrhitoid fishes. 



A typically derived pectoral fin configuration occurs in 

 the Cirrhitidae, yet the family would appear to be the least 

 derived of all cirrhitoid taxa (see p. 8). In contrast, except 

 for the condition of the pectoral fin, members of the 

 Latris-Latridopsis-Mendosoma assemblage within the 

 Latridae share with Acantholatris and Nemadactylus, and 

 with the cheilodactylids, aplodactylids and chironemids, 

 the derived condition for all the osteological and myologi- 

 cal features discussed on page 5. That being so, it is 

 unlikely that the pectoral fin form in Latris, Latridopsis 

 and Mendosoma can be interpreted as a true retention of 

 the plesiomorph condition. If that was the case, then the 

 derived condition must have evolved more than once 

 within the cirrhitoids. A more parsimonious solution to the 

 problem therefore, would, be to interpret the pectoral fin 

 form in Latris, Latridopsis and Mendosoma as a secondary 

 reversal to a seemingly more plesiomorphic condition than 

 is found in any other cirrhitiforms, including the family 

 with the greatest number of plesiomorphic features, the 

 Cirrhitidae (see p. 8). 



The geographical distribution of the two groups within 

 the Latridae has an interesting pattern. Of the taxa in the 

 long-finned assemblage, Nemadactylus (see p. 5) occurs 

 only in Australia, Tasmania and New Zealand, thus over- 

 lapping the entire range of Latridopsis a member of the 

 short-finned group and one restricted to that region; it 

 overlaps in part (New Zealand and Tasmania) that of the 

 widely distributed Mendosoma lineatum, also a member of 

 the short-finned group, and in part, that of Latris, another 

 member of the short-finned group (Australia; New 

 Zealand; Gough and Tristan da Cunha islands; Vema 

 Seamount; St Paul and Amsterdam islands). The other 

 long-finned taxon, Acantholatris , does not occur in Aus- 

 tralasia, but has a wide western distribution, including St 

 Paul, Amsterdam, and Gough Islands, Tristan da Cunha, 

 Vema Seamount Chile, Juan Fernandez and the western 

 coast of South America from Rio de Janeiro southwards. 

 This distribution thus widely overlaps that of the short- 

 finned, monotypic genus Mendosoma lineatum, viz St Paul, 

 Amsterdam and Gough islands, the coast of Chile and, as 

 noted above, New Zealand and Tasmania (the latter being 

 areas where Acantholatris does not occur); data from 

 Norman, 1937; Fowler, 1945; Mann, 1954; Smith 1984; 

 Last et. al., 1983; Gon & Heemstra, 1987; Paulin et. al. 

 1989; Andrew & Hecht, 1992; Andrew, pers.comm., 1993). 



TAXONOMIC AND PHYLOGENETIC 

 CONCLUSIONS 



Taxonomy 



The material studied indicates, on osteological and myologi- 

 cal grounds (p. 5), that the species currently named Nema- 

 dactylus macropterus (the type species of Gill's (1862) genus 

 Dactylopagrus; see Wheeler, 1986) should be classified in the 

 Latridae and not the Cheilodactylidae as it is at present. 



The genus Acantholatris Gill (1862, type species Chaetodon 

 monodactylus Carmichael, 1818), was overlooked by Regan 

 (1911) in his synoptic review of cirrhitoid families, but is 

 currently placed in the family Cheilodactylidae (see p. 5). 

 However, on the basis of its urohyal morphology, and its 

 lacking a suborbital shelf (see p. 5) the genus should be 

 classified in the Latridae. Regan (1911) differentiated the 

 Latridae from the Cheilodactylidae on the basis of the latrids 

 having feeble, unbranched pectoral rays that are not pro- 

 duced beyond the fin's margin, and by their lacking a 

 suborbital shelf; in other feaures he noted that the two 

 families are similar. With the inclusion of Acantholatris and 

 Nemadactylus in the Latridae the nature of the pectoral fin no 

 longer serves as a differentiating feature (see p. 5), the 

 principal diagnostic characters for the family now lying in the 

 form of the urohyal bone, the absence of a suborbital shelf, 

 and in the more prominent arrangement of the basal sheath- 

 ing scales of the soft dorsal fin (see p. 5). 



Gill's (1862) suprageneric classification included Nemadac- 

 tylus as a division - Nematodactyli - of his subfamily Latridi- 

 nae, in which subfamily but as another division to which he 

 gave the name Latrides he also included Latris, Latridopsis, 

 Mendosoma, Acantholatris, Chirodactylus, Cheilodactylus 

 and Goniistius. Regan (1911) on the other hand, but without 

 reference to Gill's paper, treated the latter author's four 

 subfamilies as families, and recognised a fifth, the Cheilodac- 

 tylidae, for two genera, viz Cheilodactylus and Nemadactylus; 

 no mention is made in Regan's paper of the other taxa in 

 Gill's Latridinae except for Latris and Mendosoma, which 

 Regan retained in his family Latridae. 



The evidence presented here (pp. 2-5), especially that 

 based on urohyal morphology, would support Regan's (1911) 

 classification with regard to the separation of Cheilodactylus 

 (and, although not mentioned by Regan, Chirodactylus and 

 Goniistius) from the other taxa included in Gill's Latridinae, 

 and would support the inclusion of all three taxa in one 

 family, the Cheilodactylidae. The same evidence would also 

 support Gill's inclusion of Latris, Latridopsis, Acantholatris, 

 Mendosoma and Nemadactylus in a single suprageneric 

 taxon. Since Regan's (1911) familial ranking has been 

 accepted and used since that time, and until contraindicative 

 evidence is available to suggest otherwise, that ranking (i.e. 

 Latridae) is retained. 



The anatomical and other features used in this paper 

 (pp. 2-5) would support the recognition of Gill's (1862) and 

 Regan's (1911) other suprageneric lineages, again, for the 

 reasons given above, as families and not subfamilies, viz. the 

 Cirrhitidae, Aplodactylidae (Gill's Haplodactylinae) and Chi- 

 ronemidae. 



At an intrafamilial taxonomic level, Allen & Heemstra 

 (1976) note that 'The currently accepted classification of the 

 Cheilodactylidae ... is most unsatisfactory' a sentiment I 

 would not only endorse, but would extend to other cirrhiti- 

 form families as well. In part this situation has resulted from 

 the use of mainly superficial characters, with little or no 

 attention paid to anatomical features, especially myological 

 and osteological ones. Thus on those grounds I cannot agree 

 with Allen & Heemstra's (op.cit.) treating Acantholatris as a 

 subjective synonym of Cheilodactylus and its consequent 

 placement in the Cheilodactylidae (see above, p. 5). How- 

 ever, at least on the characters and specimens I have exam- 

 ined, I would endorse their synonymy of Whitley's (1957) 

 genus Morwong (type species Cheilodactylus fuscus Castel- 

 nau, 1879) with Cheilodactylus. 





