spiny dorsal fin is prolonged anteriorly over the posterior 

 one-third of the surface of the ethmoid, and the first dor- 

 sal spine is placed over the front of the eye. In Pseuda- 

 luteres the basal pterygiophore is even further prolonged 

 anteriorly, being of such great length that it covers nearly 

 the entire length of the dorsal surface of the skull, the 

 dorsal spines being placed further anteriorly than in any 

 other monacanthid, distinctly in front of the eye. 



As discussed before, there is no counterpart in any 

 other scleroderm plectognath fish of the true or pleural 

 ribs found in Pseudaluteres. 



Oxymonacanthus and Pseudaluteres share a specializ- 

 ed condition of the sphenotic found in no other mona- 

 canthids. In other monacanthids the lower region of the 

 side of the cranium, just above the region of articulation 

 of the upper end of the hyomandibular, is formed by the 

 sphenotic anteriorly and by the pterotic posteriorly, the 

 former ending ventrally as a ventrally or anteroventrally 

 directed prong and the latter having a larger ventrally 

 directed prong that overlies the posterodorsal region of 

 the hyomandibular. In both Oxymonacanthus and 

 Pseudaluteres, as explained in detail in the section on 

 anatomical diversity, the sphenotic is displaced pos- 

 teriorly by a ventrally directed portion of the posterior 

 region of the frontal bearing its own ventrally directed 

 prong, the side of the cranium just above the ar- 

 ticulation of the hyomandibular thus being formed by 

 the frontal anteriorly and the pterotic posteriorly, with 

 the sphenotic squeezed in between and with a relatively 

 small ventral prong. Moreover, Oxymonacanthus and 

 Pseudaluteres are the only two genera of monacanthids 

 to have completely lost the prootic shelf. 



Oxymonacanthus is speculated to be an early offshoot 

 of the line leading to Pseudaluteres, an offshoot of the 

 pie-Alutera group before the dorsal lobe of the pelvis was 

 lost and the encasing scales nearly entirely lost. In Oxy- 

 monacanthus the dorsal lobe of the pelvis is only slightly 

 developed and the encasing scales are reduced to an in- 

 determinate number of scales gradually and slightly in- 

 creased in size from those of the surrounding region, 

 while the number of vertebrae is increased to about 25 

 (only 1 less than in Pseudaluteres), and the forward ex- 

 tension of the supraoccipital is absent. 



Except for its specialized small nipping mouth (with 

 reduced number of dentary teeth) and slightly elongate 

 snout, a form such as Oxymonacanthus could easily have 

 given rise to Pseudaluteres by the continued reduction in 

 the dorsal lobe and encasing scales and the further 

 migration forward of the dorsal spine and the anterior 

 elongation of the basal pterygiophore. The shared highly 

 specialized nature of the sphenotic in both genera is as- 

 surance of their close relationship. 



All of the other genera of monacanthids examined 

 here, but not yet discussed, form the middle group of 

 moderately specialized genera intermediate in many 

 ways between the basal Stephanolepis-Monacanthus- 

 Paramonacanthus group and their close relatives and the 

 group of genera of several lines thought to be derived 

 from a pre-A/u(era-like form. The middle group con- 

 tains genera without a forward extension of the supraoc- 



cipital and without a fin-ray element, but with encasing 

 scales usually present in an inflexible series or only 

 minutely developed, with 19 or 20 vertebrae and the body 

 not especially either elongate or deepened. These are 

 Amanses, Cantherhines, Rudarius, Navodon, and 

 Acanthaluteres. 



Of these, Amanses and Cantherhines have two and a 

 half segments of encasing scales, Navodon two, Rudarius 

 two or one, and Acanthaluteres only minute series when 

 present. Amanses, Cantherhines, and Acanthaluteres 

 have a deep groove posterior to the dorsal spines which is 

 not present in Navodon and Rudarius, while the first dor- 

 sal spine is placed behind the middle of the eye in 

 Navodon, Rudarius, and Acanthaluteres, but over the 

 middle or front of the eye in Amanses and Cantherhines, 

 the latter two genera also having relatively plain dorsal 

 spine ornamentation. 



Amanses and Cantherhines have long been considered 

 closely related, rightly so I think, and have been well 

 characterized by Randall (1964), who showed them to 

 differ primarily in spinulation {Amanses with coarser 

 scales and a patch of long quill-like spines on the side of 

 the body between the soft dorsal and anal fins) and a 

 deeper caudal peduncle and lower numbers of soft dor- 

 sal and anal fin rays. Amanses is obviously a slightly 

 specialized spinier scaled version of Cantherhines, and 

 Randall (1964:332) has suggested that Scobinichthys and 

 Meuschenia (neither studied here) are also closely 

 related to Cantherhines. 



I suspect that Cantherhines and its close relatives are 

 derived from a Pervagor-like stock (prior to the develop- 

 ment of the special rotation joint between the pelvis and 

 coracoid), some species of which have a groove behind 

 the dorsal spines located above the middle of the eye, by 

 a decrease in the amount of ornamentation of the first 

 dorsal spine and sometimes of its slight forward migra- 

 tion, and by the reduction of the encasing scale series 

 from a flexible three segments around a fin-ray element 

 to an inflexible two and a half segments without a fin-ray 

 element. 



Whether Acanthaluteres, with a deep groove behind 

 the dorsal spines, is related to the Cantherhines-hke 

 genera is difficult to say, for it has a more generalized 

 dorsal spine position over the rear of the eye but a more 

 specialized pelvic apparatus, the dorsal lobe being es- 

 sentially absent and the encasing scales either absent or 

 minute. I suspect that Acanthaluteres is an offshoot of 

 the line leading to the pre-/l/ufera-like group hy- 

 pothesized here as containing the Alutera-Psilocephalus 

 line and the Brachaluteres-Paraluteres and the Oxy- 

 monacanthus- Pseudaluteres lines. 



Rudarius is probably a derivative of the line leading 

 from a Pervagor-\\ke form to those of the Cantherhines- 

 like genera, but with the placement of the dorsal spine 

 remaining a little more posterior, usually just behind the 

 middle of the eye, while the trend of reducing the in- 

 flexible encasing scale series continued further to only 

 two segments in one of the species and to one segment in 

 the other, while the dorsal lobe and ventral dewlap 

 remained of moderate to small size. 



