FISHERY BULLETIN: VOL. 87, NO. 3. 1989 



mens). Since this species was first identified dur- 

 ing the course of the present study, it is possible 

 that other misidentified specimens reside in the 

 collections. They were not, however, confused 

 with G. nigricans or M. californiensis since we 

 have reexamined all identifications of these spe- 

 cies. Larvae of H. azurea occurred in Manta net 

 samples at three stations during the 1978 survey 

 and at nine stations on the 1981 survey. Occur- 

 rences ranged from Santa Monica Bay, CA 

 (cruise 8107, station 86.7.33) to Pt. San Juanico, 

 Baja California (cruise 7807, station 133.23). 

 Occurrences were shoreward of station 36 on 

 each CalCOFI Hne (4-42 km from the coast). 



Larvae of//, azurea were well represented in 

 the SONGS plankton collections of MEC Analyt- 

 ical Systems. The shallow distribution of H. 

 azurea larvae was clearly demonstrated in the 

 SONGS collections. Surface tows accounted for 

 87% of the larvae and oblique tows for 14%. 

 Larvae occurred principally in the outer section 

 of the SONGS transect (1.9-7.2 km from the 

 coast; ca. 18-75 m depth). Larval occurrence was 

 highly seasonal, with 62% of the total number 

 taken in August and >99% in July-September. 



SYSTEMATICS 



Johnson (1984) discussed kyphosid fishes in his 

 review of percoid systematics and ontogeny. He 

 considered Girellidae, Kyphosidae, and Scorpidi- 

 dae to be distinct famihes and redefined the lat- 

 ter to include only four genera (Scorpis, 

 Medialuna, Labrucoglossa, and Bathystethus). 

 Evidence from adult anatomy that led to these 

 decisions was deferred to a forthcoming review 

 (Johnson and Fritzsche, in press). Johnson 

 (1984) pointed out the similarity of girelHne and 

 scorpidine larvae and presented this as evidence 

 for considering them sister gi'oups; he now be- 

 lieves girellines, scorpidines, and kyphosines 

 form a monophyletic group (G. D. Johnson"). 



A principal problem in assessing ontogenetic 

 characters of teleost fishes is a lack of infor- 

 mation for all but a few taxa under considera- 

 tion. This is especially true for percoids and for 

 this group of percoids. In addition to our de- 

 scription of Girella nigricans, ontogenetic 

 series are known for two other species of 

 Girella (G. punctata eggs, larvae, and juveniles 

 [Mito 1958a]; G. melanichthys larvae [Okiyama 



'^G. D. Johnson, Fishery Scientist, National Museum of 

 Natural History, Washington D. C. 20560, pers. commun. 

 April 1989. 



1988]). Developmental stages of the other girel- 

 line genus (Graus) are unknown. Other than our 

 description of Medialuna californiensis, the 

 only scorpidine larval series description is that 

 of Labracoglossa argentiventris (Hattori 1964). 

 Developmental stages of two of the four kypho- 

 sine genera are known. In addition to our de- 

 scription of larvae and pelagic juveniles of 

 Hermosilla azurea there are descriptions of 

 developmental stages of at least three species of 

 Kyphosus (K. cinerascens larvae and juveniles, 

 [Mito 1958b; Okiyama 1988]); K. sectatrix trans- 

 forming larvae and juveniles [Moore 1962]; K. 

 incisor transforming larvae and juveniles 

 [Moore 1962]; K. vaigensis or bigibbus eggs 

 [Watson and Leis 1974]; K. vaigensis or bigib- 

 bus larvae [Miller et al. 1979; Leis and Rennis 

 1983]). Developmental stages of Sectator have 

 not been described. 



The literature on development stages of these 

 fishes does not provide sufficient descriptive de- 

 tail to adequately assess ontogenetic characters. 

 In this paper we present detailed descriptions of 

 representatives of the three putative families to 

 establish a basis for character comparisons. The 

 striking feature of the three larval series is their 

 similarity. The fact that the three species were 

 confused with one another during the history of 

 CalCOFI surveys attests to this in a practical 

 sense. Our more rigorous study of their morphol- 

 ogies and pigment patterns has allowed us to 

 identify each specimen correctly, while reinforc- 

 ing the similarities perceived by early CalCOFI 

 workers. These similarities are 1) a general per- 

 coid body form with a Girella- Medialuna- 

 Hermosilla grade in degree of robustness; 2) 

 dorsal and ventral midline melanophore series, 

 with unique variations for each species; 3) lateral 

 midhne melanophore series, with unique varia- 

 tion; 4) an embedded melanistic band through 

 the eye region; 5) minute melanophores at the tip 

 of the notochord which become associated with 

 the hypural margin of the caudal fin; 6) an ante- 

 riad progression of general body pigmentation 

 late in the larval period; and 7) head spination 

 with a Girella-Hermosilla-Medialuna grade in 

 degree of development. This morph is not unique 

 among percoids (see Leis and Rennis 1983; John- 

 son 1984); however, the coherence of these and 

 other more subtle characters among these three 

 eastern Pacific species supports the argument 

 that they represent sister groups. 



The pattern of lateral pigmentation is more 

 variable in kyphosine larvae than in gireUines 

 and scorpidines. We can recognize three pat- 



766 



