550 



Fishery Bulletin 98(3) 



loaded on gill rakers of the lower first and second arches 

 and upper first arch and supraorbital pores (Table 10). By 

 plotting PCI of the meristic analysis and PC2 of the mor- 

 phometric analysis (Fig. 2C), we found a distinct cluster 

 representing L. mochigarei and an overlap of clusters rep- 

 resenting L. bilineata and L. polyxystra n. sp. 



In the PCA of only the eastern North Pacific species, 

 loadings of morphometric PCI were all positive and thus 

 exhibited a strong size effect, accounting for 969< of morpho- 

 metric variation. Principal components 2 and 3 accounted 

 for33'7f and \9'/( of the remaining morphometric variation, 

 respectively. Principal component 2 loadings described a 

 gradient based primarily on interorbital width, as well as 

 on ventral orbit length and blind-side maxilla length; PCS 

 loadings described a gradient on interorbital width, caudal 

 peduncle depth, and body depth. The plot of PC2 versus 

 PC3 revealed extensive overlap in morphometric charac- 

 ters (Fig. 3A). Principal components 1 and 2 of the meris- 

 tics analysis accounted for 28'% and ll'7(, respectively, of 

 the meristic variance: PCI strongly loaded on gill rakers 

 and supraorbital pores and PC2 strongly loaded on dorsal- 

 fin rays, anal-fin rays, and cheek scales. The plot of mer- 

 istic PCI versus PC2 revealed two distinct clusters (Fig. 

 3B). By plotting PCI of the meristic analysis and PC2 of 

 the morphometric analysis (Fig. 3C), we found two distinct 

 clusters representing L. bilineata and L. polyxystra n. sp. 



When data points of eastern North Pacific species were 

 identified by geographic regions, no significant clustering 

 was apparent within groups of either L. bilineata or L. 

 polyxystra n. sp. Extensive overlap was found between 

 clusters representing each of these general geographic 



Larvae 



Raw morphometric data were subjected to an ANCOVA 

 analysis, which indicated differences in several morpho- 

 logical characters (snout-to-anus length, body depth, and 

 orbit length) between larvae of L. bilineata and L. polyxys- 

 tra n. sp. at various stages of development (Table 11). 

 Several characters showed significant differences in post- 

 flexion larvae (P<0.0001), including snout-to-anus length, 

 head length, and orbit length. Snout-to-anus length is 

 shorter in larvae of L. bilineata during the preflexion. flex- 

 ion, and postflexion stages (Fig.4). Larvae of L. bilineata 

 are deeper-bodied during the yolksac, preflexion, and flex- 

 ion stages of development (Fig. 5). The PCA plots of PC2 

 and PC3 for each developmental stage produced broadly 

 overlapping clusters. 



