12 



Fishery Bulletin 90(1). 1992 



and S. I. orientalis that may be about 2000km wide. 

 Stenella I. orientalis is found primarily in the north- 

 eastern blocks we assessed. Our analyses confirm that, 

 in general, adult spinner dolphins from this region are 

 smaller than those from areas to the south, southwest, 

 and west (for general trends, refer to block projections 

 onto canonical variable 1 in Fig. 8B). The nominate 

 subspecies, S. I. longirostris, of Perrin (1990) subsumes 

 a series of broadly distributed populations. He indicated 

 that S. I. longirostris likely includes areal entities (out- 

 side the eastern Pacific) worthy of formal taxonomic 

 recognition, but to date these have not been evaluated 

 properly because of a paucity of specimens from major 

 portions of the range. 



Extensive data on geographic variation in external 

 morphology of S. longirostris in the eastern tropical 

 Pacific were assessed by Perrin et al. (1991). They 

 evaluated color patterns, dorsal-fin shapes, and total 

 lengths for 5. longirostris from throughout the geo- 

 graphic range covered in our study. Some external 

 characters (e.g., ventral field coloration pattern) ex- 

 hibited a "radial" or concentric pattern of variation, 

 where spinners to the south, southwest and west were 

 similar, but markedly different from those to the north- 

 east. This pattern also was prevalent among cranial 

 variables (e.g., see values for Postorbital W. [Fig. 8A], 

 canonical variable 1 [Fig. 8B], and L. Braincase [Fig. 

 13A]). 



We were able to incorporate specimens into our 

 analyses from the general vicinity of the Hawaiian 

 Islands. Perrin (1975b) evaluated Hawaiian specimens 

 for cranial features and concluded that, in general, they 

 were strikingly larger than other spinners. However, 

 at the time, few specimens were available from south- 

 ern localities. When these southern blocks are incor- 

 porated into the analysis, the Hawaiian specimens are 

 not quite as extreme, although for most characters the 

 Hawaiian specimens remain the largest (see Postorbital 

 W. [Fig. 8A], W Temporal Fossa [Fig. 11 A] and L. 

 Braincase [Fig. 13A], as well as principal component 

 I [Figs. 3 and 4] and canonical variable 1 [Figs. 7 and 

 8B]). Also, for many of the characters the Hawaiian 

 specimens are more similar to far-southern ones than 

 to those from geographically closer western blocks 

 located between 5° and 10°N. When evaluating other 

 western single-specimen blocks that are situated closer 

 to the Hawaiian Islands, some analyses (e.g., canonical 

 variates analysis; see Fig. 7) indicate that spinners 

 similar to Hawaiian specimens are present; however, 

 additional specimens will be needed in order to clarify 

 the trends in variation in this part of the Pacific. 



In some descriptive analyses (e.g., see Figs. 6 and 

 7), block 0702 to the south of the Hawaiian Islands is 

 depicted as quite distinct from other blocks, including 

 the adjacent block to the north (0802). This incongruity 



is likely a statistical aberration related to the small 

 sample size for 0702 (n 2), rather than to a biological 

 difference. Checking the specimens from the two blocks 

 indicated that they were taken in relatively close prox- 

 imity, but were separated because of where the border 

 between the blocks happened to be located. This ap- 

 parent anomaly does not detract from the general con- 

 clusion that the spinners in the vicinity of the Hawaiian 

 Islands are among the largest found in the overall study 

 region. 



Schnell et al. (1986) conducted an extensive analysis 

 of geographic variation of offshore S. attenuata, a 

 similar species that broadly overlaps in range with 

 S. longirostris. The two species frequently are seen in 

 mixed schools (Au and Perryman 1985, Reilly 1990). 

 Reilly (1990) noted that 73% of the S. longirostris 

 sightings from research vessels also included S. attenu- 

 ata; 49% of the records of the more common S. attenu- 

 ata involved schools that also had S. longirostris. While 

 detailed comparisons evaluating interspecific geo- 

 graphic covariation in morphology are beyond the scope 

 of this paper, our preliminary findings indicate that 

 about one-half of the individual morphological char- 

 acters show similar geographic patterns for blocks 

 where both species are represented. However, two 

 characters involving the temporal fossa (variables 17 

 and 18) exhibit negative correlations (the length cor- 

 relation is statistically significant and the width near- 

 ly so)— localities where the fossa is larger in S. attenu- 

 ata, it is smaller in S. longirostris. The temporal fossa 

 reflects the size of muscles involved in the feeding 

 apparatus. Also, the upper tooth counts show pro- 

 nounced negative correlations interspecifically. 

 Opposite trends in the two similar species for these 

 characters may be an example of ecological character 

 displacement related to differences in feeding and the 

 types of food taken by the two species in given localities 

 (Perrin 1984). 



Genetic subdivision, management units, 

 and implications of cranial variation 



Considerable attention has been given to definition of 

 stock units with a meaningful biological basis that can 

 be employed to manage iS. longirostris in the eastern 

 Pacific (Perrin 1975a, b; Perrin et al. 1979b, 1985, 

 1991). One of the important questions with respect to 

 the effectiveness or relevance of geographic manage- 

 ment units is the degree to which the species is 

 genetically subdivided. Perrin et al. (1991) noted a com- 

 plex patchwork pattern of geographic variation in ex- 

 ternal and other characteristics in S. longirostris sug- 

 gesting "that there is not a large amount of movement 

 between the various regions." They pointed out that 

 the complex geographic pattern of variation in the 



