344 



Fishery Bulletin 92(2). 1994 



product-moment correlations were significant for 15 

 characters. Some of these associations were very 

 strong (i.e. correlations as high as 0.883; see Table 

 10). Given that we are comparing two species, enti- 

 ties that have independent gene pools, the most 

 likely explanation for common positive trends in 

 morphologic covariation is that the two species are 

 being subjected to similar forces of natural selection. 



However, we found several striking examples 

 where covariation was negative. A significant nega- 

 tive association was found for length of temporal 

 fossa, while the negative correlation for width of 

 temporal fossa was nearly significant statistically. 

 The number of upper teeth (characters 23 and 24) 

 also show significant negative correlations (Table 

 10). Muscles associated with the feeding apparatus 

 are positioned in the temporal fossa; obviously, tooth 

 numbers could be related to prey types taken. One 

 suspects that the presence of antithetical trends in 

 these particular skull characteristics is a result of 

 competitive interactions involving these two inter- 

 acting species. This may be an example of ecologi- 

 cal character displacement related to differences in 

 feeding and food types taken (Perrin, 1984). Cer- 

 tainly, the two species are found in close association 

 over much of the eastern tropical Pacific (Au and 

 Perryman, 1985; Reilly, 1990); 49% of S. attenuata 

 schools included some S. longirostris, while 73% of 

 the schools of the less common S. longirostris in- 

 cluded S. attenuata (Reilly, 1990). 



The information available to date indicates that 

 spinner and spotted dolphins may have different 

 feeding habits or preferences in areas of co-occur- 

 rence in the eastern tropical Pacific. Based on analy- 

 sis of stomach contents of spinner and spotted dol- 

 phins captured together in purse-seine hauls in the 

 tuna fishery, Perrin et al. (1973) concluded tenta- 

 tively that while some prey species are taken by 

 both, spinner dolphins in the mixed-species associa- 

 tions specialize in small mesopelagic fishes (mainly 

 myctophids and gonostomatids) and squids, whereas 

 the spotted dolphins consume larger and more epi- 

 pelagic species such as flying fishes, small scom- 

 broids (e.g. Auxis sp.), and larger squids. In addi- 

 tion, state of digestion of the stomach contents in- 

 dicted that the spinner and spotted dolphins had fed 

 at different times of the day. Stomachs of a spotted 

 dolphin from Hawaii, where the two species do not 

 school together, and from two spotted dolphins from 

 the far western portion of the range in the eastern 

 Pacific, where the mixed species associations are 

 less common than in the core area of the tuna fish- 

 ery off Mexico and Central America (Au and 

 Perryman, 1985; fig. 11), contained a large propor- 

 tion of small mesopelagic species like those eaten by 



the spinner dolphin in the mixed species associa- 

 tions farther to the east. This geographic variation 

 in feeding habits may reflect resource partitioning 

 where the two species associate closely, which in 

 turn may be manifested in morphological character 

 displacement. 



Genetic subdivision, management units, 

 and implications of cranial variation 



The results suggest that gene flow is not uniform 

 throughout the range of S. attenuata in the eastern 

 Pacific; the morphological heterogeneity probably 

 reflects genetic subdivision, a conclusion also 

 reached as a result of the earlier study by Schnell 

 et al. (1986). A similar inference was drawn by Dou- 

 glas et al. (1992) based on morphologic studies of S. 

 longirostris. For S. attenuata, we found that 93.3% 

 of the 30 morphologic characters had statistically 

 significant geographic variation, with 60.0% exhib- 

 ited regional patterning and 73.3% local patterning. 

 This geographic partitioning of morphologic variance 

 was demonstrable even with pooling of specimens 

 taken over a number of months and years, a process 

 that would tend to shroud such relationships. 



The boundaries of management units presently 

 employed (Perrin et al., 1985) are not fully consis- 

 tent with the general pattern of morphologic varia- 

 tion described here. It appears that animals west of 

 about 120°W longitude have greater affinity with 

 those in the Southern Offshore management unit 

 than they do with S. attenuata from the eastern 

 portion of the present Northern Offshore unit, the 

 unit in which they are presently included. Further, 

 the present boundary between the Northern and 

 Southern Offshore units, at 1°S latitude, is probably 

 too far south; a boundary at about 5"N would be 

 more consistent with the general pattern of cranial 

 variation. 



The proposed Northern Offshore unit bounded by 

 5°N and 120°W would be nearly congruent with the 

 "conservation zone" suggested for S. longirostris 

 (Perrin et al., 1991). This is to be expected based on 

 the correlated trends of covariation with environ- 

 mental variables; the two species, as they exist in 

 this region, apparently are parts of an endemic 

 fauna uniquely adapted to the far eastern tropical 

 Pacific and, as such, are "evolutionarily significant 

 units" (Dizon et al., 1992). 



The cranial results are only one line of evidence 

 useful for delineation of management units; others, 

 such as patterns in movements, external morphol- 

 ogy or life-history parameters, also should be taken 

 into consideration. For example, other data may 

 indicate that S. attenuata west of 120°W differ sig- 



