444 



Fishery Bulletin 92|2). 1994 



ETP cetacean ecology. The first axis contrasts 

 the habitat use of common dolphins with spot- 

 ted and spinner dolphins. The placement of com- 

 mon dolphins into cool upwelling habitat is con- 

 sistent with results reported by Au and 

 Perryman (1985) and Reilly (1990). The place- 

 ment of spotted and spinner dolphins in con- 

 trasting habitat (negative axis-1 values; essen- 

 tially warm tropical water) is also as reported 

 earlier. Consistency with results of Reilly ( 1990) 

 is not surprising, because that study shared 

 data from 1986 and 1987 with this study, but is 

 somewhat reassuring because different analyti- 

 cal techniques were used. 



The second axis separated eastern spinners 

 from whitebelly spinner dophins. This separa- 

 tion was even clearer between sightings of east- 

 ern spinners mixed with spotted dolphins and 

 whitebelly spinners mixed with spotted dol- 

 phins. The ordination placement of whitebelly 

 spinner dolphins in habitat with a deeper ther- 

 mocline ( negative axis 2 ) follows from their more 

 offshore distribution and the general tendency 

 for the thermocline to become deeper to the west 

 in the ETP. Spotted dolphins alone occurred 

 intermediate to these mixed schools. If this is a 

 general pattern it suggests that the two mixed- 

 school types of spotted and spinner dolphins are 

 utilizing habitats as different as those used by 

 separate species (e.g. common dolphins and 

 spotted dolphins on axis 1). These results are 

 consistent with the hypothesis that the morpho- 

 logical distinctness of the endemic eastern spin- 

 ner dolphin subspecies reflects adaptation to 

 local habitat conditions (Dizonetal., 1991), and 

 the recent finding that spotted dolphins north 

 of the equator and east of long. 120°W, i.e. those 

 available to school with eastern spinner dol- 

 phins, comprise a distinct 'stock' (Dizon et al., 

 in press). 



The ordination of striped dolphins near the 

 origin of both axes 1 and 2 indicates either that 

 this is near their optimum habitat or that their 

 distribution is unrelated to the environmental 

 patterns represented in the canonical axes. The 

 low "R 2 " for striped dolphins (Table 4), and their wide- 

 spread spatial distribution (Fig. 1; Reilly, 1990) sup- 

 port the latter interpretation. 



The species-environment correlations observed 

 were quite high: 0.67 for the first species and envi- 

 ronment axes, 0.42 for the second axes. However, 

 variation extracted by the canonical correspondence 

 analysis accounted for just 15% of the total encoun- 

 ter rate variance. (This was increased to over 20% 

 when fixed geographic effects were considered. ) This 



CANONICAL AXIS 1 



► Common dolphin ^Spotted dolphin 



Figure 6 



Maps of distribution of canonical axis 1 for 1986-90. Positive 

 areas are shaded. Spotted dolphin, Stenella coeruleoalba, 

 sighting localities are shown as open triangles, common dol- 

 phin, Delphinus delphis, localities as closed circles. A"+" rep 

 resents a sighting day during which neither spotted nor com- 

 mon dolphins were seen. 



modest explanatory power is in fact fairly good, given 

 the unknown but surely large sampling variability 

 inherent in daily encounter rates, and is consistent 

 with levels of explanatory power in similar CCA 

 analyses of abundance data (e.g. Ter Braak, 1986). 

 Dolphins are very mobile and patchily distributed 

 large predators, and are known to have complex so- 

 cial and behavioral interactions with their own and 

 other species. These characteristics combine to pro- 

 duce highly variable abundance indices. 



