Reilly and Fiedler: Interannual variability of dolphin habitats 



445 



CANONICAL AXIS 1 



1990 



120 110 100 90 ftO 



• Common dolphin ^Spotted dolphin 



Figure 6 (Continued) 



Explanatory power for common dolphins was 

 surprisingly high: 36% with the six oceano- 

 graphic variables, and 42% with fixed geogra- 

 phy included. This result indicates that com- 

 mon dolphins have the tightest association with 

 the environmental variables of the seven school 

 types studied here. It also demonstrates the ro- 

 bustness of CCA, considering the bimodal distri- 

 bution of common dolphins on axis 1 (Fig. 4). 



The notable increase in performance for 

 whitebelly spinner dolphins resulting from con- 

 sideration of fixed geography raises interesting 

 questions. Are they directly responding to some 

 geophysical cue, such as magnetic anomalies 

 (Kirschvink et al., 1986; Klinowska, 1985)? Or, 

 does this result simply reflect orientation to 

 oceanographic features or processes (e.g. prey 

 distribution) not represented in our data? 



Interannual variability 



Interannual environmental variability is appar- 

 ent in the geographic distributions of the ca- 

 nonical axis scores, and to a lesser extent in lo- 

 cations of dolphin sightings (Figs. 6 and 7). In 

 1986, cool upwelling habitat was found along 

 the equator to long. 130°W, north of the equa- 

 tor to about lat. 10°N along the coast of Central 

 America, and off the coast of Baja California. 

 In 1987, cool upwelling habitat south of Baja 

 California did not extend west of 110°W or north 

 of 4°N, except in the Gulf of Panama. The study 

 area was dominated by warm, low-productiv- 

 ity tropical water (negative axis-1 scores). This 

 change was caused by a moderate El Nino event 

 that began in late 1986 and lasted through 1987 

 (Kousky and Leetmaa, 1989). In 1988, cool up- 

 welling water extended far north of the equa- 

 tor and south of Baja California, considerably 

 reducing the area covered by tropical water. 

 1988 was a strong anti-El Nino or La Nina year 

 (Leetmaa, 1990; Fiedler et al., 1992). In 1989 

 and 1990, conditions represented by axis-1 

 scores returned to a state similar to 1986. 

 Interannual variation along axis 2 was not 

 strongly related to ENSO variability. The area 

 with positive axis-2 scores ("coastal tropical" 

 habitat) was small in 1986 and 1987, increased 

 in 1988 and again in 1989, and showed some 

 diminishment in 1990. 



Common dolphin distribution was previously 

 observed to show no apparent seasonal changes 

 (Reilly, 1990) but was observed here to change 

 interannually more than the other school types 

 studied, and these changes appear related to 



