Reilly and Fiedler: Interannual variability of dolphin habitats 



443 



Fixed geographic effects 



Inclusion of latitude and longitude in addition to the 

 six oceanographic variables produced a notable in- 

 crease in dolphin variance explained, from 14.7% to 

 20.5% (Table 5). The amount of additional influence 



indicated for fixed geographic effects varied substan- 

 tially among school types. The largest increases were 

 for whitebelly spinners, alone and with spotted dol- 

 phins. No improvement in explaining variance was 

 made for schools of spotted dolphins alone. 



Group size effects 



"15POT&ESPIN 



QWB SPIN 

 SPOT&WBSPtNQ 



B 



-t 



V h 



This study used encounter rates as 

 an index of abundance. This index 

 does not encompass effects of vary- 

 ing group size. There is some evi- 

 dence for geographic patterns in 

 group size for the dolphin school 

 types studied ( Gerrodette and Wade, 

 1991), so the analyses reported here 

 were also run with the dependent 

 variables modified as follows. En- 

 counters with schools were weighted 

 by the number of individuals esti- 

 mated to be in the school. The 

 weighted rate was then log-trans- 

 formed. Canonical correspondence 

 analyses run with these modified 

 dependent variables produced essen- 

 tially the same patterns as before, 

 but with a small loss of explanatory 

 power: the cumulative percent of the 

 species variance explained was 

 14.2%, down from 14.7%. 



Discussion 



Species-environment patterns 



The ordination results were gener- 

 ally consistent with past studies of 



Figure 5 



Ordination results from canonical corre- 

 spondence analysis of cetacean species/ 

 stocks and environmental conditions in 

 the eastern tropical Pacific. (A) Biplot of 

 first two canonical axes and environmen- 

 tal variables. (B) Ordination showing 

 95% confidence limits for the species 

 loadings. The environmental variables, 

 represented by arrows in 5A, are surface 

 temperature (TEMP), surface salinity 

 (SAL), thermocline depth indexed by 

 20°C isotherm depth (Z20), thermocline 

 strength, indexed by the difference in depth 

 between the 20°C and 15°C isotherms (ZD), 

 surface water density (SIGMAT), and sur- 

 face chlorophyll, log-transformed (LOGC). 

 These two axes represent 94% of the spe- 

 cies-environment variance, 15% of the to- 

 tal encounter rate variance. 



