Perrin et al.: Geographic variation in cranial morphology of Stenella attenuata 



335 



third had relatively high loadings with the 

 two characters involving solar insolation 

 (variables 4 and 5). A more detailed descrip- 

 tion of character associations with the prin- 

 cipal components is available in Douglas et al. 

 (1992). 



Projection values for environmental princi- 

 pal component I are summarized in Figure 9A 

 for the 29 blocks with larger samples of S. 

 attenuata. It reflects the fact that sea surface 

 temperatures are considerably higher in 

 northern than southern blocks, and that the 

 northern blocks exhibit relatively little an- 

 nual variation in surface temperatures. Block 

 projections on environmental principal com- 

 ponent II are portrayed in Figure 9B, which 

 summarizes the increases in thermocline 

 depth, water depth, and surface salinity as 

 one proceeds west and south. 



Correlations of morphologic variables, prin- 

 cipal components, and canonical variables 

 with environmental variables and environ- 

 mental principal components are summarized 

 in Table 7. The sea current measures (vari- 

 ables 1 and 2) have virtually no statistical 

 association with morphological characters, 

 while water depth (variable 3) has positive 

 correlations with lengths and widths of the 

 rostrum, as well as principal component I and ca- 

 nonical variable 1 (Table 7). 



Solar insolation (Jan.), the fourth variable, has 

 larger values in the south; values become smaller 

 to the north. It has significant positive correlations 

 with nine morphologic variables, and negative as- 

 sociations with six others. The negative associations 

 with the two temporal fossa measures are particu- 

 larly strong (width of temporal fossa summarized in 

 Fig. 10A). This environmental variable has rela- 

 tively high correlations with canonical variable 1 

 and principal component II (Table 7). 



Not unexpectedly, the fifth variable, solar insola- 

 tion (annual) exhibits high values at the equator. 

 Readings are lower for blocks closer to either pole. 

 It has few significant statistical associations with 

 morphologic characters, although the negative cor- 

 relations with length of braincase and width of tem- 

 poral fossa (Fig. 10A) are relatively high (Table 7). 



The sixth and seventh environmental measures 

 (surface temperatures in January and July) have 

 negative associations with a number of width mea- 

 surements, as well as with a few length variables 

 (Table 7). They have very strong positive correla- 

 tions with temporal fossa measures. Figure 10 sum- 

 marizes the values for sea surface temperature 

 (July), as well as for the closely associated width of 



a"'" 



0216- 



"m 0215 pO310 

 M.5 MhJJ"" 



Canonical Variable 1 



Figure 7 



Projections of latitude-longitude blocks onto first two canoni- 

 cal variables based on 30 characters. Solid symbols indicate 

 25 blocks on which analysis was conducted. Open symbols rep- 

 resent blocks with only single specimens, which were projected 

 onto axes generated from 29 blocks with two or more specimens. 



temporal fossa (r=0.799; the highest correlation of 

 an environmental and a morphological variable). 

 Sea surface temperature (annual variable), the 

 eighth environmental variable, has significant cor- 

 relations with relatively few morphologic characters 

 (Table 7), although its pattern has affinities with 

 those summarized by principal component II and 

 canonical variable 1. Environmental variable 9, oxy- 

 gen minimum layer (depth), shows very few statis- 

 tically significant correlations with morphological 

 measurements (Table 7). 



Surface salinity, variable 10, exhibits strong 

 covariation with numerous measurements, particu- 

 larly those involving the anterior portion of the skull 

 (Table 7). It also has high correlations with princi- 

 pal component I (Fig. 4) and canonical variable 1 

 (Fig. 8B). Salinity, which was depicted in Douglas 

 et al. (1992: fig. 13B) for S. longirostris blocks, shows 

 east-west changes from lower to higher values at a 

 given latitude, as well as a north-to-south trend of 

 increasing values (below 15°N). 



The eleventh variable, thermocline depth (winter), 

 is summarized in Figure 11B. It has positive corre- 

 lations with 12 morphological measures and a nega- 

 tive correlation with 1 character. The correlation of 

 this environmental variable with skull width (at 

 parietals), shown in Figure 11A, is 0.610. Variable 



