762 



Fishery Bulletin 91(4), 1993 



Peruvian and Chilean crania, explainable by the rela- 

 tively greater morphological similarity and geographic 

 proximity. Overall there were successfully identified 

 to geographical population 164 out of 188 (87.29c) of 

 the skulls, increasing to 179 out of 188 (95.2%) when 

 skulls from Chile and Peru were pooled. 



Six characters with high discriminatory power were 

 identified from the standardized discriminant function 

 coefficients, /-tests and ANOVA's and an additional dis- 

 criminant analysis was performed using those vari- 

 ables (Table 8). Classification results and the discrimi- 

 nant function plot (Fig. 4) demonstrate that specimen 

 prediction remains fairly effective (143 of 188 or 76.1% 

 and 171 out of 188 or 91.0% when the SE Pacific sample 

 is pooled as one group). 



Non-metrical characters 



Tooth counts — As expected, tooth counts derived 

 from skulls were consistently greater (P<0.0001) 

 than counts made on heads in the flesh because 

 small, apical teeth frequently do not erupt from 

 the gums (see Van Waerebeek, 1992b). Differences 

 of mean tooth counts between skull and fresh 

 specimens varied from 2.29 to 2.67 per jaw ar- 

 cade. Considering confidence intervals, the best 

 estimate (closest integer) for each of the four ar- 

 cades was two more teeth visible in skulls than 

 in heads in the flesh. 



Tooth counts are bilaterally symmetrical in up- 

 per and lower jaws and for all sample pairs it- 

 tests, P>0.16). The upper jaw has significantly 

 more teeth than the lower jaw, (one-sided /-tests; 

 P<0.005 in all cases); confidence intervals in- 

 cluded one (tooth) as the largest integer but not 

 zero. 



Sexual dimorphism — The 28 non-metrical cranial 

 characters (NMC) tested are, without exception, 

 independent of sex (chi-square contingency tests, 

 P>0.05). Sexual dimorphism neither exists in the 

 number of teeth in Peruvian dusky dolphins, both 

 as counted from skulls and from fresh animals 

 (/-tests, P>0.25). It is considered safe to assume 

 that this is generally true for the other popula- 

 tions of L. obscurus and therefore the sexual fac- 

 tor is disregarded in the subsequent analyses. 



Developmental variation — Twenty-two of the 28 

 NMC characters tested (78.6%) in Peruvian dusky 

 dolphins are independent of maturity status 

 (Table 9). Variables which may be correlated with 

 age to some degree (chi-square, P<0.05) 

 include position of the left premaxillary fora- 

 men at the base of the prenarial triangle (1L), 

 presence of fenestration* s) in occipitals (18L) or 

 basisphenoid (30), visibility of palatine spine (25), and 

 the number of mental foramina (33L). At least one of 

 these is expected to be due to chance fluctuation 

 (a=0.05). The more frequent occurrence of a notch in 

 the upper margin of the foramen magnum (17) in im- 

 mature animals as compared to adults proved highly 

 significant (chi-square=8.83, P<0.005). 



Geographic variation — The six growth dependent char- 

 acters were removed from the data set before varia- 

 tion between populations was tested. By doing this, 

 samples available for geographic variation analysis 

 were greatly enhanced as juvenile specimens (but not 

 neonates) could be included. Table 10 lists absolute 



