Statistical analysis indicated that individual 

 whales did not associate randomly with one another. 

 For individuals seen together at least once, 38.5% of 

 all CAI values from the initial analysis of all indi- 

 viduals in all years were significantly less than at 

 a 5% confidence level (30.1% at a 1% level) and 27.5% 

 were significantly greater than at a 5% level (23.2% 

 at a 1% level) (see Cole, 1949). Because these values 

 are far greater than expected owing to chance, we 

 rejected the null hypothesis that association patterns 

 were random. The various animals positively associ- 

 ated with a particular individual tended to be posi- 

 tively associated themselves. These groups were also 

 indicated by the agglomerative cluster analysis. A 

 dendrogram showing associations among all 202 

 individuals included in the analyses was too large to 

 exhibit here as a single figure; it appears split into 

 pods along with matrices of CAI values and proposed 

 genealogical trees for each pod (Figs. 1, and 3-12) 



In nearly all cases, intrapod groups of individuals 

 linked by association analysis at greater than 20% 

 CAI reflected groupings established by direct obser- 

 vation. Direct observation also indicated that these 

 intrapod groups virtually always traveled together. 

 Most intrapod groups were centered around a repro- 

 ductive or postreproductive female, which led us to 

 suspect that these groups reflected matrilineal ge- 

 nealogies. However, in four cases single males (AE14, 

 J02, N19, and DOl) were not joined to other intrapod 

 groups (CAI>20%) and in two cases pairs of males 

 (AD02 and AD12, and AB02 and AB29) were linked 

 to each other but not to other intrapod groups 

 (CAI>20%). The nine pods examined contained 48 

 intrapod groups, including male singles and pairs ( Figs. 

 1 and 3-12). With the exception of AB pod, pods com- 

 prised 1-9 intrapod groups. Two pods (AI and AD 16) 

 contained only one intrapod group. The three subpods 

 of AB pod contained a total of 12 intrapod groups. 

 Intrapod groups were composed of 1-9 individuals. 



Patterns of association between intrapod groups, 

 as indicated by statistical analysis, generally sup- 

 ported the pods constructed by direct observation. 

 Forty-six of the 48 intrapod groups (190 of 202 indi- 

 viduals) were linked at the zero level of association 

 into their respective pods (Fig. 2). There were two 



exceptions; repeated direct observations were used 

 to clarify separations into intrapod groups. First, the 

 ABIO subpod was linked in the dendrogram with AI 

 pod (PCC=4) before it was linked with the other 

 subpods of AB pod. Second, AK pod and AD16 pod 

 formed distinct clusters as expected but were joined 

 at the PCC = 8 level, which was greater than the 

 zero level adopted to define the other pods. 



On the basis of direct observation, ANIO pod and 

 AN20 pod were considered a single pod (AN pod) until 

 1989, when they began traveling separately a ma- 

 jority of the time. AN whales were encountered on 

 46 occasions during 1984-88; the ANIO and AN20 

 groups were observed together in 35 (76%) of these 

 encounters. In contrast, during 1989-95 they were 

 observed together in only 3 of 65 (5%) encounters 

 where AN whales were present. After 1992 they were 

 not seen traveling together and were thus designated 

 as separate pods. The association analysis supported 

 their designation as separate pods (Fig. 1). AD pod 

 was also considered a single pod early in the study 

 on the basis of direct observation during a few en- 

 counters. From more recent direct observation and 

 the results of the association analysis, it is now des- 

 ignated as two pods (AD5 and AD 16 pods). 



