Matkin et al : Association patterns of Orcinus orca 



901 



Ellis, 1987; Heise et al., 1992). The first letter in the 

 code (A) designates Alaska, followed by a letter (A- 

 Z) indicating the animal's pod as determined by ob- 

 servational data. Individuals within a pod are as- 

 signed sequential two-digit numbers. Pods that re- 

 sult from an obsei'\'ed splitting of a pod share letter 

 designations, i.e. ANIO and AN20 pod. 



In this study, we tested the hypothesis that Prince 

 William Sound resident killer whales form nonran- 

 dom associations by examining the structure of some 

 of the extensively studied pods in greater detail. Bond 

 strength was established by direct observation of the 

 proximity of whales to one another and by statisti- 

 cal analysis of the association of individuals in pho- 

 tographic sequences. We used statistical analyses to 

 evaluate whether association patterns were random 

 and used clustering techniques to describe underly- 

 ing association patterns. Subsequently, we inferred 

 genealogical relationships as an explanation for ob- 

 served association patterns. 



An understanding of social organization and ge- 

 nealogy is a prerequisite for studies of the sociology, 

 life history, and population dynamics of Prince Will- 

 iam Sound resident killer whales. 



Materials and methods 



Field methods 



Field observations and photographs used in this 

 analysis were collected from 1984 to 1995 in Prince 

 William Sound, Alaska, and adjacent waters. Data 

 were obtained throughout the Sound, although the 

 study was centered in the southwestern portion. Field 

 work occurred from April to October, but the great- 

 est effort and the majority of encounters with resi- 

 dent pods occurred in July, August, and September. 

 Vessels used to approach the whales to obtain photo- 

 graphs ranged in length from 4 to 9 m and in power 

 from 50 hp outboard motors to 185 hp diesel inboard- 

 outboard engines. 



The techniques used to approach and photograph 

 individuals were consistent over the duration of the 

 study. During each encounter, individual whales were 

 typically photographed several times from a distance 

 of 15-30 m. Individual photographs were taken of 

 the left side of each whale, showing details of the 

 dorsal fin and gray saddle patch, with the technique 

 developed by Bigg et al. ( 1986). An effort was made 

 to move systematically through the pod and to ob- 

 tain lateral-aspect photographs of individuals, such 

 that each filled at least 50'7f of the frame. Photo- 

 graphs were taken during all behavioral modes and 

 when whales were tightly grouped (i.e. resting) as 



well as when they were spread over a considerable 

 distance (i.e. feeding). Encounters lasted until all 

 whales were photographed or until weather or dark- 

 ness (or both) made photography impracticable. 



In the early years of the study we used a 35-mm 

 SLR Nikon FM2 camera with an autowinder 

 mounted on a shoulder brace. Later this setup was 

 replaced by a 35-mm SLR Nikon 8008 autofocus cam- 

 era with a shoulder brace. In all years a 300-mm lens 

 was used and Ilford HP5 35-mm film was exposed 

 and processed at ISO 1600. 



Data recorded during each encounter with killer 

 whales included date, time, duration, and location 

 of the encounter. The total number of individuals 

 present, the identity of individuals that could be 

 immediately recognized, and the general behavior 

 of the whales (i.e. feeding, resting, traveling, social- 

 izing, and milling) were also noted. Whales were con- 

 sidered associated on the basis of direct observation 

 when they were seen traveling in the same direction 

 and at the same speed for a period of greater than a 

 few minutes and for a distance less than several body 

 lengths. 



Analysis of photographs 



Throughout the study, each photographic frame was 

 examined numerous times with a dissecting micro- 

 scope to insure that all individuals were correctly 

 identified (Bigg et al., 1987). Mothers of new calves 

 were identified on the basis of consistent close asso- 

 ciation of calves with a specific female (Bigg et al., 

 1990). Identified individuals were recorded in a stan- 

 dard format to facilitate computer analyses, and 

 uncertain identifications were excluded from the 

 analysis. 



Data analysis 



In order to test the hypothesis that killer whales form 

 nonrandom associations, associations between indi- 

 viduals were examined by using statistical analysis 

 of photographic data. Photographs were placed in the 

 order in which they were taken, and the identity of 

 all individuals in each frame was tallied for each 

 encounter. The analytical program examined asso- 

 ciations between animals within photographic frames 

 and in immediately adjacent frames. Photographic 

 sequences that contained less than two individuals 

 were deleted because they provided no information 

 on association among individuals. Data from the re- 

 maining frames were accumulated in a 2x2 contin- 

 gency table for each pair of whales for all years in 

 which both individuals were photographed: 



