Olesiuk Prey consumption of Phoca vituhna 



497 



the diet. A second estimate, FR2 sll) , was obtained di- 

 rectly from Boulva and McLaren's (1979) relationship 

 between daily ingestion rates and body mass: 



FR2^ tu = 0.089-M s , 



(from Fig. 2 in Boulva 

 and McLaren, 1979) [15] 



based on the amount of undigested prey found in the 

 stomachs of harbor seals collected on the east coast of 

 Canada. 



Diet composition 



The diet of harbor seals in the study area was deter- 

 mined by scat analyses. During 1982-88, a total of 

 2,841 scats (216 collections) were collected from 58 

 sites (11 estuaries and 47 non-estuary sites) distrib- 

 uted throughout the Strait of Georgia (Fig. 1). Samples 

 were collected in all months of the year (Fig. 2). Be- 

 cause most major haulout sites were sampled, the 

 sampled sites accounted for about 45% of all seals ob- 



served during censuses of the study area both in May- 

 June and in August, 1988. The collections thus pro- 

 vided a representative sample from which regional and 

 seasonal variations in diet composition could be 

 assessed. 



Undigested prey remnants were recovered from the 

 scat samples with an elutriator (Bigg and Olesiuk, 

 1990). Elutriation recovery rates ranged between 90- 

 100% (X=98.6%) for various fish structures and 70- 

 100% (X=85.0%) for cephalopod beaks. In contrast to 

 previous scat studies which have relied almost exclu- 

 sively on otoliths to identify prey, harbor seal prey 

 were identified by using a wide array of different struc- 

 tures including otoliths, teeth, scutes and scales, as 

 well as numerous cranial, appendicular, axial and cau- 

 dal elements (see Appendix I in Olesiuk et al., 1990b). 



The relative importance of prey in the diet was mea- 

 sured by using a new index, termed split-sample fre- 

 quency of occurrence, designed specifically for scat 

 analyses (Olesiuk et al., 1990b). The index was predi- 

 cated on two assumptions: 1) prey identified in scat 

 samples represented all those consumed in the previ- 

 ous meal (i.e., 24-hour period); and 2) all prey species 

 constituting a meal had been consumed in equal vol- 

 umes. Thus, the proportion of the diet comprised of 

 the /jth prey species in the jth strata (j=E for estuaries 

 and .7=0 for outside estuaries) in the mth month (m=l 

 for January, etc.), P jkm , was estimated from the i=l,..,n 

 samples collected from that strata in that month by 



* jkn 



BO 



ijkm' 



®0„kJn k: 



ijkm 



1,..,N (N=# different 

 prey species) [16] 



where O l]km is a binary variate that indicates whether 

 the £th prey species was absent or present (0=absent 

 and l=present) in the z'th sample collected in thej'th 

 strata in the mth month, such that Y.0„ km for k=\,..JSl 

 represents the total number of prey species present in 

 the ith sample. Therefore, if only one prey species oc- 

 curred in a sample, its occurrence was scored as 1, if 

 two prey species occurred each occurrence was scored 

 as 0.5, and so forth. The split-sample index is consid- 

 ered an improvement over conventional frequency of 

 occurrence indices in that prey species that comprised 

 an entire meal, which had presumably been consumed 

 in large quantities, weighted the split-sample index 

 more than prey species consumed in a diverse meal 

 comprising many species, each of which had presum- 

 ably been consumed in smaller quantities. Estimates 

 of the diet composition on the tth date, P Ekl and P 0kl , 

 were derived by linearly interpolating between the 

 monthly estimates, P Ekni and P ok „„ plotted at the mid- 

 point of each month. 



