498 



Fishery Bulletin 91(3), 1993 



Although it was not possible to assess the assump- 

 tion that all prey in a meal had been consumed in 

 equal volumes (see Results), the maximum potential 

 biases introduced by deviations from the assumption 

 could be determined. An upper limit of the importance 

 of a particular prey species in the overall diet was 

 obtained by assuming that whenever the species oc- 

 curred in a meal, it comprised the entire meal and 

 that all other species in the same meal had been con- 

 sumed in negligible quantities. Conversely, a lower limit 

 was obtained by assuming that whenever a particular 

 species was consumed in the same meal along with 

 other species, it had been consumed in negligible quan- 

 tities. Mathematically, the upper and lower limits for 

 the /th of k=l,...JV species were calculated from Equa- 

 tion 16 by 



1) Setting O ijkm =l for k=l and 0„,„ =0 for k±l; 



2) Setting O ijk „ =0 for k=l and O ijkm =l for kd when N>1, 



respectively. 



Two corrections were applied to the split-sample in- 

 dex to account for suspected biases. First, very small 

 scat samples, which undoubtedly represented only a 

 small fraction of whole scats and probably contained 

 only a portion of all the prey species actually con- 

 sumed in meals, were weighted less than large scat 

 samples. Second, the relative proportions of hake and 

 herring, the two predominant prey species, in samples 

 that contained both species were volumetrically 

 weighted based on the mean relative number of ele- 

 ments of each species in the sample compared with 

 the relative number in samples that were composed 

 exclusively of each of these two species. Both correc- 

 tions, each of which had a relatively minor influence 

 on the results, are described in detail in Olesiuk et al. 

 (1990b). 



Results 



Abundance and distribution 



The two complete censuses of the study area in 1988, 

 when adjusted to post-pupping levels and corrected for 

 missed seals (Eqn. 3), yielded population estimates of 

 16,531 and 15,091, respectively. Their mean, 15,810, 

 was adopted in subsequent analyses. The mean was 

 considered most appropriate because the proportion of 

 the population counted during any single census may 

 have been lower or greater than the mean estimate of 

 0.884. The validity of the 1988 census estimate was 

 substantiated by a projected population estimate of 

 15,050 based on population trends during 1973-88, 

 over which period the population had been steadily 



increasing (log-linear r 2 =0.994) at a rate of 12.5% per 

 annum (Olesiuk et al., 1990a). 



Seasonal fluctuations in the size of population as a 

 result of mortality and recruitment are shown in Fig- 

 ure 3A. The total size of the population ranged from a 

 low of 12,990 just prior to the pupping season (16 June) 

 to a peak of 15,810 at the end of the pupping season (6 

 September). The mean size of the population was 

 14,270. Because the population was below carrying ca- 

 pacity and increasing at its intrinsic rate, it was 12.5% 

 larger at the end than at the beginning of the year. 



Seasonal changes in the distribution of harbor seals 

 between estuaries and non-estuaries are shown in Fig- 

 ure 3B. The overall abundance within estuaries was 

 lowest (4% of the total population) in December and 

 remained low throughout the winter and into early 

 spring. Numbers increased sharply during June- 

 August, mainly due to large influxes into the two larg- 

 est estuaries — Boundary Bay and lower Fraser River 

 (Fig. 1). These influxes coincided with the pupping sea- 

 son (Bigg, 1969; Olesiuk et al, 1990a) which might 

 indicate these two large estuaries were preferred whelp- 

 ing areas, as has been reported for other major es- 



M J J A 

 MONTH 



Figure 3 



Seasonal fluctuations in (Al the total size of the Strait of 

 Georgia harbor seal population, N„ due to recruitment and 

 mortality; and (B) the total number of seals inhabiting estu- 

 aries, N Rv Note that the two parameters are plotted on differ- 

 ent scales. 



