254 



Fishery Bulletin 98<2) 



hypothesis of a complete overlap {G0=1) can be sta- 

 tistically tested by using the V-statistic (Ludwig and 

 Reynolds, 1988). 



The SO is a pairwise nonsymmetric index which 

 evaluates the probability of obtaining the utilization 

 curve of one group from the utilization curve of the 

 other (Petraitis, 1979; Ludwig and Reynolds, 1988). 

 The probability of obtaining the utilization curve of 

 the group / from the utilization curve of group k is 

 denoted by SO,^. The null hypothesis of a complete 

 overlap (S0^.=1) can be tested with the fZ-statistic 

 (Ludwig and Reynolds, 1988). 



The comparisons made were 1 > geographical area 

 of sampling (between northern Patagonia and cen- 

 tral Patagonia), 2) period of sampling (between 

 1982-1987 and 1990-1998), 3) source of sampling 

 (between dead animals on shore and entangled ani- 

 mals in the fishery), and 4) sex (between males and 

 females). Because entangled individuals were mostly 

 males and were obtained only during the 1990-1998 

 period (Table 1), comparisons were made between 

 entangled and nonentangled males in the period 

 1990-1998. 



The data employed for this analysis were those 

 data on the occurrences of prey species that pre- 

 sented an %IRI greater than 2'7c in the pooled sample. 

 Inherent in the use of occurrences of prey species 

 is the assumption that each prey species in a stom- 

 ach was consumed independently. For this reason, 

 the correlations between the prey species used in 

 our analyses were evaluated with the Spearman 

 rank correlation coefficient (r^) (Siegel and Castel- 

 lan, 1995). The use of occurrences of prey species as 

 data also increases the sample size for these com- 

 parisons because the stomach of one sea lion usually 

 contained more than one prey species. 



Differences in prey sizes consumed were tested by 

 using the nonparametric two-sample Mann-Whitney 

 U test in those cases where differences were detected 

 (Siegel and Castellan, 1995). 



The relationship between mean length of prey in 

 each stomach and predator SL was evaluated using 

 the Tj, (Siegel and Castellan, 1995). This analysis was 

 performed by using the pooled sample and analyzed 

 by sex of predator. 



The function of gastroliths 



The role of gastroliths in eliminating stomach par- 

 asites and the potential function of gastroliths in 

 buoyancy control were investigated. The indepen- 

 dence between the presence of gastroliths and the 

 presence of parasites in the stomachs was tested 

 with Fisher's exact test. The relationship between SL 

 and total weight of gastroliths found in the stomach 



(GW) was evaluated with the Spearman rank cor- 

 relation coefficient (r^) (Siegel and Castellan, 1995). 

 This relationship was analyzed by considering each 

 sex and the pooled sample. 



Results 



Prey species 



Forty-eight of 59 stomachs analyzed contained food 

 remains (Table 1). Aproximately 37 prey species 

 were identified, mostly fishes and cephalopods (Table 

 3). Additionally, the stomach of one female found 

 dead on the beach contained two sponge species, 

 tube polychetes, nudibranchs and hagfish (Mixine 

 sp.). Because this specimen was considered sick and 

 anomalous, it was excluded from the analysis. 



The collection analyzed was composed of 1449 

 individual prey, and the total estimated weight was 

 209.9 kg. 



Males consumed a broader trophic spectrum of 32 

 prey species (Table 3), dominated by Argentine hake, 

 followed by Patagonian squid, Loligogahi, Argentine 

 shortfin squid, "raneya," Raneya brasiliensis, and 

 red octopus, Enteroctopus megalocyathus. Only the 

 Argentine hake had a %IRI greater than 10%. The 

 total number of prey found in male sea lions stom- 

 achs was 738 and the total estimated biomass was 

 91.4 kg. The five most important prey represented 

 74.0% by number and 74.6% by weight. 



Twenty-nine species were found in female sea lion 

 stomachs. The important prey were red octopus, 

 Argentine shortfin squid, Argentine hake, "raneya," 

 and Argentine anchovy (Engraulis anchoita) (Table 

 3). Only the first three species had %IRIs greater 

 than 10%. The total number of prey found in the 

 female stomachs was 711 and the estimated weight 

 of this collection was 118.5 kg. The five most impor- 

 tant prey represented 75.5%> by number and 91.3% 

 by weight. 



Homogeneity of the sample 



Six species (Argentine hake, red octopus, Argentine 

 shortfin squid, Patagonian squid, "raneya," and 

 Argentine anchovy) had an %IRI greater than 2% in 

 the pooled sample (Fig. 3). All pairwise correlations 

 for these prey species were nonsignificant (P>0.05). 

 Data on the occurrences of these species were used 

 in the overlap analysis. 



Considering the GO, no differences in diet were 

 found between geographical areas and between peri- 

 ods of sampling (Table 4). However, the SO indicates 

 differences in diet between the periods 1990-1998 and 



