FISHERY BULLETIN: VOL. 84, NO. 2 



energy budget for females and pups during the first 

 two months of the reproductive cycle. 



Although most mammals ingest more food while 

 they are lactating than they would in a nonlactating 

 state, many species of phocid seals fast during the 

 lactation period (Harrison 1969). These seals (e.g., 

 gray seal, Halichoerus grypus, and northern ele- 

 phant seal, Mirounga angustirostris) do not feed 

 from parturition to weaning of the young, and all 

 of their energy needs during lactation must be met 

 by metabolism of in situ energy such as fat reserves. 

 This behavior has been well documented for the gray 

 seal (e.g., Amoroso and Matthews 1951, 1952; Fedak 

 and Anderson 1982) and also for the harp seal, Phoca 

 groenlandica, (Lavigne et al. 1982). However, 

 metabolism of fat reserves does not reduce the 

 energetic costs of producing offspring; it merely 

 shifts the time that energy must be acquired, at 

 some energy cost for storage (Millar 1978). 



The objectives of this study were to show, using 

 both stomach content and body mass data, that lac- 

 tating female fur seals ingest more food than nonlac- 

 tating females in order to meet their increased 

 energy requirements for maintenance and milk pro- 

 duction, and to make estimates of the magnitude of 

 this difference in food ingestion. For this study, we 

 utilized data from postpartum and nonpregnant 

 adult, female northern fur seals, Callorhinus ur- 

 sinus, taken pelagically during 1958-74. 



METHODS 



Data on the contents of stomachs from the female 

 northern fur seals taken pelagically (Fig. 1) in the 

 eastern Bering Sea during 1958-74 by the United 

 States and Canada were analyzed to determine the 

 relation between lactation and food consumption 

 during the summer breeding season. Only data from 

 female fur seals (age >4 yr) which had information 

 on both body mass and stomach content mass were 

 included. Age was determined from longitudinal 

 half-sections of the upper canine teeth by counting 

 the annual growth layers in the dentine, a method 

 widely accepted by researchers during recent 

 decades to determine the age of many species of 

 mammals (Klevezal' and Kleinenberg 1967). 

 Methods used during 1958-74 to determine age, 

 reproductive status, and the different items in the 

 stomachs were discussed by Lander (1980). 



The data used in this study represented stomach 

 contents under different stages of digestion; how- 

 ever, it was not possible to make comparisons be- 

 tween stages because no data on stages of diges- 

 tion were recorded. Rates of digestion of all prey 



were assumed to be similar for all females during 

 the same time interval. In our study, all postpartum 

 females were considered lactating, and all non- 

 pregnant (not postpartum) females were considered 

 nonlactating. 



Statistical Methods 



The cumulative frequency distributions of data on 

 mass of total stomach contents for both lactating 

 and nonlactating females were compared using the 

 one-tail Kolmogorov-Smirnov two-sample test 

 (Siegel 1956). 



Data from seals with empty stomachs or stomachs 

 with only a trace of contents (i.e., <10 cc) were con- 

 sidered as zero mass and pooled with data from seals 

 with food in their stomachs. Data for different ages 

 and months were pooled to provide sufficient sam- 

 ple size for analysis because the normal approxima- 

 tion to compute confidence limits is only valid if sam- 

 ple sizes are adequate (Cochran 1977). In order to 

 use parametric statistics, and yet not seriously 

 violate basic assumptions of normality, data were 

 transformed by the modified arcsine transformation 

 discussed by Zar (1974): 



X = \/ M + 0.5 arcsin \/ (S + 0.375)/(M + 0.75) 



where M is the net body mass (excluding mass of 

 stomach contents, S) and X is the transformed value. 

 This equation was used because of its utility where 

 a large number of the data were from stomachs con- 

 taining only a trace or less. 



The transformed values on the mass of total 

 stomach contents (expressed as a percentage of net 

 body mass) obtained from the above equation were 

 transformed back to percentages to obtain means. 

 We calculated an index of the relative intake of food 

 by lactating females compared with that of non- 

 lactating females by multiplying the ratio of their 

 respective mean mass of stomach contents by 100. 

 The £-test for two independent samples, with the 

 assumption of unequal variance (Snedecor and 

 Cochran 1980), was used on the transformed data 

 to determine if any significant difference in total 

 food consumption and body mass existed between 

 females of different reproductive status. 



The relative importance of individual prey in the 

 total diet was assessed using the modified volume 

 percentage method (Bigg and Perez 1985). Only 

 foods with fleshy remains were used as evidence of 

 diet in this method, and the procedure combined the 

 traditional methods of volume and frequency of oc- 

 currence. The proportion of total fish and total squid 



372 



