344 
Fishery Bulletin 95(2), 1997 
■ 61-100% FULL 
^ 3 1 -60% FULL 
□ 0-30% FULL 
Time (h) 
Figure 4 
Percent of the sample in each of the stomach fullness catego- 
ries: 0-30%, 31-60%, and 61-100% for spotted dolphins col- 
lected between 0600-1800 h: 0600-0900 h (n=106), 0901-1200 
h (n= 87), 1201-1500 h (n = 134), and 1501-1800 h (ra = 101). 
The SFI was scaled to 100% by using the stomach of maxi- 
mum fullness in the sample (see text for details). 
1974; Sidwell, 1981; Croxall and Prince, 1982). A 
possible explanation for increased consumption of 
squid by lactating spotted dolphins is that milk pro- 
duction increases the demand for metabolic water 
(Croxall and Prince, 1982; Young and Cockcroft, 1994) 
and most squid have a higher water content than 
fish (Sidwell, 1981; Croxall and Prince, 1982). Ceta- 
cean milk has been estimated to be 40-60% water 
(Eichelberger et al., 1940; Gregory et al., 1955; 
Slijper, 1966; Best, 1982; Lockyer, 1984). 
Rather than by a change in diet composition, the 
high metabolic demands of lactation could be met by 
increasing the amount of food consumed (Baldwin, 
1978; Millar, 1979; Clutton-Brock et al., 1982; Yasui 
and Gaskin, 1986; Perez and Mooney, 1986; Kastelein 
et al., 1993). Estimates of food intake for lactating 
versus nonlactating females have been estimated to 
increase by 75-86% for lactating minke whales 
(Balaenoptera acutorostrata ) and fin whales ( Balaen - 
optera physalus) (Lockyer, 1978, 1981), by 500 g for 
lactating harbor porpoise ( Phocoena phocoena) 
(Recchia and Read, 1989), and by 30% for captive 
Commerson’s dolphins ( Cephalorhynchus com- 
mersoni) (Kastelein et al., 1993). For spotted dol- 
phins, Bernard and Hohn (1989) reported a SFI that 
was 24% higher for lactating females (Mann- Whitney, 
0.05<P<0.10). Similarly, we found that the SFI of 
lactating females was 16% higher than that for preg- 
nant females and that a higher SFI was maintained 
by lactating females throughout the day (Table 4). 
This finding parallels those of Cockcroft and Ross 
(1990) in which they estimated that lactating 
bottlenose dolphins (T hrsiops truncatus) would 
need to consume 3 or 4 stomachfuls of food per 
day to keep up with energetic demands, whereas 
males and resting females (not pregnant or lac- 
tating) had to consume only 2 stomachfuls to main- 
tain their energetic requirements. Our results also 
suggest that lactating spotted dolphins may con- 
sume more food throughout the day to meet the 
higher energetic and nutritional demands of 
lactation. 
Biases associated with analyses 
As in every food-habit study, there are inherent 
biases associated with the analyses. Differential 
prey digestibility and secondary prey can bias all 
measures. The importance of small prey species 
can be overestimated by methods that determine 
both the percent number and the percent frequency 
of occurrence. Furthermore, infrequently con- 
sumed species may be overemphasized by the per- 
cent frequency-of-occurrence method (Hyslop, 1980; 
Bigg and Fawcett, 1985; Bigg and Perez, 1985). 
Analysis of only hard parts can be biased through 
differential passage, retention, and degradation rates 
of beaks and otoliths (Hyslop, 1980; Bigg and 
Fawcett, 1985; Bigg and Perez, 1985; Pierce and 
Boyle, 1991). The importance of squid can be exag- 
gerated by using only hard parts because beaks tend 
to get stuck in the stomach rugae and accumulate 
(Ross, 1979; Shroud et al., 1981; Clarke, 1986a). 
There are few data available on the passage and re- 
tention rates of various prey in cetaceans, although 
work on captive bottlenose dolphins (TYirsiops 
aduncus) has indicated that fish otoliths were re- 
tained for up to 48 h and squid beaks for up to 72 h 
(Ross, 1979). Similarly, Clarke (1980) reported that 
sperm whales (Physeter macrocephalus ) retained 
squid beaks for 36-60 h. The only way to avoid bi- 
ases introduced by using hard parts is to use only 
fresh or whole prey recovered from stomachs. Al- 
though this method can tremendously limit the 
sample size and underestimate the importance of 
squid because squid flesh digests more quickly than 
fish flesh (Bigg and Fawcett, 1985). 
Another source of bias may have affected the in- 
terpretation of our results. Ninety-six percent (96%) 
or 411 of the stomachs were recovered from net sets 
from which more than one stomach was collected. 
We tested whether multiple samples per set would 
bias our interpretation of important prey species by 
calculating the percent number for prey families, 
