36 
Fishery Bulletin 99(1 ) 
along the California coast, they probably respond to 
seasonal or interannual oceanographic changes. 
Large swells (3 m) during our ship survey may have 
caused observers to miss more porpoise groups on the 
trackline than is accounted for by our g{0) correction 
factor. The independent observer recorded an insuffi- 
cient number of sightings during our survey; therefore 
we could not independently estimate g(0). Instead, we 
adopted theg(O) value of 0.769 (SE=0.117) estimated 
by Barlow 5 who used the same searching method and 
vessel type as we did. During our survey, the largest 
swells (3 m) occurred in northern California, where 
porpoise densities are highest. Further, swell height 
increased with latitude in California and the slope 
of the regression was significant (Zar, 1984; AN OVA, 
P«0.005). Even in calm sea states, porpoise groups 
may be easily missed because animals surface behind 
large swells and thus are not observed. Our transect 
lines frequently were oriented directly into the pre- 
vailing northwest swell or directly down-swell as a 
consequence of our sampling design, possibly contrib- 
uting to more groups being missed on the transect 
line than expected. If swells were considerably larger 
during 1995 survey than during the 1980s surveys, 
then Barlow’s g(0) value may not reflect the fraction 
of trackline groups missed and our estimate of abun- 
dance would be negatively biased by an unknown, 
but nontrivial, amount. Larger winter swells may also 
help explain why porpoise densities from winter ves- 
sel surveys are generally lower than those from sum- 
mer and autumn surveys (Goetz, 1983; Barlow, 1988; 
Sekiguchi, 1995; Calambokidis et al. 7 ). 
Abundance in northern California may be under- 
estimated because the area inshore of the 20-m isobath, 
known to have relatively high porpoise densities, was not 
routinely surveyed. Only 2% of all survey effort and 4% 
of all sightings occurred at depths shallower than 20 m. 
Shallow water effort is included in our abundance esti- 
mate, thus we implicitly assume that porpoise density 
from shore to 20 m is equal to that in the remainder of 
our study area. Depth data from aerial surveys conduct- 
ed from 1988 to 1997 support this assumption. Porpoise 
encounter rates (weighted by effort for each depth stra- 
tum) are nearly equal for the area from shore to 90 m 
(0.33 porpoise/km) and from 20 to 90 m (0.32 porpoise/ 
km) (NMFS 9 ). However, these data were collected primar- 
ily during September and October and may not reflect 
the depth distribution of harbor porpoise in November. 
During our survey, if porpoise densities were significantly 
higher from shore to 20 m than from 20 to 90 m, our es- 
timate of porpoise abundance would be negatively biased 
by an unknown amount. The same bias might result if 
significant numbers of harbor porpoise moved into waters 
deeper than 90 m between the time of the aerial and ship 
surveys. 
B 1 1 B a m _ . 
fd 
o 
o 
o 
o 
Simulated abundance 
o 
o 
o 
o 
■ SHIP N = 5686 CV = 0.29 0 AERIAL A/= 13,145 CV = 0.39 
a; 
-O 
E 
Difference in abundance (ship minus aerial) 
Figure 5 
Distribution of simulated log-normal abundance estimates for 
1995 ship and aerial surveys in northern California (A). Dif- 
ference between 5000 simulated abundance estimates for 1995 
ship and aerial survey in northern California (B). 
The distribution of harbor porpoise in northern Califor- 
nia is highly clustered, especially near Cape Mendocino, 
where densities are highest, and our level of survey effort 
may not have fully captured the spatial heterogeneity in 
porpoise densities in northern California. The bootstrap 
CV of ship abundance estimate was not extremely precise 
(0.29) and reflects the spatial variability in porpoise en- 
counter rates in northern California on small (5-km boot- 
strap unit) scales. The estimated number of transect effort 
in kilometers needed to attain higher levels of precision 
for the northern California estimate, following the method 
described in Buckland et al. (1993, p. 303-304) were ap- 
proximately 800 and 3200 km, respectively, to obtain CVs 
of 0.20 and 0.10 (we surveyed 377 km). These effort esti- 
mates assume a Poisson distribution of porpoise, and are 
probably conservative. Repeated sampling of our transect 
lines over a longer period may have yielded an estimate 
of abundance more similar to recent aerial surveys. An 
adaptive sampling method, as used by Palka and Pollard 
(1999), would be useful on future surveys to increase sam- 
pling effort in known high-density regions. 
Abundance of harbor porpoise off central California 
Region 1 has the lowest densities of harbor porpoise in 
California (Barlow, 1988; Forney et al., 1991; Dohl et al. 3 ) 
and we did not detect any porpoise during 60 km of survey 
9 NMFS (National Marine Fisheries Service). 1999. Unpubl. 
data. Protected Resources Div., Southwest Fisheries Science 
Center, P.O. Box 271, La Jolla, CA. 92038. 
