192 
Fishery Bulletin 1 14(2) 
(Fig. 4). Although vessel traffic declined in Tracy Arm, 
the number of vessels counted in nearby Endicott Arm 
increased (U.S. Forest Service 9 ; Fig. 4). 
Seal-vessel interactions 
Randomized focal observations We made 662 10-min 
observations of focal seal groups, and recorded the be- 
havior of 3250 seals (group size: 1-24, median 4). The 
probability of a seal entering the water, during a 10-min 
period when no vessels were present — an undisturbed 
9 U.S. Forest Service. 2002-2013. Unpubl. data. Juneau 
Ranger District Wilderness Program, U.S. Forest Service, Ju- 
neau, Alaska 99801. 
baseline level for the entering the water behavior — 
averaged across all other predictor variables, was 0.06 
(95% Cl: 0.05-0.08). In other words, every 10 minutes 
we would expect 6 of 100 undisturbed seals, on aver- 
age, to enter the water. The odds of a seal entering the 
water was not related to DOY or TOD (P>0.17), to ice 
type (dense versus scattered), or to any of the weather 
predictors (Fig. 5). The presence of a pup, usually in- 
dicative of a mother-pup pair, and the number of pups 
were associated with reduced probability of a seal from 
those groups entering the water (Fig. 5, pup present). 
Vessel type affected the probability of seals enter- 
ing the water in different ways (Fig. 5, vessel type). 
The odds of seals entering the water were >2 times 
higher when vessels were present in the study area 
