Mathews et at: Effects of vessel disturbance on Phoca vitulina on glacial ice in Tracy Arm, Alaska 
197 
Group size 
increase of 1 (P=0.03) 
Pup present (no) 
yes (P=0.01) 
Age category (non-pup) 
pup (P=0.15) 
Age * distance (P=0.49) 
Percent Ice cover 
increase of 1% (P=0.03) 
Sky condition (clear) 
partly cloudy (P<0,01) 
broken-overcast (P=0.04) 
Temperature 
increase of 1°C (P=0.49) 
Precipitation (heavy) 
none (P=0.80) 
light (P= 0.85) 
Vessel activity (stopped) 
moving (p<0. 01) 
Vessel type (P<0.01) 
Vessel distance (P<0 01 ) 
Vessel type * distance (P<0 01) 
0.1 1.0 10.0 
Less likely More likely 
Probability of seals entering the water 
Odds ratio 
Figure 6 
Odds ratios of a harbor seal (Phoca vitulina ) entering the water in response 
to vessels approaching or remaining near an iceberg where the seal was 
hauled out. Bars indicate 95% confidence intervals (CIs). Solid black symbols 
indicate factors that were statistically significant ( i.e. , where the Cl does 
not cross 1). The odds ratios are relative to what we used as the comparison 
category for each variable (in parentheses). Observations were made from 27 
May through 30 June 2001. 
(not shown) 
(see Figure 7) 
(see Figure 7) 
(see Figure 7) 
6 
Focal observations 
during 
vessel approaches 
upper limit 
upper limit j 73.5 
Vessels and seals 
Vessel traffic We documented high levels of vessel ac- 
tivity in Tracy Arm in 2001 (10.2 vessels/day (maxi- 
mum 33 vessels/day) during 32 days of observation), 
compared with Johns Hopkins Inlet in GBNP during 
months when vessels could enter the inlet (3.7 vessels/ 
day, 1994-2001, E. A. Mathews, unpubl. data; 2.8 ves- 
sels/day, 178 vessels during 64 days of observation in 
July 2007 and 2008, Young et al., 2014). After 2001, 
vessel traffic decreased in Tracy Arm, whereas it in- 
creased in nearby Endicott Arm (Fig. 4), likely because 
of the rapid recession of South Sawyer Glacier that 
increased the number of icebergs in the water, mak- 
ing maneuvering boats and ships in Tracy Arm more 
difficult and potentially hazardous. From 2001 through 
2013, the combined vessel traffic in Tracy Arm and 
Endicott Arm remained high and may have increased 
(Fig. 4). 
Seal counts and vessel numbers Higher vessel counts 
were not associated with reduced seal counts in 2001 
(Table 1), a pattern also noted by Jansen et al. (2015) 
in Disenchantment Bay. However, our total counts in- 
cluded seals in the water, and therefore they were less 
sensitive to vessel disturbance. Furthermore, there 
could have been an effect of vessels on seal counts that 
we were unable to detect because of the constant daily 
presence of boats. In Muir Inlet in GBNP, peak counts 
of seals on days with no vessels were, on average, 15% 
higher than counts when vessels were present (Calam- 
bokidis et al. 1 ). 
Seal-vessel interactions All of our analyses of seal-ves- 
sel interactions (i.e., randomized focal observations and 
direct observations of vessel approaches) revealed in- 
creased probabilities that seals would enter the water 
in response to at least some types of boats (Tables 1, 
Figs. 5-7). We also found that the probability of a seal 
