Figure 7. Total numbers of bald eagles on land in Race Rocks Ecological Reserve on each of 2 daily censuses
(1=morning; 2=afternoon) as observed from atop the light tower. Monitored days when the Bentinck Island
demolition range was active are indicated by vertical lines.

5.1.3 Gulls
Eight species of gull were observed during the census counts (Appendix 5). The most abundant
by far was Thayer’s gull. This species breeds in the arctic, but winters in the study area; hence
the large peaks in gull abundance during the late autumn and winter (Figure 8). The next most
abundant was the glaucous-winged gull. It was the only species present year-round and the only
one observed to breed in the study area. Maximum daily counts of all gulls combined ranged
from 39 in the late winter to approximately 9000 in late autumn (Figure 8). There was no
consistent pattern in the difference between census counts in the morning and in the afternoon.
On most days during autumn and winter, numbers in the area fluctuated greatly and repeatedly as
birds moved between Race Rocks and nearby feeding areas in Juan de Fuca Strait.

Glaucous-winged gulls were observed to breed in the grassy areas of Great Race Rock (see
Photo 1). Breeding behaviour (copulating) was first noted on 2 May 2003. By 25 May, some
birds had initiated incubation, while others continued to court. The first chicks were observed on
5 July. Broods of 2-3 chicks were typical. On 15 August, juveniles were first observed taking
“practice” flights. On 2 September, all juveniles were flying except for three chicks from a clutch
that hatched much later than the rest. The peak counts of glaucous-winged gull chicks were made
on 7 August when 77 were counted in the morning and 63 in the afternoon. We did not attempt
to quantify chick mortality, but as many as six carcasses were observed at one time.

)

5.1.5 Harbour Seal
Harbour seals were the most abundant pinniped throughout the study. Maximum daily counts
ranged from 0 in the winter to 667 in the summer (Figure 10). The peak in numbers during
summer corresponds to the breeding and moulting seasons. There were predictable differences
between census counts in the morning and in the afternoon. In winter when tides tended to be
lower during the afternoon compared to the morning, afternoon counts of harbour seals tended to
be higher. Conversely, fewer harbour seals in the afternoons during June through October
corresponded to higher tides at that time (Figure 4). The effects of tide on harbour seals are
explored further in section 5.2.1.4. The maximum differences between any pair of daily surveys
(n=49 pairs) were +167 and -472 animals, the median daily difference was -30. Sex and age class
data were not recorded for all animals, so only general conclusions can be made. Based on size
and general appearance, both sexes and all age classes were present. Pupping activity was most
apparent during the month of July. On 17 July 2003 we witnessed a birth in Sub-Area F on Great
Race Rock. The pup was born at 14:25, began nursing at 15:35 and began swimming at 16:30 as
the tide rose to where it was lying. The highest number of pups recorded on a single census was
27 on 27 July 2003.

Figure 10. Total numbers of harbour seals hauled out in Race Rocks Ecological Reserve on each of 2 daily censuses
(1=morning; 2=afternoon) as observed from atop the light tower. Monitored days when the Bentinck Island
demolition range was active are indicated by vertical lines.

5.1.6 California Sea Lion
California sea lions were the third most abundant pinniped during the study. Maximum daily
counts ranged from 0 in the winter and summer to 244 in late summer (Figure 11). There was no
consistent pattern in the difference between census counts in the morning and in the afternoon.
The maximum differences between any pair of daily surveys (n=49 pairs) were +43 and -43
animals, the median daily difference was -1. Sex and age class data were not recorded, but based
on size and general appearance, the vast majority of individuals were adult males. The peaks in
abundance observed during: (1) autumn 2002 and late summer-early autumn 2003, and (2) spring
2003, correspond to the times of year when animals are moving north into, then south out of
British Columbian waters, respectively. It is not clear what movement patterns the peak during
December 2002 might have represented.

Figure 11. Total numbers of California sea lions hauled out in Race Rocks Ecological Reserve on each of 2 daily
censuses (1=morning; 2=afternoon ) as observed from atop the light tower. Monitored days when the Bentinck
Island demolition range was active are indicated by vertical lines.

5.1.7 Northern Sea Lion
Northern sea lions were the second most abundant pinniped during the study. Maximum daily counts
ranged from 0 in the summer to 555 in late autumn (Figure 12). Except for some days when blasting
on Bentinck Island occurred, total numbers of northern sea lions were typically higher during the
afternoon than in the morning (see section 5.2.1.6). The maximum differences between any pair of
daily surveys (n=49 pairs) were +279 and -100 animals, the median daily difference was 0. Sex and
age class data were not recorded, but based on size and general appearance, both sexes and all age
classes were present during the autumn and winter; the few animals present in April 2003 were large
males. The peaks in abundance observed during late autumn and winter both years correspond to the
time of year when animals are moving away from breeding colonies. It is not clear what proportion
of the animals at Race Rocks are from breeding colonies such as those off northern Vancouver
Island versus those from colonies to the south in California and Oregon. In any event, all individuals
that occur at Race Rocks are believed to be from the non-endangered, Eastern Stock.

Figure 12. Total numbers of northern sea lions hauled out in Race Rocks Ecological Reserve on each of 2 daily
censuses (1=morning; 2=afternoon ) as observed from atop the light tower. Monitored days when the Bentinck
Island demolition range was active are indicated by vertical lines.

5.2 Natural and Human-Caused Disturbances
Effects of factors that potentially disturb birds and pinnipeds at Race Rocks are divided into two
categories, treated separately under sections 5.2.1 and 5.2.2:

• 1. Effects on animal abundance caused by factors operating over the course of a monitored
  day. These include a mix of natural and human-caused factors such as environmental
  factors, whether or not military ranges were active, and the extent of range activity.
  2. Effects on animal behaviour and abundance caused by factors operating within a
  monitored day. These include a mix of discrete factors that are natural and human-caused
  in origin. They include predators, boats, aircraft, pedestrians, foghorn, and military
  explosions.
  

5.2.1 Effects of Disturbance on Animal Abundance from Morning to Afternoon
5.2.1.1 Cormorants
According to logistic regression analysis, the relative abundance of cormorants during the day
was not significantly described by a model using the independent parameters of tide, swell
condition or activities on Bentinck Island (or a combination of these; all P >>0.050). Further,
none of the individual terms were significant within any of the models examined (all P >0.050).
A logistic regression model of the displacement responses of cormorants during demolition runs
on Bentinck Island was significant (Table 5). The model that provided the best fit to the data
consisted of the independent parameter air temperature, whereby higher air temperatures
decreased the probability of displacement.
5.2.1.2 Gulls
According to logistic regression analysis, the relative abundance of gulls during the day was best
predicted by a model with swell height as the lone term (P = 0.043; Table 4). None of the other
model terms were significant. A graphical depiction of the data partitioned by relative swell
condition on days with and without blasting is shown in Figure 13. From that chart it is apparent
that higher swell conditions in the afternoon tended to increase the abundance of gulls, and
seemingly more so on days when blasting occurred on Bentinck Island, but that interaction was
not significant (Table 5). Note, however, that some sample sizes are particularly small.

Table 4. Results of logistic regression analyses to detect the effect of tide, swell height, and Bentinck Island
demolitions on the relative change in numbers of animals in the Race Rocks census area in the afternoon versus
morning periods from October 2002 through November 2003. Refer to Table 1 for information on corresponding
dates. Dates when only the Whirl Bay Range was active are excluded. Whole Model Test indicates the overall
significance of the model; Lack of Fit indicates the significance of any lack of fit of the selected set of terms; Odds
Ratio indicates how the odds of animal displacement are affected by an increase in the model parameter. Odds ratios
>1 correspond to increased probability of higher numbers in the afternoon than in the morning. All models were
fitted with an intercept term (B0).

Table 5. Results of logistic regression analyses to detect the effect of number of projects in a run, tide height, swell
height, air temperature, wind speed, wind direction, sea state, cloud cover, precipitation class, and visibility class on
animal displacement during demolition runs on Bentinck Island from October 2002 through November 2003. Whole
Model Test indicates the overall significance of the model; Lack of Fit indicates the significance of any lack of fit of
the selected set of terms; Odds Ratio indicates how the odds of animal displacement are affected by an increase in
the model parameter. Odds ratios >1 correspond to increased probability of displacement. All models were fitted
with an intercept term (B0).

Figure 13. Boxplot summaries of daily changes in the abundance of gulls when Bentinck Island was inactive and
when blasting occurred there, grouped by net change in swell condition in the afternoon versus morning period.
Refer to Table 1 for information on corresponding dates. Dates when only the Whirl Bay Range was active are
excluded. Sample sizes are indicated.

None of the logistic regression models of the displacement responses of gulls during demolition
runs on Bentinck Island were significant (Table 5). This reflected the fact that blasts on Bentinck
Island rarely flushed gulls from the land surfaces of the study area (refer to section 5.2.2.9 for
further information).
5.2.1.3 Northern Elephant Seal
According to logistic regression analysis, the relative abundance of elephant seals during the day
was best predicted by a complex interaction term which involved the product of tide, swell and
Bentinck Island activities (P=0.028; Table 4). None of the other model terms were significant.
5.2.1.4 Harbour Seal
According to logistic regression analysis, the relative abundance of harbour seals during the day
was best predicted by a model with tide as the lone term (P=0.002; Table 4). None of the other
model terms were significant. A graphical depiction of the data partitioned by relative tide height
on days with and without blasting is shown in Figure 14. From that figure it is clear that the
greatest changes in abundance were decreases resulting from tidal effects. On days when the tide
was lower during the afternoon survey, there was also a tendency for numbers of seals to
increase on non-blasting days and decrease on blasting days, but according to logistic regression
analysis, this interaction was not significant. Flood tides regularly forced hundreds of harbour seals

into the water as haulouts were submerged (Photo 11).

Figure 14. Boxplot summaries of daily changes in the abundance of harbour seals when Bentinck Island was
inactive and when blasting occurred there, grouped by net change in tide height in the afternoon versus morning
period. Refer to Table 1 for information on corresponding dates. Dates when only the Whirl Bay Range was active
are excluded. Sample sizes are indicated.

A logistic regression model of the displacement responses of harbour seals during demolition
runs on Bentinck Island was significant (Table 5). The model which provided the best fit to the
data consisted of the independent terms of: number of projects in a run (higher numbers of
projects increased the probability of displacement); swell height (higher swell heights reduced
the probability of displacement); and air temperature (higher air temperatures increased the
probability of displacement).
5.2.1.5 California Sea Lion
Logistic regression showed that the relative abundance of California sea lions during the day was
best predicted by a model with swell height as the lone term (P=0.023; Table 4). None of the
other model terms were significant. A graphical depiction of the data partitioned by relative
swell condition on days with and without blasting is shown in Figure 15. From that chart it is
apparent that higher swell conditions in the afternoon tended to reduce the abundance of

California sea lions, but note that some sample sizes are small.

Photo 11. Example of how tide height affected haulout availability for harbour seals, including during the peak of
their pupping season. Net tidal difference of the sequence is (1.51 m – 0.52 m =) 0.99 m. Photos show time (hh:mm)
and number of harbour seals hauled out on a portion of Sub-Area F, as viewed from atop the light tower at Race
Rocks (see Appendix 2). Effects of wind-waves, swells, and human-caused disturbance were negligible. Predicted
tide heights for William Head (Hopper 2002). 17 July 2003.

Figure 15. Boxplot summaries of daily changes in the abundance of California sea lions when Bentinck Island was
inactive and when blasting occurred there, grouped by net change in swell condition in the afternoon versus morning
period. Refer to Table 1 for information on corresponding dates. Dates when only the Whirl Bay Range was active
are excluded. Sample sizes are indicated.

A logistic regression model of the displacement responses of California sea lions during
demolition runs on Bentinck Island was significant (Table 5). The model which provided the best
fit to the data consisted of the independent terms of: number of projects in a run (higher numbers
of projects increased the probability of displacement) and wind direction (wind directions from
the vicinity of Bentinck Island toward Race Rocks increased the probability of displacement).

5.2.1.6 Northern Sea Lion
According to logistic regression analysis, the relative abundance of northern sea lions during the
day was not significantly related to any model tested (Table 4). The best model (P=0.104) was
one built with swell height and Bentinck activity values. It suggested that blasting on Bentinck
Island might have significantly decreased the relative daily abundance of northern sea lions if
larger sample sizes were examined. A graphical depiction of the data partitioned by relative
swell condition on days with and without blasting is shown in Figure 16. From that chart it is
apparent that relative numbers of northern sea lions tended to be lower in the afternoons of days
when Bentinck Island was active, and especially on such days when swell conditions were
higher. Note that some sample sizes are small.

None of the logistic regression models of the displacement responses of northern sea lions during
demolition runs on Bentinck Island were significant (Table 5). This reflected the fact that blasts
on Bentinck Island caused displacement of 1 individual of this species in the vast majority of
observed instances (Photo 12; refer to section 5.2.2.9 for further information).

Figure 16. Boxplot summaries of daily changes in the abundance of Northern sea lions when Bentinck Island was
inactive and when blasting occurred there, grouped by net change in swell condition in the afternoon versus morning period. Refer to Table 1 for information on corresponding dates. Dates when only the Whirl Bay Range was active are excluded. Sample sizes are indicated.

1.Preblast (09:40:00)
(resting behaviour) 2. Immediately post-blast (09:42:10) (sea lions active and moving toward
water)
3. After third blast(09:50:00)
(only northern elephant seals remain)
Photo 12. An example of typical pinniped responses to blasting. Northern elephant seals, California sea lions and
northern sea lions in Sub-Area 2-5 (see Figure 2) on 2 May 2003. Times are shown.

5.2.2 Effects of Discrete Disturbance Events on Animal Abundance and Behaviour
Two main types of animal behaviour were monitored: (1) activity levels of pinnipeds on a
haulout, and (2) displacement from land to the air (birds) or water (birds and pinnipeds). Results
of one-way ANOVA testing for differences between the mean activity levels of animals at the
beginning of the day (i.e., prior to discrete disturbances), grouped by days with: (a) no blasting;
(b) blasting on Bentinck Island; and (c) blasting in Whirl Bay only are summarized in Table 6.
Pre-discrete disturbance pinniped activity levels did not differ significantly among the types of
monitoring days. There were no significant differences in mean activity levels of pinnipeds
during samples throughout the day when no discrete disturbance stimulus was attributed to the
observation when the Bentinck Island Range was active compared to days when that range was
inactive (Table 7).

Ecotour boat traffic levels did not differ among the types of monitoring days. Results of a t-Test
for differences between the numbers of ecotour boats observed in the study area during days
when Bentinck Island was active (mean = 4.4; n = 8) and the subset of days when the range was
inactive (refer to Table 1; mean = 6.4; n = 11) indicated no significant difference (P = 0.459).
A comprehensive summary of disturbance events is presented in Appendix 6. Table 8 and Figure
17 summarize disturbance events with potential to flush animals in the study area. The total
numbers of detonations in each of the three military ranges during monitored days are summarized in Figure 18. Summary statistics regarding the frequency of selected disturbance
events are presented in Table 9.
Table 8. Percent of potential disturbance events that caused birds to fly or birds/pinnipeds to enter the water. The total
number of times each potential disturbance event occurred during times when a given species/group was present in or
near a monitored sub-area is also given. Blank cells represent zero values. For example, of 139 potential disturbance
events involving pedestrians and harbour seals, displacement was recorded 4.3% of the time (i.e., 6 times).

Figure 17. Stacked line (cumulative total) chart of potential disturbance events (i.e., raptor, boat, pedestrian, and air traffic) during 52 monitoring sessions in the Race Rocks study area of from 6 October 2002 through 27 November 2003.

Figure 18. Number of blasts that occurred during each of the 52 monitoring sessions in the Race Rocks study area of from 6 October 2002 through 27 November 2003. Zero values and blasts during non-monitoring days are not
plotted.

5.2.2.1 Effects on Harbour Seals
In the absence of any discrete disturbance attributed to a given sample of animal activity, mean
activity levels of harbour seals were &Mac178;19%, regardless of whether or not blasting occurred during
the day (see “None” in Figure 19). Except for blasts on Bentinck Island, mean activity levels
resulting from all other potential discrete disturbances were &Mac178;21%. Blasts on Bentinck Island
were the only discrete disturbance to consistently cause a noticeable increase in harbour seal
activity, with a mean activity level of 47% observed during samples taken immediately post-
blasting (Figure 19).

Figure 19. Activity levels of harbour seals at Race Rocks expressed as the mean values observed for each
disturbance class on days when no blasting occurred (“No Range”), days when blasting occurred on Bentinck Island
(“Bentinck”), and days when blasting occurred in Whirl Bay (“Whirl Bay”) only. Error bars denote +1 standard
deviation. Values are based on samples of >9 animals and >4 observations per disturbance class per category of
blasting. “None” represents samples taken when no disturbance stimulus was attributed to observations. Sample
sizes are presented in Appendix 7.

Figure 20 approximates the extent to which selected discrete disturbance stimuli (and absence
thereof) affected the abundance of hauled-out harbour seals in monitored sub-areas. Because not
all sub-areas were monitored, and because of limitations in the sampling procedures (see section
4.1) numbers represent relative, not absolute values. Further, the plotted values cannot be
interpreted as being solely a function of a given disturbance stimuli, as changes in animal
abundance occur in the absence of discrete effects (e.g., changes under the category “none”).
Pedestrians, blasts on Bentinck Island, and unknown disturbance stimuli had the most noticeable
effects on harbour seal displacement from land to water.

Figure 20. Comparative changes in numbers of harbour seals in monitored sub-areas for selected potential
disturbance types. Change was calculated as number of animals at time = (t+1) minus number of animals at time = t
(where the given disturbance occurred at time = t+1). Note that observations at time = t were not always
disturbance-free. Mean values () represent the aggregate mean of the daily mean change in numbers for individual
monitored areas by disturbance type. The upper limit of the vertical lines represents the maximum value of any
mean change for each disturbance type by individual area by day. The lower limit of the vertical lines represents the
minimum value of any mean change for each disturbance type by individual area by day. The net daily change (y-
axis on right side) represents the sum of the mean changes by disturbance type by monitored area by day. Sample
sizes, in parentheses, indicate the number of unique area-date combinations from which the data were generated.

5.2.2.2 Effects on California Sea Lions
In the absence of discrete disturbances, mean activity levels of California sea lions were &Mac178;33%
regardless of the status of a range on a given day (Figure 21). Of samples with no disturbance,
mean activity was lowest on days when no range was active, suggesting some residual effect of
blast-induced disturbance. During days when no range was active, a 12% increase in activity
(i.e., from 19% to 31%) was noted when 1 ecotour boat was recorded as the primary
disturbance stimuli compared to the situation when no disturbance was attributed to a sample.
Blasts on Bentinck Island caused the most noticeable increases in California sea lion activity,
with a mean activity level of 54% observed during samples taken immediately post-blasting.

Figure 21. Activity levels of California sea lions at Race Rocks expressed as the mean values observed for each
disturbance class on days when no blasting occurred (“No Range”), days when blasting occurred on Bentinck Island
(“Bentinck”), and days when blasting occurred in Whirl Bay (“Whirl Bay”) only. Error bars denote +1 standard
deviation. Values are based on samples on >9 animals and >4 observations per disturbance class per category of
blasting. “None” represents samples taken when no disturbance stimulus was attributed to observations. Sample
sizes are presented in Appendix 7.

Figure 22 approximates the extent to which selected disturbance stimuli (and absence thereof)
affected the abundance of California sea lions on monitored haulouts. Because not all areas were
monitored, and because of limitations in the sampling procedures (see section 4.1) numbers
represent relative, not absolute values. Further, the plotted values cannot be interpreted as being
solely a function of a given disturbance stimuli, as changes in animal abundance occur in the
absence of discrete disturbance effects (e.g., changes under the category “none”). Pedestrians,
pleasure boats, blasts on Bentinck Island, and unknown disturbance stimuli had the most
noticeable effects on California sea lion displacement from land to water.

Figure 22. Comparative changes in numbers of California sea lions in selected monitored areas for selected potential
disturbance types. Change was calculated as number of animals at time = (t+1) minus number of animals at time = t
(where the given disturbance occurred at time = t+1). Note that observations at time = t were not always
disturbance-free. Mean values () represent the aggregate mean of the daily mean change in numbers for individual
monitored areas by disturbance type. The upper limit of the vertical lines represents the maximum value of any
mean change for each disturbance type by individual area by day. The lower limit of the vertical lines represents the
minimum value of any mean change for each disturbance type by individual area by day. The net daily change (y-
axis on right side) represents the sum of the mean changes by disturbance type by monitored area by day. Sample
sizes, in parentheses, indicate the number of unique area-date combinations from which the data were generated.

5.2.2.3 Effects on Northern Sea Lions
In the absence of any disturbance, mean activity levels of northern sea lions were &Mac178;41%,
regardless of the status of a range on a given day (Figure 23). Of samples with no disturbance
stimulus, mean activity was highest on days when the Bentinck Island Range was active,
suggesting some residual effect of blast-induced disturbance. Of all disturbance stimuli, blasts on
Bentinck Island caused the greatest increase in northern sea lion activity, with a mean activity
level of 73% observed immediately post-blast. Ecotour boats caused an increase in activity
levels; particularly on days when the Bentinck Island Range was active.
Figure 24 approximates the extent to which selected disturbance stimuli (and absence thereof)
affected the abundance of northern sea lions on monitored haulouts. Because not all areas were

monitored, and because of limitations in the sampling procedures (see section 4.1) numbers
represent relative, not absolute values. Further, the plotted values cannot be interpreted as being
solely a function of a given disturbance stimuli, as changes in animal abundance occur in the
absence of discrete disturbance effects (e.g., changes under the category “none”). Blasts on
Bentinck Island and unknown disturbance stimuli had the most noticeable effects on northern sea
lion displacement from land to water.

Figure 23. Activity levels of northern sea lions at Race Rocks expressed as the mean values observed for each
disturbance class on days when no blasting occurred (“No Range”), days when blasting occurred on Bentinck Island
(“Bentinck”), and days when blasting occurred in Whirl Bay (“Whirl Bay”) only. Error bars denote +1 standard
deviation. Values are based on samples on >9 animals and >4 observations per disturbance class per category of
blasting. “None” represents samples taken when no disturbance stimulus was attributed to observations. Sample
sizes are presented in Appendix 7.

Figure 24. Comparative changes in numbers of northern sea lions in selected monitored areas for selected potential disturbance types. Change was calculated as number of animals at time = (t+1) minus number of animals at time = t (where the given disturbance occurred at time = t+1). Note that observations at time = t were not always
disturbance-free. Mean values () represent the aggregate mean of the daily mean change in numbers for individual
monitored areas by disturbance type. The upper limit of the vertical lines represents the maximum value of any
mean change for each disturbance type by individual area by day. The lower limit of the vertical lines represents the minimum value of any mean change for each disturbance type by individual area by day. The net change (y-axis on right side) represents the sum of the mean changes by disturbance type by monitored area by day. Sample sizes, in parentheses, indicate the number of unique area-date combinations from which the data were generated.

5.2.2.4 Disturbance by Killer Whales
Approximately 21 killer whales in 4 separate group sizes of 15, 4, 1, and 1 were observed in the
monitored area during the study. Many more were observed outside the Reserve. Pinnipeds, and
in particular, harbour seals, were vigilant while killer whales were present (from 4 to 15 min per
sighting), no pinnipeds were observed to leave a haulout in response to a whale’s presence

.
5.2.2.5 Disturbance by Raptors
Overflights by raptors such as bald eagles were the most consistent (93.7% of events recorded)
disturbance that caused some animals (in this case, primarily gulls) to take flight or enter the
water. Such disturbances were usually very brief and most gulls landed shortly after the raptor had passed. Most raptor disturbance events occurred from December through February when
bald eagles were at their peak abundance locally (Appendix 5, Figure 17, Photo 13). The adverse
effect of raptors on California sea lions depicted in Figure 22 was primary the result of the sea
lions’ response to the hundreds of gulls that were flushed by a red-tailed hawk. Prior to the
disturbance initiated by the hawk, several pedestrians were in the general vicinity. The extent to
which the pedestrians might have aggravated the sea lions’ response in that instance is not
known.

Photo 13. Raptors such as this adult bald eagle frequently disturbed cormorants and gulls in the study area. This bird
is feeding on a gull it captured. 20 November 2003.