J. Zool., Lond. (2002) 256, 255±270 # 2002 The Zoological Society of London Printed in the United Kingdom DOI:10.

1017/S0952836902000298

Behavioural responses of killer whales (Orcinus orca) to

whale-watching boats: opportunistic observations and
experimental approaches

Rob Williams1*, Andrew W. Trites1 and David E. Bain2

1
Department of Zoology and Marine Mammal Research Unit, Fisheries Centre, University of British Columbia, 6248 Biological Sciences Road,
Vancouver, BC V6T 1Z4, Canada
2
D. E. Bain, Six Flags Marine World Vallejo, Vallejo, CA 94589, USA. E-mail: [email protected]
(Accepted 5 March 2001)

Abstract
Johnstone Strait provides important summer habitat for the northern resident killer whales Orcinus orca of
British Columbia. The site is also an active whale-watching area. A voluntary code of conduct requests
that boats do not approach whales closer than 100 m to address perceived, rather than demonstrated,
effects of boat traf®c on killer whales. The purpose of the study was to test the relevance of this distance
guideline. Relationships between boat traf®c and whale behaviour were studied in 1995 and 1996 by shore-
based theodolite tracking of 25 identi®able focal animals from the population of 209 whales. Individual
killer whales were repeatedly tracked in the absence of boats and during approaches by a 5.2 m motorboat
that paralleled each whale at 100 m. In addition, whales were tracked opportunistically, when no effort was
made to manipulate boat traf®c. Dive times, swim speeds, and surface-active behaviours such as breaching
and spy-hopping were recorded. On average, male killer whales swam signi®cantly faster than females.
Whales responded to experimental approaches by adopting a less predictable path than observed during
the preceding, no-boat period, although males and females used subtly different avoidance tactics. Females
responded by swimming faster and increasing the angle between successive dives, whereas males
maintained their speed and chose a smooth, but less direct, path. Canonical correlations between whale
behaviour and vessel proximity are consistent with these conclusions, which suggest that weakening whale-
watching guidelines, or not enforcing them, would result in higher levels of disturbance. High variability in
whale behaviour underscores the importance of large sample size and extensive experimentation when
assessing the impacts of human activity on killer whales.

Key words: whale, whale-watching, behaviour, canonical correlation, disturbance, Orcinus orca

INTRODUCTION targets for the growing ecotourism industry (Hoyt,

1997). In 1993, the International Whaling Commission
In recent decades, a dramatic shift has occurred in the (IWC) adopted a resolution that declared its desire `to
way that people relate to killer whales Orcinus orca. encourage the further development of whale watching
Plans to `cull' killer whale populations on the coast of as a sustainable use of cetacean resources' (IWC, 1994).
British Columbia were considered as recently as 1960 The economic bene®ts of this industry are undeniable.
(Ford, Ellis & Balcomb, 1994). Today, such plans would The whale-watching industry has exposed millions of
be unthinkable. In fact, many people are concerned that urban-dwellers to animals in their natural environment,
the killer whale is now too popular in British Columbia, which may change attitudes toward protecting critical
and may be suffering from too much attention (John- habitat and threatened populations (Barstow, 1986;
stone Strait Killer Whale Committee (JSKWC), 1991; Duffus & Dearden, 1993). However, vessel traf®c may
Kruse, 1991; Adimey, 1995; Trites, Hochachka & carry costs for whales (IWC, 1995). A suitable manage-
Carter, 1995). ment goal might be to ensure that the economic and
Cetacean populations around the world are becoming conservation value of whale-watching does not come at
the price of excessive stress to individual whales or their
*All correspondence to: Rob Williams. Sea Mammal Research Unit,
Gatty Marine Lab., University of St Andrews, Fife KY16 8LB, populations.
Scotland, UK Researchers have identi®ed four distinct populations
E-mail: [email protected] of killer whales on the British Columbia coast that have
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
256 R. Williams, A. W. Trites and D. E. Bain

overlapping ranges, but are socially and ecologically summer in Johnstone Strait. A whale might respond to
isolated (Ford et al., 1994). Whale-watching tends to boats by varying the duration of its dives (vertical
focus on the northern and southern populations of avoidance), or by swimming faster or altering the direc-
resident killer whales, the ®sh-eating types, since these tion of swimming (horizontal avoidance). Longer dives
whales are sighted more consistently than the offshores can be considered vertical avoidance, if the whale holds
or the marine-mammal-eating transients. One of the its breath longer than the attention span of a whale-
most reliable places to see killer whales in the wild is watcher. Whales may also display agonistic behaviours,
Johnstone Strait, British Columbia, Canada (Fig. 1). such as slapping ¯ukes or pectoral ®ns on the surface of
Northern resident killer whales return here each the water.
summer to socialize, to rub their bodies on smooth Our secondary goal was to describe how whale beha-
pebble beaches, and to prey on migrating salmon as viour varied across the range of traf®c conditions seen
they are funnelled through the narrow strait (Nichol & in Johnstone Strait in summer. Observing whales oppor-
Shackleton, 1996). A similar core habitat for southern tunistically, when many boats were present, and when
resident killer whales is found in Haro Strait between boats approached animals closely, allowed insights into
British Columbia and Washington State, where proxi- killer whale behaviour under traf®c conditions that
mity to urban areas makes whale-watching a much would have been dif®cult to replicate experimentally.
larger industry than in Johnstone Strait. In 1995 and This dual nature of data collection allowed the causal
1996, the core whale-watching ¯eet in Johnstone Strait relationships identi®ed by experimental approaches to
consisted of three operators, with a total of four boats. be compared with trends in whale behaviour across a
Operators typically offered one or two tours daily from wide range of traf®c conditions.
mid-June to early September. In contrast, the southern
resident whale-watching industry employed 34 Cana-
dian and American companies in 1995 and 40 in 1996, MATERIALS AND METHODS
which represents a ¯eet of 47 (1995) and 54 (1996)
vessels (Osborne & Otis, 2000). The southern resident Study area
whale-watching season is also longer and more variable,
with companies operating one to six trips daily, based Data were collected between 1 July and 31 August 1995,
on demand, from May to September (Osborne & Otis, and between 16 July and 10 September 1996, from a
2000). land-based observation site on the south shore of West
Johnstone Strait is unique for reasons other than its Cracroft Island in Johnstone Strait (50830'N, 126830'W;
designation as the core habitat for northern resident Fig. 1). This cliff-top site is an ideal vantage point to
killer whales. It is an important area for commercial view whales in the Reserve (with relatively little whale-
salmon ®shing (JSKWC, 1991). It is also home to oriented boat traf®c), and adjacent to the Reserve
people who have been successful in encouraging the (where whale-watching vessels often congregate). The
provincial government to protect Robson Bight as a shore-based nature of the study allowed observation of
critical habitat for killer whales, and who have estab- vessel activity without contributing to potential vessel
lished voluntarily a code of conduct to self-regulate effects.
behaviour around whales (JSKWC, 1996). In 1990, Data were collected using an electronic theodolite
British Columbia Parks initiated a warden and mon- (Pentax ETH-10D with a precision of ‹ 10'' of arc)
itoring programme of boat and whale usage of the connected to a laptop computer equipped with custom
Robson Bight±Michael Bigg Ecological Reserve, and software (THEOPROG: available from D. E. Bain).
the waters immediately adjacent to it. Analysis of data Cliff height was measured by stretching a rope of
from 1990±94 by Trites et al. (1995) found that the known length at the water's edge on a beach below the
likelihood of whales leaving the Ecological Reserve cliff and by using the theodolite to obtain horizontal
increased as increasing numbers of boats entered it. and vertical angle coordinates for both ends (Williams,
However, sound scienti®c management of whale- 1999). Height was then calculated using trigonometric
watching must involve more than simply closing critical relationships described by Davis et al. (1981) and
habitat to boats; it must also de®ne and promote WuÈrsig, Cipriano & WuÈrsig (1991). The theodolite was
responsible whale-watching activities outside the located c. 50 m a.s.l.
Reserve. Current self-imposed whale-watching guide- Reliability of the theodolite±computer apparatus was
lines request that boats remain 100 m from whales, but measured by stretching a rope of known length along
this may be based more on aesthetics than biological the water's edge on a beach across the strait from the
relevance. Guidelines should be based on actual impacts cliff. At a distance of 3.79 km, our mean estimate of the
of human activity on whale behaviour rather than length of a 30 m rope as measured by the theodolite±
perceived effects. Otherwise, token guidelines may give computer apparatus was 28.93 m (n = 20, se = 0.18).
the false sense that boats are not disturbing whales, This translates to a measurement error of c. 3.5% in
provided that they follow some groundless rules. terms of accuracy, and < 1% in terms of precision. Per
The primary goal of our study was to test whether a cent errors in measuring cliff height, distance travelled
vessel following whale-watching guidelines affected the and speed tend to be approximately equal (WuÈrsig
behaviour of northern resident killer whales during et al., 1991).
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Behavioural responses of killer whales to whale-watching 257

that tracks shorter than 1000 s tend to bias estimates of

British
Columbia respiration rate (Kriete, 1995). Animals were tracked
only when displaying typical foraging behaviour. This
activity is the most commonly observed activity of
resident killer whales in summer in Johnstone Strait
(Nichol & Shackleton, 1996), and is recognized when
Pacific groups are spread out and all animals are swimming
Ocean
essentially in the same direction (Ford et al., 1994). This
consistency in tracking only foraging animals prevented
any effect of activity state on respiration rate and swim
speed from masking effects of boat traf®c.
126° 35′ N

Tracking

Tracking whales

Joh The tracking team consisted of a spotter, a theodolite

Stra nston West Cracroft Island
it e operator and a computer operator. The spotter an-
nounced each time that a focal animal surfaced to
* breathe or display surface-active behaviour. The theo-
dolite operator located the position of the whale.
50° 30′ N Ecologic
al Reserv
Behaviours recorded by the computer operator in-
e cluded: breath, breach, ¯uke slap, pectoral ®n slap,
Vancouver dorsal ®n slap, unidenti®ed splash, porpoising, and spy-
Island hop (Ford et al., 1994). The computer was linked to the
theodolite to record the time that it retrieved the
km horizontal and vertical angle coordinates of a whale's
0 2 4 6 8
position. A scale, marked at 10 cm intervals, was
painted on a rock wall below the cliff. Water level was
Fig. 1. Study area in Johnstone Strait, BC, Canada, showing noted every 15 min to determine the height of the
lines of sight ( - - - ), position of theodolite ( * ), and boundaries theodolite above sea level over changing tides.
of Robson Bight±Michael Bigg Ecological Reserve.

Tracking boats
Selection of focal animals
During the whale's long dives, boat positions were
Northern resident killer whales entered the study area in recorded along with information about vessel type,
matrifocal social units called subpods (Olesiuk, Bigg & whale-watching status, orientation relative to the whale,
Ellis, 1990; Ford et al., 1994). Temporary groupings of estimated distance from the whale, and direction of
subpods ranged in size from 2 to 120 individuals. travel. The tracking team recorded the position of every
Careful selection of a focal animal was chosen over boat within 3 km of the whale, as often as possible
random selection to ensure representative sampling of without losing track of the whale's position. A vessel
the population and reliability of re-sighting an indivi- was deemed to be ignoring the whale if it made no
dual within a tracking session. We chose focal animals direction change toward the whale, continued out of the
that would not be confused easily with other members study area, or if it was engaged in a non-whale-oriented
of the group and that were likely to be consistently re- activity, such as ®shing. All others were deemed to be
sighted. A focal animal typically had a distinctive dorsal whale-oriented traf®c. Every attempt was made to
®n and saddle patch (Ford et al., 1994). Focal whales record when a vessel's status changed within a tracking
were usually within a few hundred metres of other session.
members of the group. Whales swimming mid-strait
were chosen over those swimming along the Vancouver
Island shore, since accuracy of a reading diminishes Track types
with distance from the theodolite (WuÈrsig et al., 1991).
In 1995, only males were tracked (since their dorsal ®ns Whales and boats were tracked under 3 traf®c condi-
can reach twice the height of the those of adult females). tions:
However, in 1996, we were able to consistently re-sight (1) No-boat. `No-boat' tracks were de®ned as ones
distinctive females. when no boats were seen within 3 km of the focal
Animals were selected that were likely to be visible for animal. During the 1996 season, local charter operators
a minimum of 15 min, because earlier work has shown agreed to stay away from the focal whale while its
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
258 R. Williams, A. W. Trites and D. E. Bain

movements were tracked, thereby increasing the number

--
----------------------

--
of no-boat tracks.

--
--
(2) Treatment. In 1996, an experimental boat (a 5.2 m

--
α2

--
Hourston motorboat with a 90 hp Yamaha 2-stroke

--
Deviation

--
index
outboard engine) was available to approach focal α1 +α2 α1 d3
animals. The boat operator followed local whale- α=
2
watching guidelines by paralleling the whale at 100 m d2
for a minimum of 20 min, after 20 min of observation
under control (no-boat) conditions. No sudden direc- T
tion changes were made, and the operator was d1
instructed not to place the boat in the path of the whale Directness index
(an activity referred to as `leapfrogging'). The experi-
mental boat operator was in constant VHF radio T
T
DI = 100
contact with cliff-based observers, who kept the boat (dd11 ‡
+ dd22 ‡
+dd3)†
100 m from the whale by informing the operator
periodically of the distance between boat and whale as Fig. 2. A sample swimming path with four surfacings (*) and
measured by the theodolite. three dives (di ), showing two measures of path predictability:
(3) Opportunistic. Opportunistic tracks occurred when deviation and directness. The deviation index is the mean of
at least 1 boat was present within 3 km of the focal all angles between observed dives and the straight-line paths
whale, and no effort was made to manipulate traf®c predicted (_) by preceding dives. The directness index is the
around the focal animal. ratio of the track diameter (T) to its perimeter.

Data compilation
the whale) was calculated within 100 m, 400 m, and
Calculating predictor variables 1000 m radii of each surfacing. The maximum number
of vessels (whale-oriented and non-whale-oriented) ob-
Temporal and biological variables. Whales were classi®ed served within the 3 radii was then calculated for each
as either young or old, based on life history information track.
available for individuals in this population (Olesiuk et
al., 1990). A female was classi®ed as old if her presumed
age was at least 40 years, which is the average age for Calculating response variables
the onset of reproductive senescence. A male was con-
sidered old if his presumed age was at least 30 years (the A mean dive time (i.e. average time between surfacings)
average life expectancy for male northern resident killer was calculated for each track. The average swimming
whales) (Olesiuk et al., 1990). Age estimates of young speed of the whale was obtained by dividing the total
whales are more reliable than those of old whales in this distance travelled by the duration of the tracking
population, since annual photo-identi®cation of most session. Two measures of path predictability were calcu-
individuals began in the mid-1970s (Ford et al., 1994). lated: a directness index and a deviation index (Fig. 2).
Traf®c variables. THEOPROG was used to sort and The directness index measures path predictability on
transform the series of angles, times and codes into x±y the scale of a tracking session. It is generated by dividing
coordinates and speeds. Boats were recorded less fre- the distance between end-points of a path by the
quently than the focal animal, and were assumed to cumulative surface distance covered by all dives. The
travel at constant speed between marks. This is a safe directness index can be thought of as the ratio of the
assumption in the Robson Bight area, where community diameter of a path to its perimeter, and is the inverse of
pressure discourages boats from leapfrogging. The ap- the milling index of Tyack (1982) and Kruse (1991). The
proximate location of each boat was interpolated in directness index ranges from 0 (a circular path) to 100 (a
order to determine where it was every time the whale straight line).
surfaced. Distance between whale and boat was calcu- The deviation index measures path predictability
lated for every surfacing in a track using the actual from 1 surfacing to the next (Fig. 2). It is the mean of all
position of the whale and the interpolated or extrapo- angles between adjacent dives, and can be considered an
lated position of each boat. inverse measure of a path's smoothness. For each surfa-
We measured intensity of boat traf®c in 3 ways: cing in a track, the angle between the path taken by a
(1) track type: control, treatment or opportunistic; dive and the straight-line path predicted by the dive
(2) vessel proximity: the distance between each boat before it was calculated (Williams, 1999). The deviation
and the whale was calculated for every surfacing to index is the mean of the absolute value of each of these
determine a minimum proximity within a track; discrepancies, in degrees, during the entire track. A low
(3) number of vessels: the number of whale-oriented deviation index indicates a smooth path, while a high
vessels (those that altered course toward whales) and deviation index indicates an erratic path. Indices of
non-whale-oriented vessels (those apparently ignoring directness and deviation were calculated for each track.
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Behavioural responses of killer whales to whale-watching 259

A track that shows high deviation and high directness is Table 1. Number of tracks, by gender and traf®c conditions.
described as erratic but directional, whereas a track with Bold numbers, number of observations; numerals in parenth-
low deviation and low directness is smooth but non- eses, number of subjects. For a further breakdown, see Wil-
liams (1999)
directional.
We recorded each time that surface-active events such No-boat Treatment Opportunistic Total
as spy-hopping or breaching occurred. A bout of tail-
slapping or ®n-slapping was scored as 1 event if > 1 slap Male 50 (12) 27 (11) 56 (15) 133 (16)
Female 27 (9) 13 (8) 8 (5) 48 (9)
occurred on a surfacing.
Total 77 (21) 40 (19) 64 (20) 181 (25)

Data analysis

Gender and age effects maximize correlation between predictor and response
variable sets are selected (James & McCulloch, 1990).
Relationships among age class, gender and whale beha- The result is a canonical R2, which indicates the propor-
viour were examined before considering effects of boat tion of the variance in whale behaviour that is explained
traf®c on behaviour. One value for each dependent by variance in the explanatory variable set. The
variable (mean dive time, mean swim speed, a deviation maximum number of variates possible in canonical
index, a directness index and a rate of surface-active correlation is equal to the number of variables in the
behaviour) was calculated for each track. Scarcity of smaller set.
surface-active behaviour required us to pool all events The contribution of a variable (e.g. distance to
into a single category of surface-active behaviour, which nearest boat) to its own set (e.g. boat traf®c) is indicated
was standardized to the number of surface-active by a standardized coef®cient, the canonical weight
events/h. Mean values for each dependent variable were (Milstein, 1993). A canonical loading is a coef®cient
averaged across all observations for an individual, that reveals the contribution of a variable to its opposite
regardless of traf®c conditions. Means were then calcu- set (e.g. distance to nearest boat vs whale behaviour).
lated for each gender and age class, such that each Canonical correlations are interpreted using the magni-
whale was represented only once. Two-factor analyses tude and direction of the weights and loadings, which
of variance (ANOVA) were performed on each depen- allows some ¯exibility in interpretation. Some authors
dent variable. have chosen 0.30 as an arbitrary minimum coef®cient
for interpretation (Tabachnick & Fidell, 1996). Alterna-
tively, inherent noise in cetacean behavioural data has
Experimental approaches been cited as a rationale for interpreting coef®cients as
low as ‹ 0.20 (Bauer & Herman, 1986), which is the cut-
Variables recorded under control and experimental con- off we used.
ditions were compared using two-tailed, paired t-tests.
Comparisons were made only when at least 20 min of
baseline, control observation was followed by an experi- RESULTS
mental approach of the same whale lasting at least
20 min. Sample size

Over two seasons, we spent 1416 h observing boats and

Opportunistic observations whales in the study area. This effort yielded 181 usable
tracks of 25 individuals, during which 9863 respiratory
Whale responses to experimental approaches were com- intervals were timed. Focal animals were tracked con-
pared with whale behaviour across a continuous range tinuously for 32.3 h in 1995 and 70.1 h in 1996. Sample
of boat traf®c observed in Johnstone Strait. Canonical size is listed by gender and traf®c conditions in Table 1.
correlations (STATISTICA v. 5) enabled us to investi-
gate how the combination of temporal, biological and
traf®c variables together related to whale behaviour. Gender and age effects
Canonical correlation analysis is a multivariate tech-
nique designed to describe complex relationships Mean values for each of the ®ve dependent variables
between 2 sets of variables (Tabachnick & Fidell, 1996). were calculated for each of the 25 whales observed. The
It has been used particularly well to reveal subtle trends values were normally distributed as indicated by Kol-
in the highly variable datasets common to many ceta- mogorov±Smirnov tests (P < 0.01) (Zar, 1996). Recall
cean studies (Bauer & Herman, 1986; Whitehead et al., that whales in this study were observed while the group
1998). Canonical correlation allows exploration of si- was spread out, searching for food. Observations were
multaneous variance in predictor variable sets (e.g. boat pooled across all traf®c conditions, since our goal was
traf®c) and response variable sets (e.g. whale beha- to describe variation within our sample in order to guide
viour). The linear combinations (variates) that subsequent analyses.
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
260 R. Williams, A. W. Trites and D. E. Bain

Male Female dispersed temporarily. Other activities, such as resting

n= 10 6 16 3 6 9 and socializing, require group cohesion, during which
50
pods regroup.
(a)
The most common surface-active behaviour for both
Dive time

age and sex classes was tail-slapping, with spy-hops and

(s)

pectoral ®n-slaps accounting for most of the remaining

40
activity. Rarity of surface-active events required us to
pool observations to an expected rate of any surface-
active event/h, even though these activities undoubtedly
(b) serve different purposes. No signi®cant differences were
7 found between mean rates of surface-active behaviour
Speed
(km/h)
(km/h)

of males (0.84 h-1) and females (1.17 h71) (F1,23 = 0.20,

speed

P = 0.66). Similarly, no signi®cant relationship was

5
found between age class and whale behaviour, nor were
there any signi®cant interactions between gender and
age class.
30 (c)
(degrees)
Deviation

Experimental approaches

20 The experimental boat approached whales on 40 occa-

sions. Of these, experimental approaches were preceded
32 times by at least 20 min of observation under control
90 (d) conditions. Paired, two-tailed t-tests were performed
on the ®ve response variables for the 32 paired
Directness

80 observations (Fig. 3). Examples of four experimental

approaches of male and female whales are shown in
Fig. 4 and Fig. 5, respectively. Separate analyses were
70
performed for experimental approaches of males and
females, since gender-based differences in swim speed
2
(e) (Fig. 6) indicated potential for different responses to
boat traf®c.
events (/h)
Surface

Increased probability of Type I errors is a concern

1 with these analyses. Greater concern about Type II
error rates as well as other arguments given by Stewart-
0 Oaten (1995) justi®ed the avoidance of a multiple
Young Old All Young Old All
comparison technique at this stage of the analysis.

Fig. 3. Relationship of gender and age class to whale beha-

viour (mean ‹ se), averaged across all traf®c conditions. Each Male response
whale is represented only once. (a) Dive time; (b) swimming
speed; (c) deviation index; (d) directness index; (e) surface- When approached by the experimental boat, the paths
active behaviour. of male whales became less direct than during the
preceding control conditions (t23 = 2.25, P = 0.03). The
reduction in directness can best be understood in terms
of distance covered. A directness index of 83.6 (the
The 16 male northern resident killer whales that were average directness during control sessions) translates to
tracked swam c. 34% faster than the nine females when a whale swimming 119.6 m along a circuitous path to
engaged in typical foraging behaviour (Fig. 3) end up 100 m from his original position. The same
(F1,23 = 6.43, P = 0.02). Average swim speeds were whale, following a path with a directness index of 74.1
6.32 km/h for males and 4.71 km/h for females. No (the average directness during treatment sessions),
signi®cant gender differences were observed in dive time would need to cover 135 m to make 100 m headway.
(males 41.62 s, females 43.97 s; F1,23 = 0.14, P = 0.19), Thus, the average male responded to the experimental
deviation index (males 20.86, females 25.99; F1,23 = 2.20, boat by covering 13% more distance along a circuitous
P = 0.15), or directness index (males 84.80, females path than it covered before the boat arrived. No sig-
77.26; F1,23 = 1.47, P = 0.24). With males swimming ni®cant changes in dive time (t23 = 1.55, P = 0.13), swim
faster than females, and with no difference seen in speed (t23 = 0.45, P = 0.66), deviation index (t23 = 0.56,
direction and deviation indices, it seems that males P = 0.58) or rate of surface-active behaviour (t23 = 1.17,
always outdistance females. Recall that whales were P = 0.25) were observed during experimental
tracked only during foraging activity, when groups approaches.
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Behavioural responses of killer whales to whale-watching 261

2
Distance (km)

0
6

0 2 4 8 0 2 4 8
Distance (km)

Fig. 4. Four examples of paths taken by male focal whales (_) when approached by the experimental boat (7). Each dot, a
surfacing by the whale; arrow, the whale's original direction of travel. Letters are placed at 10-minute intervals: x, the position of
the whale; o, the interpolated position of the boat. The ®rst synchronous pair of boat and whale positions is joined by an oval. In
general, paths of male whales during experimental approaches were smooth, but less directional than during control conditions.

Female response tolerance (Fig. 6). The whale behaviour variable set
included dive time, swim speed, directness and deviation
When approached by the experimental boat, female indices, and the rate of surface-active behaviour. Three
whales responded by swimming 25% faster (t7 = 3.29, traf®c variables were included in the explanatory vari-
P = 0.01) and increasing the mean angle deviation able set: (1) the minimum distance in a track between
between surfacings by 29% (t7 = 2.90, P = 0.02). No any boat and the whale; (2) the maximum number of
signi®cant changes in mean dive time (t7 = 0.29, whale-oriented vessels within 1000 m of the whale; (3)
P = 0.78), directness index (t7 = 0.40, P = 0.70) or rate of the maximum number of non-whale-oriented vessels
surface-active behaviour (t7 = 1.34, P = 0.22) were within 1000 m of the whale. The explanatory variables
observed. also included the day of year, start time of each track,
and age in years. Additional traf®c variables were
eliminated to avoid concerns of multicollinearity (Ta-
Opportunistic observations bachnick & Fidell, 1996). No correlation within a
variable set was > 0.5 after reducing the number of
Canonical correlations were calculated between the set variables.
of whale behaviour variables and a set of explanatory Figures 7 and 8 show scatterplot matrices of relation-
variables. Separate canonical correlations were per- ships among the original variables included in canonical
formed for males and females, since experimental tracks correlations for male and female whales, respectively.
indicated potential for gender-based difference in boat Table 2 shows how linear combinations of some of these
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
262 R. Williams, A. W. Trites and D. E. Bain

2
Distance (km)

0
6

0 2 4 8 0 2 4 8
Distance (km)

Fig. 5. Four examples of paths taken by female focal whales (_) when approached by the experimental boat (7). Each dot, a
surfacing by the whale; arrow, the whale's original direction of travel. Letters are placed at 10-minute intervals: x, marks the
interpolated position of the whale; o, the interpolated position of the boat. The ®rst synchronous pair of boat and whale
positions is joined by an oval. In general, when approached by an experimental boat, the path of female whales became more
erratic while retaining directionality.

relationships were synthesized in one multivariate de- canonical correlation (rc) for males was 0.44, indicating
scription of male behaviour, and another for females. 19% overlapping variance between the two sets, with all
Standardized correlation coef®cients (weights) between ®ve pairs of canonical variates included (w230 = 50.121,
the original variables and the ®rst pair of canonical P = 0.012). After removing the ®rst pair of canonical
variates are listed in Table 2, along with a signi®cance variates, subsequent w2 tests were not signi®cant. There-
level for the correlation with all ®ve pairs of canonical fore the ®rst pair of canonical variates accounted for the
variates included. This w2 test has (kx)(ky) degrees of signi®cant canonical correlation between the two sets of
freedom, where kx is the number of variables in the variables.
explanatory set, and ky the number of variables in the Results of the canonical correlation should be inter-
response set (Tabachnick & Fidell, 1996). preted with caution. The technique describes trends
based on linear combinations of variables (variates),
rather than the original variables. The correlation
Male behaviour between the ®rst pair of canonical variates is statistically
signi®cant, and suggests potentially important relation-
Using 133 tracks, a signi®cant relationship was found ships based on the strength of linear correlations.
between the set of explanatory variables and behaviour However, it does not imply causality of those relation-
of male whales. The signi®cant explanatory variables ships. The following statements about pairwise
were: date and time of the observation; age; maximum relationships describe trends that contributed most to
number of whale-oriented vessels within 1000 m; proxi- the signi®cant canonical correlation, rather than sug-
mity of the nearest boat. Signi®cant behavioural gesting a statistically signi®cant relationship between
variables for male whales were swimming speed, direct- any pair of variables. Similarly, it is unknown whether
ness index, and rate of surface-active behaviour. The the relationships presented here can be extrapolated
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Behavioural responses of killer whales to whale-watching 263

Male Female Table 2. Correlations, canonical correlations, per cents of

n= 24 8 variance extracted and redundancies, between explanatory and
behavioural variables and corresponding canonical variates
45 (a) for 133 observations of males, and 48 observations of females.
Numbers in bold, variables that were interpreted in the text
Dive time

40
(s)

Standardized correlation
coef®cients between original
35 variables and ®rst canonical
variates
(b) Male (n = 133) Female (n = 48)
6
Explanatory set
(km/h)
speed
Speed
(km/h)

Date 70.47 0.58

Time 0.26 0.02
4 Age 0.79 70.40
Whale-orienteda 0.21 70.35
Non-whale-orientedb 70.15 70.56
(c)
40 Minimum distance 0.47 71.06
Per cent of variance 16 10
(degrees)
Deviation

Redundancy (%) 3 5
30
Behavioural set
Mean dive time 70.19 70.20
Swimming speed 70.90 0.13
(d)
Deviation 70.16 1.21
Directness 0.36 0.65
80 Surface behaviour 0.48 70.05
Directness

Per cent of variance 20 17

Redundancy (%) 4 8
70 Canonical correlation (rc) 0.44 0.69
wc230 (subsequent pairs of
variates not signi®cant)c 50.121, 49.744,
(e)
P = 0.012 P = 0.013
2
a
Maximum number of whale-oriented boats within 1000 m of
the whale.
events (/h)

0
Surface

b
Maximum number of non-whale-oriented boats within 1000 m
of the whale.
c
degrees of freedom in chi-square refers to the product of the
No-boat Boat No-boat Boat number of variables in each set, not number of observations.
Fig. 6. Behavioural responses (mean ‹ se) of whales to experi-
mental approach by one boat paralleling the whale at 100 m active behaviour declined. Male whales tended to swim
for 20 min. (a) Dive time; (b) swimming speed; (c) deviation slower, more directly and with more surface-active
index; (d) directness index; (e) surface-active behaviour. behaviour as the day progressed. These potential
diurnal and seasonal effects are unlikely to have affected
our experiment, since treatment observations immedi-
directly to other regions, or seasons, or populations of ately followed control observations, and thus, date and
killer whales. hour were constant.
All other things being equal, paths of male killer While the canonical correlation between the two sets
whales tended to be less direct as boats got closer is signi®cant, the proportion of variance extracted by
(Table 2), just as the experimental tracks would predict the ®rst pair of variates is moderate. The ®rst canonical
(Fig. 6). However, paths were more direct when the variate of the explanatory set extracts 16% of the
number of whale-oriented boats increased. Whales variance in its own set. In addition, it accounts for 4%
tended to swim faster as boats got closer, and to slow of the variance in its opposite set of behavioural vari-
down as number of boats increased. Rates of surface- ables (the so-called redundancy of the behavioural set).
active behaviour decreased as boats moved closer to the The ®rst canonical variate of the behavioural set ex-
whales, but increased as the number of whale-oriented tracts 20% of the variance of its own set, and 3% of the
vessels increased. variance in the explanatory set.
These trends are confounded by the fact that older
whales tended to swim more slowly, more directly, and
with more surface-active behaviour than younger Female behaviour
whales. Similarly, there seems to be a seasonal compo-
nent to whale behaviour, since speeds increased during Using 48 tracks, we found a signi®cant relationship
the study period as path directness and rates of surface- between the set of explanatory variables and behaviour
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
264
Date Time (h) Age (years) Whale-oriented Non-whale- Minimum
boats oriented boats distance (m)

60
Dive time
(s)
40

12

R. Williams, A. W. Trites and D. E. Bain

Swimmimg
speed(km/h) 8

Deviation 40
(degrees)
20

90

Directness 60
index
30

Surface 6
events (/h)
3

Aug Sep 12p 4p 20 30 2 5 8 2 5 100 1000 2000

Fig. 7. Scatterplot matrix showing relationships between the set of explanatory variables and behaviour of male whales (n = 133). Each plot is smoothed with a lowess function. Solid
lowess line, relationships that were interpreted in the canonical correlation; dashed line, relationships that were not interpreted.
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Date Time (h) Age (years) Whale-oriented Non-whale- Minimum
boats oriented boats distance (m)

60
Dive time
(s) 40

Behavioural responses of killer whales to whale-watching

8
Swimmimg
speed(km/h)
4

40
Deviation
(degrees)
20

90
Directness
index 60

30

Surface 6
events (/h)
3

Aug Sep 12p 4p 20 40 60 0 2 40 1 100 1000 2000

Fig. 8. Scatterplot matrix showing relationships between the set of explanatory variables and behaviour of female whales (n = 48). Each plot is smoothed with a lowess function.
Solid lowess line, relationships that were interpreted in the canonical correlation; dashed line, relationships that were not interpreted.

265
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
266 R. Williams, A. W. Trites and D. E. Bain

of female killer whales. The signi®cant explanatory Horizontal avoidance tactics

variables were date, age, maximum numbers of whale-
oriented and non-whale-oriented vessels within 1000 m, The tendency for the paths of whales to become less
and proximity of the nearest boat. Signi®cant predictable when approached by the experimental boat
behavioural variables for female whales were mean dive is consistent with horizontal avoidance patterns. This
time, and deviation and directness indices. The offers observers new information when interpreting the
canonical correlation (rc) was 0.687 for female whales, behaviour of these animals, since the study also mea-
indicating 47% overlapping variance between the sured parameters such as dive time and surface-active
two sets, with all ®ve pairs of canonical variates in- events, that showed less consistent variation with boat
cluded (w230 = 49.744, P = 0.013). After removing activity. Intriguingly, the trend toward less predictable
the ®rst pair of canonical variates, subsequent tests paths was detected on two different spatial scales.
were not signi®cant, therefore the ®rst pair of canon- Female killer whales tended to reduce predictability
ical variates accounted for the signi®cant canonical from one surfacing to the next, while males reduced
correlation. path predictability on the scale of an entire tracking
As boats got closer to female whales, the deviation session. Essentially, females tended to evade a pursuing
index tended to increase (Table 2). This is consistent boat by adopting an erratic but directional path,
with results from experimental tracks (Fig. 6). Thus, as whereas males adopted a smooth, non-directional path.
boats got closer, tracks tended to be erratic but direc- Swim speeds also increased when the experimental boat
tional, and dives tended to be shorter. Once again, approached female whales (Fig. 6) and as boats got
however, the relationship between whale behaviour and closer to male whales (Table 2). Although gender-based
proximity shows the opposite trends as the ones differences in vessel avoidance are interesting, the key
between behaviour and boat number. As the number of point is that both males and females responded to
boats (both whale-oriented and non-whale-oriented) experimental approaches by adopting less predictable
increased, the deviation index decreased, dives got paths.
shorter and paths became less direct. Thus, with many Howland (1974) and Weihs & Webb (1984) described
boats, female whales tended to adopt a smooth, non- efforts to model optimal strategies for evading preda-
directional path. As the season and day progressed, tors. In both models, successful escape is linked to the
dives tended to get shorter and paths tended to become simultaneous variation of velocity and turning radius. In
more erratic (Table 2). order for this simple form of horizontal avoidance to be
successful, prey vary their speed and the extent to which
they turn away from the path of the predator. Prey may
compensate for a larger turning radius by increasing
DISCUSSION their speed, or may increase manoeuvrability to com-
pensate for slower movement. Thus, slower prey can
Our land-based study of killer whales, combining ex- escape faster predators if prey are able to turn more
perimental approaches and opportunistic observations, sharply (Howland, 1974). The response of a killer whale
reveals a complex relationship between whale behaviour to a boat that follows it may be considered loosely
and vessel activity. It has shown that a single vessel analogous to a predator±prey interaction. In fact, some
following whale-watching guidelines affects the move- tracks of killer whales and the experimental boat (Figs 4
ment patterns of northern resident killer whales. & 5) are reminiscent of long-exposure photographs of
Furthermore, our study provides a useful description of moths evading bats (Roeder, 1967). This resemblance
how whale behaviour varied across a range of traf®c creates an opportunity to compare behaviour of whales
conditions in Johnstone Strait. around a boat to the tactics that some prey use to
The utility of our study can be judged in three ways. escape predation.
First, the study identi®es how the behaviour of focal The analogy between the response of a killer whale to
animals changed when they were approached by the a boat and a typical predator±avoidance strategy
experimental boat. Thus, it de®nes the nature of the becomes more apparent when one recalls that killer
effect in this population at this time. Second, the study whales (prey) tended to increase both swim speed and
identi®es how much the animals' behaviour changed deviation from a straight-line path as a boat (predator)
during experimental approaches. This provides informa- approached closely. While gender-based differences in
tion about effect size at the treatment level of current avoidance tactics are present, they are merely variations
whale-watching guidelines, with a given number and on a common theme of evading boats by adopting an
community of whale-watching operators. Finally, op- irregular path. The difference in surfacing patterns seen
portunistic observations suggest that effect size is between males and females may be accounted for by
related to proximity of vessels. Therefore, the study Howland's (1974) trade-off paradigm. Perhaps female
cautions that weakening guidelines, by allowing boats whales compensate for slower swimming speeds (Fig. 3)
to approach whales closer than 100 m, will yield higher by increasing the `escape angle' in Howland's model ±
levels of disturbance. Of course, adopting more conser- the deviation index (Fig. 6). While there is no evidence
vative distance guidelines would confer even greater that female killer whales cannot swim at the speeds
bene®t to the animals. observed for males in this study, female killer whales do
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Behavioural responses of killer whales to whale-watching 267

seem to swim less ef®ciently at higher speeds than males singly or in small groups. She found that `milling
(Kriete, 1995). The existence of gender-based differences indexes were about the same for both disturbed and
in boat tolerance is not altogether surprising. Ma- undisturbed whales.' (Of the observations in Kruse's
triarchs in this population have been described as more `disturbed' category, 68% contained only one boat.) She
dif®cult than males to approach closely for biopsy also found a tendency for swimming speeds to increase
(Barrett±Lennard, Smith & Ellis, 1996) and humpback with an increasing number of boats. In our study, males
whales demonstrate differential boat responses among (from canonical correlation) and females (from experi-
age±sex classes when on the winter breeding and calving mental approaches) tended to speed up as a vessel
grounds off Hawaii (Bauer & Herman, 1986). approached closely, but neither group swam faster as
A predator±prey analogy also offers a plausible the number of boats increased. In fact, speeds of males
framework for context-speci®c avoidance tactics. were negatively correlated with the number of vessels.
Whales would be expected to display a variety of Does this apparent change in behaviour indicate habi-
responses to a variety of traf®c scenarios, depending on tuation, or does it re¯ect differences in study design?
the speed and manoeuvrability of the whale and A gender-based sampling bias may have existed in
vessel(s) involved. Certainly, the avoidance responses Kruse's study. Mean speed of Kruse's `undisturbed'
generated by the experimental boat are reaf®rmed by whales (4.6 km/h) matches closely with that of female
canonical correlation between close boats and less pre- whales in our study (4.7 km/h), and speed of `disturbed'
dictable swim paths. However, as number of vessels whales (6.4 km/h) approximates that of male whales in
increased, swim paths became more predictable (i.e. the our study (6.3 km/h). It is conceivable that no-boat
paths of male whales tended to be more direct, and the tracks in the earlier study were biased toward females,
paths of females tended to be less erratic). with a disproportionate number of male tracks con-
The trade-off between number and proximity of boats taining boats.
suggests either that whales were less disturbed by other Habituation is the second explanation for the discre-
boats than by the experimental boat, or that whales pancy. This would suggest that, in addition to
abandon this avoidance tactic when many boats ap- avoidance being context-speci®c, the response could
proach. An irregular path may be a useful avoidance also change over time. Certainly, whales would have
tactic with a single boat but ineffective with more than several incentives to abandon a fast-swimming avoid-
one. In a multiple-vessel scenario, a dive that takes a ance tactic. As swim speed increases, breathing rate of
whale farther from one boat may bring it closer to grey whales (Sumich, 1983) and metabolic rate of killer
another. Perhaps the positive correlation between vessel whales (Kriete, 1995) have been shown to increase
number and dive time seen in female whales (Table 2) exponentially. Thus, the shift away from Kruse's ob-
suggests that these animals shifted from horizontal served avoidance response may indicate that animals
avoidance of a single boat to vertical avoidance of many have shifted away from avoidance behaviour that
boats. This compromise deserves further attention in the carries relatively high energetic costs. In addition, the
form of a multiple-vessel experiment. Similarly, the corresponding increase in surfacing rate (Sumich, 1983)
relationships among date, time and behaviour (Table 2) as whales swim faster may actually serve to make the
serve as a reminder that distribution of salmon may be animal more conspicuous. Finally, and most plausibly,
the best determinant of whale distribution and activity swimming faster would be simply an ineffective avoid-
in Johnstone Strait (Nichol & Shackleton, 1996). ance strategy around motorboats.
A more rigorous comparison of available data from
the two periods is warranted. Our choice between these
Effect ± size competing explanations, sampling bias and habituation,
illustrates the key barrier to sound scienti®c manage-
One characteristic of many interactions between ment of whale-watching: uncertainty. In the context of
humans and wildlife populations is that behavioural ambiguous and often apparently contradictory ®ndings,
responses to human activity diminish over time. Habi- managers are faced with a choice between maximizing
tuation has been shown in chimpanzees exposed to immediate recreational bene®ts to humans and a pre-
long-term ecotourism ventures (Johns, 1996). Bighorn cautionary approach that withholds interactions to
sheep show reduced response to predictable human mitigate perceived impacts on whales (Duffus & Baird,
activity (MacArthur, Geist & Johnston, 1982). It may 1995). If the discrepancy between our study and Kruse's
be that after two decades of commercial whale-watching (1991) study indicates habituation to whale-watching,
in Johnstone Strait, killer whales have reduced or then this lends support to managing for human bene®t.
altered their response to boat traf®c. Indeed, perhaps If, however, the apparent discrepancy simply re¯ects
the most intriguing aspect of these ®ndings is the differences in sampling protocol, then it suggests that
apparent change in avoidance strategies since these northern resident killer whales have yet to grow accus-
whales were tracked in 1983, near the beginning of tomed to sharing the Strait, even after a killer-whale-
commercial whale-watching in Johnstone Strait (Kruse, generation of commercial whale-watching activity.
1991). Boat noise can mask communication signals used by
Kruse (1991) measured the swimming speeds of killer whales (Bain & Dahlheim, 1994). Bain & Dahl-
northern resident killer whales that were travelling heim (1994) tested the ability of captive killer whales to
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
268 R. Williams, A. W. Trites and D. E. Bain

detect pure tones, discrete calls of conspeci®cs, and larly while the issue of habituation remains unresolved.
echolocation click trains under varying levels of back- Clearly, southern residents are exposed to more boat
ground vessel noise. They found that low-frequency traf®c, for a greater part of the year, than northern
components of calls, which are omnidirectional, were residents (Osborne & Otis, 2000). Our ®nding that focal
masked by noise, and that masking was strongest when male whales travel a more circuitous route when fol-
the source was placed directly in front of the whale. In lowed than when no boats are present certainly bears
addition, a higher level of boat noise elicited a stronger consideration in the context of southern resident killer
masking effect. This study has several implications for whale conservation. If boat traf®c generally forces
wild killer whales. Bain & Dahlheim (1994) argue that animals to swim further to ®nd food, then this undoubt-
the key consequence of masking is to reduce the distance edly carries a metabolic cost. However, additional
over which killer whales can effectively search for food studies should test the applicability of our ®ndings to
by masking the lateral, low-frequency components of the southern resident community. In the meantime, we
calls. This hypothesis is critical for linking short-term view boat traf®c as one factor that can in¯uence the
behavioural responses to human activity and long-term `cost of living' for whales. In other words, low levels of
implications for the health of individuals and popula- disturbance may not be problematic in a thriving popu-
tions. Such a link is dif®cult to establish, but examples lation, but when coupled with reduced prey availability
from many disparate studies on a variety of taxa reveal and increased contaminant load, short-term behavioural
some recurring themes. responses should not be dismissed lightly.
Whales tend to respond to boat traf®c with the
stereotyped, short-term avoidance tactics (Howland,
1974; Weihs & Webb, 1984) of increasing swim speed Recommendations
(Kruse, 1991; Green, 1998) and varying the time and/or
position of surfacings (Bauer & Herman, 1986; Fraker, The northern resident killer whale population has
Richardson & WuÈrsig, 1995; Notarbartolo di Sciara et served as a useful subject of many studies over the last
al., 1996). On a larger spatial scale, northern resident 20 years. Long-term photo-identi®cation studies of this
killer whales are more likely to leave Robson Bight as population reveal no evidence of population decline
boat traf®c enters that reserve (Trites et al., 1995), just (Ford et al., 1994; Baird, 1999). Similar focus on
as mule deer abandoned portions of their range when sighting records will reveal whether these whales are
military activity intensi®ed (Stephenson, Vaughan & using Johnstone Strait less today than in previous years.
Andersen, 1996). Repeated disturbance in other taxa Careful comparison of our results to other datasets (e.g.
can force animals to be increasingly alert (MacArthur et Kruse, 1991), where appropriate, could offer valuable
al., 1982; Weisenberger et al., 1996), and may cost them insights into other processes, such as habituation,
foraging opportunities (Stephenson et al., 1996; Burger changes in habitat use, and disruption of foraging and
& Goch®eld, 1997; Galicia & Balassarre, 1997). Re- resting activity. Bain (1986) found diurnal patterns in
peated disturbance can cause lowered immune function the behaviour of captive killer whales. This pattern has
(Kraabel & Miller, 1997), abandonment of microhabi- not been observed in northern resident killer whales
tats (Eckstein et al., 1979) and disruption of sleep (Ford et al., 1994). If circadian rhythms of killer whales
patterns (OÈ hrstroÈm, BjoÈrkman & Rylander, 1990). can be changed in captivity, perhaps this can forge a
Currently, any proposed link between short-term link between short-term responses to boats and longer-
response and long-term effects is tenuous. Although no term implications. Existing datasets should be examined
study to date can address the underlying concern that to see whether the behaviour of northern resident killer
short-term disruptions may have a cumulative, long- whales has begun to re¯ect the diurnal patterns of
term effect that has not been measured yet, we do know whale-watching traf®c.
that the whales continue to return each year, and that Opportunistic tracking of boats and killer whales
this population continues to grow (Ford et al., 1994). should continue. Initially, these tracks should target
Therefore, the only fair assessment of large-scale effects observations poorly represented in our dataset (Figs 7 &
of boat traf®c on northern resident killer whales is a 8), namely those of females between 25 and 50 years of
qualitative one. The results presented here indicate that age, and of boat traf®c between 500 and 3000 m.
boat traf®c can disrupt short-term behaviour of indivi- Similarly, experimental studies with a single boat should
duals; however, there is no convincing evidence that continue, in order to identify effects at closer and
human disturbance is adversely affecting northern resi- further distances than 100 m, in order to determine
dent killer whales on the level of the population. whether the relationship between boat proximity and
Ultimately, though, managers must consider whether whale avoidance is linear, as canonical correlation ana-
whale-watching traf®c can in¯uence reproductive lyses assume. Ideally, these experiments should be
success. Recently, there has been concern about whether conducted in parallel with studies on southern residents.
whale-watching traf®c may play a role in the recent Our study measured behavioural responses of killer
decline of southern resident killer whales (Baird, 1999). whales to a vessel that actually followed whale-watching
We believe that it is unwise simply to assume that guidelines. It is important to remember that distance is
southern residents are disturbed to the same extent as very dif®cult to judge at sea, so many boats may be
northern residents by whale-watching traf®c, particu- violating whale-watching guidelines unintentionally.
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Behavioural responses of killer whales to whale-watching 269

Baird and Burkhart (2000) note, however, that humans Hanishewski, Erin Hildebrand, Fae Logie and Michelle
have a tendency to underestimate distance. Thus, in a Tung. The whale-watching and research community of
community where whale-watchers make a concerted Johnstone Strait sets an extraordinary example, and we
effort to stay 100 m away from whales, they may be thank David Arcese, Jim Borrowman, Mike and Judy
staying even further away than that. Durban, John Gasner, Bill and Donna Mackay, Dave
The results of studies on masking sounds suggest that and Maureen Towers, Paul Spong, Anna Spong, Helena
respectful whale-watching involves slow, parallel ap- Symonds and Dean and Kathy Wyatt. For assistance at
proaches. Leapfrogging may be inappropriate, since the data compilation, analysis and writing stages, we
speeding up to overtake the whale increases the intensity thank John Ford, Lee Gass and Daniel Pauly, Robin
of cavitation noise (Richardson et al., 1995). Increasing Baird, Lance Barrett-Lennard, Alistair Blachford,
propeller rotation rate also shifts engine noise to higher Volker Deecke, Ed Gregr, Kathy Heise, Susan Kruse,
frequencies (Richardson et al., 1995), which would have Rich Osborne, Bob Otis, John Pritchard, Pamela Ro-
greater potential for masking killer whale communica- senbaum, Cory Warren, Harald Yurk and two
tion signals (Bain & Dahlheim, 1994). Furthermore, anonymous reviewers. Graeme Ellis (Department of
placing a boat directly ahead of the whale's path puts Fisheries and Oceans) provided use of the experimental
the source of masking noise in the most disruptive boat. Shell Canada's Environment Fund and Mountain
position (Bain & Dahlheim, 1994). These factors may Equipment Co-op supported ®eld work. RW was sup-
explain why avoidance responses to leapfrogging vessels ported by an NSERC Industrial Post-graduate
seem to be more dramatic than to vessels travelling Scholarship and a grant from the Gae Weber Marine
parallel to the whale (R. Williams, pers. obs.). In our Mammals Fund.
predator-avoidance model, leapfrogging implies a fast,
manoeuvrable, dif®cult-to-evade `predator'. This sug-
gests that a ®ght, rather than ¯ight, response may be REFERENCES
more effective, as indicated by increased rates of ago-
nistic behaviour. This may be easier for us to detect Adimey, N. M. (1995). A descriptive study of the percussive
subjectively against baseline than changes in travel behavior of orcas, Orcinus orca, in Johnstone Strait, British
patterns. Further studies of this type should also record Columbia. MSc thesis, Nova Southeastern University, Fort
group size during observation sessions. This will result Lauderdale, Florida.
in a clearer understanding of the relationship between Bain, D. E. (1986). Acoustic behavior of Orcinus: Sequences,
periodicity, behavioral correlates and an automated technique
boat traf®c and rates of surface-active behaviour, which for call classi®cation. In Behavioral biology of killer whales:
vary with group size in southern residents (Hoelzel, 335±371. Kirkevold, B. C. & Lockard, J. S. (Eds). New York:
1993). Alan R. Liss.
The value of an experimental approach for detecting Bain, D. E. & Dahlheim, M. E. (1994). Effects of masking noise
subtle behavioural responses cannot be overempha- on detection thresholds of killer whales. In Marine mammals
sized. The growing desire to see animals in their and the Exxon Valdez: 243±256. Loughlin, T. R. (Ed.). San
natural habitat is prompting enormous changes in how Diego: Academic Press.
Baird, R. W. (1999). Status of killer whales in Canada. Species
society values whales. Indeed, many identify whale- Status Report. Ottawa: Committee on the Status of Endangered
watching as a sustainable alternative to whaling Wildlife in Canada (COSEWIC).
(Barstow, 1986; IWC 1994; IFAW, Tethys Research Baird, R. W. & Burkhart, S. M. (2000). Bias and variability in
Institute & Europe Conservation, 1995). However, distance estimation on the water: implications for the manage-
such efforts must be tempered by assurance that these ment of whale watching. International Whaling Commission
encounters do not cost the health of individual whales Meeting Document SC/52/WW1. (Available from http://
is.dal.ca/~whitelab/rwb/iwcvers.pdf )
or their populations (IWC, 1995). If we do not remain
Barrett-Lennard, L. G., Smith, T. G. & Ellis, G. M. (1996). A
sceptical of the apparently benign nature of ecotourism, cetacean biopsy system using lightweight pneumatic darts, and
we risk enforcing token whale-watching guidelines that its effect on the behavior of killer whales. Mar. Mamm. Sci. 12:
may be counterproductive. Furthermore, if we do not 14±27.
test the biological relevance of these guidelines, we may Barstow, R. (1986). Non-consumptive utilization of whales.
end up harming animals with our desire to appreciate Ambio 15: 155±163.
them. Bauer, G. B. & Herman, L. M. (1986). Effects of vessel traf®c on
the behavior of humpback whales in Hawaii. Unpublished report.
Honolulu: U.S. National Marine Fisheries Service.
Burger, J. & Goch®eld, M. (1997). Effects of ecotourists on bird
Acknowledgements behavior at Loxahatchee National Wildlife Refuge, Florida.
Environ. Conserv. 25: 13±21.
This study was funded largely by a BC Parks contract to Davis, R. E., Foote, F. S., Anderson, J. & Mikhail, E. (1981).
Bion Research Inc. The authors acknowledge the in- Surveying theory and practice. New York: McGraw Hill.
valuable assistance of Rik Simmons (BC Parks) and Duffus, D. A. & Baird, R. W. (1995). Killer whales, whale
watching and management: a status report. Whalewatcher
Marke Wong, David Briggs and Colin Mahony (Bion). (Fall/Winter): 14±17.
For additional ®eld assistance, we thank Peter Bauer, Duffus, D. A. & Dearden, P. (1993). Recreational use, valuation,
Darren Borrowman, Cheryl Ciccone, Carolyn Cornish, and management, of killer whales (Orcinus orca) on Canada's
Dorothy Dick, Phyllis Edgar, Graeme Ellis, Beth Paci®c coast. Environ. Conserv. 20: 149±156.
 14697998, 2002, 2, Downloaded from https://zslpublications.onlinelibrary.wiley.com/doi/10.1017/S0952836902000298 by Universidad Politecnica De Madrid, Wiley Online Library on [17/08/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
270 R. Williams, A. W. Trites and D. E. Bain

Eckstein, R. G., O'Brien, T. F., Rongstad, O. J. & Bollinger, J. G. and foraging behaviour of northern resident killer whales
(1979). Snowmobile effects on movements of white-tailed deer: (Orcinus orca) in relation to the inshore distribution of salmon
a case study. Environ. Conserv. 6: 45±51. (Oncorhynchus spp.) in British Columbia. Can. J. Zool. 74:
Ford, J. K. B., Ellis, G. M. & Balcomb, K. C. (1994). Killer whales: 983±991.
the natural history and genealogy of Orcinus orca in British Notarbartolo di Sciara, G., Jahoda, M., Biassoni, N. &
Columbia and Washington State. Vancouver: UBC Press. Lafortuna, C. (1996). Reactions of ®n whales to approaching
Fraker, M. A., Richardson, W. J. & WuÈrsig, B. (1995). Distur- vessels assessed by means of a laser range ®nder. In Proceedings
bance responses of bowheads. In Behavior, disturbance re- of the Tenth Annual Meeting of the European Cetacean Society,
sponses and feeding of bowhead whales Balaena mysticetus in the Lisbon, 11±13 March 1996. Kiel: ECS.
Beaufort Sea, 1980±1981: 145±248. Unpublished report by LGL È hrstroÈm, E., BjoÈrkman, M. & Rylander, R. (1990). Effects of
O
Ecological Research Associates, Inc., Bryan, TX. Washington: noise during sleep with reference to noise sensitivity and
U.S. Bureau of Land Management. habituation. Environ. Int. 16: 477±482.
Galicia, E. & Balassarre, G. A. (1997). Effects of motorized Olesiuk, P. F., Bigg, M. A. & Ellis, G. M. (1990). Life history and
tourboats on the behavior of nonbreeding American ¯amingos population dynamics of resident killer whales (Orcinus orca) in
in Yucatan, Mexico. Conserv. Biol. 11: 1159±1165. the coastal waters of British Columbia and Washington State.
Green, M. L. (1998). The impact of vessels on the Hawaiian Rep. Int. Whaling Comm. Spec. Issue 12: 209±243.
humpback whale (Megaptera novaeangliae): an experimental Osborne, R. W. & Otis, R. E. (2000). Growth trends in vessel-
study. Unpublished report. Washington: U.S. National Marine based killer whale watching in Haro Strait along the boundary
Fisheries Service. of British Columbia and Washington State. In Abstracts of the
Hoelzel, A. R. (1993). Foraging behaviour and social group 7th International Conference of the American Cetacean Society,
dynamics in Puget Sound killer whales. Anim. Behav. 45: Monterey CA, November 2000. San Pedro, CA: ACS.
581±591. Richardson, W. J., Greene, C. R., Malme, C. I. & Thomson,
Howland, H. C. (1974). Optimal strategies for predator avoidance: D. H. (1995). Marine mammals and noise. San Diego: Academic
the relative importance of speed and manoeuverability. J. Press.
theor. Biol. 47: 333±350. Roeder, K. D. (1967). Nerve cells and insect behavior. Cambridge,
Hoyt, E. (1997). The potential of whale watching in Europe. Bath: MA: Harvard University Press.
Whale and Dolphin Conservation Society. Stephenson, T. R., Vaughan, M. R. & Andersen, D. E. (1996).
IFAW, Tethys Research Institute and Europe Conservation Mule deer movements in response to military activity in south-
(1995). Report of the workshop on the scienti®c aspects of east Colorado. J. Wildl. Manage. 60: 777±787.
managing whale watching. Montecastello di Vibio, Italy. East Stewart-Oaten, A. (1995). Rules and judgements in statistics: three
Sussex: IFAW. examples. Ecology 76: 2001±2009.
International Whaling Commission (1994). Forty-fourth report of Sumich, J. L. (1983). Swimming velocities, breathing patterns, and
the International Whaling Commission. Cambridge: IWC. estimated costs of locomotion in migrating gray whales, Es-
International Whaling Commission (1995). Forty-®fth report of chrichtius robustus. Can. J. Zool. 61: 647±652.
the International Whaling Commission. Cambridge: IWC. Tabachnick, B. G. & Fidell, L. S. (1996). Using multivariate
James, F. C. & McCulloch, C. E. (1990). Multivariate analysis in statistics. New York: Harper & Row.
ecology and systematics: panacea or Pandora's box? Annu. Rev. Trites, A. W., Hochachka, W. & Carter, S. K. (1995). Killer
Ecol. Syst. 21: 129±166. whales and vessel activity in Robson Bight from 1991 to 1994.
Johns, B. G. (1996). Responses of chimpanzees to habituation and Report. British Columbia: Ministry of Environment, Land and
tourism in the Kibale Forest, Uganda. Biol. Conserv. 78: Parks.
257±262. Tyack, P. L. (1982). Humpback whales respond to the sounds of
Johnstone Strait Killer Whale Committee (1991). Background their neighbors. PhD thesis, Rockefeller University, New York.
report. Vancouver: JSKWC. Weihs, D. & Webb, P. W. (1984). Optimal avoidance and evasion
Johnstone Strait Killer Whale Committee (1996). Johnstone Strait tactics in predator±prey interactions. J. theor. Biol. 106:
whale watching guide. Vancouver: JSKWC. 189±206.
Kraabel, B. J. & Miller, M. W. (1997). Effect of simulated stress Weisenberger, M. A., Krausman, P. R., Wallace, M. C., De
on susceptibility of bighorn sheep neutrophils to Pasteurella Young, D. W. & Maughan, O. E. (1996). Effects of simulated
haemolytica leukotoxin. J. Wildl. Dis. 33: 558±566. jet aircraft noise on heart rate and behavior of desert ungulates.
Kriete, B. (1995). Bioenergetics in the killer whale, Orcinus orca. J. Wildl. Manage. 60: 52±61.
PhD thesis, University of British Columbia, Vancouver. Whitehead, H., Dillon, M., Dufault, S., Weilgart, L. & Wright, J.
Kruse, S. (1991). The interactions between killer whales and boats (1998). Non-geographically based population structure of
in Johnstone Strait, B.C. In Dolphin societies: discoveries and South Paci®c sperm whales: dialects, ¯uke-markings and ge-
puzzles: 149±159. Pryor, K. & Norris, K. S. Berkeley: Univer- netics. J. Anim. Ecol. 67: 253±262.
sity of California Press. Williams, R. M. (1999). Behavioural responses of killer whales to
MacArthur, R. A., Geist, V. & Johnston, R. H. (1982). Cardiac whale-watching: opportunistic observations and experimental
and behavioral responses of mountain sheep to human distur- approaches. MSc thesis, University of British Columbia. Avail-
bance. J. Wildl. Manage. 46: 351±358. able from http://www.marinemammal.org/pub.htm
Milstein, A. (1993). Factor and canonical correlation analyses: WuÈrsig, B., Cipriano, F. & WuÈrsig, M. (1991). Dolphin movement
basic concepts, data requirements and recommended proce- patterns: information from radio and theodolite tracking
dures. In Multivariate methods in aquaculture research: case studies. In Dolphin societies: discoveries and puzzles: 79±111.
studies of tilapias in experimental and commercial systems. Pryor, K. & Norris, K. S. (Eds). Berkeley: University of
ICLARM (Int. Cent. Living Aquat. Resour. Manage.) Stud. California Press.
Rev. 20: 24±31. Prein, M., Hulata, G. & Pauly, D. (Eds). Zar, J. H. (1996). Biostatistical analysis. Upper Saddle River, NJ:
Nichol, L. M. & Shackleton, D. M. (1996). Seasonal movements Prentice Hall.