AN ANALYSIS OF HARBOR SEAL (PHOCA WTULINA)AND
GRAY SEAL (NALICHOERUS GRYPUS) HAUL-OUT PATTERNS,
BEHAVIOR BUDGETS, AND AGGRESSIVE INTERACTIONS
ON MOUNT DESERT ROCK, MAINE
BY
Steven C. Renner
B.S. Allentown College of Saint Francis de Sales, 1995
A THESIS
Submitted in Partial Fulfillment of the
Requirements for the Degree of
Master of Science
(in Wildlife Ecology)
The Graduate School
The University of Maine
May, 2005
Advisory Committee:
James R. Gilbert, Professor of Wildlife Ecology, Advisor
Sean Todd, Professor of Marine Science, College of the Atlantic, Bar Harbor, ME
William E. Glanz, Professor of Biological Sciences
Frederick A. Servello, Professor of Wildlife Ecology
AN ANALYSIS OF HARBOR SEAL (PHOCA WTULINA) AND
GRAY SEAL (HALICHOERUS GRYPUS) HAUL-OUT PATTERNS,
BEHAVIOR BUDGETS, AND AGGRESSIVE INTERACTIONS
ON MOUNT DESERT ROCK, MAINE
By Steven C. Renner
Thesis Advisor: Dr. James R. Gilbert
An Abstract of the Thesis Presented
in Partial Fulfillment of the Requirements for the
Degree of Master of Science
(in Wildlife Ecology)
May, 2005
As gray seal (Halichoerus gypus) populations continue to grow in the Gulf of Maine,
it is necessary to quantie changes to the regional ecology for both management and
conservation purposes. T h s study compares haul-out patterns, presents summer activity
budgets, and contrasts intra- and interspecific aggressive interactions among the harbor
seal (Phoca vitulina) and gray seal population on Mount Desert Rock, Maine. These data
were collected using Altman's scan method and focal animal sampling.
Time to low tide and day of year influenced both harbor seal and gray seal haul-out
patterns on Mount Desert Rock. For both species, more seals hauled-out closer to low
tide. Day of year was correlated positively with numbers of harbor seals and negatively
with numbers of gray seals.
Overall activity budgets are presented for both species; the harbor seal budget agrees
well with previously published data and the gray seal budget is the first documented for
this species during summer months, but is generally similar to winter breeding season
budgets. Intrinsic and environmental factors influenced harbor seal behavior patterns.
Gray seals had little, if any, effect on harbor seal behavior patterns and were only
important in one of three models describing harbor seal sleep budgets. Harbor seal
behavior budgets and rates were most broadly-affectedby number of adjacent seals. At
moderate densities of 3-5 adjacent seals, focal harbor seals slept more in longer, fewer
bouts, whle reducing overall time scanning via fewer bouts. Sleep budgets were also
positively correlated with total number of seals on a haul-out ledge, day of year, wind
speed, and cloud cover and negatively correlated with absolute time to low tide and air
temperature. Sleep bouts occurred more frequently during higher tides and during calm
wind conditions. Scanning budgets were positively correlated with hgher tide states and
the absence of fog and negatively correlated with number of seals on a ledge, wind speed,
day of year, and cloud cover.
Harbor seal sleep and scan behaviors varied over the course of a day. The activity
cycle was independent of tide state and was characterized by alternating peaks of sleep
s
and scan behavior. Sleep peaked near midday in both years and t h ~ may explain why
maximum haul-out numbers occur near midday in numerous other studies.
Although gray seals did not affect harbor seal behavior patterns, male gray seals
clearly were dominant to harbor seals during aggressive interactions. Intra- and
interspecific interactions involving only two individuals were similar in duration. Harbor
seals responded aggressively less frequently when a gray seal changed body position than
when another harbor seal performed the same behavior. Harbor seals may recognize and
avoid confrontations with significantly larger opponents.
ACKNOWLEDGMENTS
I would to thank the students and staff of Allied Whale, particularly Dan Dendanto,
Judy Allen, Ozlem Uz, Courtney Vashro, Nikolai Klibansky, Jessica Damon, Keri
Barber, and Kate Doan for their logistical and moral support, scientific advice, and
assistance collecting data.
I acknowledge Dr. Bill Halteman for his statistical advice and Dr. James Gilbert and
Dr. Sean Todd for their analytical guidance and editorial comments during the writing
stages of this thesis. I also recognize Dr. Fred Serve110 for his guidance and support, and
Dr. Bill Glanz for sharing his expertise in mammalian ecology and behavior.
Thanks to my parents and in-laws for watchng the luds for days at a time while I
stared at the computer or had my nose buried in a statistics book or journal article.
Special thanks to my wife, Tuesday, and our daughters, Lara Anne and Mackenzie Paige,
for their support and understanhng while we endured this seemingly never-ending
process.
Without the financial support of The Green Endowment Fund, The University of
Maine Alumni Association Student Travel Support Program, the Maine Forest and
Agriculture Experiment Station, and partially-subsidized travel and living expenses from
Allied Whale, this research would not have been possible.
REFERENCES ............................... .....................................-.........................................,..-60
APPENDIX: Effects of human-induced disturbances on number and distribution
of hauled-out harbor seals (Phoca vitulina) and gray seals
(Halichoerusgrypus) on Mount Desert Rock..... ........................ ................65
...-....
BIOGRAPHY OF THE AUTHOR ......................................... ........................ ...-......
80
LIST OF TABLES
Table 1. Variables that influenced number of harbor seals, gray seals, and total
seals hauled-out on Mount Desert Rock during summer from 1999-2001........17
Table 2. Definitions of behaviors observed among harbor seals and gray seals
while collecting scan and video data on Mount Desert Rock in 2000 and
2001. ..................................................................................................................18
Table 3. Average activity budgets (percent time) for harbor seals and gray seals
on Mount Desert Rock from l u n e - ~ u ~ u2000 and 200 1. ............................... 18
st
Table 4. Variables that were significantly related to proportion of time harbor
seals spent sleeping on Mount Desert Rock in 2000 and 200 1 based on
scan and video data. .......................................................................................... .2 1
Table 5. Variables that affected number of sleep bouts/minute for harbor seals on
Mount Desert Rock based on focal animal video data from 2000 and
200 1. .................................................................................................................
.26
Table 6. Variables that were sipficantly related to proportion of time harbor
seals spent scanning their surroundmgs on Mount Desert Rock in 2000
and 200 1 based on scan and video data. ............................................................28
Table 7. Variables that affected number of scan bouts/minute of focal harbor seals
on Mount Desert Rock in 2000 and 200 1 based on video data. ........................30
Table 8. Outcomes of interactions for all species, age class, and gender
combinations observed on Mount Desert Rock in 2000 and 200 1....................36
Table 9. Differences in interaction duration among observed participant
combinations on Mount Desert Rock in 2000 and 200 1 using
Kruskal-Wal lis multiple comparison procedures. ............................................
-36
Table 10. Total 2 values for response patterns to all initial behaviors during
interactions observed on Mount Desert Rock in 2000 and 2001. ......................37
Table 11. X2 contingency table for harbor seal responses to opponents' change of
body position as a function of participants in the interaction............................39
Table A. 1. Average ranks of net change in total number of seals hauled-out on
Mount Desert Rock in 200 1 before and 40 minutes after lsturbance
events for all combinations of tide state and treatment type..............................72
Table A.2. Pairwise c o m p s o n s of net change in total number of seals
hauled-out on Mount Desert Rock in 200 1 before and 40 minutes after
disturbance events for all combinations of tide state and treatment type
using Kruskal-Wallis procedures. .....................................................................
.73
Table A.3. ANOVA table for ranked change in number of seals hauled-out on
each ledge before and 40 minutes after a lsturbance event on Mount
Desert Rock in 200 1. .................................................................................. .73
Table A.4. Comparison of control and disturbance treatments within tide states
observed on Mount Desert Rock in 200 1 using the one-tailed Dunnett's
Test.................................................................................................................... .75
LIST OF FIGURES
Figure 1. (a) Smoothed plot of total number of seals hauled-out on Mount Desert
Rock in 1999,2000, and 2001 as a function of relative number of
minutes from low tide. (b) Smoothed plot of total number of seals
hauled-out on Mount Desert Rock in 1999,2000, and 200 1 as a function
of absolute number of minutes from low tide. ................................................. 16
Figure 2. Smoothed plots of sleep (a) and scan (b) budgets as a function of
relative time to low tide for 2000 and 2001 scan data. Smoothed plots
of sleep (c) and scan (d) budgets as a function of absolute time to low
tide for 2000 and 2001 scan data. ....................................................................
.19
Figure 3. Mean sleep budgets (logt transformed) for all genders and age classes
of focal harbor seals on Mount Desert Rock based on video data from
2000. .................................................................................................................-23
Figure 4. Mean sleep and scan budgets (a, b), rates (c, d bars), and durations
(c, d 0) for focal harbor seals on Mount Desert Rock in 200 1 as a
function of number of seals within one seal-length of the focal animal.. ..........24
Figure 5. Mean number of sleep boutslminute (bars) and mean duration of sleep
bouts (points) for all genders and age classes of focal harbor seals
observed on Mount Desert Rock in 200 1 video data.........................................27
Figure 6. Hourly sleep patterns among focal harbor seals on Mount Desert Rock
in 2000 (0) and 200 1 (x)...................................................................................
33
Figure 7. Hourly variation in scanning budgets (x) and rates in 2000 (0) and
2001 (m) among focal harbor seals on Mount Desert Rock................................
34
Figure A. 1. Number of seals hauled-out on Mount Desert Rock during
disturbance and control counts. .........,..............................................................-69
INTRODUCTION
Harbor seals (Phoca vitulina) and gray seals (Halichoerusgypus) are the most
common phocids in the Gulf of Maine. Both species were historically prevalent in this
region and remains have been reported in numerous middens along the New England
coast (Speiss and Lewis 2001, Borque 1973, Waters 1967). As the fisheries industry
grew, so did human interactions with marine mammals. Bounty hunting and illegal
killing negatively affected harbor and gray seal populations until the Marine Mammal
Protection Act in 1972 protected all marine mammal species. Despite hunting pressure,
harbor seal populations persisted in the Gulf of Maine; gray seals, however, were nearly
eradicated from U.S. waters. A remnant gray seal breeding colony in Massachusetts has
grown significantly since the mid-1970's (Northeast Fisheries Science Center 2003, S.
Wood, pers. cornm., University of Massachusetts). Small breeding colonies have
recently been documented on several remote islands in the Gulf of Maine (J.R. Gilbert,
pers. cornm., University of Maine). The core of the Western Atlantic gray seal breeding
population on Sable Island in Canada has been increasing 13% annually (Bowen et al.
2003, Lesage and Harmnill2001, Mohn and Bowen 1994). Population growth is 3.4 *
1.5% per year in other breeding areas in the region (Lesage and Hammill 2001). During
summer months, breeding colonies disperse and several thousand gray seals haul-out and
feed in the northern Gulf of Maine.
Because gray seals have continued to become more numerous in the Gulf of Maine
throughout the year, it has become necessary to develop a more complete understanding
of their biology and ecology in the region for both conservation and management
purposes. Regulations and policies regarding seal interactions with fisheries and
aquaculture may need to be revised pending results of current research efforts. Multiple
studies are presently ongoing to determine prey overlap (Williams 1999, D. Kopec, pers.
cornm., University of Maine), fisheries and aquaculture interactions (Nelson 2004 ), bi-
catch mortality (Williams 1999), and population genetics for both species in the Gulf of
Maine (S. Wood, pers. comm., University of Massachusetts). The objectives of thls
study were to describe haul-out patterns, to calculate activity budgets for hauled-out
harbor seals and gray seals during summer, to characterize effects of gray seals on harbor
seal behavior patterns, and to contrast intra- and interspecific agonistic interactions
among seals on Mount Desert Rock, Maine.
Pinnipeds leave the water, or haul-out, for several purposes incluchng, but not limited
to, thermoregulation, reproduction, rest, and predator avoidance. Although haul-out
behavior, duration, and seasonal variability differ among Phocidae, harbor seals (Frost et
al. 2001) and gray seals (Sjiiberg et al. 1995, Cameron 1970) haul-out throughout the
year.
Harbor seals haul-out on various substrates including sand beaches, flat rock ledges,
pebble beaches, ice floes, glacial drift, man-made floats, and tidal flats, ledges, and
shelves (Stewart 1984). Harbor seals prefer sites or site clusters that have a Iarge total
landing area, are close together, and are accessible over a wide range of tidal heights
(Kneber and Barrette 1984). When a variety of substrates are available, harbor seals are
likely to select large, offshore, sub-tidal rocks exposed earlier during the outgoing tide
over nearby sand and pebble beaches (Nordstrom 2002, Kneber and Barrette 1984,
Schneider and Payne 1983).
Gray seals also haul-out on numerous substrates, although very few descriptions of
non-breehng colonies have been published. Krieber and Barrette (1984) anecdotally
described haul-out areas used by non-breeding gray seals near their harbor seal study site
in Forillon National Park, Canada. Rock outcroppings used by gray seals were generally
large, dome shaped, and relatively far from shore, although there was little similarity
between ledges (Krieber and Barrette 1984).
Number of hauled-out harbor seals is heavily influenced by time of day, with peak
counts occurring near mid-day (Boveng et al. 2003, Reder et al. 2003, Simpluns et al.
2003, Frost et al. 1999, Sjoberg at al. 1999, Watts 1996, Thompson et al. 1989,
Calambokidis et al. 1987, Pauli and Terhune 1987a, Pauli and Terhune 1987b, Stewart
1984). Tidal phase may also be important, particularly where haul-out site availability is
limited by water depth (Boveng et al. 2003, Henry and Hammill 2001, Frost et al. 1999,
Thompson 1989, Pauli and Terhune 1987%Schneider and Payne 1983). Lunar phase
(Watts 1993), weather conditions (Boveng et al. 2003, Henry and Hammill 2001, Pauli
and Terhune 1987), dsturbance regimes (Lelli and Harris 200 1, Schneider and Payne
1983, Renouf et al. 1981), distance from shore (Nordstrom 2002), and availability of ice
(Boveng et al. 2003, Calambokidis et al. 1987) may affect haul-out trends in specific
situations.
Gray seals often remain hauled-out for the entire mid-winter pupping, nursing, and
mating season (Anderson et al. 1975). Data on gray seal haul-out behavior during
summer months, however, are scarce. Cameron (1971) reported that gray seals in eastern
Canada exhibit a diurnal activity rhythm attuned to tide cycle, hauling-out at low tide and
foraging during high tide. Sjoberg et al. (1999), however, found that gray seals followed
a die1 haul-out pattern with peak numbers present on the haul-out at night and
hypothesized the observed gray seal activity cycle was lmked to prey behavior.
While hauled-out, harbor and gray seals exhibit numerous alert, rest, and social
behaviors. Several factors influence the frequency and duration of seal behaviors on a
particular haul-out site, including group size, haul-out duration, and relative body
position. DaSiIva and Terhune (1988) and Terhune (1985) reported that as group size
increased, individual harbor seals spent less total time in alert scan behaviors, scan
durations were shorter, and intervals between scans were longer. Vigilance, a function of
scan rate and duration, also decreased over elapsed time since haul-out (Terhune and
Brillant 1996). These studies, however, do not claim that reduction in scanning results in
a converse increase in sleep time. Position and orientation of individuals on the haul-out
are also important. Terhune and Brillant (1996) demonstrated that seals on the periphery
more frequently oriented toward the water and spent more time scanning than seals in the
interior. Sullivan (1982) reported a general activity budget for a small haul-out group,
but did not statistically analyze these patterns as a function of group demographics or
environmental factors. Thompson et al. (1995), Godsell (1988), Davis and Renouf
(1987), Kneber and Barrette 1984, and Ling et al. (1974) identified gray seals on or near
their study sites, but did not include their presence as a variable in any analyses of harbor
seal behavior.
Because of the colonial nature of haul-out groups, social behaviors frequently include
agonistic exchanges between two or more seals. These interactions result from incidental
contact between adjacent individuals, group landings on the haul-out site, and
competition for preferred resting areas (Sullivan 1982). Frequency of agonistic conflicts
is positively correlated with density on the haul-out; duration and outcomes of
interactions are a function of gender, size, and age class of participants (Neumann 1999,
Sullivan 1982). Although harbor seals mate in water, adult males may frequently scan
(Renouf and Lawson 1986) and change location (Sullivan 1981) to follow females in
estrous, often disrupting other seals on the ledge, thereby precipitating aggressive
behavior. Agonistic interactions among harbor seals are particularly evident when haul-
out space is limited (Neumann 1999).
Detailed infoxmation regarding interactions among gray seals has not been
documented outside of the breeding season. Cameron (1971) noted that identifiable
males and females returned to a particular ledge daily and defended it against all other
seals in a small non-breeding colony. Descriptions of agonistic behavior between gray
seals and harbor seals are also absent fiom the literature.
Unlike most other islands in the Gulf of Maine, Mount Desert Rock is far from the
madand, is a large, relatively undisturbed haul-out site for both harbor seals and gray
seals, and is seasonally accessible to researchers for long durations. These unique
characteristics of Mount Desert Rock permitted me to study three major components of
harbor seal and gray seal ecology: (1) haul-out patterns, (2) terrestrial behavior budgets,
and (3) intra- and interspecific interactions.
First, I determined predictor variables for haul-out patterns of both harbor seals and
gray seals. Because haul-out space is only somewhat limited by tide on Mount Desert
Rock, I expected time of day and weather variables to play an important role in
explaining haul-out patterns. Secondly, I measured activity budgets for harbor seals and
gray seals during summer months on Mount Desert Rock. To the best of my knowledge,
this is the first study incorporating group size, number of adjacent seals, species
composition, and environmental variables in a model to predict sleep and scan behavior
patterns at multiple spatial scales. If presence of gray seals on a mixed-species haul-out
affects harbor seal behavior patterns, I expected harbor seals to sleep less and to scan
more frequently, particularly when in close proximity to gray seals. Lastly, I compared
duration, outcomes, and progression of specific aggressive behaviors of intra- and
interspecific interactions. Given the marked difference in size between the two species
(male gray seals are 3 my288 kg, female gray seals are 2.3 m, 248 kg, and adult harbor
seals are 1.5 my115 kg), I hypothesized that gray seals would be dominant to harbor seals
and that interspecific interactions would be shorter and include more fiequent strongly
aggressive behaviors than encounters involving only harbor seals.
STUDY AREA
Mount Desert Rock (43O58'North 68O08'West) is a 1.82 hectare granite outcropping
in the Gulf of Maine. Overall, the island is dome-shaped, although it is divided into
numerous smooth, rounded ledges separated by fissures that extend to the low tide line.
Aside from sparse terrestrial vegetation introduced by human residents, the island is bare
rock above the high-tide line. Seals predominantly haul-out in intertidal areas covered by
numerous algal species, barnacles, and mussels. High and low tides alternate
approximately every 6.2 hours.
Mount Desert Rock was continuously inhabited for nearly 100 years by families
ih
tasked wt maintaining lighthouse facilities, a responsibility later transferred to the
United States Coast Guard. In the 19707s,the College of the Atlantic (Bar Harbor,
Maine) and Allied Whale (Bar Harbor, Maine) began using the island as a research
station during summer months in cooperation with the Coast Guard. The College of the
Atlantic acquired the island in 1996 and presently operates it as the Edward Blair Mount
Desert Rock Marine Research Station. The remote facility consists of 4 historic
structures including a lighthouse, whrch provided an overview of all haul-out ledges, a
light keeper's house, a boathouse, and a utility shed.
METHODS
Several types of data were collected to meet the objectives of t h s study. I used count
data fiom 1999,2000, and 200 1 to identifj haul-out patterns for both harbor seals and
gray seals. I used Altmann's (1 973) scan method and focal animal sampling in 2000 and
200 1 to generate activity budgets and to assess factors influencing behavior patterns. I
observed interactions opportunistically during both sampling methods in 2000 and 200 1.
Haul-out Patterns
To determine which environmental factors contributed to number of seals hauled-out
on Mount Desert Rock, I counted total number of harbor seals and gray seals on the
entire island at 4 tide states (low, high, mid-rising, mid-falling) and at all times of day in
1999 and 2000. In 2001, I did not perform counts specifically for this purpose, but
utilized total island counts in conjunction with disturbance data collection (see
Appendix).
I combined data fiom all years and used forward stepwise regression (a 50.05 to
include/remove) in SYSTAT's Generalized Linear Model (GLM) to determine predictor
variables for harbor seals, gray seals, and combined counts. Independent variables
recorded at the time of data collection and included in the analysis were day of year, time
of day (1 hour blocks), relative time to low tide (minutes before (-) or after (+) low tide),
cloud cover (0 if 50%), and presencelabsence of fog. At a later date, I matched wind
direction (eight 45" blocks beginning with 0-44O), wind speed (knots), and air
temperature ("C) f o hourly Mount Desert Rock NOAA weather station reports to each
rm
data collection period.
For each scan, activity budgets equaled the proportion of seals engaged in each
behavior. I averaged these results across all scans to create overall activity budgets for
both harbor seals and gray seals. Because most observed behaviors occurred
infrequently, I only analyzed sleep and scan budgets.
I used SYSTAT's GLM protocol to perform a forward stepwise regression (a 5 0.05
to include/remove) to determine predictor variables for sleep and scan activity patterns. I
selected the GLM because it efficiently handled both categorical and continuous
variables. I included day of year, year, time of day, time to low tide, observation ledge,
total number of seals on the observation ledge, presencelabsence of gray seals on the
observation ledge, wind direction, wind speed, air temperature, cloud cover, and
presencelabsence of fog as independent variables in the analyses. I compared means
withn sigruficant variables using Scheffe procedures for multiple comparisons.
To meet the assumptions of the GLM, I transformed three continuous variables. I
square-root transformed total number of seals and logit-transformed
(Y' = log(e) [Y/(l-Y)]) sleep and scan proportions to normalize these data. Because the
relationship between logit-transformed sleep and scan budgets and relative time to low
tide appeared non-linear and symmetrical around low tide, I linearized t h s relationship
by substituting absolute time to low tide for relative time to low tide prior to additional
analyses.
Prior to analyses, I removed 73 data points from the sleep and scan data sets. Twelve
cases (Y=1.O) in sleep data and 55 cases (Y=O.O) in scan data were not entered into the
initial regression analyses because logit transformations were not possible. After creating
initial regression models, I identified outliers using normal probability plots of residuals.
I removed these cases prior to fitting the final models. To determine whether deleted
cases would have affected the final model and to confirm that other excluded cases were
outliers, I retrofitted all deleted cases into their respective models and compared
Bonferroni corrected confidence intervals of resulting estimates to observed values.
To assess factors affecting harbor seal behavior at a smaller scale, T used focal animal
sampling. Focal animal sampling occurred in conjunction with scan sampling, so I
implemented identical schedules among designated ledges and times of day for both
methods. I videotaped focal harbor seals on the designated study ledge for 5 minutes at
30 minute intervals over a 4 hour period, provided seals were present and visibility was
adequate. I selected focal harbor seals using two criteria: (1) most or all of the selected
animal's body was visible and (2) its position on the ledge was approximately midway
between the water and the seal farthest from water. Previous research indicated behavior
patterns, particularly scanning rates, differ with an individual seal's position on the ledge
(Terhune and Brillant 1997). I also recorded day of year, year, time of day (1 hour
blocks), relative time to low tide, cloud cover (0 if 50%), and
presencelabsence of fog. At a later date, I added wind drection (eight 45" blocks
beginning with 0-44"), wind speed (knots), and air temperature ("C) data from the Mount
Desert Rock NOAA weather station.
Between March and May 2002, I reviewed 19.5 hours of video containing 223
samples from 2000 and 10.9 hours containing 128 samples from 2001. I recorded the
duration of each behavior performed by the focal seal. From these data, I calculated an
activity budget (percent time with each behavior), frequency of each behavior (rate per
minute), and average duration of each behavior (average time in behavior). In addition, I
noted species, gender (male, female, undetermined), and age class (pup, juvenile, adult)
of the focal harbor seal and of all adjacent seals in the video sample. I defined "adjacent"
as any seal within one adult harbor seal-length (approximately 1.5m) fiom the focal
animal, without another seal or other physical barrier in between. Whenever possible, I
sampled a different individual in each 5-minute segment. However, when few seals were
hauled-out or visibility was limited during a particular observation period, I sampled the
same individual in no more than 3 consecutive video segments. I believe that other
independent variables changed adequately during the 90 minute time span between the
first and third sample to justifl this and to prevent having a "case study" of a particular
seal.
I used SYSTAT's GLM to perform a backward stepwise regression (a=0.05 to
includelremove) to determine predictor variables for sleep and scan activity budgets and
frequency. After combining data fiom both years in initial analyses, 2 of 3 models
included 'year' as a sipficant predictor variable, so I removed year as a variable and
proceeded to analyze each year independently. Independent variables included in the
model were observation ledge, gender and age class of focal harbor seal (adult male,
adult female, or juvenile), total number of adjacent seals (1 through 6), presence/absence
of an adjacent gray seal, day of year, time of day, absolute time to low tide, and all
measured weather variables. Because harbor seals accounted for most of the observed
adjacent seals and gray seals were absent in 66% of samples, I included only total number
of adjacent seals and presence/absence of an adjacent gray seal, rather than the specific
number of each species adjacent to the focal animal.
Prior to regression analysis, I logit-transformed sleep and scan proportions to
normalize these data. I also removed cases where logit transformation was not possible
(Y = 0.0 or 1.0) and identified outliers via studentized residuals and normal probability
plots of residuals. After generating final models, I retrofitted all deleted cases into the
final models. I used Bonferroni adjusted confidence intervals to compare estimates with
expected values.
Interactions
During the intervals between scan and video sampling, I opportunistically documented
aggressive interactions between 2 or more seals. I noted the participants' species, gender,
and age class, which seal and event precipitated the interaction, progression of behaviors,
and outcome. In addition to these data, samples from videotaped focal animals with
aggressive interactions were included in analyses.
I used X2 analysis to d e t m i n e whether some classes of seals were dominant to others
based on outcomes of agonistic interactions. "Winners" were defined as seals whose
behavior resulted in "losers" moving to a new location or changing body position. If all
species, age classes, and genders were equally likely to win, one would expect a 1:1 ratio
of outcomes. Although I observed interactions involving 3 or more seals, I limited
analysis of outcomes to interactions including only 2 seals because "winners" were
clearly defined. I also limited analyses to participant combinations with at least 10
samples to meet statistical assumptions. For remaining data, I generated 1x2 contingency
tables with participants in columns and frequency each participant "won" in the row and
used X2 analysis to identifjr departures from a 1:1 ratio for all types of interactions.
Video samples also permitted calculation of interaction duration. I analyzed duration
as a function of number (2 or 3) and demographics (age class and species) of participants
using Kruskal-Wallis One-Way ANOVA. I identified one outlier (an interaction among
3 harbor seals that persisted for 86 seconds) via box plots and removed it fiom the data
set prior to analysis. Four treatments included (1) adult harbor seal vs. adult harbor seal,
(2) adult harbor seal vs. adult gray seal, (3) adult harbor seal vs. juvenile gray seal, and
(4) three adult harbor seals. I compared treatments means using Kruskal-Wallis multiple
comparisons procedures for significant results.
I used X2 analyses to determine if frequency of harbor seal responses to each
aggressive behavior was contingent upon the species of seal performing the initial
behavior. I adjusted alpha to account for five X2 tables. I partitioned sigmficant 2 values
post-hoe to make general comparisons between harbor seal-harbor seal and harbor seal-
gray seal interactions.
All sampling occurred on 15 days between 4 July-2 1 July 1999,32 days between 15
June- 16 August, 2000, and 17 days between 22 June- 19 August, 200 1. In 1999 and 2000
I observed seals from a portable blind, fiom behind rocks and ledges, and from several
buildings. In 2001, construction of a semi-permanent platform on the roof of the
boathouse afforded me excellent visual access to a majority of study sites and greatly
reduced the risk of disturbing seals. I used either a tripod mounted 500-rnrn Orion
achromatic refractor teIescope with 26-mm eyepiece, Minolta 10x25 binoculars, or a
Sony Digital 8 Handycam video camera to collect all data.
RESULTS
Haul-out Patterns
Approximately 1,550 seals (1,300 harbor seals and 250 gray seals) hauled-out on
Mount Desert Rock's numerous ledges exposed by low tide in 1999,2000, and 200 1.
Some seals often remained hauled-out throughout the tide cycle, although I observed
maximum numbers near low tide (Fig. 1).
In both the harbor seal and gray seal models, day of year and absolute time to low tide
were the strongest predictors of number of hauled-out seals on Mount Desert Rock (Table
1). Use of absolute over relative time to low tide improved the fit of all three haul-out
models. Number of gray seals was also positively correlated with wind speed. Number
of seals in both the harbor seal and the combined-species models was inversely related to
air temperature. Time of day, cloud cover, wind direction, and fog were not significantly
related to numbers of gray or harbor seals observed.
Behavior Budgets
I observed numerous resting and non-resting behaviors among both species of seals on
Mount Desert Rock (Table 2). Both harbor seaIs and gray seals spent most of their time
in resting behaviors (Table 3). I restricted further analyses to sleep and scan budgets
because all other behaviors occurred infrequently. Use of absolute rather than relative
time to low tide improved fit of all models and appeared to be justified based on a
comparison of individual regression models (Fig. 2).
Several intrinsic and environmental factors affected harbor seal sleep budgets on
Mount Desert Rock haul-out ledges (Table 4). Separate regression models from scan and
Figure 1. (a) Smoothed plot of total number of seals hauled-out on Mount Desert
Rock in 1999,2000, and 2001 as a function of relative number of minutes from low
tide. (b) Smoothed plot of total number of seals hauled-out on Mount Desert Rock
in 1999,2000, and 2001 as a function of absolute number of minutes from Iow tide.
Polynomial plot (solid) approximates a linear function (dashed). F o r both plots, p-
values a r e from least-squares linear regression models and n = 120.
0 1 .
I I I I I I
-400 -300 -200 -100 0 100 200 300 400
Time from low tide (minutes)
Absolute time to low tide (minutes)
Table 1. Variables that influenced number of harbor seals, gray seals, and total seals hauled-out on Mount Desert
Rock during summer from 1999-2001 (n = 120).
Model (F,p) Variable Coefficient df F P Cumulative 2
Harbor Seals Day of year 9.533 1 28.083 13 indicates an increase in number of seals; average rank 4 3 indicates a decrease in number
of seals.
Tide Treatment (n) Rank Sum Average Rank Corresponding Net Change
Falling Control (4) 14.25 Slight increase in number
Short Disturbance (4) 7.75 Lost -330
Rising Control (4) 8.25 Lost -300
Short Disturbance (1) 8.00 Lost -325
Long; Disturbance (3) 2.33 Lost -100
Table A.2. Pairwise comparisons of net change in total number of seals hauled-out
on Mount Desert Rock in 2001 before and 40 minutes after disturbance events for
all combinations of tide state and treatment type using Kruskal-Wallis procedures.
N = 1 6 , p = 22.667, k = 5, T = 10.989, t (o.ws,~l) 2.201.
=
Treatments
1 - Falling tide, Control
2 - Falling tide, Short hsturbance
3 - &sing tide, Control
4 - Rising tide, Short disturbance
5 - Rising Tide, Long Disturbance
Comparisons IRilni- Ri /nil t(o.975.48 [Y(N-1 ~ / ~ - k ) l ' ~ ( l+n i
- / llni)'" Conclusion
1 and 2 6.50 4.474 1# 2
1 and 3
1 and 4
1 and 5
2 and 3
2 and 4
2and 5
3 and 4
3and 5
4 and 5
Table A.3. ANOVA table for ranked change in number of seals hauled-out on each
ledge before and 40 minutes after a disturbance event on Mount Desert Rock in
2001. (N = 360).
Source Sum of Squares d Mean-Square
f F P
Tide 206018.178 1 206018.178 49.780 0.05). However, short duration
~
disturbances during falling tides resulted in a smaller net change in number of seals than
during rising tides (Tukey T = 10.442,p < 0.00 1).
Table A.4. Comparison of control and disturbance treatments within tide states observed on Mount Desert Rock in 2001
using the one-tailed Dunnett's Test. (SE = 11.745, q(o.os,60,3) = 1.95). Zero net change is equivalent to rank = 262.5.
Tide Treatment Mean Rank Comparison JC Conclusion
Falling Control 280.375 vs. Long Disturbance 15.809 Control # Long Disturbance
vs. Short Disturbance 3.592 Control Short Disturbance
Long Disturbance 94.700
Short Disturbance 238.192
Rising Control 199.625 vs. Long Disturbance 6.895 Control # Long Disturbance
vs. Short Disturbance 4.100 Control # Short Disturbance
Long Disturbance 118.642
Short Disturbance 151.467
DISCUSSION
Tide state and duration of disturbance events were both important in characterizing
changes in seal numbers at both the scale of the entire island and of individual haul-out
ledges. At the larger scale, short disturbances, regardless of tide state, had similar effects
on the number of seals on Mount Desert Rock. When compared to control counts
however, effects of short disturbances were dependent on tide. If a short disturbance
occurred during a rising tide, net change in number of seals on the island 40 minutes after
the disturbance event was similar to control counts. Short disturbances during falling
tides resulted in fewer hauled-out seals relative to control counts.
These results are counterintuitive. I would have expected seals that experienced short
less
disturbances during the rising tide to have e ~ b i t e d recovery because higher tides
reduced available haul-out space on Mount Desert Rock. Conversely, as more haul-out
area became available during falling tides, I would have expected recovery to numbers
close to control counts within 40 minutes after a short disturbance. To maximize rest
duration, harbor seals should haul-out as soon as ledges are available on the falling tide
and remain hauled as long as tide height permits. Because short disturbance effects were
*
temporally and spatially localized, seals that recently hauled-out and were closest to the
water may have been most affected. After re-entering the water, these seals may have
been less likely to haul-out again either in the same location or within the 40 minute
window observed because of reduced availability or because of behavioral changes. I
observed harbor seals scanning potential haul-out locations from the water and it may be
interesting to assess aquatic scanning rates relative to conditions and frequency of
disturbance. Aquatic scans of potential haul-out sites were also noted by Terhune and
Brillant (1 996), although data were not analyzed.
I observed 4 types of short duration disturbance. This sample size did not permit
isolation of hfferential effects among specific short duration disturbances that may be
important. For example, vessel type, speed and direction can Influence harbor seal
reactions (Henry and Hammill 200 1, Lelli and Hams 200 1, Suryan and Harvey 1999).
Seals on Mount Desert Rock, however, did not flush when lobster boats approached haul-
out ledges, indicating some degree of habituation to regularly occurring disturbance types
is possible (Terhune and Almon 1983, Greenwood 1976).
Long duration disturbances only occurred on rising tides immediately following low
tide to maximize efficiency of scat collection. The effkct of long disturbances on number
of seals hauled-out was highly variable because duration, area, and number of
participating researchers in scat collection were not consistent among samples. For
example, scat collection on 19 August (Fig. A. 1f) included 6 areas and lasted 135
minutes resulted in more seals leaving haul-outs relative to scat collections on 17 and 18
August (Figs. A. Id and e) that included only 3 areas and lasted 75 minutes.
In an attempt to isolate specific ledges affected by disturbances, the ANOVA model
failed to show differences within areas among treatments. These results imply that seals
do not immediately haul-out in alternate locations following disturbances on their
preferred site. Subsequent observations over consecutive low tide cycles may be
necessary to determine if seals shift haul-out locations over longer time scales following
disturbances. Anecdotal evidence, based on observations from boats upon first arriving
at Mount Desert Rock each field season and during early morning observations, indicated
shifts in haul-out patterns away fiom areas repeatedly disturbed by human activity.
These general observations, as well as existing literature (Suryan and Harvey 1999, Traut
1999, DaSilva and Terhune 1988, Terhune 1985, Sullivan 1980), show that disturbance
frequency may also be important in determining behavioral effects.
The ANOVA model identified a significant interaction effect between area and tide
state. Painvise comparisons showed that changes in haul-out numbers were different in
only four areas across tide states. Although the ANOVA model identified areas that
experienced the greatest fluctuation in number of seals throughout the tide cycle, this
interaction effect is probably not important to understanding disturbance effects. Terhune
and Almon (1983) showed that when more than one distinct group of seals was present
on a site, not all groups would react to man-made disturbances in a similar way. It is
possible that the four areas that experienced sipficant changes in haul-out number were
more susceptible to disturbances. Increased susceptibility may have resulted from more
juvenile or "non-habituated" seals in these areas than on other ledges or spatial and visual
isolation from other haul-outs. Regardless, disturbances that occur in theses four areas,
particularly during falling tides, may affect larger numbers of seals than other areas.
Estimates for long duration disturbances during falling tides were similar to data for
rising tides and did not reveal area-specific trends that would indicate shifts to alternate
haul-out locations. However, counts were not performed continuously during scat
collections. Post-disturbance counts were performed only after scat collection was
completed. Future studies should include continuous counts through long duration
disturbances to better quantify potential shifts in haul-out locations.
In summary, tide state and disturbance duration influenced seal haul-out behavior.
Long duration disturbances resulted in few seals remaining hauled-out and little recovery
regardless of tide state. Unexpectedly, seals showed more recovery following short
duration disturbances during rising tides than during falling tides. No areas were
identified as secondary haul-out sites following disturbances. Future research should (1)
include changes in numbers throughout long duration disturbances, (2) account for
specific location and orientation of seals (adjacent to water versus mland) affected by
short, localized disturbances, (3) include several days of observations to explore post-
disturbance haul-out trends over longer time scales, and (4) evaluate trends relative to
repeat disturbances.
BIOGRAPHY OF THE AUTHOR
Steven C. Renner was born in Fort Meade, Maryland on September 22, 1973. He was
raised in Philadelphia, Pennsylvania and graduated from Cardma1 Dougherty High
School in 1991. He attended Allentown College of Saint Francis de Sales and graduated
in 1995 with a Bachelor of Science degree in Biology. During his last semester, Steve
interned as an environmental educator at Hawk Mountain Sanctuary in Kempton,
Pennsylvania. He then went on to attend medical school at the Medical College of
Pennsylvania in PhiladeIphia. After two years of coursework, he returned to the
education field as an interpreter at the Philadelphia Academy of Natural Sciences. While
employed there, Steve developed several new interpretive programs including one on
marine mammal ecology.
Steve married Tuesday Mxhele Lester in 1997 and they moved to Eddington, Maine
in 1998. Upon arriving, Steve spent his first of two summers teaching at the College of
the Atlantic Natural History Museum in Bar Harbor, Maine prior to beginning graduate
courses at The University of Maine in fall of 1998. He completed his coursework and
data collection in 2001, then relocated to Edwards Air Force Base, California to support
h s wife's military obligation. Steve spent the last four years completing his thesis and
raising two amazing daughters. HISfamily hopes to return to Maine in 2005.
After completing his degree, Steve will continue to be a stay-at-home Dad and hopes
to pursue a career as an independent wildlife and scenic photographer and part-time
educator and researcher. Steve is a candidate for the Master of Science degree in
Wildlife Ecology from The University of Maine in May, 2005.