Seasonal, meteorological, tidal and diurnal effects on ... · Introduction Monitoring pinniped...
Transcript of Seasonal, meteorological, tidal and diurnal effects on ... · Introduction Monitoring pinniped...
ORIGINAL PAPER
Seasonal, meteorological, tidal and diurnal effects on haul-outpatterns of harbour seals (Phoca vitulina) in Iceland
Sandra Magdalena Granquist1,2 • Erlingur Hauksson3
Received: 6 March 2015 / Revised: 5 February 2016 / Accepted: 7 February 2016
� Springer-Verlag Berlin Heidelberg 2016
Abstract It is of critical importance to identify factors
that affect harbour seal haul-out patterns to improve the
accuracy of harbour seal censuses. In this study, haul-out
patterns of harbour seals (Phoca vitulina) were investigated
during different conditions at several major haul-out sites
on Vatnsnes peninsula, NW Iceland (65�400N and
20�480W), over the 2008–2011 period. A seasonal haul-out
pattern was detected among the seals, with the maximum
number of seals on land found in July for most of the sites.
Analyses of data for harbour seals on Vatnsnes indicate
that the main pupping period occurs during late May to the
beginning of June and moulting during late July to early
August. Abundance at the sites increased with rising air
temperature and decreased with increased windspeed and
rising tides. However, no evidence that precipitation or
cloud cover affected haul-out behaviour of the seals was
detected. The diel haul-out pattern was investigated thor-
oughly in one of the haul-out sites and the results under-
lined the effect of tidal flucturation, air temperature and
wind direction on the haul-out behaviour of harbour seals
in the area. Results from this study can be used to improve
the survey design when estimating the population size of
harbour seals in Iceland and applied more broadly to the
study of haul-out behaviour of harbour seals.
Keywords Harbour seal � Haul-out � Diurnal pattern �GLM � Iceland
Introduction
Monitoring pinniped populations is of great importance for
their management and conservation. Aerial counting sur-
veys of harbour seals (Phoca vitulina) provide information
about the total number of seals on land, and the number of
counted individuals may serve as a population index for
between-year comparisons (Gilbert et al. 2005; Lonegran
et al. 2013). Stakeholders such as government agencies,
however, often demand an estimate of the whole popula-
tion. Since pinnipeds only spend part of their time on land,
the total number of seals in the population has to be esti-
mated by models incorporating the number of counted
individuals together with the probability of seals spending
time on land (Simpkins et al. 2003; ICES 2007; Lowry
et al. 2008; Bjørge et al. 2010; Hauksson and Einarsson
2010; Harvey and Goley 2011). To improve the survey
design and increase the accuracy of harbour seal censuses
made from aerial surveys, it is of critical importance to
identify those conditions under which the seals are most
likely to haul out of the water (i.e., to rest on land).
One reason for hauling out may be to reduce the pre-
dation risk from marine predators such as killer whales
(Orcinus orca) (Watts 1992; London et al. 2012). Since
harbour seals usually haul out in groups, the individual
predation risk is decreased further (Hamilton 1971). Har-
bour seals give birth and nurse their pups on land, and it
has been frequently described that harbour seals haul out to
a larger extent during the pupping and nursing periods
(Boulva and McLaren 1979; Hauksson 1993). Another
reason for hauling out may be that seals save energy by
& Sandra Magdalena Granquist
1 The Icelandic Seal Center, Brekkugata 2, 530 Hvammstangi,
Iceland
2 Institute of Freshwater Fisheries, Keldnaholt, 112 Reykjavık,
Iceland
3 Vor Marine Research Center at Breiðafjorður, Norðurtangi 3,
355 Olafsvık, Iceland
123
Polar Biol
DOI 10.1007/s00300-016-1904-3
spending time on land during periods of warm and calm
weather compared with spending time in cold waters (Pauli
and Terhune 1987a; Watts 1992). Harbour seals also haul
out more during the moulting period when they shed their
fur since being on land elevates the skin temperature,
which can induce a faster moult (Feltz and Fay 1966; Ling
1970; Reder et al. 2003; Cronin et al. 2009; Paterson et al.
2012). Because of the largest fraction of harbour seals
spending time on land during the pupping and/or moulting
periods, these periods are usually chosen for aerial count-
ing surveys (Mogren et al. 2010).
A number of studies have explored the effects of various
environmental factors on pinniped haul-out behaviour
(Gaspari 1994; Bondo-Harders 2003). Several meterologi-
cal factors that affect the number of harbour seals hauling
out have been identified, for example, windspeed and wind
direction (Brasseur et al. 1996; Bondo-Harders 2003;
Simpkins et al. 2003), temperature (Pauli and Terhune
1987a; Brasseur et al. 1996; Reder et al. 2003; Simpkins
et al. 2003), cloud cover (Pauli and Terhune 1987a; Grel-
lier et al. 1996) and precipitation (Pauli and Terhune
1987a; Grellier et al. 1996; Simpkins et al. 2003). Fur-
thermore, authors have also described diurnal haul-out
patterns, with more seals hauling out during low tide. This
may be explained by haul-out sites becoming more
exposed and easier to access for the seals during low tide.
Other studies have suggested that it is more effective for
seals to forage during high tide when possible littoral
organism prey species visit the watercovered littoral zones
(e.g. Pauli and Terhune 1987b; Reder et al. 2003; Renner
2005). Several authors have also shown that the time of day
can affect the harbour seal haul-out pattern, with a haul-out
peak occurring at midday, which might partly be explained
by the air temperature usually peaking at midday. Age and
gender may also explain some of the variability in haul-out
behaviour (Thompson and Rothery 1987; Thompson et al.
1989; Kovacs et al. 1990; Harkonen et al. 1999).
Relationships between conditions affecting the propor-
tion of hauled-out seals in a population are complex and
hence tend to be quite variable not only between years, but
also between locations because of site-specific differences
(e.g. Terhune and Almon 1983; Hauksson 1985; Gaspari
1994; Grellier et al. 1996; Watts 1996; Reder et al. 2003;
Patterson and Acevedo-Gutierrez 2008; Mogren et al.
2010). Such local variation in harbour seal haul-out pat-
terns indicates that population- or site-specific analyses of
conditions affecting haul-out patterns may be required to
obtain the highest possible accuracy in population esti-
mates (Mogren et al. 2010).
In Iceland, 10 harbour seal censuses were carried out
between 1980 and 2011 (Hauksson and Einarsson 2010;
Granquist et al. 2011). The latest survey conducted in 2011
resulted in an estimated population size of 11,000
individuals (95 % confidence interval 8000–16,000 indi-
viduals) (Granquist et al. 2011; Marine Research Institute
of Iceland 2013). Day of the year, windspeed, tidal height
and hours to solar noon have been suggested as factors that
may affect the haul-out probability of harbour seals in
Icelandic conditions (Hauksson 1993, 2010). Based on
these indications all Icelandic abundance surveys have
been conducted mainly during August, with counts made
within ±3 h of low tide. However, previous studies of the
factors affecting the haul-out probability in Iceland were
restriced to the summer time and based on a rather limited
number of observations. In addition, previous Icelandic
studies were based on aerial surveys, meaning that data
were only recorded during days when the weather was
good enough to manoeuvre a small airplane. There is
therefore a lack of information regarding the seasonal
variation in harbour seal haul-out behaviour for Icelandic
environmental conditions and further there is very little
understanding of how meterological factors may affect
haul-out patterns of harbour seals living under Icelandic
conditions.
The aim of the study was to test factors affecting the
haul-out behaviour of Icelandic harbour seals. Through
observation of harbour seals on seven haul-out sites on
Vatnsnes peninsula, NW Iceland, we invstigated the pos-
sible effects of season, height of tide, time of day and
meteorologic factors (air temperature, cloud cover, precip-
itation, wind speed and wind direction) on haul-out patterns.
In addition, we tested the diurnal effect more thoroughly at
one of the haul-out sites (Illugasstadir). An increased
knowledge regarding what factors need to be taken into
consideration when estimating population sizes from counts
can improve aerial survey design and population estimation
models. Results from the present study could hence be
applied for harbour seal management and conservation
purposes in both Icelandic conditions and elsewhere.
Methods
Study area
Vatnsnes peninsula, NW Iceland (65�400N, 20�480W)
(Fig. 1), is one of the areas in Iceland with the highest
density of harbour seals (Hauksson 2010). During the
2008–2011 period, harbour seal abundance was monitored
in seven breeding and haul-out sites on Vatnsnes pennin-
sula: Svalbard, Naggur, Illugastadir, Hindisvik, Krossanes,
Osar and Bjargaos (Fig. 1; Table 1). Seal-watching sites
have recently been established at some of the sites (Gran-
quist and Nilsson 2013; Granquist and Sigurjonsdottir
2014), while Hindisvik is a former seal-watching site
where seal watching has been prohibited since 2008. The
Polar Biol
123
Bjargaos estuary is an inlet for some of the richest salmon
rivers in the northwestern part of Iceland. In the estuary,
harbour seal pups have been hunted in nets or shot since
several years before the study to protect salmon fishing
interests, although not between 2008 and 2010. In 2011,
however, several seals were shot in the estuary (Fig. 1;
Table 1).
Monitoring of seals
Seals on land were counted by a land-based observer
equipped with 7 9 50 binoculars and situated approximately
100 m from the seals with the exception of Bjargaos,
where the seals were counted from a distance of 1.5 km
using a field-scope Leica Televidd 77 20—60 9 77 zoom.
At Vatnsnes tides are diurnal with an interval of 12 h and
25 min between low tides. To investigate effects of sea-
son and meterological variables on haul-out behaviour,
harbour seals were monitored at the seven haul-out sites
during low tide (±2.5 h) of every spring tide between
January and November throughout the years 2008–2011
(Table 1). Due to bad weather conditions and visibility, it
was not possible to conduct observations during the
December spring tide in any of the years. To investigate
GREENLAND
ICELAND
VATNSNES
NORWAY
Hindisvik
Illugastadir
BjargaosSvalbard
Osar
Naggur
Krossanes
Arctic circle
Fig. 1 Map of Iceland indicating the location of the Vatnsnes peninsula and the Vatnsnes peninsula with the seven sites included in the study
marked
Polar Biol
123
the effects of time of day and tide height on the haul-out
pattern, more frequent observations were made between
June and August (three times per week at various times
and tidal states for comparison). All observations were
carried out during the day and none were recorded at
night. The data were somewhat unbalanced, with obser-
vations lacking for some sites in 2008, 2009 and 2011. At
Krossanes observations were only made in 2010 and
2011. Furthermore, no observations were made at Osar
and Bjargaos estuaries in 2008, and the data from Bjar-
gaos collected in 2011 were excluded from the analyses
because of seal hunting in the area (Table 1).
Possible diurnal patterns in harbour seal haul-out
behaviour were further investigated in one of the sites
(Illugastadir). From June to August in the period
2008–2011 2-h observation sessions were performed
twice every observation day and the seals were counted
with 15-min intervals during the observation sessions.
The observation sessions were initiated at different times
and distributed between 08:00 and 21:00 over 113 days.
The observations were made from a fixed point between
two cliffs, where the observer could not be easily detected
by the seals. To minimise observer bias, the recordings
were standardised by regularly performed inter-observer
reliability tests (agreement between observers [95 %).
All observations made at Illugastadir were used to assess
the diurnal effects; however, when analyses of the sea-
sonality and effect of meteorological and tidal variables
were carried out (see above), only two randomly chosen
counts from each day were included for Illugastadir in
order to match the data collection to the six other haul-out
sites.
Tide height and time to low tide were calculated using
the method presented in the Tide Almanac published by the
Icelandic Coast Guard (Icelandic Coast Guard 2009, 2010,
2011). Cloud cover (% of sky covered with clouds)
and precipitation (raining = 1; not raining = 0) were
recorded during every observation. Information about air
temperature and wind speed were obtained from the Ice-
landic Meteorological Institute.
Statistical analyses
The relationship between the total number of harbour seals
at the different sites and possible explanatory variables was
analysed with generalised linear models (GLMs), or gen-
eralised additive modelling (GAM), generalised linear
mixed modelling (GLMM) and generalised additive mixed
modelling (GAMM) in cases where correlations between
the observations were found (Zuur et al. 2009). All possible
prediction variables, as well as interactions between vari-
ables, were included in the initial models (two models;
seasonal and environmental). We examined whether the
following variables affected the haul-out pattern of the
seals: year, season (measured as day of the year), time of
day, height of tide and meterological variables (precipita-
tion, cloud cover, temperature, wind speed and wind
direction). The importance of the association of each
variable with the number of hauled-out seals was deter-
mined by individually deleting one variable at a time and
examining changes in the Akaike information criteria
(AIC), deviance and maximum likelihood of the models.
Only the variables that were found to have a significant
(p\ 0.05) relationship with the number of hauled-out seals
were kept in the final models. In cases of correlations
between different possible predictor variables [estimated
with Pearson’s product moment correlation (rp) or Spear-
man’s rank correlation rho (rs)], only one of the variables
was included in the final model. Due to warmer tempera-
tures during the summer, there was a correlation between
the day of the year and temperature (rp = 0.32, t = 18.61,
df = 2990, p\ 0.001); hence, the temperature data were
excluded from the model where seasonality (day of the
year) was investigated. Further, time to midday was highly
correlated with height of tide (rp = -0.13, t = -7.04,
df = 2991, p\ 0.001); time to midday was correlated with
Table 1 Haul-out sites, habitat
descriptions, seal watching
occurrences and total numbers
of observations (n) for the
different years
Haul-out site Habitat description Seal watching n2008 n2009 n2010 n2011
Svalbard Rocky skerries Yes 54 101 57 39
Naggur A small skerry No 34 100 90 43
Illugastadir Rocky skerries Yesa 450 834 641 38
Hindisvik Rocky shore and skerries Nob 29 49 51 39
Krossanes Rocky shore and skerries No – – 40 35
Osar Lagoon estuary, sandy shore Yes – 44 49 41
Bjargaos River estuary, sandy shore Noc – 45 46 –
a Except May 20–June 20b Seal watching was permanently prohibited in 2008c Disturbances due to salmon fishing and seal hunting
Polar Biol
123
time to low water (rp = 0.05, t = 3.23, df = 2991,
p = 0.001) and tide height and time to low water were
correlated (rp = 0.05, t = 2.50, df = 2991, p = 0.01)
since observations were only carried out during the day.
The effect of time of day on haul-out behaviour was
therefore excluded from the models investigating season
and meterological effects on haul-out behaviour.
The negative binomial distribution was used as the basis
for the GLM analysis, which turned out to give a better fit
than the Poisson distribution. The negative binomial dis-
tribution’s GLM routine estimates the dispersion parameter
theta (/) and standard errors of the parameters. In GAM,
GAMM and GLMM the value of the calculated / from the
negative binomial GLM was used in fitting of the data. The
fit of models was investigated with residual deviance (RD),
null deviance (ND), deviance explained [(ND - RD)/ND]
(AIC) and the 2 9 log likelihood (Zuur et al. 2009).
Neither the seasonal nor diurnal (Illugastadir) data
showed a significant trend in abundances of harbour seals
in relation to years in the 2008–2011 period. Therefore,
data were pooled for all years in the analysis. Since the size
of the different haul-out sites varied substantially with
regard to the average number of seals observed, the site
was included as a variable in all models: a fixed or random
variable depending on the model type.
At Illugastadir, where several daily data points were
available, diurnal effects on haul-out behaviour were fur-
ther investigated by plotting haul-out data in relation to
hours from low tide (TLW) using a locally weighted
regression procedure for fitting a regression curve by
smoothing the dependent variable (TLW) as a function of
the independent variable (harbour seal counts)—loess
smoothing. All statistical analyses were performed in R (R
Development Core Team 2009) and using the MASS
library (Venables and Ripley 2002).
Results
Effect of seasonal factors
The mean number of seals hauling out in each month on
the different sites is described in Table 2. At all sites,
seasonality in haul-out patterns was detected, with the
maximum average for most sites observed in July. At
Illugastadir, the average number of seals hauling out was
similar for both July and August, while in the estuary
(Bjargaos) the maximum number of seals hauled out in
June (Table 2).
Results of the GAMM seasonal model using the value
of the theta from the negative binomial GLM on the sea-
sonal data, with sites and day of the year as a smoothing
term, explained 66.3 % of the variability in the data Table
2M
ean
nu
mb
er(S
D)
of
har
bo
ur
seal
s(Phoca
vitulina
)fo
rea
chm
on
th,
po
ole
dfo
rth
ey
ears
20
09
–2
01
1fo
rS
val
bar
d(n
=2
51
),N
agg
ur
(n=
26
7),
Illu
gas
tad
ir(n
=4
05
),H
ind
isv
ik
(n=
16
8),
Kro
ssan
es(n
=7
5),
Osa
r(n
=1
34
)an
dB
jarg
aos
(n=
91
)
Jan
Feb
Mar
Ap
rM
ayJu
nJu
lA
ug
Sep
Oct
No
v
Sv
alb
ard
2.0
(2.8
)6
.0(–
)4
.0(0
.0)
4.8
(7.0
)7
.1(7
.1)
15
.5(9
.6)
19
.2(1
0.8
)7
.8(6
.0)
4.6
(7.8
)1
0.0
(14
.1)
1.3
(2.3
)
Nag
gu
r0
.0(0
.0)
0.0
(–)
0.0
(0.0
)0
.0(0
.0)
5.6
(4.5
)3
.9(4
.0)
5.4
(5.6
)1
.7(3
.5)
0.0
(0.0
)0
.0(–
)0
.0(–
)
Illu
gas
tad
ir1
.0(1
.4)
28
.0(–
)2
1.0
(18
.6)
35
.8(3
3.5
)2
9.1
(20
.0)
44
.0(1
8.9
)7
0.8
(25
.8)
71
.2(2
6.9
)6
1.6
(25
.9)
67
.5(1
6.3
)1
8.0
(21
.6)
Hin
dis
vik
11
.5(9
.2)
32
.0(–
)1
8.0
(18
.4)
53
.4(5
4.7
)7
5.1
(49
.2)
77
.5(3
6.1
)1
12
.4(4
8.6
)6
5.8
(37
.9)
69
.7(4
6.6
)3
6.5
(41
.7)
5.3
(5.5
)
Kro
ssan
es0
.0(–
)–
2.0
(–)
–2
0.0
(6.9
)1
8.1
(9.9
)2
5.9
(13
.6)
12
.0(9
.0)
0.0
(–)
––
Osa
r5
1.0
(–)
75
.0(–
)1
2.0
(16
.7)
21
2.8
(15
4.4
)1
08
.6(8
6.6
)1
11
.4(7
0.9
)2
51
.3(7
9.6
)1
59
.9(7
1.9
)1
32
.7(6
1.0
)1
57
.5(1
73
.2)
58
.3(3
1.3
)
Bja
rgao
s1
.0(–
)3
.0(–
)0
.0(0
.0)
0.0
(0.0
)3
4.0
(36
.7)
14
.8(1
5.6
)6
.3(1
1.5
)0
.0(0
.2)
0.0
(–)
0.0
(–)
0.0
(0.0
)
Cal
cula
tio
ns
for
Kro
ssan
esar
ela
ckin
gfo
r4
mo
nth
sb
ecau
seo
fsc
arce
ob
serv
atio
ns
Polar Biol
123
(Rad2 = 0.663, scale est. = 0.964 and n = 2228). The dis-
persion parameter for the negative binomial distribution (/)
was 1.76. All sites were significantly different from the
reference site Bjargaos (p\ 0.001) and all sites except
Naggur generally had a greater abundance of harbour seals
than the reference site Bjargaos (Table 2). The smoothing
term, day of the year, had an approximate significance of
p\ 0.001. Across all sites harbour seals were less likely to
haul out during winter than spring, summer and autumn.
Figure 2 shows that in general for the whole Vatnsnes area,
there was an indication of a bimodal haul-out pattern
during the summer months, one peak at the end of May/
beginning of June and a second in the end of July (Fig. 2).
When the sites were investigated individually, the curve of
hauling-out seals to the day of the year was bimodal, with
the first peak usually occurring at the end of May/begin-
ning of June and the second in late July/early August.
Bjargos was an exception, however, with only a single
peak observed in June (Fig. 3).
Effect of environmental factors
To test the dependency of hauling out on temperature and
other environmental variables, only data collected during
the summer months (June–September), when the temper-
ature was relatively constant, were used. The results of the
GLMM with / equal to 1.7 from the analysis of negative
binomial distribution using GLM indicated that harbour
seals were overall more likely to haul out when the air
temperature was warmer, windspeed was lower and heights
of tides were lower (Table 3). There was, however, varia-
tion between sites and all sites showed significantly dif-
ferent abundances of harbour seals compared to the
reference site of Bjargaos (p\ 0.001). The harbour seal
abundance was greater at most sites compared to the
reference site, the exceptions being Naggur and Svalbard
where the abundance was lower. The interaction between
sites and air temperature showed the highest variance
(Table 3). The GLMM model with sites as fixed effects and
site, air temperature, windspeed and tide height as random
effects gave an AIC of 18,281.8, a maximum log likelihood
of -9128.9 and a deviance of 18,257.8. A GLMM model
with the same variables, but with the sites as a random
effect, did not converge (Figs. 4, 5, 6). None of the other
meterological variables included in the initial model (cloud
cover, precipitation or wind direction) had an effect on the
model (Table 3).
There was a correlation between the abundance of har-
bour seals in many sites and the highest correlation was
found between Illugastaðir and Svalbard (Table 4). In the
GAMM used for estimating the effect of the variables of air
temperature, windspeed and tide height on the number of
harbour seals hauling out, sites were the fixed variables and
the former variables were used as smoothing terms. Sites
were all significantly different from the reference site
Bjargaos (p\ 0.001); the smoothing terms were significant
too: air temperature (p\ 0.001), windspeed (p = 0.04)
and tide height (p\ 0.001). The GAMM explained 62.5 %
of the variability of the data (Rad2 = 0.625, scale esti-
mate = 0.849 and n = 2055).
Diurnal effect and environmental variables
At Illugastadir, where several daily observations were
made, a clear diurnal pattern was found with the average
highest number of seals hauling out around low tide: i.e.
from 2 h before to 3 h after maximum low tide (Fig. 6).
Further, more seals hauled out with increasing air tem-
perature, at wind directions of 50� and 180�, and in the
summer (Fig. 7). The GAM results, indicating this,
Fig. 2 Estimated smoother for
the day of the year obtained by
the GAMM applied on the
number of harbour seals (Phoca
vitulina) hauling out; all data
combined for the Vatnsnes area
(indicated with an unbroken
line), 95 % pointwise
confidence bands (indicated
with broken lines) and residuals
from the GAMM (indicated
with dots)
Polar Biol
123
explained 38.5 % of the deviance, Rad2 = 0.5, unbiased risk
estimator (UBRE) = -0.18712, scale estimate = 1 and
n = 1921. All smooth terms were highly significant
(p\ 0.001) and the intercept was significantly different
from zero (p\ 0.001). The / for the negative binomial
distribution used in the GAM was 2.7.
Discussion
The results presented in this study are derived from a large
data set, spanning 4 full years, and are geographically
limited to Vatnsnes, NW Iceland. It is therefore extremely
valuable for elucidating the effect of both time of year and
environmental variables on the haul-out behaviour of har-
bour seals in the area.
The haul-out pattern of harbour seals in Vantsnes
depends on the time of year and we found a bimodal
temporal distribution during the summer in all haul-out
sites except for one. In addition, we observed a clear
positive effect of air temperature and negative effect of
Fig. 3 Ln (abundance) of harbour seals (Phoca vitulina) hauling out each decimal day of the year on the seven sites; data from all years
combined. Unbroken lines indicate the trend estimated with a loess smoother
Table 3 Results of the generalised linear mixed model (GLMM) fit
for abundance of harbour seals (Phoca vitulina) in relation to sites in
the Vatnsnes area and environmental variables by maximum likeli-
hood, with negative binomial / = 1.7, tide height (TH), windspeed
(WS) and air temperature (AT)
Random effects Variance SD Z value (p)
TH:WS:AT:site 4.379 2.093 –
WS:AT:site 0.0 0.0 –
AT:site 4.453 2.110 –
Site 0.0 0.0 –
Fixed effects Estimate SE Z value (p)
Intercept 1.8298 0.1070 17.101 (p\ 0.001)
Bjargaos Reference site – –
Hindisvık 2.6151 0.1264 20.685 (p\ 0.001)
Illugastadir 2.0908 0.1084 19.284 (p\ 0.001)
Krossanes 1.0713 0.1459 7.343 (p\ 0.001)
Naggur -0.5543 0.1191 -4.655 (p\ 0.001)
Osar 3.3092 0.1315 25.171 (p\ 0.001)
Svalbard 0.8307 0.1195 6.954 (p\ 0.001)
Polar Biol
123
windspeed on the number of harbour seals hauling out
during summer. We also found evidence that lower tide
heights resulted in more seals hauling out. Summer data
from Illugastadir further confirmed that seal haul-out pat-
terns follow changes in tide height over the course of each
day. The direction of the wind also influenced the number
of harbour seals hauling out and seals were more likely to
haul out when the wind direction was northeasterly (50�) or
southerly (180�). The northeasterly wind at Illugastadir is
directed landwards, and due to the topography at Illugas-
tadir the seals are sheltered when the wind blows in this
specific direction. Southerly winds blow along the coast
and the seals are also quite sheltered when the wind blows
in this direction.
At all sites some evidence of seasonality was observed
(Table 2; Figs. 2, 3). The overall minimum counts of
hauling-out harbour seals was observed in the winter
months, while the highest counts for most sites were
observed in July. This is to some extent in agreement with
earlier studies (Hauksson 1993, 2010). Although the stud-
ies of Hauksson (1993, 2010) were built on fewer obser-
vations, he found that at Vatnsnes (Hauksson 1993) and in
other areas of Iceland (Hauksson 2010) the maximum
number of harbour seals hauled out in July and August.
Similar patterns have also been found among harbour seals
in other areas of the world (e.g. Sullivan 1980; Renner
2005; Cronin et al. 2009; London et al. 2012). Previous
studies have suggested that harbour seals are more likely to
spend time on land during the pupping season while
nursing their young and again later during the moulting
period (Boulva and McLaren 1979; Hauksson 1993; Cronin
et al. 2009). Research on harbour seal females in Scotland
using radio-tracking has also shown that lactating females
spend more time feeding during the latter part of the lac-
tation period and less time hauling out (Thompson et al.
1994). The bimodal nature of the smoothing curve of
harbour seal abundance against the day of the year indi-
cates therefore that pupping of harbour seals on Vatnsnes
peninsula occurs in late May to early June, while the
moulting period occurs in late July to early August
(Figs. 2, 3) (Granquist and Hauksson unpublished).
The Bjargaos river estuary, leading to several important
salmon rivers (Gudbergsson 2012), was the only site in the
present study where a bimodal seasonal pattern was not
found. This may be due to the limited the number of
observations, or it may be that harbour seals use this area as
a feeding site rather than a resting site and hence spent
more time in this area feeding between the periods of
pupping and moulting. The estuary was moreover the only
area where the average maximum number of seals hauling
out peaked in June and a single peak found in late June for
this area supports this hypothesis, since it coincides with
salmon abundance in the estuary (Fig. 2). Prey availability
was however not further investigated in this study, pri-
marily because of the lack of data. A positive correlation
between the numbers of seals resting on land with air
temperature has also been described in some previous
studies (e.g. Watts 1996; Mogren et al. 2010, but see
Grellier et al. 1996), which suggests that seals save energy
by resting on land during warm periods (Pauli and Terhune
1987a; Watts 1992).
Neither precipitation nor cloud cover affected the har-
bour seal haul-out pattern in our models. Windspeed was,
however, found to affect the seals on Vatnsnes and at
Illugastadir wind direction affected haul-out patterns over
the summer months. The weather stations logging wind
speed and direction data were located between 20 and
40 km from the different haul-out sites investigated in the
study; therefore local variation in wind speed might have
affected the results. Further, effects of wind direction
Fig. 4 Estimated smoother for
air temperature obtained by the
GAMM applied on the number
of harbour seals (Phoca
vitulina) hauling out in the
Vatnsnes area (indicated with
an unbroken line), 95 %
pointwise confidence bands
(indicated with broken lines)
and residuals (indicated by dots)
Polar Biol
123
depend on site-specific topography and might hence have
been important at other sites than Illugastadir, although not
detected for methodological reasons.
All of the environmental factors tested in this study have
frequently been described to affect haul-out patterns in
other areas. Mogren et al. (2010), studying sites in Ves-
teralen, Norway, found that increased cloud cover resulted
in fewer seals hauled out in July (Mogren et al. 2010). A
relationship between cloud coverage and haul-out fre-
quencies has also been found in other studies (e.g. Pauli
and Terhune 1987a; Grellier et al. 1996). Wind speed (e.g.
Harders 2003; Simpkins et al. 2003) as well as wave action
(Pauli and Terhune 1987a; Thompson 1989) has been
suggested to affect the haul-out frequency of harbour seals
by several authors and Simpkins et al. (2003) found that
maximum haul-out occurred at a wind speed of 10 miles/h.
Fig. 5 Estimated smoother for
combined effects of air
temperature and windspeed
obtained by the GAMM applied
on the number of harbour seals
(Phoca vitulina) hauling out in
the Vatnsnes area (indicated by
unbroken lines), 95 % pointwise
confidence bands (indicated by
broken lines) and residuals
(indicated by dots)
Fig. 6 Estimated smoother for
tide height obtained by the
GAMM applied to the number
of harbour seals (Phoca
vitulina) hauling out in the
Vatnsnes area (unbroken line),
95 % pointwise confidence
bands are indicated by broken
lines, and residuals of the
GAMM are indicated by dots
Table 4 Correlation of fixed
effects (sites) in the GLMM fitSites Intercept Hindisvık Illugastaðir Krossanes Naggur Osar
Hindisvık -0.828
Illugastadir -0.954 0.791
Krossanes -0.711 0.590 0.680
Naggur -0.828 0.689 0.796 0.593
Osar -0.792 0.657 0.757 0.564 0.660
Svalbard -0.865 0.718 0.828 0.617 0.723 0.687
Polar Biol
123
However, Brasseur et al. (1996) observed that air temper-
ature had a larger effect than wind speed on the haul-out
behaviour of harbour seals in captivity and Grellier et al.
(1996) did not find any consistent effect of wind speed or
wind chill-adjusted temperatures on the number of harbour
seals hauling out in Morey Firth, Scotland. Evidence that
harbour seals haul out less during percipitation has also
been reported (Pauli and Terhune 1987a; Grellier et al.
1996; Simpkins et al. 2003).
The effect of time of day could not be assessed sepa-
rately in this study because it was highly correlated with
tide height. A negative correlation between increasing tide
height and harbour seal abundance was found for all sites
in the study (Fig. 6). At Illugastadir, where several daily
observations were made during the summer months to
investigate diurnal effects more thoroughly, the highest
number of seals was on average observed on land around
low tide, from 2 h before to 3 h after maximum low tide.
These findings therefore further underline a diurnal haul-
out pattern (Figs. 7, 8). Tidal state was previously sug-
gested to be an important factor influencing haul-out pat-
terns of harbour seals on the coast of Iceland (Hauksson
1985). Hauksson (2010) observed that tide height, time of
day and wind force significantly affected the haul-out
behaviour of Icelandic harbour seals in the 1980–2006
period. Harbour seals following a diel tidal haul-out cycle
is in agreement with previous studies and often numbers on
shore were found to peak near midday (e.g. Yochem et al.
1987; Reder et al. 2003; Simpkins et al. 2003; Renner
2005; Cronin et al. 2009; London et al. 2012). Further,
Cronin et al. (2009) found individual variation in the
influence of time of day on haul-out behaviour.
The variation in haul-out behaviour found between sites
in the present study was greater than anticipated, taking
into consideration their geographical proximity. However,
local and between-year variation in haul-out behaviour has
been reported by several authors (e.g. Terhune and Almon
1983; Hauksson 1985; Watts 1996; Reder et al. 2003;
Patterson and Acevedo-Gutierrez 2008; Mogren et al.
2010) and suggested reasons for the variation are demo-
graphic changes within the population (Thompson 1989;
Harkonen et al. 1999; Reder et al. 2003), differences in
topography and habitat types (Hauksson 2010), ice cover-
age (Calambokidis et al. 1987; Lesage et al. 2004), fluc-
turations in prey availability (Thompson 1988; London
et al. 2012) or anthropogenic disturbance (Henry and
Hammill 2001; Granquist and Sigurjonsdottir 2014;
Andersen et al. 2014). Furthermore, although harbour seals
Fig. 7 Total number of harbour seals (Phoca vitulina) in relation to hours from low tide, fitted with a linear loess smoother. The 12 observation
days with the highest frequency of observations are shown as examples in the figure
Polar Biol
123
often have been described to show high site fidelity to their
haul-out sites (e.g. Yochem et al. 1987; Dietz et al. 2013;
Andersen et al. 2014), these factors may also cause annual
variation in preferred sites. In some cases harbour seals
also move between feeding and breeding areas (Thompson
1989), which are nevertheless often in similar locations
(Suryan and Harvey 1998).
Testing individual variation due to gender or age was not
possible in this study because of the large distance from
which the animals were observed. Such factors may never-
theless affect the haul-out behaviour of harbour seals (e.g.
Thompson and Rothery 1987; Kovacs et al. 1990; Harkonen
et al. 1999; Reder et al. 2003); hence, extrapolating infor-
mation regarding haul-out behaviour from a few haul-out
areas to a whole population is unsuitable (Mogren et al.
2010). As an example, Harkonen et al. (1999) showed in
their study that no age/sex group of harbour seals was rep-
resentative for the whole population in terms of haul-out
behaviour.
It has been suggested that differences in habitat types
can be a reason for between-site variation in the haul-out
pattern (Thompson 1989; Hauksson 2010) and this may
partly explain the between-site variation that we found in
this study despite the closeness of the haul-out sites. In
haul-out sites covered by water at high tide, it is impossible
for the seals to haul out as the water rises, while at other
locations seals might be able to move higher up when the
water starts to rise and hence haul out longer, and in the
latter case the state of tide height will have less effect
(Thompson 1989). All of the sites included in the present
study are partly covered during high tide, although Hin-
disvık and Krossanes are the sites least affected by tides in
this study. Some areas are more exposed to wind or waves
than others, causing weather-dependent variation in haul-
out feasibility between different sites (Pauli and Terhune
1987a; Thompson 1989), despite geographical closeness,
although this was not tested in the present study.
In cases where predator abundance is high, harbour seals
may prolong their haul-out bouts to avoid predation (London
et al. 2012), but on Vatnsnes the predation risk for harbour
seals is considered to be low (Granquist and Sigurjonsdottir
2014). However, in several studies harbour seals have been
described to change their haul-out pattern because of
anthropogenic disturbance (Watts 1996; Henry and Hammill
2001; Acevedo-Guiterrez and Cendejas-Zarelli 2011;
Granquist and Sigurjonsdottir 2014). Acevedo-Guiterrez and
Cendejas-Zarelli (2011) found that in areas where artificial
noise levels are high, seals chose to haul out during the night
when the disturbance was smaller. Tourism may also affect
haul-out behaviour, both when it comes to choosing the site
and the time to haul out. London et al. (2012) showed that
harbour seals changed their haul-out pattern because of
human disturbance and preferred to haul out during the night
in August and September because of high human distur-
bance, while this pattern was reversed in October and
November. Other studies have reported changes in preferred
haul-out areas due to anthropogenic disturbance associated
with tourism, with seals moving to more remote skerries
(Henry and Hammill 2001; Granquist and Sigurjonsdottir
2014). Although disturbance due to humans was not
Fig. 8 The Illugastadir site, west coast of Vatnsnes, W-Hun., NW
Iceland. Estimated smoothers for environmental variables: air tem-
perature (left top), wind direction (right top), hours to low water (left
bottom) and day of the year (right bottom) obtained by the GAM
applied on the abundance of harbour seals (Phoca vitulina) indicated
by unbroken lines. Broken lines indicate 95 % pointwise confidence
bands and dots show residuals
Polar Biol
123
included in the study, Vatnsnes is a popular tourist area in
Iceland, including activities such as seal watching, and with
increasing numbers of tourists in the area, it is important that
human disturbance is taken into consideration in calcula-
tions. Granquist and Sigurjonsdottir (2014) found that har-
bour seals on Illugastadir, Vatnsnes, moved to skerries
further away from the mainland during periods of the
summer when the highest number of visits are made by
tourists to the area, although there was no evidence to
suggest that the seals moved away from the area. Further,
the majority of seals were hauling out during the peak of the
tourist season, indicating that the effects of tourists on the
haul-out site at Illugastadir were small (Granquist and Sig-
urjonsdottir 2014). At the other sites in the present study, the
effect of tourism on harbour seal haul-out behaviour has
not been investigated and it is therefore not possible to say
whether or not haul-out patterns are affected.
Conclusions
In this study, we found evidence for a seasonal haul-out
pattern among the seals, with the maximum number of
seals on land found in July for most of the sites. Bimodal
smoothing curves of abundance on a day-of-the-year basis
indicate that the main pupping period is in late May and
early June for harbour seals at Vatnsnes while the main
moulting period is in late July to early August. Abundance
at the sites increased with rising air temperature and
decreased with rising tides during summer. These findings
strengthen the already enforced method used in the aerial
census of harbour seals in Iceland to count at ±3 h of low
tide and to carry out the census during the moulting period.
However, earlier censuses have been carried out during
August and since the present study indicates that the
moulting period starts earlier, in the middle of July at most
sites, starting to conduct the harbour seal census in Iceland
in the middle of July should be considered. Practically,
counting early in the summer may be complicated since
many areas in Iceland are protected during June because of
eiderduck (Somateria mollissima) nesting and flying over
nesting sites is prohibited. During years with normal
weather conditions, eiderduck nesting is finished by the
middle of July. Notably, the variation in haul-out behaviour
found between the sites, despite their close proximity,
underlines the need for care when extrapolating results
obtained from a few sites to investigate a larger area.
Acknowledgments We would like to thank Helgi Gudjonsson,
Ester Cacho Sanchez, Laila Arranda Romero and Eva Haunss for
assistance in the field. Thanks to the landsowners at Vatnsnes, espe-
cially at Illugastadir and Hindisvık. Earlier versions of this manuscript
were improved by comments from Anders Angerbjorn, Ian Bytheway
and anonymous reviewers.
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict
of interest.
Ethical standards This article does not contain any studies with
human participants performed by any of the authors. All applicable
international, national and/or institutional guidelines for the care and
use of animals were followed.
References
Acevedo-Guiterrez A, Cendejas-Zarelli S (2011) Nocturnal haul-out
patterns of harbor seals (Phoca vitulina) related to airborne noise
levels in Bellingham, Washington, USA. Aquatic Mamm
37:167–174
Andersen SM, Teilmann J, Dietz R, Schmidt NM, Miller LA (2014)
Disturbance-induced responses of VHF and satellite tagged
harbour seals. Aquat Conserv 24:712–723
Bjørge A, Desportes G, Waring GT, Rossing-Asvid A (2010)
Introduction: the harbour seal (Phoca vitulina)—a global
perspective. NAMMCO Sci Publ 8:7–14
Boulva J, McLaren IA (1979) Biology of the harbour seal, Phoca
vitulina, in eastern Canada. Bull Fish Res Bd Can 200:1–24
Brasseur S, Creuwels J, vd Werf B, Reijnders R (1996) Deprivation
indicates necessity for haul-out in harbour seals. Mar Mamm Sci
12:619–624
Calambokidis J, Taylor BL, Carter SD, Steiger GH, Dawson PK,
Antrim LD (1987) Distribution and haul-out behavior of harbor
seals in Glacier Bay, Alaska. Can J Zool 65:1391–1396
Cronin MA, Zuur AF, Rogan E, McConnell BJ (2009) Using mobile
phone telemetry to investigate the haul-out behaviour of harbour
seals Phoca vitulina vitulina. Endanger Species Res 14:255–267
Dietz R, Teilmann J, Andersen SM, Riget F, Olsen MT (2013)
Movements and site fidelity of harbour seals (Phoca vitulina) in
Kattegat, Denmark, with implications for the epidemiology of
the phocine distemper virus. ICES J Mar Sci 70:186–195
Feltz ET, Fay FH (1966) Thermal requirements in vitro of epidermal
cells from seals. Cryobiology 3:261–264
Gaspari S (1994) Haul-out behaviour, site fidelity and vigilance of
common seals (Phoca vitulina) and grey seals (Halichoerus
grypus) in the tees Estuary. Master thesis, Durham University,
UK
Gilbert JR, Waring GT, Wynne KM, Guldager N (2005) Changes in
abundance of harbour seals in Maine, 1981–2001. Mar Mamm
Sci 21:519–535
Granquist S, Nilsson PA (2013) The wild north: network cooperation
for sustainable tourism in a fragile marine environment in the
Arctic region. In: Muller D, Lundmark L, Lemelin R (eds) New
issues in polar tourism. Springer, Netherlands, pp 123–132
Granquist SM, Sigurjonsdottir H (2014) The effect of land based seal
watching tourism on the haul-out behaviour of harbour seals
(Phocavitulina) in Iceland. Appl Anim Behav Sci 156:85–93
Granquist SM, Hauksson E, Arnadottir AB, Kasper J (2011)
Landselstalning ur lofti arið 2011. Framvinda og niðurstoður
[Aerial survey of the Icelandic harbour seal population in 2011].
Institute of Freshwater Fisheries, VMST/11051. Reykjavık,
Iceland
Grellier K, Thompson PM, Corpe HM (1996) The effect of weather
condition on harbour seal (Phoca vitulina) haulout behaviour in
the Morey Firth, northeast Scotland. Can J Zool 74:1806–1811
Gudbergsson G (2012) Lax- og silungsveiðin 2011 [Salmon- and trout
angling in 2011]. Institute for Freshwater Fisheries, VMST/
12032. [In Icelandic]
Polar Biol
123
Hamilton WD (1971) Geometry for the selfish herd. J Theor Biol
31:295–311
Harders PB (2003) Ophold pa land, forstyrrelser og fodevalg hos
spættet sæl (Phoca vitulina) og grasæl (Halichoerus grypus) pa
Rosand. Syddansk Universitet og Danmarks Miljoundersogelser,
Denmark, Biologiska Institutet, Afdelning for Arktisk Miljo
Harkonen T, Harding KC, Lunneryd SG (1999) Age- and sex-specific
behaviour in harbour seals (Phoca vitulina) leads to biased
estimates of vital population parameters. J Appl Ecol 36:
825–841
Harvey JT, Goley D (2011) Determining a correction factor for aerial
surveys of harbor seals in California. Mar Mamm Sci 27:
719–735
Hauksson E (1985) Fylgst með landselum ı latrum [Monitoring
harbour seals on haul-out sites]. Natturufræðingurinn 55:
119–131 [In Icelandic]
Hauksson E (1993) Seasonal pattern of hauling-out of common seals
(Phoca vitulina L.) on Vatnsnes, NW-Iceland. Natturufræðin-
gurinn 62:37–41
Hauksson E (2010) Monitoring trends in the abundance of harbour
seals (Phoca vitulina) in Icelandic waters. NAMMCO Sci Publ
8:227–244
Hauksson E, Einarsson ST (2010) Historical trend in harbour seal
(Phoca vitulina) abundance in Iceland back to the year 1912.
NAMMCO Sci Publ 8:147–159
Henry E, Hammill MO (2001) Impact of small boats on the
hauloutactivity of harbour seals (Phoca vitulina) in Metis Bay,
Saint LawrenceEstuary, Quebec, Canada. Aquat Mamm
27:140–148
Icelandic Coast Guard (2009) Tide Almanac. Hydrographic Depart-
ment, Reykjavik
Icelandic Coast Guard (2010) Tide Almanac. Hydrographic Depart-
ment, Reykjavik
Icelandic Coast Guard (2011) Tide Almanac. Hydrographic Depart-
ment, Reykjavik
ICES (2007) Report of the Working Group on Marine Mammal
Ecology (WGMME). ICES CM 2007/ACE:03
Kovacs KM, Jonas KM, Welke SE (1990) Sex and age segregation by
Phoca vitulina concolor at haul-out sites during the breeding
season in the Passamaquoddy Bay region, New Brunswick. Mar
Mamm Sci 6:204–214
Lesage V, Hammill MO, Kovacs KM (2004) Long-distance move-
ments of harbour seals (Phoca vitulina) from a seasonally ice-
covered area, the St. Lawrence River estuary, Canada. Can J
Zool 82:1070–1081
Ling JK (1970) Pelage and molting in wild mammals with special
reference to aquatic forms. Q Rev Biol 45:16–54
London JM, Ver Hoef JM, Jeffries SJ, Lance MM, Boveng PL (2012)
Haul-out behaviour of harbor seals (Phoca vitulina) in Hood
Canal, Washington. PLoS ONE 7:e38180. doi:10.1371/journal.
pone.0038180
Lonegran M, Duck C, Moss S, Morris C, Thompson D (2013)
Rescaling of aerial survey data with information from small
numbers of telemetry tags to estimate the size of a declining
harbour seal population. Aquat Conserv 23:135–144
Lowry MS, Caretta JW, Forney KA (2008) Pacific harbor seal censusin California during May–July 2002 and 2004. Calif Fish Game
94:180–193
Marine research Institute of Iceland (2013) State of stocks 2012/2013.
nr. 169. Reykjavık, Iceland
Mogren HG, Lindstrom U, Nilssen KT, Haug T (2010) Haulout
behaviour of harbour seals (Phoca vitulina) during breeding and
moult in Vesteralen, Norway. NAMMCO Sci Publ 8:302–312
Paterson W, Sparling CE, Thompson D, Pomeroy PP, Currie JJ,
McCafferty DJ (2012) Seals like it hot: changes in surface
temperature of harbour seals (Phoca vitulina) from late preg-
nancy to moult. J Therm Biol 37:454–461
Patterson J, Acevedo-Gutierrez A (2008) Tidal influence on the haul-
out behavior of harbor seals (Phoca vitulina) at a site available at
all tide levels. Northwestern Nat 89:17–23
Pauli BD, Terhune JM (1987a) Meteorological influences on harbour
seal haul-out. Aquatic Mamm 13:114–118
Pauli BD, Terhune JM (1987b) Tidal and temporal interaction on
harbour seal haul-out patterns. Aquatic Mamm 13:93–95
R Development Core Team (2009) R: a language and environment for
statistical computing. Vienna, Austria: R Foundation for Statis-
tical Computing. http://www.R-project.org. Accessed 15 May
2015
Reder S, Lydersen C, Arnold W, Kovacs KM (2003) Haulout
behaviour of high Arctic harbour seals (Phoca vitulina vitulina)
in Svalbard, Norway. Polar Biol 27:6–16
Renner SC (2005) An analysis of harbour seal (Phoca vitulina) and
grey seal (Halichoerus grypus) haul-out patterns, behavior
budgets, and agressive interactions on Mount Desert Rock,
Maine. Master thesis. The University of Maine, USA
Simpkins MA, Withrow DE, Cesarone JC, Boveng PL (2003)
Stability in the proportion of harbor seals hauled out under
locally ideal conditions. Mar Mamm Sci 19:791–805
Sullivan RM (1980) Seasonal occurrence and haul-out use in
pinnipeds along Humbolt county. Calif J Mamm 61:754–760
Suryan RM, Harvey JH (1998) Tracking harbor seals (Phoca vitulina
richardsi) to determine dive behavior, foraging activity, and
haul-out site use. Mar Mamm Sci 14:361–372
Terhune JM, Almon M (1983) Variability of harbour seal numbers on
haul-out sites. Aquatic Mamm 10:71–78
Thompson PM (1988) Timing of the mating in the common seal
(Phoca vitulina). Mamm Rev 18:105–112
Thompson PM (1989) Seasonal changes in the distribution and
composition of common seal (Phoca vitulina) haul-out groups.
J Zool Lond 217:223–294
Thompson P, Rothery P (1987) Age and sex differences in the timing
of moult in the common seal, Phoca vitulina. J Zool Lond
212:597–603
Thompson PM, Fedak MA, McConnell BJ, Nicholas KS (1989)
Seasonal and sex-related variation in the activity patterns of
common seals (Phoca vitulina). J Appl Ecol 26:521–535
Thompson PM, Miller D, Cooper R, Hammond PS (1994) Changes in
the distribution ad activity of female harbour seals during the
breeding season: implications for their lactation strategy and
mating patterns. J Anim Ecol 63:24–30
Venables WN, Ripley BD (2002) Modern applied statistics with S,
4th edn. Springer, New York
Watts P (1992) Thermal constraints on hauling out by harbour seals
(Phoca vitulina). Can J Zool 70:553–560
Watts P (1996) The diel hauling-out cycle of harbour seals in an open
marine environment: correlates and constraints. J Zool Lond
240:175–200
Yochem PK, Stewart BS, DeLong RL, DeMaster DP (1987) Diel
haul-out patterns and site fidelity of harbour seals (Phoca
vitulina Richardsi) on San Miguel Island, California, in autumn.
Mar Mammal Sci 3:323–332
Zuur A, Ieno EN, Walker N, Saveliev AA, Smith GM (2009) Mixed
effects models and extensions in ecology with R. Springer, NewYork
Polar Biol
123