Content uploaded by Thomas J. Evans
Author content
All content in this area was uploaded by Thomas J. Evans on Mar 20, 2018
Content may be subject to copyright.
Ibis
(2008), doi: 10.1111/j.1474-919x.2008.00805.x
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
Blackwell Publishing Ltd
GPS tracking of the foraging movements of
Manx Shearwaters
Puffinus puffinus
breeding on
Skomer Island, Wales
T. C. GUILFORD,
1,
* J. MEADE,
1
R. FREEMAN,
1,2
D. BIRO,
1
T. EVANS,
1
F. BONADONNA,
3
D. BOYLE,
4
S. ROBERTS
5
& C. M. PERRINS
4
1
Animal Behaviour Research Group, Department of Zoology, South Parks Road, Oxford, OX1 3PS, UK
2
Microsoft Research Ltd, Roger Needham Building, J J Thomson Avenue, Cambridge, CB3 0FB, UK
3
CNRS – CEFE, Behavioural Ecology Group, 1919, route de Mende, F-34293 Montpellier, Cedex 5, France
4
Edward Grey Institute of Field Ornithology, Department of Zoology, South Parks Road, Oxford, OX1 3PS, UK
5
Department of Engineering Science, Parks Road, Oxford, OX1 3PJ, UK
We report the first successful use of miniature Global Positioning System loggers to track
the ocean-going behaviour of a
c
. 400 g seabird, the Manx Shearwater
Puffinus puffinus
.
Breeding birds were tracked over three field seasons during the incubation and chick-rearing
periods on their foraging excursions from the large colony on Skomer Island, Pembrokeshire,
UK. Foraging effort was concentrated in the Irish Sea. Likely foraging areas were identified
to the north, and more diffusely to the west of the colony. No foraging excursions were
recorded significantly to the south of the colony, conflicting with the conclusions of earlier
studies based on ringing recoveries and observations. We discuss several explanations including
the hypothesis that foraging may have shifted substantially northwards in recent decades.
We found no obvious relationship between birds’ positions and water depth, although there
was a suggestion that observations at night were in shallower water than those during the
day. We also found that, despite the fact that Shearwaters can be observed rafting off-shore
from their colonies in the hours prior to making landfall at night, breeding birds are usually
located much further from the colony in the last 8 h before arrival, a finding that has
significance for the likely effectiveness of marine protection areas if they are only local to
the colony. Short sequences of precise second-by-second fixes showed that movement
speeds were bimodal, corresponding to sitting on the water (most common at night and
around midday) and flying (most common in the morning and evening), with flight behaviour
separable into erratic (indicative of searching for food) and directional (indicative of
travelling). We also provide a first direct measurement of mean flight speed during directional
flight (
c
. 40 km/h), slower than a Shearwater’s predicted maximum range velocity, suggesting
that birds are exploiting wave or dynamic soaring during long-distance travel.
Keywords:
climate change, navigation, procellariiform, satellite telemetry, seabird.
Many procellariiforms have an inherently precarious
life-history, with breeding concentrated at a relatively
small number of islands or other isolated locations,
coupled with central-place foraging strategies involving
long-distance movements over the open ocean.
Understanding their foraging movements is crucial
to assessing their conservation status, and indirectly
the health of the marine resources on which they
depend, but their ocean-wandering lifestyle makes
these hard to study with any precision. Satellite
tracking has brought huge advances in our under-
standing of the precise foraging movements of larger
species such as albatrosses, bringing into focus their
vulnerability to modern fishing practices, but the size
of loggers has precluded this for smaller species
*Corresponding author.
Email: tim.guilford@zoo.ac.uk
2
T. C. Guilford
et al.
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
(reviewed in BirdLife International 2004). In this
paper we report the first use of precision GPS
logging technology to study the individual foraging
movements during breeding of a 400 g procellarii-
form, the Manx Shearwater
Puffinus puffinus
.
Britain and Ireland hold the majority of the world
population of Manx Shearwaters (Lloyd
et al
. 1991,
Hamer 2003), with just two small Welsh islands,
Skokholm and Skomer, holding in excess of 150 000
breeding pairs, a substantial proportion of the global
breeding population (Smith
et al
. 2001). Manx
Shearwaters, which shallow dive from the surface,
predominantly for small clupeids and some squid
(Thompson 1987, Brooke 1990), are known to fly
considerable distances during the breeding season;
for example, incubating birds of both sexes are away
from the colony for an average of about a week at a
time (Brooke 1990). Sea surveys have revealed that
there are concentrations of birds at various places in
the Irish Sea and its vicinity (Begg & Reid 1997).
Distribution maps produced from such surveys are
invaluable in pin-pointing the areas where the main
feeding concentrations of seabirds occur. However,
where there is more than one feeding concentration
and more than a single breeding colony, they cannot
shed light on the feeding ranges of the birds from the
different colonies, nor on the foraging behaviour of
individuals. Furthermore, in a species as long-lived as
the Manx Shearwater, in any year possibly as much
as half of the population are not breeding; these are
mainly immature birds, not yet of breeding age (up
to 7 years; Brooke 1990), but also include divorced
or widowed birds searching for a new mate. While
each of these birds (without breeding duties) will be
associated with a particular colony, they may not
return there with the regularity of breeders. Hence
their foraging patterns at sea may well be different
from those of breeders.
Early attempts to determine the foraging move-
ments of breeding Manx Shearwaters involved
interpretation of ringing recoveries, together with off-
shore and coastal sightings. These led to the hypothesis
that many birds utilize the sardine fisheries originating
in the Bay of Biscay, which move northwards up the
bay during the northern summer (Lockley 1953).
Lockley’s hypothesis implies a substantial long-
distance southwards component to the Shearwater’s
foraging movements. Later analyses, however (Perrins
& Brooke 1976), found little support for this
hypothesis. In this paper, we use miniature GPS
loggers to track the movements of a number of breed-
ing birds while they are away from Skomer Island,
during both the incubation and the chick-rearing
periods. This has allowed us to make a first direct test
of the current validity of Lockley’s hypothesis, and to
investigate in some spatial detail how breeding birds
utilize marine resources accessible from their colony.
METHODS
Subjects
Between June and August, in the three seasons
2004–6, breeding birds were located in their burrows
using call playback at two densely occupied sites on
Skomer Island, Pembrokeshire, Wales (54
°
44
′
N
5
°
17
′
W): one close to the Warden’s House at North
Haven (2004, 2005), the other close to the Old
Farm (2006). Burrows were chosen for the accessi-
bility of the nest chamber via the burrow entrance or
an inspection hatch. In 2004, a small wire gate, with
a one-way trip, was constructed in the mouth of each
burrow to allow capture of returning birds, or, during
incubation, retention of departing birds immediately
after changeover. Frequent inspections (every 20–
30 min) ensured that we were able to capture focal
study birds by hand promptly. In 2005 and 2006, a
simpler system of small marker pegs placed vertically
in the entrance and frequent inspection (every 15–
20 min) was used to identify the comings and goings
of focal birds. To determine whether adults had fed
at sea, we weighed them with a spring balance before
and after each deployment where it was expedient
and would cause minimal extra disturbance. Only
birds weighing 400 g or more were fitted with trackers.
GPS tracking
The global positioning devices (GPDs, adapted from
original designs by von Hünerbein
et al
. 2000,
Steiner
et al
. 2000) consisted of an integrated GPS
receiver, flash memory and ground plane (µ-blox Co.,
Thalwil, Switzerland), with 4 V, 150 mAh, lithium
polymer rechargeable battery (Ultralife Batteries Inc.,
Newark, USA), and 12.5
×
12.5
×
3 mm ceramic
insulated antennae (Compotron Ltd, Swindon, UK).
Total active operation time was about 55 min. The
GPDs were configured to attempt to obtain posi-
tional fixes for up to 200 s at 2-h intervals. This
meant that, when successful, GPDs recorded a brief
burst of positional fixes (at 1-s intervals) every 2 h,
allowing the type and direction of movement to be
determined, whilst maximizing the proportion of
the foraging journey recorded. GPDs were then
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
GPS tracking of breeding Manx Shearwaters
3
waterproofed using lightweight heat-sealed plastic
sleeves immediately prior to deployment. The
GPDs (which had a flat, streamlined profile of
8
×
30
×
60 mm and weighed
c
. 17 g including
attachment material) were attached dorsally between
the shoulders using three to five thin strips of black
TESA marine cloth tape, each anchored beneath a
small bunch of back feathers and closed over the top
of the device (tape attachments have less impact
than harnesses; Phillips
et al
. 2003). On return, the
device was removed by peeling the tape away from
the feathers and the bird was returned to its burrow.
Data were downloaded onto a field laptop using pro-
prietary µ-
BLOX
software and visualized and analysed
using F
UGAWI
, M
AT
L
AB
, and A
RC
GIS 9 software. A
British Geological Survey DigBath250 data licence
was purchased for conducting bathymetry analysis.
RESULTS
Effectiveness of the tracking technique
In total, we obtained 50 individual datasets of
varying degrees of completeness, constituting a
success rate of about 50% per deployment. Tracks
were recorded from 34 different individuals. Eight
birds were tracked twice, one bird was tracked three
times and two birds were tracked on four occasions
but across 2 years. No bird was tracked more than
three times in a season. We obtained six tracks in
2004, five from chick-rearing birds; in 2005 we
obtained 18 tracks, 16 from chick-rearing birds; and,
in 2006 we obtained 26 tracks, 16 from chick-rearing
birds. No tracked bird failed to return to its burrow.
However, in many cases trackers failed to acquire
positions, ran out of battery whilst the bird remained
underground, returned waterlogged, or were lost at
sea because the tape attachments eventually fail in
seawater (an important welfare failsafe in case we
did not manage to recapture the bird). The mean
duration of the foraging trips studied was 71.8 h,
but the distribution was heavily skewed, with 68%
lasting just 1 or 2 days and the longest trip lasting
12 days. Mean recording duration of the devices was
31.2 h, so that most trips (69%) of 2 days or less
were fully recorded, but only the early part of longer
trips was recorded before the battery was exhausted.
In all, 46% of all trips were recorded completely or
almost so (up to 4 h before recovery), and 89% of all
fix bursts contained fixes generated by four or more
satellites, providing highly accurate position (±
c
. 4 m),
speed and direction data.
Evidence for foraging
The duration of foraging excursions varies greatly in
unmanipulated birds, especially with breeding stage
and, to a lesser extent, sex. Nevertheless, trip durations
can give an indication of whether our manipulated
birds were indulging in normal foraging. Whilst a
systematic comparison between manipulated and
unmanipulated birds was not possible because we
did not measure controls at the same time, the duration
of tracked excursions was broadly consistent with
previous observations. In our study, trips during
incubation lasted a mean of 5.46 days (± 0.92 se,
n
= 13), and during chick feeding a mean of 1.65 days
(± 0.15,
n
= 37). Brooke (1990) measured mean trip
durations of 6.88 ± 0.29 for males and 5.42 ± 0.23
for females during later incubation (Skokholm 1975
and 1976 breeding seasons), and reported means of
1.63 days for males and 2.0 days for females during
chick feeding (data from Rum, Thompson 1987).
Gray and Hamer (2001, data from Skomer 1999)
measured trips during chick feeding at 1.5 ± 0.2 sd
days for males and 1.8 ± 0.2 sd days for females. In
addition, in 33 cases (66%) there was clear evidence
from an adult’s weight measured before and after
deployment that it had fed during the tracked excursion.
In 13 cases (26%), one or other adult measurement
was missed and we were unable to judge. In four
(8%) cases, a tracked adult was sufficiently lighter
when weighed the second time that it is possible
it had not fed (although in all four cases it is possible
that it had fed its chick before we weighed it, and
this accounted for the weight reduction). Hence,
most or all trips tracked represent successful foraging
excursions, and this also suggests that birds were able
to cope with the drag or the increased wing-loading
resulting from carrying the devices.
Distribution and direction of trips
All tracks recorded were plotted in Figure 1 to show
the overall distribution of foraging movements
during incubation and chick-rearing. Plotting the
data as tracks connecting each approximately 2-h
position estimate also allowed some visualization
of any pseudoreplication effects (inherent in
tracking data of this sort) in the following analyses.
Nevertheless, it was clear that birds could travel long
distances between each 2-h fix, and hence such
fixes provided a powerful description of the overall
distribution of the birds’ activity when away from
the colony.
4
T. C. Guilford
et al.
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
Foraging movements were concentrated northwards
and westwards into the Irish Sea, and not south-
wards. This was true both during incubation (red),
when birds were often away a week or longer, and
during the shorter trips of chick-rearing birds (blue).
Furthermore, birds did not frequent all areas within
reach given the actual trip distances measured.
Several areas of activity could be identified, with
particularly dense activity around Skomer, in Car-
digan Bay, and at locations in the Irish Sea further
north (off Dundalk and the Mull of Galloway) and
west of the colony. Though sample sizes between
years were too small to determine definitively whether
or not there were inter-annual effects, informal
inspection suggests there was substantial overlap
between locations visited in different years.
As the logging devices were programmed to obtain
a short series or cluster of consecutive fixes at 1-s
intervals during each 2-h duty cycle, we obtained not
just a position every 2 h, but a snapshot of the bird’s
movement behaviour (principally surface speed and
direction of movement) at the same time. First, we
assigned each cluster of fixes a single position
represented by the median latitude and median
longitude of the series, hereafter called simply a ‘location’,
and plotted these for all excursions in relation to for-
aging trip length. Figure 2a shows that despite our
being unable to track birds for the duration of longer
trips, there was a clear tendency for longer trips to start
in a more northwards direction. Removing all pseudo-
replication effects by representing each trip with just
a single location taken from the middle of the first day
of the trip shows essentially the same result (Fig. 2b).
We then assigned an estimate of quality to each fix
based on the number of satellites contributing to the
fix. Our devices recorded fixes whenever three or
more satellites were successfully accessed, but with
only three satellites, a position fix can be several tens
of metres off true position. This has no significance
for the general location estimates presented here,
but can greatly affect speed estimates derived from
movement between fixes within a cluster. We therefore
filtered the clusters by quality and calculated mean
movement speeds (‘speed’) whenever there were
sufficient fixes based on at least four satellites. Next,
we used the data within each cluster to calculate
both the direction of movement from one second to
the next (‘direction’), and the variability in that
direction across the cluster (‘directionality’ using the
circular statistic
r
, measured from 0 to 1, where 1
equates to unidirectional and 0 equates to randomly
oriented; Batchelet 1981). Directionality was then
plotted in Figure 3 against speed, for every location.
Surface speeds were bimodal. We therefore fitted
a mixture model of two Gaussian curves to the
distributions shown, estimating the mean and
variance of the two movement modes, which clearly
correspond to locations where the bird was either
sitting (0.85 ± 0.24 m/s) or flying (11.13 ± 9.55 m/s).
The higher variance characteristic of the fast move-
ment mode indicated that birds sometimes fly with
the wind and sometimes against it, whereas sitting
birds may drift with the current and probably make
no attempt to counteract their drift. Nevertheless, the
central tendency of the fast distribution provided a
direct measure of flight speed, if we assume that
into-wind and down-wind effects are roughly balanced
across the dataset.
Figure 1. Plots of all fixes with each trip’s fixes connected in
sequence by lines to show approximate course of foraging
excursions. Data are classified according to whether the bird
was, at the start of the trip, incubating an egg (in red), or rearing
a chick (in blue). The position of the colony at Skomer Island is
immediately off the extreme western end of the headland marked
Skomer.
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
GPS tracking of breeding Manx Shearwaters
5
Figure 2. Locations classified by foraging trip length: (a) all locations plotted (white = 1 day, light grey = 2 days, dark grey = 2–4 days,
black ≥ 4 days): (b) each excursion represented by a single location, that closest to 1300 h, and not outside 1100–1500 h, on the first
day, to remove effects of pseudoreplication.
Figure 3. Gaussian mixture model of speeds during all fix clusters based on four or more satellites, overlaid on a scatter plot of speed
against directionality for each location, with a box showing values taken to indicate ‘directional flight’ (r > 0.85). The optimal separation
boundary between the two distributions is shown as a dashed line at 2.5 m/s.
6
T. C. Guilford
et al.
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
Using the optimal separation boundary of the two
fitted distributions (2.5 m/s) we classified each
location as either sitting or flying with maximum
probability. There was also considerable variation in
directionality. For slow moving (sitting) birds, this
variation was almost certainly attributable to increasing
dominance of GPS position error as actual speed
approaches zero, so it was ignored. For fast moving
(flying) birds, variation in directionality allowed us
to distinguish birds in strongly oriented flight, and
therefore probably travelling, from those in erratic
flight that were more likely to be involved in localized
behaviour such as active searching for food. Informal
inspection of the tracks suggested that below a
threshold of about 0.85 (normalized directionality),
birds were often involved in making major turns,
whereas above this threshold flight paths showed
undulations, perhaps indicative of wave- or shear-
soaring, but no major alterations in overall orientation.
In an attempt to distinguish approximately localized
activity (foraging, resting) from travelling behaviour,
Figure 4 plots the locations assigned by whether
birds are sitting or flying erratically (Fig. 4a), or
flying directionally (Fig. 4b). Fix clusters that did not
contain accurate position data were excluded. Figure 4
also plots the British Geological Survey DigBath250
bathymetric database for the Irish Sea to provide a
detailed visualization of the birds’ activity in relation
to underlying water depth.
The patchy distribution across the region clearly
shows several important areas where activity is
concentrated. Comparing the distributions suggests
that activity on the water and erratic flight may have
been more locally concentrated than travelling behaviour,
although there was no clear separation in most areas.
Around the colony itself, the concentration of activity
could have been partially explained by birds waiting
to come onshore (which they only do after dark) at
the end of their fishing trips, or preparing themselves
for a foraging trip immediately after leaving the
colony. To remove activity specifically associated with
leaving or returning to the colony, we plotted in Figure 5
the distribution of locations after removal of all
points up to 8 h after deployment or 6 h before return.
Although several relatively shallow areas (e.g.
Bristol Channel east of Lundy, and the area north and
Figure 4. Locations, filtered for fix quality, classified as: (a) sitting (red points) or erratic flying (red arrows); or, (b) travelling (green
arrows). Locations are overlain on bathymetry.
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
GPS tracking of breeding Manx Shearwaters
7
east of Anglesey) contained no activity, there was no
clear overall relationship with water depth. Figure 6
(top two panels) compares the distribution of locations
with, as a null hypothesis, a random distribution of
points (not on land) generated for the area within
the minimum convex polygon surrounding these
observations. A Mann-Whitney
U
test showed no
significant difference between the depths of actual
locations and randomly created locations (
Z
= –1.09,
n
random
= 600,
n
bird
= 549,
P
= 0.28). We then used
time of Civil Twilight (U.S. Naval Observatory,
Astronomical Applications Department website –
http://aa.usno.navy.mil/faq/, accessed February 2008)
for the actual times and dates of each observation to
determine whether it was night or day, and segmented
the data again by sitting versus flying. A Kruskal–
Wallis test showed that the depth distributions dif-
fered significantly (
χ
2
= 32.84, d.f. = 3,
n
day
= 458,
n
night
= 91,
P
< 0.001), with night-time locations more
likely to be shallow. Further post-hoc comparisons
showed that there was a significant depth difference
between night-time and day-time locations (
Z
= 5.54,
n
night-time
= 91,
n
day-time
= 458,
P
< 0.001), shown in
the bottom two panels of Figure 6, but not between
flying and sitting (
Z
= –0.94,
n
flying
= 169,
n
sitting
= 380,
P
= 0.35). It is important to remember that since
each bird and each track contributed a series of
locations within the actual data (though not the
simulated random data) there was some inherent
pseudoreplication in the tests. However, since loca-
tions within a bird’s track were always at least 2 h
apart there was in general ample time for birds to
have moved across the entire depth range between
locations. Nevertheless, we followed the one signifi-
cant result here with a more conservative Wilcoxon
matched pairs test using a single day and a single
night location (those closest to midday or midnight)
for each track. Although the median depth difference
was negative, and hence night locations were shallower
than day, this was not significant (
U
= 91.0,
n
= 23
pairs,
P
= 0.157; median depth difference = –14.74 m).
We plotted flying versus sitting activity with time
of day in Figure 7. There was a clear diurnal cycle.
Sitting activity was approximately bimodal, with
peaks during the night and in the middle of the day.
Flying activity also appeared to be bimodal, with two
peaks dovetailing quite well with the pattern of
sitting activity. Finally, in an attempt to determine
which locations were most used by birds in the hours
before returning to the nest at the end of the foraging
trip, we plotted activity distributions during the final
8 h before recovery (Fig. 8).
DISCUSSION
Foraging movements during the breeding
season
Early attempts to determine the foraging movements
of birds breeding in the Pembrokeshire colonies
involved interpretation of ringing recoveries, together
with off-shore and coastal sightings. Ringing recov-
eries led to the hypothesis that many birds utilize the
sardine fisheries originating in the Bay of Biscay
which move northwards up the bay during the northern
summer (Lockley 1953). Later, Perrins and Brooke
(1976) recognized that Lockley had underestimated
the age at first breeding, and applied more stringent
criteria for whether birds recovered in Biscay were
likely to be breeders. They found no evidence for
foraging excursions this far south after laying, and
suggested that from the breeding population only
females absent from the colony for the 2 weeks or so
Figure 5. Locations at least 8 h after deployment and more than
6 h before recovery.
8
T. C. Guilford
et al.
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
of egg formation were likely to make the journey to
Biscay. Because departure from the burrow of such
‘honey-mooning’ females is unpredictable, we were
not able to track such birds using GPS and so this
hypothesis remains untested. Nevertheless, Lockley
(1953) also reported observations of mass bird
movement off-shore and particularly past the head-
lands of Cornwall and Finistere (which he termed
the ‘Manx Seaway’). Although the Manx Seaway
was reportedly less intense in June and July than in
April and May, it still appears to provide evidence for
a substantial long-distance southwards component
to the Pembrokeshire birds’ foraging movements
even long after laying. Our findings are strikingly at
odds with this. We found no evidence for southerly
foraging excursions other than very locally, even
when birds were absent from their burrows for many
days. As we were unable to track birds for more than
the first day or two of each excursion, we cannot
know exactly what happens beyond this time on
longer journeys. Nevertheless, Figure 2 shows that
there was a strong tendency for birds to start heading
in a northerly, and to a lesser extent westerly, direction
when they were embarking on what turn out to be
longer journeys. It is possible that southerly move-
ments might have been observed had we tracked birds
earlier during incubation (our tracks of incubating
birds are from late June and July), but Lockley still
reported strong southwards mass movements along
the Manx Seaway during July, and this does not
Figure 6. Histogram of locations with depth. Random and actual locations are compared (top two histograms), and actual locations,
classified as light or dark using nautical night-time, are compared in the bottom two histograms. Note that there are fewer night-time
observations because the night is very short at these latitudes during summer.
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
GPS tracking of breeding Manx Shearwaters
9
match our findings. Even our earliest birds did not fly
southwards. There would appear to be three feasible
explanations.
First, southwards moving birds may be non-breeders.
Up to half of the birds ‘belonging’ to the colonies
were immature, partnerless, or failed breeders.
Presumably, such directional specialization would
require either that breeders and non-breeders seek
different resources, or that breeders lack time
between nest-tending duties to fly sufficiently far
south to exploit the sardine fisheries. In this latter
case, we expect that non-breeders’ excursions could
be longer than those of breeders, but this has never
been measured.
Secondly, if there is strong local partitioning in the
Pembrokeshire colonies, and sub-colonies utilize
their own resource areas, then it might be possible
that our samples of breeders happen to be taken
from two areas that only forage to the west and
north. A division between the neighbouring islands
of Skomer and Skokholm is a related possibility, but
the exchange of ringed birds between the two
(Brooke 1990) suggests that mixing across the
colonies might make the formation of stable and
different traditional foraging routes unlikely. At
present, data are too sparse to test this interesting
possibility.
Thirdly, the species’ foraging ecology may have
changed in the half century since Lockley’s observations,
and the birds are no longer utilizing foraging grounds
to the south. Contemporary headland sightings
could potentially address this hypothesis. Certainly,
Brooke’s (1990) analysis of ringing recoveries since
Figure 7. A histogram showing the frequency of locations containing flying and sitting activity over the 24-h period.
Figure 8. A plot of locations before birds arrive onshore at the
end of their foraging trip (white 0–2 h before arrival, light grey 2–
4 h before arrival, dark grey 4–6 h before arrival, black 6–8 h
before arrival).
10
T. C. Guilford
et al.
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
the 1940s suggested that birds were even then
penetrating less far south than in Lockley’s time, and
it is possible that this trend has continued. Nevertheless,
if a northwards shift has occurred, perhaps because
of warming seas, then it is surprising that we found
no evidence of excursions further north than the
Mull of Galloway, which is about the limit of north-
wards movement proposed by Lockley on the basis
of his original data. As the Pembrokeshire population
is probably increasing, we should expect to see a
dramatic increase in the density of birds in the Irish
Sea if they are no longer travelling south, and still
travelling no further north.
Activity within the Irish Sea
Within the Irish Sea itself, observations at sea (e.g.
Pollock
et al
. 1997) have shown that Manx Shearwaters
are not particularly abundant in March and April,
become more common during May and June, and
peak during July and August. Although hard to
generalize, the peak numbers tend to occur in the
south Irish Sea (including Cardigan Bay and extending
as far west as the south-eastern tip of Ireland), in the
North Channel (Mull of Kintyre to Mull of Galloway)
and in the Irish Sea close to the Irish coast from
about Dublin north to Dundalk (
c
. 53
°
15
′
N to
54
°
0
′
N). The first two of these are close to breeding
colonies (Pembrokeshire islands and Copeland,
respectively) and may, at least in part, be associated
with local movements of birds from these populations.
The third is in some ways of more interest because
there are no significant colonies nearby: Bardsey
with about 7000 pairs is the nearest; the colony on
the Calf of Man is insignificant (Newton
et al
. 2004).
Yet throughout the summer, large numbers of birds
are found in this area and high numbers are main-
tained into September, even after the numbers at the
colonies (and the two other sites of concentration
mentioned above) have started to diminish. This
area lies to the north and west of the Irish Sea front
(Pollock
et al
. 1997) where high seabird density has
been observed (Begg & Reid 1997), presumably in
response to high marine productivity associated with
the sea front and the stratified waters west of it.
Our tracking data matched these distributions
well, demonstrating that birds breeding at the
Pembrokeshire colonies are likely to be a major con-
tributor to these observed concentrations. The dis-
tributions of birds on the sea or in non-directional
flight, activities most likely to be associated with feed-
ing, were clearly concentrated in these three areas
(see Fig. 4), whilst birds in travelling flight showed a
less concentrated distribution. The concentration in
Cardigan Bay was sufficiently close to Skomer to be
in range for birds waiting to come ashore at night.
Nevertheless, considerable activity here remained
even if we excluded activity in the 8 h after deployment
and 6 h before return (Fig. 5). It is most likely therefore
that Cardigan Bay is also an important foraging
destination. In fact, Figure 5 also shows that many
birds not about to make landfall, and therefore
presumed to be engaged in foraging activity, could
also be found relatively close to the colony, and to
the west in particular, suggesting that the colonies
are themselves located close to another important
foraging area, possibly associated with the Celtic Sea
Front (Pollock
et al
. 1997). In contrast, in the final
8 h prior to returning from a foraging trip, birds were
to be found not only concentrated close to the island,
but also dispersed much further afield, until very
close to the time of making landfall (Fig. 8).
Diurnal activity patterns
Birds were most likely to be recorded in flight in the
morning from around dawn until around 10:00 h, and
then again, less obviously, during the afternoon and
early evening. Perhaps the most obvious explanation
for this is that they were commuting between fishing
and rafting areas in the morning and evening, and
then sitting at night and fishing in the middle of the
day. Inspection of the spatial distribution of recorded
positions during actual subjective daylight and darkness
suggests that night-time activity is more clustered
than day-time activity. Furthermore, whilst there
was no obvious overall relationship with water
depth, other than an apparent avoidance of shallow
areas in the Bristol Channel and north of Anglesey,
there was a suggestion of a shift from deeper to
shallower water at night.
Flight parameters
The mean surface speed for our birds showing direc-
tional travel was about 11 m/s (40 km/h), which is
slightly slower than, but still in reasonable agreement
with, earlier estimates of flight speed derived from
observation (approximately 40–55 km/h, Lockley
1953). Calculation suggests that Vmp (minimum
power velocity) should be about 7.5 m/s (27 km/h)
and Vmr (maximum range velocity) about 14 m/s
(50 km/h) for the Manx Shearwater (Pennycuick
1969), so we suggest that birds are flying in the
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
GPS tracking of breeding Manx Shearwaters 11
middle of their theoretical speed range for purely
powered flight. As long-distance movement might
be expected to favour flying close to Vmr, it seems
likely here that birds may be flying closer than
expected to Vmp as they exploit non-powered flight
advantages in the form of wave or dynamic soaring.
Birds flying north to the Mull of Galloway to feed
travelled a straight-line distance from Skomer of
more than 330 km each way, involving a total predicted
minimum direct commuting flight time of around
16.5 h. This foraging area is much closer to other
breeding colonies, most notably Copeland situated
about 30 km away. The foraging destinations of
Copeland birds are currently unknown, but it seems
that they would stand to gain a 15-h (600 km)
travelling advantage over Pembrokeshire birds by
utilizing the Mull of Galloway area. The second
major northerly foraging area is also approximately
75 km closer to Copeland than Skomer, again giving
a potential 3.75 h travelling advantage to Copeland
birds. Such an advantage might be expected to result
in increased provisioning and higher breeding success.
Significance for conservation at sea
The Birds Directive (79/409/EEC) requires each
member state of the EU to set up marine special
protection areas to protect birds, such as the Manx
Shearwater, that are considered rare or vulnerable
within the European Community. Such seabirds
have two key requirements: safe nesting sites and
safe and food-rich feeding. The former are, in some
ways, easier to provide; most current large colonies
are designated nature reserves where people are on
high alert for threats, particularly those from introduced
mammalian predators. Even some formerly successful
colonies, such as Lundy and Canna, where rats have
almost exterminated the birds, have recently been
cleared of these predators and may become thriving
colonies again.
The provision of safe feeding areas at sea is,
however, much more difficult. Our findings emphasize
the problems associated with trying to provide these
places for such wide-ranging species. Marine Nature
Reserves provide some protection for the birds when
they are immediately offshore. In the case of the
Manx Shearwater, these are mainly beneficial to
evening assemblies of birds waiting to come onshore.
Our data indicate that the majority of the breeding
adults may not join these assemblies, being much
farther from the colony when dusk falls (Fig. 8),
although it is important to consider that birds carrying
trackers may have behaved differently to unmanipu-
lated birds and may therefore have reached the
vicinity of the colony later on their day of return,
simply because they were carrying an additional
encumbrance. Further, as with most other seabirds,
the areas used for feeding are much more distant
from the colony and more diffuse, making it difficult
or impossible to provide them with anything near
complete protection while they are foraging. The
future conservation of vulnerable species with such
wide-ranging habits as the Manx Shearwater will
require a fuller understanding of the areas of sea they
use and this will require integrating movement and
activity information from a variety of techniques,
including precision GPS tracking. Further, we should
not ignore the possibility that resources, and hence
resource use, may be changing.
Impacts of tracking on behaviour
The Manx Shearwater is currently one of the small-
est seabirds to have been tracked using GPS technol-
ogy. We deployed the smallest devices available, used
the generally agreed lowest impact attachment
methods (e.g. Phillips et al. 2003), and restricted
deployments to single excursions (occasionally, with
one or more subsequent deployments only after an
interval). At approximately 4% body weight, and
with a very low profile, we believe our devices
deployed in this way had minimal lasting impact. It
is of course possible that behaviour during tracked
excursions was not fully representative of normal
behaviour, and this caveat must always be considered
when drawing inferences from this and similar data.
Telemetry advances are revolutionizing the study of
wild birds, but improvements must always strive to
reduce impact both for ethical reasons and so that
data remain representative of normal behaviour.
We thank the Wildlife Trust of South and West Wales, the
Countryside Council for Wales, Juan Brown (Warden) and
many other staff of Skomer Island National Nature
Reserve. We also thank Louise Maurice and Christine
Nicol for assistance in the field, and Mike Brooke, Kerry
Leonard, Neville McKee and two anonymous referees for
comments on the manuscript.
REFERENCES
Batchelet, E. 1981. Circular Statistics in Biology. London:
Academic Press.
Begg, G. & Reid, J.B. 1997. Spatial variation in seabird density
at a shallow sea tidal mixing front in the Irish Sea. ICES J.
Marine Sci. Symp. Edn 54: 552–565.
12 T. C. Guilford et al.
© 2008 The Authors
Journal compilation © 2008 British Ornithologists’ Union
BirdLife International. 2004. Tr acking Ocean Wanderers: The
Global Distribution of Albatrosses and Petrels. Results from
the Global Procellariiform Tracking Workshop, 1–5 September;
2003, Gordon’s Bay, South Africa. Cambridge: BirdLife
International.
Brooke, M. 1990. The Manx Shearwater. London: T. & A.D.
Poyser.
Gray, C.M. & Hamer, K.C. 2001. Food-provisioning behaviour of
male and female Manx Shearwaters, Puffinus puffinus. Anim.
Behav. 62: 117–121.
Hamer, K.C. 2003. Puffinus puffinus Manx Shearwater. BWP
Update (The Journal of the Birds of the Western Palearctic)
3: 203–213.
von Hünerbein, K., Hamann, H.-J., Rüter, E. & Wiltschko, W.
2000. A GPS-based system for recording the flight paths of
birds. Naturwissenschaften 87: 278–279.
Lloyd, C., Tasker, M.L. & Partridge, K. 1991. The Status of
Seabirds in Britain and Ireland. London: T. & A.D. Poyser.
Lockley, R.M. 1953. On the movements of the Manx Shearwater
at sea during the breeding season. Br. Birds 46 (Suppl): 1–48.
Newton, S.F., Thompson, K. & Mitchell, P.I. 2004. Manx
Shearwater Puffinus puffinus. In Mitchell, P.I., Newton, S.F.,
Ratcliffe, N. & Dunn, T.E. (eds) Seabird Populations of Britain
& Ireland: 63 – 80. London: T. & A.D. Poyser.
Pennycuick, C.J. 1969. The mechanics of bird migration. Ibis
111: 525 – 556.
Perrins, C.M. & Brooke, M. de L. 1976. Manx Shearwaters in
the Bay of Biscay. Bird Study 23: 295 –300.
Phillips, R.A., Xavier, J.C. & Croxall, J.P. 2003. Effects of satellite
transmitters on albatrosses and petrels. Auk 120: 1082–1090.
Pollock, C.M., Reid, J.B., Webb, A. & Tasker, M.L. 1997. The
Distribution of Seabirds and Cetaceans in the Waters around
Ireland. JNCC Report, No. 267. Aberdeen: Joint Nature
Conservation Committee.
Smith, S., Thompson, G. & Perrins, C.M. 2001. A census of the
Manx Shearwater on Skomer, Skokholm and Middleholm,
West Wales. Bird Study 48: 330 –340.
Steiner, I., Bürgi, C., Werffeli, S., Dell’Omo, G., Valenti, P.,
Tröster, G., Wolfer, D. P. & Lipp, H.-P. 2000. A GPS-logger
and software for analysis of homing in pigeons and small
mammals. Physiol. Behav. 71: 589– 596.
Thompson, K.R. 1987. The ecology of the Manx Shearwater
(Puffinus puffinus) on Rhum, West Scotland. PhD Thesis,
University of Glasgow.
Received 21 September 2007;
revision accepted 21 January 2008.
