Content uploaded by Simeon Lisovski
Author content
All content in this area was uploaded by Simeon Lisovski on Oct 15, 2017
Content may be subject to copyright.
Movement patterns of Sanderling (Calidris alba) in the East
Asian–Australasian Flyway and a comparison of methods
for identification of crucial areas for conservation
Simeon Lisovski
A,D
, Ken Gosbell
B
, Maureen Christie
B
, Bethany J. Hoye
A
, Marcel Klaassen
A
,
Iain D. Stewart
B
, Alice J. Taysom
C
and Clive Minton
B
A
Deakin University, School of Life and Environmental Sciences, Centre for Integrative Ecology,
75 Pigdons Road, Geelong, Vic. 3220, Australia.
B
Victorian Wader Study Group, c/o 165 Dalgetty Road, Beaumaris, Vic. 3193, Australia.
C
Applied Ecology Research Group, College of Engineering and Science, Victoria University –Footscray Park
Campus, PO Box 14428, Melbourne MC, Vic. 8001, Australia.
D
Corresponding author. Email: simeon.lisovski@gmail.com
Abstract. Worldwide, most populations of migratory shorebirds are in jeopardy, none more so than those of the East
Asian–Australasian Flyway (EAAF). In order to preserve these highly mobile species a detailed understanding of their use
of feeding and resting sites along the flyway is required. In this study we used light-level geolocators and new analytical
tools to reveal individual breeding locations and migration routes of 13 Sanderlings (Calidris alba) that spend their non-
breeding season in South Australia. We then used these individual migration routes to identify the timing and location of
important stopping areas and compared this with assessments based on resightings of leg-flagged birds and count data.
During both northward and southward migration, Sanderlings were found to make extensive use of five main areas of the
Chinese coastline, the Yellow Sea and the northern end of the Sakhalin Peninsula. Insights gained from the individual
migration routes highlight inherent biases in using only count and resighting data to identify important feeding and resting
sites along the Flyway. These findings suggest that data on individual movements may be crucial to effective conservation
planning of shorebirds in the EAAF and elsewhere in the world.
Additional keywords: banding data, bird counts, bird migration, conservation planning, light-level geolocation, MCMC
path estimation, migratory connectivity, resightings.
Received 20 April 2015, accepted 1 December 2015, published online 10 March 2016
Introduction
Animal migration is thought to have evolved in response to
spatiotemporal variation in the abundance of resources and
threats (e.g. Alerstam et al.2003; Dingle and Drake 2007;
McKinnon et al.2010). Despite the immense costs involved in
performing regular, and often long-distance, journeys, migration
is a common and widespread phenomenon in the animal kingdom
(Dingle and Drake 2007). In particular, most species of shorebirds
are migratory (Kirby et al.2008), some of which undertake
migratory movements of up to 30000 km per year (e.g. Harring-
ton 2001; Gill et al.2009; Minton et al.2010).
In order to meet the physiological demands of migration
most migratory birds partition their journey into a series of flights
interrupted by periods of intense feeding and resting (Piersma
1987; Colwell 2010). Consequently most migratory birds depend
on a network of suitable feeding and resting sites along their
flyway for successful migration and survival. This reliance on
multiple sites is thought to render migratory species highly
susceptible to changes on a local and global scale (Piersma and
Baker 2000; Runge et al.2014) and, as a result, migratory
populations across a wide range of taxa have declined in recent
decades (Wilcove and Wikelski 2008). Migratory shorebirds in
all global flyways, and especially the East Asian–Australasian
Flyway (EAAF), are emblematic of these declines (Bamford
et al.2008; Amano et al.2010).
Effective protection of highly mobile species is predicated
on detailed understanding of how migrants make use of their
flyway, and hence the areas and sites that are crucial for conser-
vation (Holdo and Roach 2013; Runge et al.2014). Counts of
birds and bird-banding programs, including leg-flagging and
subsequent resightings, have traditionally been used to develop
an initial understanding of migratory movements and to initiate
conservation measures (e.g. Bamford et al.2008; Minton et al.
2011). The effectiveness of these methods is highly dependent
on the spatiotemporal distribution of observers. Accurate tracks
from individual migratory shorebirds equipped with satellite
transmitters have revealed considerably more detail on migration
schedules and routes than traditional methods (Gill et al.2009;
SPECIAL ISSUE
CSIRO PUBLISHING
Emu,2016, 116, 168–177
http://dx.doi.org/10.1071/MU15042
Journal compilation BirdLife Australia 2016 www.publish.csiro.au/journals/emu
Battley et al.2012). Despite the ongoing development of in-
creasingly lighter devices, the use of satellite transmitters is
restricted to a small number of species of larger body mass.
Light-level geolocators (hereafter geolocators) are considerably
lighter than satellite transmitters (<1 g; Bridge et al.2011), and
therefore provide the opportunity to track smaller migratory
birds and greatly expanding the number of species that may
be studied. Initial application of geolocators to shorebirds has
provided valuable information on the migratory movements of
several species using the EAAF (e.g. Minton et al.2010,2013).
However, our ability to make detailed and accurate descriptions
of migratory movements based on geolocator data has, so far,
been hindered by coarse spatial resolution (Lisovski et al.2012)
and the inability to determine spatial location during the equinox –
owing to equal daylength across the globe –and during the
breeding season –owing to constant daylight on the Arctic
breeding grounds of these shorebirds. Consequently, the identi-
fication of critical sites used by large proportions of a population
has thus far remained a key challenge to conservation along the
EAAF, but the ongoing development of geolocator devices and
processing tools offers significant potential for progress.
In this study, we used new-generation light-level geolocators
that record the full light range, in combination with recently
developed analytical tools, to determine individual movement
tracks and associated uncertainty estimates, providing detailed
understanding of migration patterns of Sanderlings (Calidris
alba) in the EAAF. Sanderlings have a worldwide distribution
and perform some of the longest distance migrations yet
recorded (Lanting 1984; Reneerkens et al.2009; Minton et al.
2013). The species breeds in the Arctic along the northern coast
of European and Siberian Russia, in parts of Alaska, in the
Canadian Arctic and in northern Greenland. During the non-
breeding season Sanderlings use a wide range of wintering sites,
including around 40–45N in western Europe, North America
and Asia (MacWhirter et al.2002), and to the southern tips of
South America, Africa and Australia.
Our aims were twofold: (1) to provide a detailed description
of the biannual migration of Sanderlings in the EAAF; and
(2) to compare three different methods –count data, leg-flag
resightings, and individual geolocator tracks –for identifying
critical areas for shorebird conservation in the EAAF. Geolocator
tracks were processed and compared using the most frequently
used method, hereafter referred to as the ‘simple threshold
method’(Lisovski and Hahn 2012), and a recently developed
and more sophisticated Bayesian framework using Markov Chain
Monte Carlo methods to estimate location posterior, hereafter
referred to as ‘MCMC path estimates’.In describing the move-
ments of Sanderlings along the EAAF, we further aimed to show
that the recent advances in light-level geolocation allow us to
(1) reveal detailed migration schedules and routes; (2) estimate
the location of the high Arctic breeding sites of the species;
and (3) describe and quantify the degree of within-population
migratory connectivity over the annual cycle.
Methods
Animal capture and tracking data
A total of 44 light-level geolocators (Intigeo-W65, Migrate
Technology Ltd, Cambridge, UK) were deployed in March
2012 at Canunda National Park, in south-eastern South Australia
(140110E, 37370S) under approval from the South Australian
Department of Environment, Water and Natural Resources.
Canunda National Park is an important wintering site for Sander-
lings during the boreal winter (austral summer), with typically
between 200 and 400 individuals present from October to April
each year (Bamford et al.2008). Sanderlings were caught, using
cannon nets, in a single catch, on the ocean beach where they
forage and roost. A geolocater was placed on the left tibia of
each bird. The geolocater was fastened to a leg-flag (made from
a‘darvik’PCV sheet) using Kevlar thread reinforced with
Araldite resin cement. To comply with the Australian Bird and
Bat Banding Scheme for South Australia, two additional flags
were employed (orange above yellow). In order to individually
identify birds on which we deployed geolocators, we used an
engraved flag from the orange (upper) flag. This flag was 0.6 mm
thick, bi-coloured material with the orange engraved through to
reveal the black below. Geolocators weighed 0.65 g, making the
total weight, including the flag, ~1 g. This represents <2% of the
mean (lean) body mass of Sanderlings (for measurements, see
supplementary material S1, available online). Based on multiple
reports it appears that shorebirds readily adapt to carrying geo-
locators on their legs and that the device has no significant effect
on annual survival (e.g. Conklin et al.2010; Niles et al.2010). A
total of 14 geolocators were retrieved (32%), 13 from birds caught
at the site where they had been first caught and geolocators
deployed, and one from a bird shot in Sakhalin, eastern Siberia,
in May 2013. Retrieval rates of geolocators in other shorebird
studies conducted by the Victorian Wader Study Group and
the Australasian Wader Studies Group varied between 10%
(Great Knot (Calidris tenuirostris)) and 50% (Ruddy Turnstone
(Arenaria interpres)). These retrieval rates reflect the ability to
detect and catch the individuals as well as their site-fidelity
and not their actual probability of survival. Other individuals
carrying geolocators have been seen occasionally at or near this
location, including one as recently as September 2016. The bird
shot in Siberia during northward migration in 2013 also indicates
that the detection rate is not 100% because the geolocator data
revealed that this individual spent the non-breeding season
of 2012 near the deployment site. All individuals, except the
bird shot in Siberia (B013), were sexed molecularly, using the
primers P8 and P2 according to the method described by
Griffiths et al.(1998). We used a principal component analysis
based on morphological measurements to assign B013 the status
of female with a probability of 0.7 (see S1 in Supplementary
material).
Analysis of geolocator data
Light-intensity recordings from geolocators were used to first
estimate the breeding sites of each individual and subsequently,
using the derived breeding site position, to estimate the full
migration path.
Breeding sites
Sanderlings breed at high Arctic latitudes (Lappo et al.2012)
and thus experience constant daylight during this part of their
annual cycle. Conventional methods to estimate positions from
light-intensity recordings over time (i.e. geolocation by light)
Sanderling migration in the EAAF Emu 169
usually fail to produce reliable position estimates under condi-
tions of 24-h daylight as the light sensor generally does not
record any variation in light intensity through the day (Lisovski
et al.2012). However, the light sensor in the geolocators used
in this study recorded the full range of light intensities for
each day, allowing the breeding location to be estimated. We
developed a template-fit analysis to estimate the positons of the
breeding sites. For each individual, light-intensity records from
the deployment site in Australia, recorded on the bird during
the stationary period after deployment or before retrieval of
the device, were used to generate a calibration curve of light
intensity as a function of zenith angles, using astronomical
functions within the R package SGAT (Wotherspoon et al.
2013). This calibration curve allowed generation of expected
light at any location for a given time or zenith angle. Using the
individual, geolocator-specific calibration curves, predictions
were made on the temporal variations in daily light intensity for
the entire breeding season for every 50 50-km grid-cell across
the entire Russian Arctic. Next, the predicted light values were
compared with the observed light data. We calculated the per-
centage of single light-intensity recordings that exceeded the
predicted values within each grid-cell. Observations can only
have been recorded within grid-cells where 100% of the observed
light-intensity values were below the predicted values. This
approach assumes that birds were resident during the entire
breeding period. To correct for potential travel after arrival at or
before departure from the Arctic we excluded observations
from three days after arriving in the region experiencing
24-h daylight and three days before departure from that region.
In the best-case scenario, when light-intensity measurements
are little influenced by shading of any kind, the 100% likelihood
contour line plotted across the potential breeding area is char-
acterised by a v-shape (or u-shape) with the highest likelihood
slightly above the minima of this contour line (see supplementary
material S2). We selected the closest position on land to this
lowest latitudinal position above the 100% contour line. From
data recorded at the deployment and retrieval site in South
Australia we estimated geolocation error to be <200 km. How-
ever, since the variation in elevation angle of the sun at the
breeding grounds is low and signal-to-noise ratio high (owing
to incubation and habitat) we expect location accuracy of the
estimated breeding sites to be up 100–300 km lower (for more
details, code and explanations see S2 in Supplementary material).
Migration pathway
Daily positions, and hence migration pathways, for each
individual were estimated from raw light-level data using the
threshold method of estimating positions based on sunrise and
sunset events (Lisovski et al.2012). Daily sunrise and sunset
times as well as initial positions based on the simple threshold
method were calculated using the R package GeoLight (Lisovski
and Hahn 2012). A light-intensity threshold of 0.8 was used for
all individuals. The corresponding zenith angle was defined
from sunrise and sunset times recorded while the birds were at
the deployment site. The defined zenith angle varied between
individuals and ranged from 93.4 to 96.5. To derive more
accurate positions, we used a Bayesian framework that incorpo-
rates the observed sunrise and sunset times together with
prior knowledge of Sanderling behaviour to provide location
estimates with associated measurements of uncertainty. The R
package SGAT (Sumner et al.2009; Wotherspoon et al.2013)
uses MCMC simulations that permit a spatial probability mask,
prior definition of the error distribution of twilight events
(twilight model) and plausible flight-speed values (behavioural
model), which collectively allowed us to refine the tracks derived
from the sunrise and sunset times (for detailed description of
the model assumptions, see Sumner et al.2009). The ‘spatial
probability mask’is based on the premise that, during migration,
Sanderlings are most commonly found on coastal sandy
beaches, although they may also occur on tidal mudflats and the
shores of lakes and rivers. Estimated positions were therefore
considered to be more likely if close to a shoreline and indepen-
dent of the habitat type: the relative probability was assumed to
decrease exponentially (from 4 to 1) with increasing distance
from the shoreline, using the equation
1þ5expððd=50 000Þ3Þ
We used a spatial shoreline dataset with a 1:75000 scale (http://
www.ngdc.noaa.gov/mgg/shorelines/shorelines.html, accessed
1 September 2014). For the ‘twilight model’, the discrepancy
between observed and expected times of twilight was assumed
to follow a log-normal distribution. For sunrise, positive values
correspond to an observed sunrise occurring after the expected
time of sunrise, whereas for sunset, positive values correspond
to an observed sunset occurring before the expected time of
sunset. We chose a conservative prior (log-normal distribution,
with meanlog = 1.65, sdlog = 0.9) since error in twilight detection
potentially varies greatly over the annual cycle. For the
‘behavioural model’, we assumed that migratory shorebirds
perform stepwise migrations, with fairly long staging periods
between periods of movement (Piersma 1987; Warnock 2010).
We modelled flight-speed (ground-speed) using a gamma distri-
bution (shape = 0.7, scale = 0.05) assuming that the speed with
the highest probability was <1 (i.e. the bird is most likely to be
stationary at any given time), and that maximum flight speeds up
to 80 km h
1
were likely to occur during migration (Pennycuick
et al.2013). For each individual we used these parameters and
started by drawing an initial 10000 samples for burn-in and
tuning of the proposal distribution. One sample reflects one set
of positions between each twilight event along the migration
path. The proposal distribution is the conditional probability –
here the spatial probability distribution of the individual –that is
calculated after all available information was taken into account.
A further 40000 samples were drawn to visually evaluate chain
convergence. A final draw of 5000 samples was then generated
to describe the posterior distribution. (R-code for individual
location estimation is available from the authors upon request.)
Analysis of migratory movements
We used the median of the posterior distribution as our estimate
for the most likely daily position of each individual and hence
their most likely migration path. We evaluated the timing of
migration and whether the individual was moving on a given day
during migration using a first-passage-time(FPT) analysis:
the FPT describes the time required for an individual to cross
a circle with a radius of 500 km (Fauchald and Tveraa 2003). In
170 Emu S. Lisovski et al.
a second step, FTP was used to identify periods of residency –
periods of stable FTP within the defined radius. All posterior
distributions, 5000 positions per day per individual, were com-
bined to calculate the aggregated time the tracked population
spent in each grid-cell. Furthermore, those posterior distributions
were used to analyse migratory connectivity and, in particular,
the spatial spread of the individuals from the tracked population
over time resulting from the temporal synchrony within the
population (Bauer et al.2015). To quantify the within-population
connectivity, a minimum convex hull was generated around
the 0.6 and 0.95 quartile contour lines –the space that includes
60% and 95% of the samples forming the posterior distribution –
across five-day intervals using the mcp function of the R package
adehabitatHR (Calenge 2006). The area where the convex hulls
and the Flyway, defined as the 0.4 quartile contour line of all
posterior distributions from all individuals, overlap were used
to quantify the spatial spread of the population for each five-day
period.
Resightings of leg-flags
To compare leg-flag resightings with our tracking results we
used resightings of Sanderlings originally flagged on the coast
of south-eastern South Australia, where the geolocator devices
were deployed. Leg-flags placed on shorebirds at this site are
orange over yellow, and a total of 3638 Sanderlings have been
flagged. Resightings of leg-flags from across the Flyway have
been drawn from the Australasian Wader Studies Group
database (http://www.awsg.org.au/flagging.php, accessed 1
September 2014). We further limited our use of resightings to
those reported during the migration periods (15 April –1 July
for northward migration,; 1 July –1 November for southward
migration). Flag sightings were aggregated on a 500 500-km
spatial grid.
Bird counts
Counts of Sanderling were extracted from estimates of shorebird
populations within the EAAF based on a review of count data in
Bamford et al.(2008, pp. 91–93). This report provides spatial
information on counts per species and for the periods of north-
ward migration, southward migration, breeding and non-breed-
ing, between 1979 and 2003. Here we used the count data from
identified internationally important sites that regularly support
1% of the individuals of a population of one species or subspecies
(Criterion 6 of the Ramsar Convention (Ramsar Convention
Bureau 1971)). Maximum counts were aggregated on a 500
500-km spatial grid.
Results
Breeding sites
The estimated positions of breeding sites spanned an area from
the New Siberian Islands of eastern Siberia, 300 km south to the
Siberian mainland of Russia and 1300 km west to the Taimyr
Peninsula (Fig. 1). The highest aggregation of estimated
breeding sites was on the New Siberian Islands.
Migration pathways
Thirteen individuals were recorded performing a complete mi-
gration cycle from the deployment site in South Australia to the
Arctic breeding grounds and back. Detailed individual migration
itineraries are shown in S3 in Supplementary material. Only
one individual (ID2030) did not perform a complete migration;
this individual left on 15 May 2012, made an extended stopover
of 35 days in Borneo and another of 35 days in the area of
Hainan Island, southern China before returning to Australia.
We excluded this individual from all further analyses.
2003
2006
2008
2020
2027
2028
2036
2038
B013
2007
2009
2019 2018
New Siberian
Islands
70˚N
110˚E 150˚E
Male
Female
200 km
R
u
s
s
i
a
n
A
r
c
t
i
c
Fig. 1. Estimated breeding locations of tracked Sanderlings (black dots = females; white dots = males;
numbers = bird IDs), based on light-intensity records (light-level geolocators). The locations are estimates
and are therefore associated with an error (see Methods and Supplementary material S2).
Sanderling migration in the EAAF Emu 171
Northward migration
After deployment with geolocators, the Sanderlings remained
at Canunda National Park until departure for their first migratory
leg between 26 April and 10 May 2012 (mean date of departure
s.d.: population, 2 May 4.5 days; females, 3 May 5 days;
males, 1 May 4 days; Fig. 2b). In general, individuals made
a single long-distance flight from South Australia, across the
Equator, to the coasts of Vietnam and China. The coasts of
Hainan Island, central China, Taiwan and the Yellow Sea were
used intensively for extended stopovers (Fig. 3a). All but one
individual (ID2007) also made an additional stopover along
the coast of the Sea of Okhotsk, at the northern end of Sakhalin
Island. Sanderlings arrived on their breeding sites between 5 and
16 June 2012 (mean date s.d.: population, 11 June 6 days;
females, 11 June 5 days; males, 10 June 6 days; Fig. 2b).
Northward migration, on average, was completed within 40 days
of departure, ranging from a maximum of 48 days (ID2003) to
a minimum of 35 days (ID2007).
Southward migration
All tracked individuals left their breeding sites between
13 July and 22 August 2012 (mean dates s.d.: population,
23 July 11 days; females, 1 Aug 19 days; males, 19
July 6 days; Fig. 2c) after staying between 32 days (ID2009)
and 66 days (ID2018) in their breeding areas. Like northward
migration, all but one individual (ID2006) used the coast of the
Sea of Okhotsk as the first major stopover site after crossing the
Arctic Circle (Fig. 3); in contrast, ID2006 used an inland route
via Mongolia before stopping on the coast of central China and
Taiwan. Most individuals had subsequent stops along the coasts
of China, Taiwan and Korea, although considerably fewer indi-
viduals visited the Yellow Sea during southward compared
with northward migration. In contrast to northward migration,
all individuals made at least one additional stopover in tropical
or subtropical regions (Philippines, Indonesia and Malaysia)
before returning to Australia. Furthermore, more than half of
the individuals (7) used at least one stopover site on the
Australian continent before returning to, or close to, Canunda
National Park. Sanderlings arrived at these wintering sites be-
tween 20 September and 12 November 2012 (mean date s.d.:
population, 9 October 27 days; females, 8 October 28 days;
males, 9 October 17 days; Fig. 2c). The timing of return
of males and females was almost the same even though, on
average, females left the breeding grounds c. 2 weeks later than
males. Southward migration, on average, was completed in
78 days, ranging from a maximum of 108 days (ID2038) to a
minimum of 57 days (ID2019).
Resightings of leg-flags
Until September 2014, there had been 488 resightings of the
3638 individual Sanderlings banded in south-eastern South
Australia reported during the migratory period. Of these, 208
were recorded during northward migration and 280 during south-
ward migration. On northward migration, most resightings
were in the Yellow Sea region (n= 137; 66%). Another 26
resightings (12%) were from the northern part of Sakhalin
Island, Russia, and 25 (12%) from the coast of central China
and Taiwan. During southward migration only one resighting
was recorded within the Yellow Sea. During southward migra-
tion, most resightings were instead from Japan (n= 145; 55%),
along with resightings from the northern end of Sakhalin Island,
Russia, (n= 31; 11%) and the coasts of central China and Taiwan
(n= 32; 11%).
Count data
After collating the maximum counts of Sanderlings from all
internationally important sites on a 500 500-km grid, we
retained a total of 12 grid-cells: seven from the period of north-
ward migration and 11 from the period of southward migration.
Two areas adjacent to the geolocator deployment site in south-
eastern Australia, two areas within the Yellow Sea (Yenchenk
National Nature Reserve and Linghekou, China) and three
areas in Japan (multiple sites in central and southern Japan) and
southern South Korea (Nakdong Estuary) were identified for
northward migration based on counts. For southward migration,
five additional areas of importance were identified: the northern
end of Sakhalin Island (Sakhalinsky Bay, Russia), the south-
eastern Yellow Sea (Kum Estuary, South Korea), two in north-
western Australia (Roebuck Bay and Eighty Mile Beach) and
one in Tasmania (Blanchet Point).
–40
0
40
80
26 Apr 16 May 5 Jun 25 Jun 10 Jul 30 Jul 19 Aug 8 Sep 28 Sep 18 Oct 17 Nov
–40
0
40
80
Latitude (°)
Date (2012)
(b) Northward migration (c) Southward migration
Breeding
140100
Lon
g
itude (°)
(a) Flyway Female
Male
Fig. 2. (a) The East–Asian Australasian Flyway and the latitudinal movement of the 13 Sanderlings tracked using light-level geolocators over time for
(b) northward migration and (c) southward migration.
172 Emu S. Lisovski et al.
Fig. 3. Comparison of methods for identifying important stopover areas during Sanderling migration. Days spent at
each area (bold numbers in Markov Chain Monte Carlo (MCMC) path estimation map) during northward (upper panel)
and southward migration (lower panel) are tabulated for two analyses of 13 individual birds tracked using light-level
geolocators, as well as the sum of maximum counts (numbers in population count maps) and leg-flag resightings (numbers in
leg-flag resighting maps). Percentage values indicate the relative proportion of time spent, all resightings or all counts across all
areas, with bold values representing the maximum value for each method. Maps represent time spent for geolocation methods
on a 100 100-km grid, and the sum of leg-flag resightings or maximum counts on a 500 500-km grid.
Sanderling migration in the EAAF Emu 173
Comparison of methods for identifying important sites
Based on the merged posterior distributions of all individual
MCMC path estimates, five areas could be identified as being
used extensively by the tracked individuals and are therefore
classified as areas of major importance. The number of days
spent within these areas based on the MCMC path estimates,
simple threshold estimates, leg-flag resightings and the sum of
maximum Sanderling counts are shown Fig. 2(within included
tables). Although the relative number of days spent between the
important areas are concordant between the two different geolo-
cator analysis methods, the leg-flag resightings and bird counts
overestimated the use of certain areas by the population (e.g.
southern and central Japan) and underestimated or even omitted
the use of other areas (e.g. the coastlines of central China and
Taiwan; Fig. 3).
Spatiotemporal patterns of migratory connectivity
Before departure on northward migration, the spatial spread
across the population of tracked individuals was low (60%
convex hull: 1.5
11
–2.5
11
km
2
; 90% convex hull: 1.3
12
–1.9
12
km
2
)
and hence the connectivity within the population was high.
With the onset of northward migration, spatial spread increased
considerably, reaching a maximum of 1.4
13
–1.9
13
km
2
(60% and
90% convex hulls) between 3 and 8 May (Fig. 4a). The area
subsequently decreased, to 1.6
12
–7.8
12
km
2
, between 23 and
28 May (Fig. 4b), before increasing again to 5.8
12
–1
13
km
2
between 2 and 7 June, shortly before the arrival of Sanderlings
on the breeding grounds (where the population spread over
5
11
–1.9
12
km
2
). Soon after the onset of southward migration,
the area used by the Sanderlings increased, peaking at 1.6
13
–
2.3
13
km
2
between 25 and 30 September (Fig. 4b). Thereafter,
the area used steadily decreased until the population returned
to the fairly small area around the deployment site in South
Australia.
Discussion
Comparison of methods for identifying important sites
When identifying areas of importance for the conservation of
highly mobile species, several methods are frequently used.
Our results for Sanderlings suggest that although individual
tracking methods, leg-flag resightings and bird counts show
considerable overlap, critical detail may be lost when relying on
the latter two more traditional methods alone. All four methods
assessed here invariably identified the coastline of the Yellow
Sea as the major stopover area for Sanderlings during northward
migration, and both geolocator data and leg-flag resightings
indicated extensive use of a large swathe of the central Asian
coastline. All methods also highlighted the use of the West
Australian coastline as Sanderlings returned to their wintering
sites, and that these areas were skipped during northward migra-
tion. Indeed, our results suggest that areas of importance to the
population may by underestimated or even missed when assess-
ments are based solely on on-the-ground observations. Although
several clear areas of importance coincided between both geo-
locator analysis methods, the population counts and leg-flag
resightings showed substantial differences. Critically, the impor-
tance of five areas, notably those in tropical and subtropical
regions, was generally underestimated using leg-flag resightings,
and even omitted when relying on count data. Strikingly, the
coasts of central and southern Japan were prominent in the
count data for both migration legs, and for southward migration
in the leg-flag resightings, yet only one of the tracked Sanderlings
(B013) showed a migration route via Japan.
Clearly, all methods discussed here have limitations and
some of the discrepancies between the count data and the other
three methods may be partly explained by the fact that we
obtained tracking data from a small number of individuals from
a single wintering population only. Additional tracking data
would undoubtedly improve the accuracy of our estimates, but
leg-flag resightings from Sanderlings caught and flagged at
other wintering sites along the EAAF show very similar spatial
patterns to the individuals flagged in South Australia (Minton
0.0
5.012
1.013
1.513
2.013
2.513
Apr May Jun Jul Aug Sep Oct Nov Dec
Area (km2)
25 May–30 May 27 Sep–2 Oct
(a)
(b)
Fig. 4. Migratory connectivity within the Sanderling population over time.
The spatial spread of the population for a given period was calculated as
the area of the minimum convex hull enclosing 60% (grey bars and dark-
grey polygon) or 95% (line bars and light-grey polygons) of the combined
MCMC-estimated paths from all 13 individuals. The maps show two
extremes: periods of the greatest (left) and least (right) connectivity within
the population. The dashed line indicates the boundaries of the Flyway used
by the tracked Sanderlings.
174 Emu S. Lisovski et al.
et al.2011). Finally, the count data presumably include indivi-
duals from several wintering populations, and yet key sites used
by our tracked individuals were not recorded in these data.
Those sites might not have been counted or do not appear in
Bamford et al. (2008) for other reasons, such as the counts may
have been too small to meet the 1% of the Flyway population
criterion. We also recognise that studies of population counts
and leg-flag resightings are primarily intended for purposes
other than identifying crucial areas for conservation, including
the study of population dynamics and estimating mortality rates.
However, these data resources constitute the most extensive
information on species-specific movements and are therefore
frequently used to identify areas of importance and to inform
conservation planning. The discrepancy between counts, leg-
flag resightings and individual movement data shown here
highlights the value of integrating multiple data sources in order
to inform conservation planning, determining priorities and
resource allocation on ground.
Geolocator analysis
The correspondence between both geolocator methods used
here is initially surprising, given the low precision and accuracy
of the simple threshold method (Lisovski et al.2012). However,
most shorebird species have a preference for open habitats,
resulting in little noise in light-intensity recordings. Moreover,
the equinoxes –periods with comparatively low accuracy and
precision in the simple threshold estimates (Lisovski et al.2012)–
did not coincide with periods in which the Sanderlings visited
the five important foraging or resting areas. The two methods
might thus yield very different results in other species, especially
in those using more vegetated or heterogeneous habitats. Addi-
tionally the MCMC path estimates feature important advantages
over the simple threshold estimates in that the Bayesian method
provides a framework that enables additional information, such
as species-specific habitat preferences, movement behaviour
and home-ranges, to be incorporated. This greatly reduces the
probability of erroneous, incompatible location estimates. More-
over, the method allows estimation of confidence intervals for
location estimates that reflect the quality of the data. Finally,
the method permits the estimation of a continuous path, whereby
each observation is used and evaluated in relation to all other
observations, rather than the rather arbitrary qualification of
each position in isolation and hence the potential omission or
inclusion of positions derived by simple threshold estimates
(Sumner et al.2009).
Our template-fit analysis allowed for the breeding areas of
the tracked Sanderlings to be identified (Fig. 1). Supporting the
analysis, the 100%- and 99%-likelihood contour lines for the
individual breeding areas showed identifiable and concurrent
minima on or close to locations on land (for more details, see
supplementary material S2). Six individuals migrated and prob-
ably bred on the New Siberian Islands, well known as an area of
breeding shorebirds, including Sanderlings (Lappo et al.2012).
The breeding locations of two individuals were estimated to be
just below the New Siberian Islands in the coastal Siberian
mainland. Although Sanderlings are not recorded as breeding in
this area (Lappo et al.2012), it should to be noted that based on
our template-fit analysis, locations on the Islands are equally
likely and are well within the latitudinal accuracy of the position
estimates. The location estimates for the remaining five indivi-
duals were west of the New Siberian Islands, as far as the eastern
part of Taimyr Peninsula (Fig. 2). Both the eastern coastal
areas of Taimyr Peninsula and the Lena Delta are confirmed
breeding areas of Sanderlings (Lappo et al.2012).
The rapid and highly-synchronised northward migration of
Sanderlings (Fig. 3) concurs with other observations on migratory
birds (reviewed by Nilsson et al.2013). In fact, the tracked
population spent twice as long on southward migration as north-
ward migration. Such patterns are thought to be related to high
selection pressure on a timely arrival at the breeding grounds (e.g.
McNamara et al.1998; Kokko 1999; Both and Visser 2001)
with northward migration therefore considered a period of con-
siderable time constraint. Similarly, the migratory connectivity
within the population varied over time and between the two
migration legs. As a result of the highly synchronised northward
migration, the area used by the population shrank by an order
of magnitude once all individuals arrived on the central Asian
coastline (Fig. 4b). In contrast, spatial connectivity was very low
during southward migration, to the extent that as some Sander-
lings arrived at wintering sites in South Australia others had
only just arrived at the Yellow Sea, lagging ~9000 km behind
(Fig. 4c).
We observed no differences between males and females in
their timing of northward migration. However, the schedule of
departure from the breeding grounds seemed to be sex-specific.
All males left within a short time period (within 14 days of
one another), whereas all but one female left later, after the last
male, and with considerably more variation between individuals
(Fig. 2c). Although very little is known about sex-specific
parental investment in Sanderlings, Tomkovich and Soloviev
(2001) and Reneerkens et al.(2014) observed that only single
birds cared for hatchlings and that more males attended broods
from earlier clutches, whereas females predominantly cared for
late clutches. This could explain our observed differences in
departure dates between the sexes and suggests that some
females may have had two clutches, as observed in other areas
(Reneerkens et al.2009), with females caring for the second
clutch.
During northward migration, Sanderlings made an initial,
long, migratory flight across the Equator, after which they all
made three to five stops before crossing the Arctic Circle and
arriving at their breeding grounds. In contrast, most Sanderlings
used as many as six or seven stops during southward migration.
This hopping between sites during southward migration and the
latter stage of northward migration is similar to that observed in
Ruddy Turnstones (Minton et al.2010) but contrasts with what
we know from tracking studies of other species within the EAAF
(Driscoll and Ueta 2002; Battley et al.2012; Minton et al.
2010,2013). These other species (Bar-tailed Godwit (Limosa
lapponica), Greater Sand Plover (Chardrius leschenaultia) and
Eastern Curlew (Numenius madagascariensis)) use one or two
major stopover sites during both northward and southward
migrations. Sanderlings and Ruddy Turnstones are generalist
feeders (Piersma et al.1996) and may thus be less restricted in
their habitat use than the other species. Alternatively, the rather
high number of stops observed in Sanderlings and Ruddy Turn-
stones may be related to their body size (cf. Piersma 1987;
Sanderling migration in the EAAF Emu 175
Warnock 2010), both species being among the smallest species
tracked within the EAAF thus far.
Conclusions
Given the apparent value of integrating tracking studies with
existing leg-flag resightings and count data to identify crucial
areas for conservation, we urgently require more detailed indi-
vidual migration tracks for the entire range of migratory shorebird
species and populations using the EAAF. Study designs should
emphasise acquisition of sufficient individual tracks to infer
distributions both within and between populations throughout
the migration period. Increased understanding of how species
and populations use the network of sites along the Flyway
would also assist predictions of how shorebirds are likely to
respond to the rapid changes to their habitats within the Flyway.
Tracking studies, combined with the systematic monitoring of
the population through marking, resighting and counting, there-
fore form an essential part of the empirical research fundamental
to conserving the many threatened migratory populations in the
EAAF and elsewhere in the world.
Acknowledgements
We would like to thank the late Ren de Garais who (together with Ian
D. Steward) first drew our attention to the presence of Sanderlingswith
leg-flags on the shores of the south-eastern coast of South Australia. We
also thank all members of the Victorian Wader Study Group (VWSG) for
deploying and retrieving geolocators from Sanderlings. We specifically
thank Roger Standen for providing the leg-flag resightings of Sanderlings,
and Simon Wotherspoon and Michael Sumner for their work on the SGAT
software (Wotherspoon et al.2013) and their support in the analyses. Heiko
Schmaljohann and Theunis Piersma provided valuable comments on a
former draft of the manuscript. Several individuals and organisations,
notably the Norman Wettenhall Foundation, provided funding to the VWSG
to conduct this geolocator project. Friends of Shorebirds South East also
raised significant funding to pay for geolocators. Banding permits were
supplied by the Australian Bird and Bat Banding Scheme, Canberra. Animal
ethics (D0001404067) and state approvals to undertake scientific research
(M23554-24) were provided by the South Australian authorities.
References
Alerstam, T., Hedenström, A., and Åkesson, S. (2003). Long-distance
migration: evolution and determinants. Oikos 103, 247–260. doi:10.1034/
j.1600-0706.2003.12559.x
Amano, T., Szekely, T., Koyama, K., Amano, H., and Sutherland, W. J.
(2010). A framework for monitoring the status of populations: an example
from wader populations in the East Asian–Australasian Flyway. Biolog-
ical Conservation 143, 2238–2247. doi:10.1016/j.biocon.2010.06.010
Bamford, M., Watkins, D., Bancroft, W., Tischler, G., and Wahl, J. (2008).
Migratory shorebirds of the East Asian–Australasian Flyway: population
estimates and internationally important sites. Available at http://www.
environment.gov.au/resource/migratory-shorebirds-east-asian-australa-
sian-flyway-population-estimates-and [Verified 8 February 2016].
Battley, P. F., Warnock, N., Tibbitts, T. L., Gill, R. E., Piersma, T., Hassell,
C. J., Douglas, D. C., Mulcahy, D. M., Gartrell, B. D., Schuckard, R.,
Melville, D. S., and Riegen, A. C. (2012). Contrasting extreme long-
distance migration patterns in Bar-tailed Godwits Limosa lapponica.
Journal of Avian Biology 43,21–32. doi:10.1111/j.1600-048X.2011.
05473.x
Bauer, S.,Lisovski, S., andHahn, S. (2015). Timing is crucial for consequences
of migratory connectivity. Oikos [Online]. doi:10.1111/oik.02706
Both, C., and Visser, M. E. (2001). Adjustment to climate change is con-
strained by arrival date in a long-distance migrant bird. Nature 411,
296–298. doi:10.1038/35077063
Bridge, E. S., Thorup, K., Bowlin, M. S., Chilson, P. B., Diehl, R. H., Fleron,
R. W., Hartl, P., Kays, R., Kelly, J. F., Robinson, W. D., and Wikelski, M.
(2011). Technology on the move: recent and forthcoming innovations
for tracking migratory birds. Bioscience 61, 689–698. doi:10.1525/
bio.2011.61.9.7
Calenge, C. (2006). The package ‘adehabitat’for the R software: a tool for
the analysis of space and habitat use by animals. Ecological Modelling
197, 516–519. doi:10.1016/j.ecolmodel.2006.03.017
Colwell, M. A. (2010). ‘Shorebird Ecology, Conservation and Manage-
ment.’(University of California Press: Los Angeles, CA.)
Conklin, J. R., Battley, P. F., Potter, M. A., and Fox, J. W. (2010). Breeding
latitude drives individual schedules in a trans-hemispheric migrant
bird. Nature Communications 1,67.
Dingle, H., and Drake, V. A. (2007). What is migration? Bioscience 57,
113–121. doi:10.1641/B570206
Driscoll, P. V., and Ueta, M. (2002). The migration route and behaviour
of Eastern Curlews Numenius madagascariensis. Ibis 144,E119–E130.
doi:10.1046/j.1474-919X.2002.00081.x
Fauchald, P., and Tveraa, T. (2003). Using first-passage time in the analysis
of area-restricted search and habitat selection. Ecology 84, 282–288.
doi:10.1890/0012-9658(2003)084[0282:UFPTIT]2.0.CO;2
Gill, R. E., Tibbitts, T. L., Douglas, D. C., Handel, C. M., Mulcahy, D. M.,
Gottschalck, J. C., Warnock, N., McCaffery, B. J., Battley, P. F., and
Piersma, T. (2009). Extreme endurance flights by landbirds crossing
the Pacific Ocean: ecological corridor rather than barrier? Proceedings
of the Royal Society B: Biological Sciences 276, 447–457. doi:10.1098/
rspb.2008.1142
Griffiths, R., Double, M. C., Orr, K., and Dawson, R. J. G. (1998). A DNA
test to sex most birds. Molecular Ecology 7, 1071–1075. doi:10.1046/
j.1365-294x.1998.00389.x
Harrington, B. A. (2001). Red Knot (Calidris canutus). In ‘The Birds of
North America’, no. 563. (Eds A. Poole and F. Gill.) pp. 1–32. (The Birds
of North America, Inc.: Philadelphia, PA.)
Holdo, R. M., and Roach, R. R. (2013). Inferring animal population distribu-
tions from individual tracking data: theoretical insights and potential
pitfalls. Journal of Animal Ecology 82, 175–181. doi:10.1111/j.1365-
2656.2012.02031.x
Kirby, J. S., Stattersfield, A. J., Butchart, S. H. M., Evans, M. I., Grimmett,
R. F. A., Jones, V. R., O’Sullivan, J., Tucker,G. M., and Newton, I. (2008).
Key conservation issues for migratory land- and waterbird species on the
world’s major flyways. Bird Conservation International 18, S49–S73.
doi:10.1017/S0959270908000439
Kokko, H. (1999). Competition for early arrival in migratory birds. Journal of
Animal Ecology 68, 940–950. doi:10.1046/j.1365-2656.1999.00343.x
Lanting, F. (1984). Sanderlings, globe trotting shorebirds of the Pacific.
Pacific Discovery 37,9–14.
Lappo, E. G., Tomkovich, P. S., and Syroechkovskiy, E. (2012). ‘Atlas of
Breeding Waders in the Russian Arctic.’(Institute of Geography, Russian
Academy of Sciences: Moscow, Russia.)
Lisovski, S., and Hahn, S. (2012). GeoLight –processing and analysing light-
based geolocator data in R. Methods in Ecology and Evolution 3,
1055–1059. doi:10.1111/j.2041-210X.2012.00248.x
Lisovski, S., Hewson, C. M., Klaassen, R. H. G., Korner-Nievergelt, F.,
Kristensen, M. W., and Hahn, S. (2012). Geolocation by light: accuracy
and precision affected by environmental factors. Methods in Ecology and
Evolution 3, 603–612. doi:10.1111/j.2041-210X.2012.00185.x
MacWhirter, B. P., Austin-Smith, P., Jr, and Kroodsma, D. E. (2002). Sand-
erling (Calidris alba). In ‘The Birds of North America’, no. 653. (Eds
A. Poole and F. Gill.) (The Birds of North America Online: Ithaca, NY.)
McKinnon, L., Smith, P. A., Nol, E., Martin, J. L., Doyle, F. I., Abraham, K. F.,
Gilchrist, H. G., Morrison, R. I. G., and Bety, J. (2010). Lower predation
176 Emu S. Lisovski et al.
risk for migratory birds at high latitudes. Science 327, 326–327.
doi:10.1126/science.1183010
McNamara, J. M., Welham, R. K., and Houston, A. I. (1998). The timing of
migration within the context of an annual routine. Journal of Avian
Biology 29, 416–423. doi:10.2307/3677160
Minton, C., Gosbell, K., Johns, P., Fox, J. W., and Afanasyev, V. (2010).
Initial results from light level geolocator trials on Ruddy Turnstone
Arenaria interpres reveal unexpected migration route. Wader Study
Group Bulletin 117,9–14.
Minton, C., Wahl, J., Gibbs, H., Jessop, R., Hassell, C., and Boyle, A. (2011).
Recoveries and flag sightings of waders which spend the non-breeding
season in Australia. Stilt 50,17–43.
Minton, C., Gosbell, K., Johns, P., Christie, M., Klaassen, M., Hassell, C.,
Boyle, A., Jessop, R., and Fox, J. (2013). New insights from geolocators
deployed on waders in Australia. Wader Study Group Bulletin 120,37–46.
Niles, L. J., Burger, J., Porter, R. R., Dey, A. D., Minton, C. D. T., Gonzales,
P. M., Baker, A. J., Fox, J. W., and Gordon, C. (2010). First results using
light level geolocators to track Red Knots in the Western Hemisphere
show rapid and long intercontinental flights and new details of migration
pathways. Wader Study Group Bulletin 117, 123–130.
Nilsson, C., Klaassen, R. H. G., and Alerstam, T. (2013). Differences in speed
and duration of bird migration between spring and autumn. American
Naturalist 181, 837–845. doi:10.1086/670335
Pennycuick, C. J., Åkesson, S., and Hedenström, A. (2013). Air speeds of
migrating birds observed by ornithodolite and compared with predictions
from flight theory. Journal of the Royal Society: Interface 10, 20130419.
doi:10.1098/rsif.2013.0419
Piersma, T. (1987). Hop, skip or jump? Constraints on migration of Arctic
waders by feeding, fattening, and flight speed. Limosa 60, 185–194.
Piersma, T., van Gils, J., and Wiersma, P. (1996) Family Scolopacidae
(snipes, sandpipers and phalaropes). In ‘Handbook of the Birds of the
World. Vol. 3: Hoatzin to Auks.’(Eds J. del Hoyo, A. Elliott and
J. Sargatal) pp. 444–533. (Lynx Edicions: Barcelona.)
Piersma, T., and Baker, A. J. (2000). ‘Life history characteristics and the
conservation of migratory shorebirds. In ‘Behaviour and Conservation’.
(Eds L. M. Gosling and W. J. Sutherland). pp. 104–124. (Cambridge
University Press: Cambridge, UK.)
Ramsar Convention Bureau (1971). Convention on wetlands of international
importance especially as waterfowl habitat 1971. Available at http://
portal.unesco.org/en/ev.php-URL_ID=15398&URL_DO=DO_TOPIC&
URL_SECTION=201.html [Verified 8 February 2016].
Reneerkens, J., Benhoussa, A., Boland, H., Collier, M., Grond, K., Gunther,
K., Hallgrimsson, G.T., Hansen, J., Meissner, W., de Meulenaer, B.,
Ntiamoa-Baidu, Y., Piersma, T., Poot, M., van Roomen, M., Summers, R.
W., Tomkovich, P.S., and Underhill, L.G. (2009). Sanderling using
African–Eurasian flyways: a review of current knowledge. Wader Study
Group Bulletin 116,2–20.
Reneerkens, J., van Veelen, P., van der Velde, M., Luttikhuizen, P., and
Piersma, T. (2014). Within-population variation in mating system and
parental care patterns in the Sanderling (Calidris alba) in northeast
Greenland. Auk 131, 235–247. doi:10.1642/AUK-13-247.1
Runge, C. A., Martin, T. G., Possingham, H. P., Willis, S. G., and Fuller, R. A.
(2014). Conserving mobile species. Frontiers in Ecology and the
Environment 12, 395–402. doi:10.1890/130237
Sumner, M. D., Wotherspoon, S. J., and Hindell, M. A. (2009). Bayesian
estimation of animal movement from archival and satellite tags. PLoS
One 4, e7324. doi:10.1371/journal.pone.0007324
Tomkovich, P. S., and Soloviev, M. Y. (2001). Social organisation
of Sanderlings breeding at northern Taimyr, Siberia. Ornithologia
(Moscow) 29, 125–136.
Warnock, N. (2010). Stopping vs. staging: the difference between a hop
and a jump. Journal of Avian Biology 41, 621–626. doi:10.1111/j.1600-
048X.2010.05155.x
Wilcove, D. S., and Wikelski, M. (2008). Going, going, gone: is animal
migration disappearing? PLoS Biology 6, e188. doi:10.1371/journal.pbio.
0060188
Wotherspoon, S.J., Sumner, M.D., and Lisovski, S. (2013). R package
SGAT: solar/satellite geolocation for animal tracking. GitHub repository,
available at https://github.com/SWotherspoon/SGAT [Verified 15
January 2016].
Sanderling migration in the EAAF Emu 177
www.publish.csiro.au/journals/emu
