Consistent variation in individual migration strategies of

MARINE ECOLOGY PROGRESS SERIESMar Ecol Prog Ser

Vol 578 213ndash225 2017httpsdoiorg103354meps11932

Published August 31sect

INTRODUCTION

Migratory seabirds spend much of the year at seafar from their breeding grounds yet have tradition-ally been studied much more intensively during thebreeding than the non-breeding season However agrowing body of literature demonstrates that thesetemporally and often geographically distinct periodsof the annual cycle are inextricably linked (Harrisonet al 2011) This bias in research towards studiesfocussing on the breeding season limits our under-

standing of the ecology of many species and howindividuals and populations are affected by majorstressors such as rapid environmental change (Aringdahlet al 2006 Small-Lorenz et al 2013)

Recent advances in tracking technologies (Bridgeet al 2011) have facilitated numerous studies invol -ving tracking individuals over an extended periodof time This development allows investigation ofspatiotemporal consistency in migration strategieswithin and among individuals and populations (egPhillips et al 2005 Dias et al 2011) Although the

copy The authors 2017 Open Access under Creative Commons byAttribution Licence Use distribution and reproduction are un -restricted Authors and original publication must be credited

Publisher Inter-Research middot wwwint-rescom

Corresponding author simeonlisovskigmailcomsectAdvance View was available online January 5 2017

Consistent variation in individual migrationstrategies of brown skuas

Johannes Krietsch1 Steffen Hahn2 Matthias Kopp1 Richard A Phillips3 Hans-Ulrich Peter1 Simeon Lisovski45

1Friedrich Schiller University Jena Institute of Ecology Polar and Bird Ecology Group 07743 Jena Germany2Swiss Ornithological Institute Department of Bird Migration 6204 Sempach Switzerland

3British Antarctic Survey Natural Environment Research Council Cambridge CB3 0ET UK4Deakin University School of Life and Environmental Sciences Centre for Integrative Ecology Geelong Victoria 3220

Australia5University of California Department of Neurobiology Physiology and Behavior One Shields Avenue Davis

California 95616 USA

ABSTRACT Seabirds show remarkable variability in migration strategies among individuals andpopulations In this study we analysed 47 migrations of 28 brown skuas Catharacta antarcticalonnbergi breeding on King George Island in the Maritime Antarctic Brown skuas from this pop-ulation used a large area during the non-breeding period north of 55deg S including parts of thePatagonian Shelf Argentine Basin and South Brazil Shelf areas which are characterised by highlevels of marine productivity However individual birds utilised only a subset of these areasadopting 1 of 4 distinct migration strategies to which they were highly faithful between years andshowed high repeatability in departure and arrival dates at the breeding ground Although theyspent the majority of the non-breeding season within a particular region almost all individualsused the same area in the late winter exploiting its seasonal peak in productivity Overall theseresults indicate consistent individual variation in migration strategies that may reflect a combina-tion of genetic control and individual experience but with considerable flexibility to shift distribu-tion in response to prevailing environmental conditions

KEY WORDS Catharacta antarctica lonnbergi middot Seabird ecology middot Light-level geolocation middotNon-breeding distribution middot Individual consistency middot Ocean primary productivity middot Migratory connectivity

Contribution to the Theme Section lsquoIndividual varability in seabird foraging and migrationrsquo OPENPEN ACCESSCCESS

Mar Ecol Prog Ser 578 213ndash225 2017

general direction of migration seems to be largelydetermined by genetics or in some species culturalinheritance (Berthold 2001 Mueller et al 2013) move-ment patterns of individuals and populations alsorespond to factors such as food availability (Shealer2002 Karnovsky et al 2003) However prey avail-ability in marine ecosystems is generally charac-terised by varying degrees of temporal and spatialpredictability (Weimerskirch 2007) and hence it can-not be assumed that all areas will be favourable for aparticular species in every year This could ultimatelylead to variation in individual migration patterns ofseabirds at some spatial scale across years Indeedalthough most species show high individual consis-tency in non-breeding destinations at a large spatialscale (Phillips et al 2005 Fifield et al 2014 Muumlller etal 2014) there are exceptions in addition in almostall species there is extensive variation both amongand within individuals in routes use of staging areasand timing (Quillfeldt et al 2010 Dias et al 2011McFarlane Tranquilla et al 2014)

In this study we analysed the migration strategiesof individual brown skuas Catharacta antarcticalonnbergi breeding on the South Shetland Islands inthe Maritime Antarctic Migration routes and non-breeding areas were derived using light-level geolo-cators (also termed global location sensors or GLSloggers) and our dataset included repeated tracksfrom individuals over 2 to 3 yr Brown skuas are long-lived and highly opportunistic top predators with acircumpolar breeding distribution on subantarcticislands and the Antarctic Peninsula (Furness 1987Ritz et al 2008) To date the only detailed distribu-tion data available for brown skuas during the non-breeding period are for birds from the population atSouth Georgia migrating to waters between thenorthern extent of the Subtropical Front and thesouthern boundary of the Antarctic CircumpolarCurrent and between the Argentine and Agulhas(Phillips et al 2007 Carneiro et al 2016) In contrastthe limited data for the closely related Falklandskuas C a antarctica tracked in different years sug-gest a non-breeding distribution mainly in sub-antarctic waters around the central Patagonian shelf-break (Phillips et al 2007)

Given the lack of knowledge of the non-breedingranges of brown skuas from the South ShetlandIslands the adjacent southern population and theinclusion in our dataset of repeated migration tracksfrom the same individuals in multiple years the aimsof the study were 2-fold Firstly we aimed to revealthe spatiotemporal non-breeding distribution of thetracked population the degree of variation among

individuals and the effects of sex year and previousbreeding performance Secondly we aimed to quan-tify individual consistency in annual migration strate-gies In addition we used activity (immersion) datarecorded by the loggers and remotely sensed dataon net primary production to examine the correlationbetween this proxy for food availability and themovement and activity patterns of individual birds

MATERIALS AND METHODS

Logger deployment and retrieval

Fieldwork was carried out on adult brown skuasat King George Island (Fildes Peninsula 62deg 19rsquo S58deg 95rsquo W) in the Maritime Antarctic A total of 46geolocator-immersion loggers were deployed on 33individuals (which at the time were of unknown sex)over the course of 3 breeding seasons (20062007 to2008 2009) Three types of loggers (manufactured bythe British Antarctic Survey) were used in this studyMK5 (n = 20) MK9 (n = 22) and MK15 (n = 4) Totalweights were 64 53 and 53 g respectively whichincluded the device aluminium ring and cable tiesused for attachment (together with metal ring of 3 gused for identification corresponding to ~06 of themean body mass) Besides recording light intensityover time all loggers tested for saltwater immersionevery 3 s and stored the sum of positive tests at10 min intervals resulting in values between 0(entirely dry) and 200 (entirely wet) Birds wererecaptured in the subsequent season and in 11 casesthe loggers were replaced by a new device thesedevices and others from the initial deployments wereretrieved in the third season A blood sample wastaken from each bird (~50 microl) stored at minus20degC andlater used to determine sex from DNA (Fridolfsson ampEllegren 1999) All individuals were monitored regu-larly in the pre- and post-migratory season (fromearly December to March) to determine breedingstatus and performance (ie success vs failure)

Departure and arrival date

Since brown skuas switch from a predominantlyterrestrial lifestyle to almost exclusively marine habi-tat after leaving the breeding site (Phillips et al2007) departure and arrival dates at King GeorgeIsland were identified by visual inspection of immer-sion data For individuals lacking immersion databecause of logger malfunction these dates were

214

Krietsch et al Brown skua migration strategies

identified from the length and frequency of shadingof the light sensor during daylight which is substan-tially higher when skuas spend time sitting on land atthe breeding ground The duration of the non-breed-ing period was based on departure and arrival datesSome individuals also went on a pre-laying exoduswithin a few weeks of first return to the colony

Movement pathway analysis

Positions during the non-breeding season wereestimated from raw light intensity data using thethreshold method (Lisovski et al 2012) Twilightevents (ie sunrise and sunset transitions) weredefined using the R package lsquoBAStagrsquo (Wotherspoonet al 2013a) based on a light intensity threshold of25 Twilight times that were clearly suspect becauseof shading of the sensor (ie gt30 min difference fromthe previous or subsequent day) were discarded andthe time interpolated with respect to the surroundingtwilights This approach was applied to between 5and 10 of all twilights during the annual migrationof each individual Locations from the breedingperiod were excluded from subsequent analyses Weused a Bayesian framework to refine the initialrough positions estimated from the threshold methodand to derive uncertainty estimates The R packagelsquoSGATrsquo (Wotherspoon et al 2013b) uses MarkovChain Monte Carlo (MCMC) simulations allowingthe incorporation of a spatial probability mask priordefinition of the error distribution of twilight events(twilight model) and a flight speed distribution torefine location estimates (for detailed information seeSumner et al 2009 and Lisovski et al 2016) The twi-light model should reflect the expected error indetecting the real time of sunrise and sunset Sincebrown skuas spend a substantial amount of time sit-ting at the breeding site which obscures the lightsensor we could not use twilight times from a knownlocation (ie breeding site) to parameterise the twi-light model We therefore used a rather conservativeprior (log-normal distribution meanlog = 2 sdlog =12) describing a large variation in the discrepancybetween the real and recorded twilight events Themovement behaviour was modelled assuming thatover the course of the non-breeding periods brownskuas are sedentary for the majority of the time (highlikelihood of very slow movement speeds) whileallowing for occasional fast movements (~80 km hminus1)during migration (gamma distribution shape = 07scale = 005) The spatial mask was based on theassumption that the tracked individuals avoid land

(10 times lower probability of occurrence on landcompared to sea) and that the spatial range wasbetween 10 and 70deg S and 85 and 20deg W (based onlocations derived using the threshold method)

The threshold method requires a zenith angle toestimate locations which is usually derived usinglight intensity recordings from tracked animals ofknown distributions or fixed loggers As these datawere unreliable or unavailable for this study weused an alternative approach the lsquoHill-Ekstrom cali-brationrsquo (see Lisovski et al 2012) to estimate the rightzenith angle for each annual migration track Firstfor all tracks initial locations were estimated using azenith angle of 95deg (ie sun elevation angle of minus5deg)Next using the initial locations a set of MCMC sim-ulations (drawing 2400 samples) was performedusing a range of zenith angles between 94 and 96degThe median path for each zenith angle and a total of2400 chains were then calculated We then choosethe zenith angle that minimised the variance in lati-tude estimates during periods when the tracked birdswere largely sedentary within their non-breedingrange The derived zenith angles varied betweenindividual tracks and ranged from 945 to 957deg

Using those zenith angles a complete MCMC sim-ulation was performed on each individual annualtrack An initial 2000 samples were drawn and dis-carded to allow for both burn-in and tuning of theproposal distribution ie to find an initial path thatmatches the model assumptions A final 4000 sam-ples were than drawn to describe the posterior distri-bution Convergence of chains ie whether 2 inde-pendent simulations produce the same result wasevaluated by visual inspection by comparing themedian tracks The final run provided 4000 chainsof possible migration pathways that satisfied thedefined sunrise and sunset times their error distribu-tion the movement behaviour and the spatial maskThe set of chains were used to generate time-spentmaps illustrating the relative probability distributionof an individual at a given time or period and to cal-culate most likely tracks

Activity analysis

Saltwater immersion data recorded by the loggerwere used to determine activity patterns of brownskuas using the web-based program Actavenet(Mattern et al 2015) Values of 0 (entirely dry) 1 to199 and 200 (entirely wet) in each 10 min periodwere categorised as either lsquoflightrsquo (as skuas remain atsea during the non-breeding period) lsquoforagingrsquo or

215


lsquositting on waterrsquo respectively assigned to daylightor darkness periods based on nautical twilight hoursand summarised accordingly (see Mattern et al2015) Although some intermediate values (from 1 to199) will reflect non-foraging behaviour in generalthis categorisation is assumed to provide a reason-able indication of foraging activity among seabirds(McKnight et al 2011 Cherel et al 2016)

Spatial data analysis

We used R (R Core Team 2015) to manipulate andanalyse all data A Lambert Azimuthal Equal Areaprojection was used for all spatial analyses and map-ping Due to the large error distribution of the twi-light times the MCMC simulation did not performwell in correcting location estimates during theperiod of the equinox (1 March to 22 April and 22August to 9 October) and these locations were there-fore excluded from all analyses

Movement patterns within the non-breeding rangewere analysed in the context of the Marine Ecore-gions of the World (MEOW) biogeographic areas ofrelatively homogenous and distinct species composi-tion (for details see Spalding et al 2007) HoweverMEOW only characterises costal and shelf areas andwe therefore added the lsquoArgentine Basinrsquo to be ableto categorise the entire non-breeding range of thetracked brown skuas For each position on the indi-vidual median tracks (ie the most likely track) weextracted the corresponding MEOW To quantify therelative use of the various MEOW by each individualthe proportions of time spent inside each regionbetween 22 April and 22 August was calculatedThese proportions were used to group individualtracks based on the similarity in the use of eachMEOW using a cluster analysis with Euclidean dis-tance (R package lsquoveganrsquo Oksanen et al 2015) Toexclude individual effects the analysis was initiallyperformed using the first track of each individualonly Subsequently and to evaluate the robustness ofthe groups all 47 annual tracks were analysed to -gether followed by a repeated analysis using Wardrsquosmethod (Oksanen et al 2015)

The relative probability distributions (ie time-spent maps) between 22 April and 22 August wereused to investigate the spatial overlap of movementpaths among and within individuals Each individualrelative probability distribution (DXY) correspondedto a raster with XY grid cells and a resolution of 393times 556 km The values were normalised such that ΣxΣy

DXY = 1 The degree of overlap (O) between 2 tracks

(a and b) was defined as the sum of the minimal valueover all shared (overlapping) grid cells according to

Oab = ΣxΣy min(DXYa | DXYb) (1)

This calculation results in 0 if the 2 tracks share nocommon grid cell and in 1 if the 2 probability distri-butions were 100 identical All combinations of the47 annual tracks were calculated The resultingdegrees of overlap were arcsine square root trans-formed to meet statistical assumptions

To reveal temporal trends in marine productivity(as a proxy of the seasonal dynamics of food avail-ability at a mesoscale level) and to test for relation-ships with individual distributions timing of move-ments and activity patterns of the tracked skuas wedownloaded net primary production data (NPP) fromwww science oregonstate edu ocean productivityindex php (accessed 5 August 2014) in 8 d intervalsand a resolution of 017deg Individual brown skuaprobability distributions were pooled into 8 d inter-vals to match NPP data format and normalised suchthat ΣxΣy DXY = 1 If the matched NPP data had miss-ing values corresponding to gt50 of the probabilitydistribution the 8 d interval was excluded from theanalysis The relative probability distributions werethen used to weight the NPP data of each correspon-ding grid cell according to

NPPsum = ΣxΣy (DXY times NPPXY) (2)

Statistical analysis

We used linear mixed-effect models (R packagelsquolme4rsquo Bates et al 2014) to quantify the effects of sexbreeding performance (ie successful [at least onechick fledged] or unsuccessful in the current season)and migration strategy on the subsequent timing ofmigration (departure date arrival date duration) andindividual activity patterns (foraging activity num-ber of dry bouts and duration of dry bouts) To avoidpseudoreplication individual identity was includedas a random intercept Two individuals with behav-iours very different to the others were excluded fromspecific analysis ID 108276 departed exceptionallyearly (15 January) in 2008 and was excluded fromthe analysis of departure date and duration and ID139514 spend several (dry) nights on a vessel or onland biasing the activity data To correct for tempo-ral autocorrelation in the model testing for an effectof prey predictability (ie NPP) on activity patternthe random effect was specified as days since thestart of the non-breeding season each year (hereafter

216


lsquoday of the yearrsquo) nested within individuals p-valueswere calculated using the lsquolmerTestrsquo package (Kuz -netsova et al 2015) A linear mixed-effect model withthe binary fixed factors (1 as lsquosamersquo and 0 as lsquodiffer-entrsquo) lsquosame sexrsquo lsquosame yearrsquo lsquosame breeding per-formancersquo and lsquosame individualrsquo and the random fac-tor lsquoindividualrsquo was used to test for differences amongand within groups in the spatial overlaps of the prob-ability distributions

Individual repeatability ie the intraclass correla-tion coefficient which allows the quantification ofvariance among and within individuals (Lessells ampBoag 1987) was calculated for timing of migrationand mean activity metrics obtained in different yearsLinear mixed-effect models with individual as ran-dom effect and year as fixed factor were fitted foreach sex Due to the small sample size models fittedto the activity metrics were not separated by sexand sex was instead included as a fixed factor Theadjusted repeatability value corresponding confi-dence interval and p-value were calculated using theR package lsquorptRrsquo (Nakagawa amp Schielzeth 2010)

RESULTS

Logger retrieval details

DNA sexing revealed that the loggers were de -ployed on 33 females and 13 males A total of 42(91 31 females 11 males) of the loggers wereretrieved which provided data on 47 annual tracks of28 individuals (20 females 8 males) between 2007and 2010 including migrations of the same indivi -duals over 2 (n = 13) or 3 (n = 3) non-breeding sea-sons Additionally saltwater immersion data were re -corded for 35 of the 47 annual tracks

Spatiotemporal distribution

During the non-breeding period the trackedbrown skuas were widely distributed north of theirbreeding sites over parts of the Patagonian Shelf theArgentine Basin and to a lesser extent the South-ern Brazil Shelf (Fig 1a) This distribution includesregions with heterogeneous levels of productivitysubantarctic mixed subantarcticminussubtropical and sub -tropical and open shelf waters The core area of thedistribution overlapped with the Patagonian shelf-break front the confluence zone of the Falkland andBrazil currents and offshore of the Rio de la Plataestuary

Single individuals used distinct portions of theentire area revealing characteristic spatiotemporalpatterns (Fig 2) Based on the time spent in eachMEOW the annual tracks could be grouped in 4 mi-gration strategies Applying cluster analyses with dif-ferent linking methods on all annual tracks or only

217

40degW70degW

STF

SAF

PF

60degW 50degW 60degW 50degW

4

6

8

log NPPMar-Aug Sep-Oct

a

b c

lt20

20minus40

40minus60

60minus80

80minus90

90minus95

Density contour ()

30degS

50degS

40degS

60degS

Fig 1 (a) Density distribution of 28 brown skuas Catharactaantarctica lonnbergi from King George Island (black star)during the non-breeding period from 2007 to 2010 Blacklines approximate locations of the Subtropical Front (STF)Subantarctic Front (SAF) and Polar Front (PF) based on Orsiet al (1995) grey line 500 m bathymetric contour indica-ting the Patagonian shelf-break (b) Mean net primary pro-duction (NPP in mg C mminus2 dminus1) between the mean departuredate of brown skuas (17 March) and August from 2007 to2010 and (c) NPP (in mg C mminus2 dminus1) between September andthe mean arrival date of brown skuas (31 October) from 2007to 2010 both overlaid by the density distribution (contourlines 20 40 and 60) of brown skuas in the same period


the first from each individual distinguished up to5 groups However 3 groups were always identifiedas clearly distinct independent of the methods anddata background (1) a group of individuals thatutilised the lsquoArgentine Basinrsquo at the end of April andin May and for a shorter period the lsquoUruguayminusBuenos Aires Shelfrsquo which corresponds to the use ofthe BrazilminusFalklands confluence Subsequently at thebeginning of June these birds moved to the SouthernBrazil Shelf where they primarily utilised the lsquoRioGrandersquo and partly lsquoSoutheastern Brazilrsquo These an-nual tracks were assigned to the migration strategylsquoSouth Brazil Shelfrsquo (SouthBrazil) and consisted of 3individuals (11) (2) Individuals that mainly usedthe lsquoUruguayminusBuenos Aires Shelfrsquo between April and

September as well as the lsquoRio de la Platarsquo Thesetracks were assigned to the migration strategy lsquoRio dela Platarsquo (RioPlata) including tracks of 3 individuals(11) (3) Individuals grouped across the southernPatagonian Shelf eg the lsquoNorth Patagonian Gulfsrsquoand lsquoPatagonian Shelfrsquo between April and July to Au-gust These tracks were assigned to the migrationstrategy lsquoSouth Patagonian Shelfrsquo (South Pata) andconsisted of 4 individuals (14) The cluster analysiscould not clearly separate the remaining 2 groupsand assignment of tracks was dependent on methodand data background (4) Individuals that mainly usedthe lsquoArgentine Basinrsquo and the lsquoUruguayminusBuenos AiresShelfrsquo Most of these birds moved frequently betweenthese 2 major re gions and some were also distributed

218

123689123839

127377139674

108276

108294

139514

030210030256

030271

066607108271108280

108285108336

123836

127245138520

138530

139679139685156143156155157779162922

108272

108291

110388

Sou

thP

ata

Rio

Pla

taN

orth

Pat

aS

outh

Bra

zil

a

May Jun Jul Aug Sep Oct Nov Dec


Marine Ecoregions of the World (MEOW)

Southeastern BrazilRio GrandeUruguayminusBuenos Aires ShelfRio de la PlataNorth Patagonian GulfsPatagonian ShelfFalklandsArgentine Basin

Equinox

c

Sum

in M

EO

W

70degW 40degW

40degS

60degS

b

Fig 2 (a) Location of the most likely position within the Mar-ine Ecoregions of the World (MEOW following Spalding etal 2007 supplemented with the lsquoArgentine Basinrsquo to coverthe whole non-breeding range) of 28 brown skuas Cathar-acta antarctica lonnbergi from King George Island duringthe non-breeding period Brown skuas were tracked in 4 yr(2007 to 2010) 16 individuals in 2 or 3 consecutive years In-dividuals are grouped by 4 migration strategies based oncluster analysis including the proportions of annual tracksin the MEOW (b) Proportional distribution of all trackswithin the MEOW Note the use of different MEOW be-tween April and August whereas almost all individualsused the same region in October and November No reliablepositions could be calculated around the equinoxes (1 Marchto 22 April and 22 August to 9 October) Migration strate-gies lsquoSouth Brazil Shelfrsquo (SouthBrazil) lsquoNorth Pata gonianShelfArgentine Basinrsquo (NorthPata) lsquoRio de la Platarsquo (Rio-Plata) lsquoSouth Patagonian Shelfrsquo (SouthPata) (c) Extent of

the MEOW used by the tracked skuas


partly over the southern Patagonian Shelf Thesetracks of 18 individuals (64) were assigned to mi-gration strategy lsquoNorth Patagonian ShelfArgentineBasinrsquo (NorthPata)

In October the differences between individualstrategies diminished (Fig 2b) and almost all skuasmoved into a highly seasonal and productive areathe transition zone between the lsquoArgentine Basinrsquothe lsquoPatagonian Shelfrsquo and the lsquoFalklandsrsquo (Fig 1c)However 3 individuals (IDs 127245 108272 and156143) did not use this area during this period butrather were distributed east from the FalklandIslands over mixed water masses of the AntarcticPolar Front and the Subantarctic Front One otherindividual (ID 156143) travelled as far east as thewaters north of South Georgia

On 9 occasions 8 individuals (5 females 3 males)performed a pre-laying exodus These birds de -parted from King George Island after a median of 6 d(min = 1 d max = 36 d) and their first arrival dateswere on average 1 wk before the arrival of birds thatdid not make a pre-breeding exodus Seven individ-uals flew back to the north or east of the FalklandIslands whereas the others remained in the proxim-ity of King George Island or performed an 8 d trip tothe Drake Passage

The degree of overlap of the probability distribu-tions was significantly larger (22 plusmn 03 SE p lt 0001)within individuals than among individuals (Fig 3)Only 1 individual (ID 139679) showed an overlap ofjust 44 whereas the within-individual overlap forall the other tracked skuas was 60 to 95 This signif-icant overlap in movement paths was reflected in thevery high consistency within individuals with respectto their migration strategy (Fig 2) This was higher inindividuals using lsquoSouth Patagonian Shelfrsquo lsquoRio de laPlatarsquo and lsquoSouth Brazil Shelfrsquo than in individuals thatadopted the lsquoNorth Patagonian ShelfArgentine Basinrsquostrategy There was no significant effect of sex (16 plusmn10 SE p = 0117) year (03 plusmn 10 SE p = 0796) orprevious breeding performance (04 plusmn 10 SE p =0691) on the distribution of the tracked birds

Timing of migration

The mean departure date from King George Islandwas 17 March (plusmn12 d SD 25 February to 13 April n =28) Females departed around 2 to 3 wk earlier thanmales (1738 plusmn 275 d SE t = 632 p lt 0001 Fig 4a)lsquoSouthPatarsquo individuals departed significantly earlierthan lsquoNorthPatarsquo (177 plusmn 347 d SE t = 508 p lt 0001)lsquoRioPlatarsquo (110 plusmn 496 d SE t = 222 p = 003) and

219

100

75

50

25

0Among

Pro

bab

ility

dis

trib

utio

n ov

erla

p (

)individuals

Within individuals

oo

Fig 3 Overlap among and within individual probability dis-tributions of 28 brown skuas Catharacta antarctica lonn -bergi from King George Island during the non-breeding pe-riods from 2007 to 2010 including annual migrations of thesame individuals over 2 (n = 13) or 3 (n = 3) yr Line medianbox 25thndash75th percentiles whiskers 5thndash95th percentiles

circles outliers

oo

o

ooFemale

MaleDeparture | Arrival date

Fora

ging

act

ivtiy

(h d

ndash1)

2

a

b

6

10

Feb Apr Jun Aug Oct Dec

Fig 4 (a) Departure and return dates of brown skuas Cath ar -acta antarctica lonnbergi (20 females 8 males) from KingGeorge Island during 2007 to 2010 Boxplot limits as in Fig 3(b) Marine foraging activity (20 d moving average and 95CI) of 16 female (red) and 7 male (blue) brown skuasthroughout the years 2007 to 2010 Foraging activity was ap-proximated by saltwater immersion data with intermediate(wet and dry) 10 min intervals Note that brown skuas fedmainly on terrestrial prey during the breeding season andswitched to a pelagic distribution after they left the breeding

site Dashed lines median departure and return dates


lsquoSouthBrazilrsquo individuals (118 plusmn 450 d SE t = 26 p =001) Besides the significant differences betweenthese particular strategies departure date was notinfluenced by previous breeding performance (1 or 2chicks fledged vs unsuccessful 341 plusmn 239 d SE t =142 p = 016) or year (minus061 plusmn 121 d SE t = 05 p =061) The repeatability value of the departure datewas moderate in both sexes (Table 1)

Brown skuas returned to King George Islandaround 31 October (plusmn10 d 17 October to 23 Novem-ber n = 28) The arrival date did not differ significantlybetween sexes (146 plusmn 430 d SE t = 034 p = 073) butwas highly repeatable at the individual level (Table 1)and there were significant differences between years(2007minus2010 207 plusmn 080 d SE t = 257 p = 0016)

The mean (plusmnSD) duration of the non-breedingperiod was 228 d (plusmn17 d 197 to 272 d n = 28) andwas highly repeatable within individuals (Table 1)Males spent less time away from the breeding sitethan females (1837 plusmn 521 d t = 352 p = 0001) inde-pendent off year (253 plusmn 17 d t = 148 p = 014) andbreeding performance (604 plusmn 37 d t = 162 p =011) Additionally lsquoSouthPatarsquo individuals had a sig-nificantly shorter non-breeding period (202 plusmn 668 dp = 0006) than lsquoNorthPatarsquo individuals

Activity patterns

During the non-breeding period tracked brownskuas spent a large proportion of the day sitting on

water (females 57 plusmn 440 SD n = 16 males 5538 plusmn608 SD n = 7) and a much smaller proportion inflight (females 1471 plusmn 345 SD n = 16 males1246 plusmn 410 SD n = 7) The length of time cate-gorised as foraging was significantly higher in malesthan females (093 plusmn 030 h SE t = 306 p = 0006Table 1) and was concentrated mainly during theday (females 6721 plusmn 808 SD n = 16 males 6945plusmn 845 SD n = 7) rather than at night (females2234 plusmn 598 SD n = 16 males 2137 plusmn 470 SDn = 7) Foraging activity differed significantly be -tween years (minus020 plusmn 007 h SE t = 261 p = 0009)and over the non-breeding period (007 plusmn 0009 h SEt = 806 p lt 0001) but there was no significantrepeatability within individuals (Table 1) Within thenon-breeding period the foraging activity peaked inearly October (Fig 4b) before the brown skuasreturned to King George Island

The number of flight bouts per day did not differsignificantly between males (416 plusmn 102 SD) and females (386 plusmn 065 SD) (035 plusmn 024 SE t = 144 p =016) Day of the year had a significant effect (016 plusmn001 h SE t = 902 p lt 0001) on the number of flightbouts but there was no effect of year (minus007 plusmn 008 hSE t = 083 p = 040) However the number of flightbouts was not repeatable within individuals (Table 1)In contrast there was a significant effect of sex (fe-males 6353 plusmn 1904 min SD males 4597 plusmn 845 minSD difference 1510 plusmn 710 min SE t = 211 p = 0049)and year (minus581 plusmn 216 min SE t = 268 p = 0009) onthe duration of flight bouts which was independent of

220

Sex Nind Value plusmn SD NindNrep R Lower CI Upper CI p-value

Departure date Female 20 13 March plusmn 10 d 1123 0477 0009 0821 0002from breeding site Male 8 26 March plusmn 13 d 410 0490 0035 0747 lt0001

Arrival date at Female 20 31 October plusmn 8 d 1225 0871 0729 0995 lt0001breeding site Male 8 30 October plusmn 15 d 410 0972 0939 0991 lt0001

Duration of the non- Female 20 232 plusmn 12 d 1225 0814 0590 0937 lt0001breeding period Male 8 218 plusmn 24 d 410 0859 0542 0974 lt0001

Daily foraging Female 16 672 plusmn 075 716 0623 0080 0895 0066time in h dminus1 () (2800 plusmn 314)

Male 7 771 plusmn 080(3216 plusmn 334)

Daily flight bouts (n) Female 16 386 plusmn 065 716 0207 0000 0750 1000Male 7 416 plusmn 102

Flight bout duration Female 16 6353 plusmn 1904 716 0656 0178 0916 lt0001(min) Male 7 45 97 plusmn 845

Table 1 Migration characteristics (mean departure date arrival date and duration) and activity patterns (foraging activitynumber and duration of flight bouts) of brown skuas Catharacta antarctica lonnbergi during the non-breeding periods from2007 to 2010 and their individual repeatability (R lower and upper 95 CIs and p-values) Mean values were calculated us-ing only data from the first migration track from every individual (Nind) and repeatability values using repeated migrations

(Nrep) over 2 or 3 yr


day of the year and significantly repeatable within in-dividuals (Table 1) NPP and breeding performancehad no effect on the foraging activity or the numberand duration of flight bouts

DISCUSSION

During the non-breeding season the trackedbrown skuas from King George Island (MaritimeAntarctic) were widely distributed over the Patagon-ian Shelf and shelf-break and the Argentine Basinparticularly in the area of the BrazilminusFalklands Con-fluence The northern end of this range is substan-tially further north than the distribution indicated forthis species in Furness (1987) but more consistentwith subsequent at-sea observations (Olmos 2002)The use by some individuals of the Southern BrazilShelf contrasts the tracking data from brown skuas atSouth Georgia which mostly spent the non-breedingseason further south in the Argentine Basin (Phillipset al 2007 Carneiro et al 2016) Hence the distribu-tions of the 2 brown skua populations overlap onlyat the BrazilminusFalklands Confluence mdash and indeedthere seems to be greater overlap of the birds fromSouth Georgia with those of the closely-related Falk-land skua in the area of the southern Patagonianshelf-break (Phillips et al 2007) However given thefew Falkland skuas (n = 4) that have been trackedand the different study periods (2002) this conclu-sion should be viewed with some caution

Within the entire (population-level) non-breedingrange individuals were continuously distributedacross space suggesting a large contiguous area ofsuitable habitat (Fig 1a) However particular indi-viduals only used distinct portions of the overallrange and in a rather consistent manner within andacross years (Figs 2 amp 3) Individual consistency inmigration strategies was also recorded for 17 southpolar skuas Catharacta maccormicki and 3 greatskuas C skua tracked in 2 and 3 consecutive non-breeding seasons (Kopp et al 2011 Magnuacutesdoacutettir etal 2012 Weimerskirch et al 2015) and suggeststhat individual consistency in migration strategies iswidespread in skuas

The 4 migration strategies identified within thisbrown skua population matched the seasonal shift inproductivity in the wintering area The majority ofthe individuals (lsquoNorthPatarsquo strategy) utilised theyear-round highly productive BrazilminusFalklands Confluence (Garcia et al 2004) Moreover a smallgroup of 3 individuals (lsquoRioPlatarsquo strategy) regularlyswitched between the highly productive region influ-

enced by the outflow of Rio de la Plata (Acha et al2008) and the more open waters towards the Pata -gonian shelf-break Individuals from the southernand northern end of the range (lsquoSouthPatarsquo andlsquoSouthBrazilrsquo strategies) used the highly productivebut seasonal frontal systems of the southern Patagon-ian Shelf (Acha et al 2004 Rivas et al 2006) and theSouth Brazil Shelf (Acha et al 2004) respectivelyDuring this period the tracked individuals spent onlya small proportion of the day flying a pattern foundin brown great and south polar skuas (Phillips et al2007 Magnuacutesdoacutettir et al 2014 Weimerskirch et al2015 Carneiro et al 2016) A clear diurnal patternwas apparent with multiple landings during the dayinterspersed by 3 to 4 short flight bouts of approxi-mately 1 h On average females made longer flightbouts than males and spend less time foraging(Table 1) The sex-specific differences might be ex -plained by the reversed sexual size dimorphism inbrown skuas (Phillips et al 2002) Males are likely tohave lower wing loading greater manoeuvrabilityand a lower cost of take-off which can lead to sex-specific differences in habitat use and behaviour(Phillips et al 2004) The level of foraging activityand number of flight bouts was not repeatable at theindividual level suggesting that brown skuas adjusttheir feeding behaviour depending on local foodavailability However there was a degree of individ-ual consistency in the duration of flight bouts whichmight also relate to differences between winteringregions in the distance between prey patches or beassociated with the animalsrsquo personality since roam-ing behaviour and exploration is often consideredas a repeatable individual trait (eg Dingemanse etal 2002 Reacuteale et al 2007 Patrick amp Weimerskirch2014) The lack of correlation between activity pat-terns and absolute NPP is most likely attributed tothe coarse spatial scale and the time lag for changesin productivity to propagate through trophic levels toaffect prey abundance for higher predators such asskuas (Frederiksen et al 2006) However we stillconsider NPP values in combination with frontal sys-tems to be a valuable indirect measure of large-scalefood predictability and availability that can be linkedto population-level distributions (see also Zainuddinet al 2006 Pinaud amp Weimerskirch 2007 Humphrieset al 2010 Thompson et al 2012)

Following breeding the tracked skuas travelled toone of several alternative wintering areas within theoverall range according to their particular migrationstrategy In contrast in the late non-breeding seasonmost tracked birds moved towards the same areaaround the central Patagonian shelf-break where

221


they remained for several weeks before returning tothe breeding ground (Fig 2) This area is particularlyknown for its strong seasonality (Signorini et al 2006)and the increased foraging activity of the brown skuassuggests they exploit the local spring peak in primaryproduction (Figs 1 amp 4b) The diversity of migrationstrategies means that birds experience different envi-ronmental conditions during the winter which couldpresumably affect body condition laying date breed-ing probability and success (eg Bogdanova et al2011 Fayet et al 2016) and the degree of exposure topollutants (eg Leat et al 2013) The high spatial andtemporal migratory connectivity (here defined as thespatial extent of one population at any given time seeBauer et al 2016 and Lisovski et al 2016) shown bythe tracked birds particularly the aggregation of mostindividuals in the same area at the end of the wintermakes the population susceptible to oceanographic orother changes within the region

It should be noted that 3 of the tracked skuas didnot join the others on the central Patagonian shelf-break but moved further offshore before returning toKing George Island Based on a discriminant analysis(including bill depth at the gonys along with tarsusculmen wing and head length) these 3 individualswere significantly smaller than the others (authorsrsquounpubl data) This suggests that they might havebeen hybrids between brown skuas and the smallersouth polar skua as hybridisation between theseclosely-related species occurs frequently (Ritz et al2006) This might explain their distinctive migrationpattern as south polar skuas are trans-equatorialmigrants (Kopp et al 2011 Weimerskirch et al 2015)and hybridisation is known to alter migration behav-iour in other species (eg Helbig 1991)

Timing of migration like distribution differedbetween sexes and was consistent within individuals(Table 1) Repeatability in the arrival date at thebreeding grounds was notably high particularlygiven the extensive variation among individuals(within a 48 d range) This could reflect the varyingcosts and benefits of the timing of migration betweenindividuals (Moslashller 1994) or among birds of differentage-classes or experience levels (Jaeger et al 2014)For example competitive individuals can evict weakcompetitors from territories even if they arrive latterand might consequently benefit from a shorter over-all attendance period at the breeding grounds(Forstmeier 2002) There was some variation be -tween years indicating a degree of flexibility inresponse to local environmental conditions as hasbeen demonstrated across a large range of taxa (egMarra et al 1998 Gill et al 2001 Norris et al 2004)

As arrival date is subject to much stronger selectionpressures (Both amp Visser 2001 Brown et al 2005) weexpected that individual repeatability in departuredates from King George Island would be lower Pre-vious studies of brown skuas as well as other seabirdspecies have shown that non-breeders or failedbreeders depart earlier because they are not con-strained by reproductive duties (Phillips et al 20052007 Bogdanova et al 2011 Fifield et al 2014)However in our data there was no such relationshipalthough the sample size was high (18 successful and29 unsuccessful individuals) The later departure ofmale brown skuas in comparison with females canbe explained with their higher degree of nest-sitefidelity (Parmelee amp Pietz 1987) and the benefits of alonger defence period that might increase theirchance of retaining the same territory in consecutivebreeding seasons

Eight tracked individuals most of which returnedrelatively early to the breeding grounds went on apre-laying exodus of ca 1000 to 1500 km back totheir non-breeding ranges The majority of brownskuas tracked from South Georgia particularlyfemales also went on a pre-laying exodus (Phillips etal 2007 Carneiro et al 2016) There are obviousbenefits of early arrival brown skuas are highly ter-ritorial and their reproductive success depends onthe quality of the acquired territory (Hahn amp Bauer2008) However there might also be costs such asthe increased risk of encountering adverse weatherduring the early season (Moslashller 1994) It appears thata pre-laying exodus is discretionary presumablydepending on conditions at the breeding colony asonly 1 of the 5 individuals tracked in multiple yearsperformed such a trip twice

In conclusion we recorded highly consistent indi-vidual migration strategies in brown skuas fromKing George Island this reflected considerable vari-ation in timing of migration non-breeding distribu-tions and activity patterns The tracked birds dif-fered extensively in their arrival dates at thebreeding ground a migratory trait that is supposedto be under strong selection (eg Kokko 1999)whereas the arrival dates within individuals werehighly repeatable Based on the high levels of pri-mary production the tracked brown skuas mainlyexploit one of a number of alternative winteringareas within the overall non-breeding range How-ever almost all individuals moved to take advantageof a seasonal peak in marine productivity in a par-ticular area for several weeks before the final returnto the colony these birds are presumably returningto this area of high resource abundance that they

222


experienced during an initial early-life explorationminusrefinement phase (Guilford et al 2011) The 3 indi-viduals that showed a different late-winter distribu-tion may not have visited this otherwise commonarea in previous years or may be hybrids exhibitingan alternative migration strategy that reflectsgenetic differences We were unable to disentanglethe relative contribution of genetic control versuspast experience in determining individual migrationstrategies To this end we would need to trackmovements of juvenile brown skuas during theirfirst years at sea and ideally also track their par-ents Such data would also be extremely valuablefor determining the flexibility in migration strategieswithin and across generations and provide an indi-cation of how quickly seabirds can adapt to rapidchanges in the environment

Acknowledgements We thank the following former mem-bers of the Polar amp Bird Ecology Group for their instrumentalhelp in the field andor lab Markus Ritz Anja Ruszlig AnneFroumlhlich Christina Braun Tobias Guumltter Michel Stelter andJan Esefeld and 3 anonymous referees for helpful com-ments on the manuscript The research was supported bythe Deutsche Forschungsgemeinschaft (PE 4541 ff) andthe German Federal Environment Agency (FKZ 3708 91 1023708 91 102 amp 3712 87 100) which also approved the fieldwork (I 24 - 94003-3151)

LITERATURE CITED

Acha EM Mianzan HW Guerrero RA Favero M Bava J(2004) Marine fronts at the continental shelves of australSouth America physical and ecological processes J MarSyst 44 83minus105

Acha EM Mianzan HW Guerrero RA Carreto J Giberto DMontoya N Carignan M (2008) An overview of physicaland ecological processes in the Rio de la Plata EstuaryCont Shelf Res 28 1579minus1588

Aringdahl E Lundberg PER Jonzeacuten N (2006) From climatechange to population change the need to considerannual life cycles Glob Change Biol 12 1627minus1633

Bates D Maumlchler M Bolker B Walker S (2014) Fitting linearmixed-effects models using lme4 J Stat Softw 67 1minus48

Bauer S Lisovski S Hahn S (2016) Timing is crucial for con-sequences of migratory connectivity Oikos 125 605minus612

Berthold P (2001) Bird migration a general survey 2nd ednOxford University Press Oxford

Bogdanova MI Daunt F Newell M Phillips RA Harris MPWanless S (2011) Seasonal interactions in the black-legged kittiwake Rissa tridactyla links between breed-ing performance and winter distribution Proc R Soc B278 2412minus2418

Both C Visser ME (2001) Adjustment to climate change isconstrained by arrival date in a long-distance migrantbird Nature 411 296minus298

Bridge ES Thorup K Bowlin MS Chilson PB and others(2011) Technology on the move recent and forthcominginnovations for tracking migratory birds Bioscience 61 689minus698

Brown CR Brown MB Raouf SA Smith LC Wingfield JC(2005) Effects of endogenous steroid hormone levelson annual survival in cliff swallows Ecology 86 1034minus1046

Carneiro APB Manica A Clay TA Silk JRD King MPhillips RA (2016) Consistency in migration strategiesand habitat preferences of brown skuas over two win-ters a decade apart Mar Ecol Prog Ser 553 267minus281

Cherel Y Quillfeldt P Delord K Weimerskirch H (2016)Combination of at-sea activity geolocation and featherstable isotopes documents where and when seabirdsmoult Front Ecol Evol 4 3

Dias MP Granadeiro JP Phillips RA Alonso H Catry P(2011) Breaking the routine individual Coryrsquos shear -waters shift winter destinations between hemispheresand across ocean basins Proc R Soc B 278 1786minus1793

Dingemanse NJ Both C Drent PJ van Oers K van Noord-wijk AJ (2002) Repeatability and heritability of ex -ploratory behaviour in great tits from the wild AnimBehav 64 929minus938

Fayet AL Freeman R Shoji A Boyle D and others (2016)Drivers and fitness consequences of dispersive migrationin a pelagic seabird Behav Ecol 27 1061minus1072

Fifield DA Montevecchi WA Garthe S Robertson GJKubetzki U Rail JF (2014) Migratory tactics and winter-ing areas of northern gannets (Morus bassanus) breed-ing in North America Ornithol Monogr 79 1minus63

Forstmeier W (2002) Benefits of early arrival at breedinggrounds vary between males J Anim Ecol 71 1minus9

Frederiksen M Edwards M Richardson AJ Halliday NCWanless S (2006) From plankton to top predators bot-tom-up control of a marine food web across four trophiclevels J Anim Ecol 75 1259minus1268

Fridolfsson AK Ellegren H (1999) A simple and universalmethod for molecular sexing of non-ratite birds J AvianBiol 30 116minus121

Furness RW (1987) The skuas 1st edn T amp AD Poyser CaltonGarcia CA Sarma Y Mata MM Garcia VM (2004) Chloro-

phyll variability and eddies in the BrazilminusMalvinas Con-fluence region Deep-Sea Res II 51 159minus172

Gill JA Norris K Potts PM Gunnarsson TG Atkinson PWSutherland WJ (2001) The buffer effect and large-scalepopulation regulation in migratory birds Nature 412 436minus438

Guilford T Freeman R Boyle D Dean B Kirk H Phillips RAPerrins C (2011) A dispersive migration in the Atlanticpuffin and its implications for migratory navigationPLOS ONE 6 e21336

Hahn S Bauer S (2008) Dominance in feeding territoriesrelates to foraging success and offspring growth inbrown skuas Catharacta antarctica lonnbergi BehavEcol Sociobiol 62 1149minus1157

Harrison XA Blount JD Inger R Norris DR Bearhop S(2011) Carry-over effects as drivers of fitness differencesin animals J Anim Ecol 80 4minus18

Helbig AJ (1991) Inheritance of migratory direction in a birdspecies a cross-breeding experiment with SE-and SW-migrating blackcaps (Sylvia atricapilla) Behav EcolSociobiol 28 9minus12

Humphries NE Queiroz N Dyer JRM Pade NG and others(2010) Environmental context explains Leacutevy and Brown-ian movement patterns of marine predators Nature 465 1066minus1069

Jaeger A Goutte A Lecomte VJ Richard P and others(2014) Age sex and breeding status shape a complex

223


foraging pattern in an extremely long-lived seabirdEcology 95 2324minus2333

Karnovsky NJ Kwasniewski S Marcin Wesławski JWalkusz W Beszczynska-Moumlller A (2003) Foragingbehavior of little auks in a heterogeneous environmentMar Ecol Prog Ser 253 289minus303

Kokko H (1999) Competition for early arrival in migratorybirds J Anim Ecol 68 940minus950

Kopp M Peter HU Mustafa O Lisovski S Ritz MS PhillipsRA Hahn S (2011) South polar skuas from a singlebreeding population overwinter in different oceansthough show similar migration patterns Mar Ecol ProgSer 435 263minus267

Kuznetsova A Brockhoff PB Christensen RHB (2015)lmerTest tests in linear mixed effects models R packageversion 20-29 https CRAN R-project org package =lmer Test

Leat EHK Bourgeon S Magnusdottir E Gabrielsen GW andothers (2013) Influence of wintering area on persistentorganic pollutants in a breeding migratory seabird MarEcol Prog Ser 491 277minus293

Lessells C Boag PT (1987) Unrepeatable repeatabilities acommon mistake Auk 104 116minus121

Lisovski S Hewson CM Klaassen RH Korner-Nievergelt FKristensen MW Hahn S (2012) Geolocation by light accuracy and precision affected by environmental fac-tors Methods Ecol Evol 3 603minus612

Lisovski S Gosbell K Christie M Hoye BJ and others (2016)Movement patterns of sanderling (Calidris alba) in theEast AsianminusAustralasian flyway and a comparison ofmethods for identification of crucial areas for conserva-tion Emu 116 168minus177

Magnuacutesdoacutettir E Leat EH Bourgeon S Stroslashm H and others(2012) Wintering areas of great skuas Stercorarius skuabreeding in Scotland Iceland and Norway Bird Study59 1minus9

Magnuacutesdoacutettir E Leat EH Bourgeon S Joacutensson JE and others (2014) Activity patterns of wintering great skuasStercorarius skua Bird Study 61 301minus308

Marra PP Hobson KA Holmes RT (1998) Linking winter andsummer events in a migratory bird by using stable-car-bon isotopes Science 282 1884minus1886

Mattern T Masello JF Ellenberg U Quillfeldt P (2015)Actavenet minus a web-based tool for the analysis of seabirdactivity patterns from saltwater immersion geolocatorsMethods Ecol Evol 6 859minus864

McFarlane Tranquilla LA Montevecchi WA Fifield DAHedd A Gaston AJ Robertson GJ Phillips RA (2014)Individual winter movement strategies in two species ofmurre (Uria spp) in the northwest Atlantic PLOS ONE 9 e90583

McKnight A Irons DB Allyn AJ Sullivan KM Suryan RM(2011) Winter dispersal and activity patterns of post-breeding black-legged kittiwakes Rissa tridactyla fromPrince William Sound Alaska Mar Ecol Prog Ser 442 241minus253

Moslashller AP (1994) Phenotype-dependent arrival time and itsconsequences in a migratory bird Behav Ecol Sociobiol35 115minus122

Mueller T OrsquoHara RB Converse SJ Urbanek RP Fagan WF(2013) Social learning of migratory performance Science341 999minus1002

Muumlller MS Massa B Phillips RA DellrsquoOmo G (2014)Individual consistency and sex differences in migra-tion strategies of Scopolirsquos shearwaters Calonectris

diomedea despite year differences Curr Zool 60 631minus641

Nakagawa S Schielzeth H (2010) Repeatability for Gaussianand non-Gaussian data a practical guide for biologistsBiol Rev Camb Philos Soc 85 935minus956

Norris DR Marra PP Kyser TK Sherry TW Ratcliffe LM(2004) Tropical winter habitat limits reproductive successon the temperate breeding grounds in a migratory birdProc R Soc B 271 59minus64

Oksanen J Blanchet FG Kindt R Legendre P and others(2015) vegan community ecology package R packageversion 23-1 https cran r-project org web packagesvegan index html

Olmos F (2002) Non-breeding seabirds in Brazil a review ofband recoveries Ararajuba 10 31minus42

Orsi AH Whitworth T Nowlin WD (1995) On the meridionalextent and fronts of the Antarctic Circumpolar CurrentDeep-Sea Res I 42 641minus673

Parmelee DF Pietz PJ (1987) Philopatry mate and nest-sitefidelity in the brown skuas of Anvers Island AntarcticaCondor 89 916minus919

Patrick SC Weimerskirch H (2014) Personality foraging andfitness consequences in a long lived seabird PLOS ONE9 e87269

Phillips RA Dawson DA Ross DJ (2002) Mating patternsand reversed size dimorphism in southern skuas (Sterco-rarius skua lonnbergi) Auk 119 858minus863

Phillips RA Silk JRD Phalan B Catry P Croxall JP (2004)Seasonal sexual segregation in two Thalassarche alba-tross species Competitive exclusion reproductive rolespecialization or foraging niche divergence Proc R Soc B271 1283minus1291

Phillips RA Silk JRD Croxall JP Afanasyev V Bennett VJ(2005) Summer distribution and migration of nonbreed-ing albatrosses individual consistencies and implicationsfor conservation Ecology 86 2386minus2396

Phillips RA Catry P Silk JRD Bearhop S McGill RAfanasyev V Strange IJ (2007) Movements winter distri-bution and activity patterns of Falkland and brownskuas insights from loggers and isotopes Mar Ecol ProgSer 345 281minus291

Pinaud D Weimerskirch H (2007) At-sea distribution andscale-dependent foraging behaviour of petrels and alba-trosses a comparative study J Anim Ecol 76 9minus19

Quillfeldt P Voigt CC Masello JF (2010) Plasticity versusrepeatability in seabird migratory behaviour Behav EcolSociobiol 64 1157minus1164

R Core Team (2015) R a language and environment for sta-tistical computing R Foundation for Statistical Comput-ing Vienna

Reacuteale D Reader SM Sol D McDougall PT Dingemanse NJ(2007) Integrating animal temperament within ecologyand evolution Biol Rev Camb Philos Soc 82 291minus318

Ritz MS Hahn S Janicke T Peter HU (2006) Hybridisationbetween south polar skua (Catharacta maccormicki) andbrown skua (C antarctica lonnbergi) in the AntarcticPeninsula region Polar Biol 29 153minus159

Ritz MS Millar C Miller GD Phillips RA and others (2008)Phylogeography of the southern skua complexmdashrapidcolonization of the southern hemisphere during a glacialperiod and reticulate evolution Mol Phylogenet Evol 49 292minus303

Rivas AL Dogliotti AI Gagliardini DA (2006) Seasonal vari-ability in satellite-measured surface chlorophyll in thePatagonian Shelf Cont Shelf Res 26 703minus720

224


Shealer DA (2002) Foraging behavior and food of seabirdsIn Shreiber EA Burger J (eds) Biology of marine birdsCRC Press Boca Raton FL p 137minus177

Signorini SR Garcia VMT Piola AR Garcia CAE MataMM McClain CR (2006) Seasonal and interannual vari-ability of calcite in the vicinity of the Patagonian shelfbreak (38degSminus52degS) Geophys Res Lett 33 L16610

Small-Lorenz SL Culp LA Ryder TB Will TC Marra PP(2013) A blind spot in climate change vulnerabilityassessments Nat Clim Change 3 91minus93

Spalding MD Fox HE Allen GR Davidson N and others(2007) Marine ecoregions of the world a bioregionaliza-tion of coastal and shelf areas Bioscience 57 573minus583

Sumner MD Wotherspoon SJ Hindell MA (2009) Bayesianestimation of animal movement from archival and satel-lite tags PLOS ONE 4 e7324

Thompson SA Sydeman WJ Santora JA Black BA and others (2012) Linking predators to seasonality of up -welling using food web indicators and path analysis

to infer trophic connections Prog Oceanogr 101 106minus120Weimerskirch H (2007) Are seabirds foraging for unpre-

dictable resources Deep-Sea Res II 54 211minus223Weimerskirch H Tarroux A Chastel O Delord K Cherel Y

Descamps S (2015) Population-specific wintering distri-butions of adult south polar skuas over three oceans MarEcol Prog Ser 538 229minus237

Wotherspoon SJ Sumner MD Lisovski S (2013a) R packageBAStag basic data processing for light based geoloca-tion archival tags GitHub Repository https githubcom SWotherspoon BAStag

Wotherspoon SJ Sumner MD Lisovski S (2013b) R packageSGAT solarsatellite geolocation for animal trackingGitHub Repository http github com swotherspoon sgat

Zainuddin M Kiyofuji H Saitoh K Saitoh SI (2006) Usingmulti-sensor satellite remote sensing and catch data todetect ocean hot spots for albacore (Thunnus alalunga)in the northwestern North Pacific Deep-Sea Res II 53 419minus431

225

Editorial responsibility Jacob Gonzaacutelez-Soliacutes (Guest Editor)Barcelona Spain

Submitted May 10 2016 Accepted October 12 2016Proofs received from author(s) November 27 2016


general direction of migration seems to be largelydetermined by genetics or in some species culturalinheritance (Berthold 2001 Mueller et al 2013) move-ment patterns of individuals and populations alsorespond to factors such as food availability (Shealer2002 Karnovsky et al 2003) However prey avail-ability in marine ecosystems is generally charac-terised by varying degrees of temporal and spatialpredictability (Weimerskirch 2007) and hence it can-not be assumed that all areas will be favourable for aparticular species in every year This could ultimatelylead to variation in individual migration patterns ofseabirds at some spatial scale across years Indeedalthough most species show high individual consis-tency in non-breeding destinations at a large spatialscale (Phillips et al 2005 Fifield et al 2014 Muumlller etal 2014) there are exceptions in addition in almostall species there is extensive variation both amongand within individuals in routes use of staging areasand timing (Quillfeldt et al 2010 Dias et al 2011McFarlane Tranquilla et al 2014)

In this study we analysed the migration strategiesof individual brown skuas Catharacta antarcticalonnbergi breeding on the South Shetland Islands inthe Maritime Antarctic Migration routes and non-breeding areas were derived using light-level geolo-cators (also termed global location sensors or GLSloggers) and our dataset included repeated tracksfrom individuals over 2 to 3 yr Brown skuas are long-lived and highly opportunistic top predators with acircumpolar breeding distribution on subantarcticislands and the Antarctic Peninsula (Furness 1987Ritz et al 2008) To date the only detailed distribu-tion data available for brown skuas during the non-breeding period are for birds from the population atSouth Georgia migrating to waters between thenorthern extent of the Subtropical Front and thesouthern boundary of the Antarctic CircumpolarCurrent and between the Argentine and Agulhas(Phillips et al 2007 Carneiro et al 2016) In contrastthe limited data for the closely related Falklandskuas C a antarctica tracked in different years sug-gest a non-breeding distribution mainly in sub-antarctic waters around the central Patagonian shelf-break (Phillips et al 2007)

Given the lack of knowledge of the non-breedingranges of brown skuas from the South ShetlandIslands the adjacent southern population and theinclusion in our dataset of repeated migration tracksfrom the same individuals in multiple years the aimsof the study were 2-fold Firstly we aimed to revealthe spatiotemporal non-breeding distribution of thetracked population the degree of variation among

individuals and the effects of sex year and previousbreeding performance Secondly we aimed to quan-tify individual consistency in annual migration strate-gies In addition we used activity (immersion) datarecorded by the loggers and remotely sensed dataon net primary production to examine the correlationbetween this proxy for food availability and themovement and activity patterns of individual birds

MATERIALS AND METHODS

Logger deployment and retrieval

Fieldwork was carried out on adult brown skuasat King George Island (Fildes Peninsula 62deg 19rsquo S58deg 95rsquo W) in the Maritime Antarctic A total of 46geolocator-immersion loggers were deployed on 33individuals (which at the time were of unknown sex)over the course of 3 breeding seasons (20062007 to2008 2009) Three types of loggers (manufactured bythe British Antarctic Survey) were used in this studyMK5 (n = 20) MK9 (n = 22) and MK15 (n = 4) Totalweights were 64 53 and 53 g respectively whichincluded the device aluminium ring and cable tiesused for attachment (together with metal ring of 3 gused for identification corresponding to ~06 of themean body mass) Besides recording light intensityover time all loggers tested for saltwater immersionevery 3 s and stored the sum of positive tests at10 min intervals resulting in values between 0(entirely dry) and 200 (entirely wet) Birds wererecaptured in the subsequent season and in 11 casesthe loggers were replaced by a new device thesedevices and others from the initial deployments wereretrieved in the third season A blood sample wastaken from each bird (~50 microl) stored at minus20degC andlater used to determine sex from DNA (Fridolfsson ampEllegren 1999) All individuals were monitored regu-larly in the pre- and post-migratory season (fromearly December to March) to determine breedingstatus and performance (ie success vs failure)

Departure and arrival date

Since brown skuas switch from a predominantlyterrestrial lifestyle to almost exclusively marine habi-tat after leaving the breeding site (Phillips et al2007) departure and arrival dates at King GeorgeIsland were identified by visual inspection of immer-sion data For individuals lacking immersion databecause of logger malfunction these dates were

214








Activity analysis


215















216




RESULTS






217

40degW70degW

STF

SAF

PF


4

6

8


a

b c

lt20

20minus40

40minus60

60minus80

80minus90

90minus95

Density contour ()

30degS

50degS

40degS

60degS





218

123689123839

127377139674

108276

108294

139514

030210030256

030271

066607108271108280

108285108336

123836

127245138520

138530

139679139685156143156155157779162922

108272

108291

110388

Sou

thP

ata

Rio

Pla

taN

orth

Pat

aS

outh

Bra

zil

a





Equinox

c

Sum

in M

EO

W

70degW 40degW

40degS

60degS

b








Timing of migration


219

100

75

50

25

0Among

Pro

bab

ility

dis

trib

utio

n ov

erla

p (

)individuals

Within individuals

oo


circles outliers

oo

o

ooFemale


Fora

ging

act

ivtiy

(h d

ndash1)

2

a

b

6

10








Activity patterns




220













DISCUSSION






221








222




LITERATURE CITED





























223







































224














225










Activity analysis


215















216




RESULTS






217

40degW70degW

STF

SAF

PF


4

6

8


a

b c

lt20

20minus40

40minus60

60minus80

80minus90

90minus95

Density contour ()

30degS

50degS

40degS

60degS





218

123689123839

127377139674

108276

108294

139514

030210030256

030271

066607108271108280

108285108336

123836

127245138520

138530

139679139685156143156155157779162922

108272

108291

110388

Sou

thP

ata

Rio

Pla

taN

orth

Pat

aS

outh

Bra

zil

a





Equinox

c

Sum

in M

EO

W

70degW 40degW

40degS

60degS

b








Timing of migration


219

100

75

50

25

0Among

Pro

bab

ility

dis

trib

utio

n ov

erla

p (

)individuals

Within individuals

oo


circles outliers

oo

o

ooFemale


Fora

ging

act

ivtiy

(h d

ndash1)

2

a

b

6

10








Activity patterns




220













DISCUSSION






221








222




LITERATURE CITED





























223







































224














225

















216




RESULTS






217

40degW70degW

STF

SAF

PF


4

6

8


a

b c

lt20

20minus40

40minus60

60minus80

80minus90

90minus95

Density contour ()

30degS

50degS

40degS

60degS





218

123689123839

127377139674

108276

108294

139514

030210030256

030271

066607108271108280

108285108336

123836

127245138520

138530

139679139685156143156155157779162922

108272

108291

110388

Sou

thP

ata

Rio

Pla

taN

orth

Pat

aS

outh

Bra

zil

a





Equinox

c

Sum

in M

EO

W

70degW 40degW

40degS

60degS

b








Timing of migration


219

100

75

50

25

0Among

Pro

bab

ility

dis

trib

utio

n ov

erla

p (

)individuals

Within individuals

oo


circles outliers

oo

o

ooFemale


Fora

ging

act

ivtiy

(h d

ndash1)

2

a

b

6

10








Activity patterns




220













DISCUSSION






221








222




LITERATURE CITED





























223







































224














225






RESULTS






217

40degW70degW

STF

SAF

PF


4

6

8


a

b c

lt20

20minus40

40minus60

60minus80

80minus90

90minus95

Density contour ()

30degS

50degS

40degS

60degS





218

123689123839

127377139674

108276

108294

139514

030210030256

030271

066607108271108280

108285108336

123836

127245138520

138530

139679139685156143156155157779162922

108272

108291

110388

Sou

thP

ata

Rio

Pla

taN

orth

Pat

aS

outh

Bra

zil

a





Equinox

c

Sum

in M

EO

W

70degW 40degW

40degS

60degS

b








Timing of migration


219

100

75

50

25

0Among

Pro

bab

ility

dis

trib

utio

n ov

erla

p (

)individuals

Within individuals

oo


circles outliers

oo

o

ooFemale


Fora

ging

act

ivtiy

(h d

ndash1)

2

a

b

6

10








Activity patterns




220













DISCUSSION






221








222




LITERATURE CITED





























223







































224














225






218

123689123839

127377139674

108276

108294

139514

030210030256

030271

066607108271108280

108285108336

123836

127245138520

138530

139679139685156143156155157779162922

108272

108291

110388

Sou

thP

ata

Rio

Pla

taN

orth

Pat

aS

outh

Bra

zil

a





Equinox

c

Sum

in M

EO

W

70degW 40degW

40degS

60degS

b








Timing of migration


219

100

75

50

25

0Among

Pro

bab

ility

dis

trib

utio

n ov

erla

p (

)individuals

Within individuals

oo


circles outliers

oo

o

ooFemale


Fora

ging

act

ivtiy

(h d

ndash1)

2

a

b

6

10








Activity patterns




220













DISCUSSION






221








222




LITERATURE CITED





























223







































224














225








Timing of migration


219

100

75

50

25

0Among

Pro

bab

ility

dis

trib

utio

n ov

erla

p (

)individuals

Within individuals

oo


circles outliers

oo

o

ooFemale


Fora

ging

act

ivtiy

(h d

ndash1)

2

a

b

6

10








Activity patterns




220













DISCUSSION






221








222




LITERATURE CITED





























223







































224














225







Activity patterns




220













DISCUSSION






221








222




LITERATURE CITED





























223







































224














225





DISCUSSION






221








222




LITERATURE CITED





























223







































224














225










222




LITERATURE CITED





























223







































224














225






LITERATURE CITED





























223







































224














225









































224














225
















225



Documents

Consistent variation in individual migration strategies of