Page 1
Lettershttps://doi.org/10.1038/s41558-018-0312-9
Long-distance migratory birds threatened by multiple independent risks from global changeDamaris Zurell 1,2*, Catherine H. Graham3, Laure Gallien 4,5, Wilfried Thuiller 5 and Niklaus E. Zimmermann 2,6
1Geography Department, Humboldt-Universität zu Berlin, Berlin, Germany. 2Dynamic Macroecology, Department of Landscape Dynamics, Swiss Federal Research Institute WSL, Birmensdorf, Switzerland. 3Department of Biodiversity and Conservation Biology, Swiss Federal Research Institute WSL, Birmensdorf, Switzerland. 4Centre for Invasion Biology, Department of Botany and Zoology, Stellenbosch University, Matieland, South Africa. 5Univ. Grenoble Alpes, CNRS, Univ. Savoie Mont-Blanc, LECA-Laboratoire d’Écologie Alpine, Grenoble, France. 6Department of Environmental Systems Science, Swiss Federal Institute of Technology ETH, Zurich, Switzerland. *e-mail: [email protected]
SUPPLEMENTARY INFORMATION
In the format provided by the authors and unedited.
NaTurE CLimaTE CHaNGE | www.nature.com/natureclimatechange
Page 2
1
SupplementaryInformation
Long-distancemigratorybirdsthreatenedbymultiple
independentrisksfromglobalchange
DamarisZurell,CatherineH.Graham,LaureGallien,WilfriedThuillerandNiklausE.
Zimmermann.
ThisSupplementaryInformationcontainsthefollowinginformation:
- SupplementaryFigureS1.Projectedchangesinseasonalspeciesrichnessoflong-
distancemigratorybirdsfor2050.
- SupplementaryFigureS2.Relationshipbetweenprojectedchangeinmigration
distanceandcurrent-daymigrationdistance.
- SupplementaryFigureS3.Projectedchangesinseasonalrangesizesand
migrationdistancefordifferentglobalchangescenariosanddispersalbuffers.
- SupplementaryFigureS4.Speciesrichnessinbreedingareasfordifferent
projectedrisksandriskcombinations.
- SupplementaryFigureS5.Projectedchangeinsummerandwinterrangesizeand
inmigratorydistancefordifferentIUCNriskcategories.
- SupplementaryFigureS6.Mapsofmultipleglobalchangerisksfordifferent
emissionscenarios.
- SupplementaryFigureS7.Sensitivityanalysesformultipleandsingleglobal
changerisksfordifferentIUCNcategories.
Page 3
2
- SupplementaryFigureS8.PredictionaccuracyofSDMscalibratedinsummerand
winterranges.
- SupplementaryFigureS9.Effectofdifferentresolutionsandthinningapproaches
onresidualspatialautocorrelationandonestimatesofglobalchangeimpacts.
- SupplementaryFigureS10.Goodness-of-fitbetweenobservedandpredicted
rangepropertiesfordifferentdispersalbufferdistances.
Page 4
3
FigureS1.Projectedchangesinseasonalspeciesrichnessoflong-distancemigratorybirdsfor2050.We
presentprojectedchangesinsummer(a,e,i)andwinter(b,f,j)richnessfortheclimateandlandcoverchange
scenariosRCP4.5-SSP1(a-d),RCP8.5-SSP3(e-h)andRCP8.5-SSP5(i-l).Additionally,wehighlightareaswherethe
projectedeffectsfromlandcoverchangewerestrongerthanfromclimatechangeforthedifferentscenarios
(c,d,g,h,k,l).Theprojectedchangeswerederivedfromtheensemblemeansforeachscenario,withamaximum
dispersaldistanceof1000km.
-150 -100 -50 0 50 100 150
-150-100-50050100150
-120
-93.3333333333333
-66.6666666666667
-40
-13.3333333333333
13.3333333333333
40
66.6666666666667
93.3333333333333
120
-110
+110Δspecies
a b
c d
e f
g h
i j
k l
LUCC > CCLUCC > CCLUCC > CCLUCC>CC
Page 5
4
FigureS2.Relationshipbetweenprojectedchangeinmigrationdistanceandcurrent-daymigrationdistance.
Linesshowthetrendsasestimatedfromphylogeneticregression.Projectionscorrespondtotheensemblemeansfor
thescenarioRCP4.5-SSP1,withamaximumdispersaldistanceof1000km.
Current migration distance [°]
Δ d
ista
nce
[°]
10 30 60 90 120
-10
010
NearcticW PalearcticE Palearctic
Page 6
5
FigureS3.Projectedchangesinseasonalrangesizesandmigrationdistancefordifferentglobalchange
scenariosanddispersalbuffers.Climateandlandcoverscenariosarecodedasnumbers1-8(1:RCP4.5,2:RCP8.5,
3:RCP4.5+SSP1,4:RCP8.5+SSP3,5:RCP8.5+SSP5,6:SSP1,7:SSP3,8:SSP5).
%Δ
sum
mer
are
a-100
-50
050
100
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Dispersal 500 km Dispersal 1000 km Dispersal 2000 km Full dispersal
%Δ
win
ter a
rea
-100
-50
050
100
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Dispersal 500 km Dispersal 1000 km Dispersal 2000 km Full dispersal
%Δ
mig
ratio
n di
stan
ce-50
050
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8
Dispersal 500 km Dispersal 1000 km Dispersal 2000 km Full dispersal
Page 7
6
FigureS4.Speciesrichnessinbreedingareasfordifferentprojectedrisksandriskcombinations.We
distinguishspeciesprojectedtobeatnorisk(a),atriskfromsinglethreats(b-d;S=summerpopulationreduction>
10%,W=winterpopulationreduction>10%,M=migrationdistanceincrease>10%)andatriskfrommultiple
threats(e-h).ProjectionscorrespondtotheensemblemeansfortheRCP4.5-SSP1scenario,withamaximumdispersal
distanceof1000km.
(a)Norisk (b)Risk:S
(c)Risk:W (d)Risk:M
(e)Risks:S+W (f)Risks:S+M
(g)Risks:W+M (h)Risks:S+W+M
102030405060
5101520253035
51015202530
246810
10203040
1234567
123456
12345
Page 8
7
FigureS5.ProjectedchangeinsummerandwinterrangesizeandinmigratorydistancefordifferentIUCN
riskcategories.IUCNrisksare:LCleastconcern,NTnearthreatened,VUvulnerable,ENendangered,CRcritically
endangered.Outliersarenotshown.Projectionscorrespondtotheensemblemeansforthedifferentemission
scenarios,withamaximumdispersaldistanceof1000km.
RCP4.5-SSP1
RCP8.5-SSP3
RCP8.5-SSP5
%Δ
sum
mer
are
a-75
075
LC NT VU EN CR
%Δ
win
ter a
rea
-75
075
LC NT VU EN CR
%Δ
dis
tanc
e-25
025
LC NT VU EN CR
%Δ
sum
mer
are
a-75
075
LC NT VU EN CR
%Δ
win
ter a
rea
-75
075
LC NT VU EN CR
%Δ
dis
tanc
e-25
025
LC NT VU EN CR
%Δ
sum
mer
are
a-75
075
LC NT VU EN CR
%Δ
win
ter a
rea
-75
075
LC NT VU EN CR
%Δ
dis
tanc
e-25
025
LC NT VU EN CR
Page 9
8
FigureS6.Mapsofmultipleglobalchangerisksfordifferentemissionscenarios.Venndiagramsillustrate
numberofspeciesprojectedtoexperiencesingleandmultiplerisks(forlegendpleaseseeFig.4andFig.S7).We
classifiedspeciesasatriskifpopulationreductionormigrationdistanceincreaseexceeded10%(a,c,e)or20%
(b,d,f).RGBmapsillustratetherelativenumberofspeciesfacingmultiplerisks(thecolouredareasoftheVenn
diagrams).Darktolightcoloursindicateincreasingspeciesnumbers.Colourbandsrepresentspecificrisk
combinations;mixedcoloursindicatethatspecieswithdifferentriskcombinationsarepresent.Projections
correspondtotheensemblemeansandamaximumdispersaldistanceof1000km.
305 114 71725 317
211 170 1101624 920
290 146 62933 419
(a)RCP4.5-SSP1+10%threshold
(c)RCP8.5-SSP3+10%threshold
(e)RCP8.5-SSP5+10%threshold
(b)RCP4.5-SSP1+20%threshold
(d)RCP8.5-SSP3+20%threshold
(f)RCP8.5-SSP5+20%threshold
S+WS+MW+M
n
222 47 7022 02
173 57 11015 23
209 39 7222 02
Page 10
9
FigureS7.SensitivityanalysesformultipleandsingleglobalchangerisksfordifferentIUCNcategories.a-d
Numberofspeciesprojectedtoexperiencemultiplerisksfromglobalchangewhena20%thresholdisusedfor
classifyingspeciesasatrisk.e-hNumberofspeciesprojectedtoexperiencesingle(butnotmultiple)risks,thus
concentratingonthoseareasoftheVenndiagramswhererisksdonotoverlap(outsideblackborder).Here,a10%
thresholdwasusedforclassifyingspeciesasatrisk.ProjectionscorrespondtotheensemblemeansfortheRCP4.5-
SSP1scenarioandamaximumdispersaldistanceof1000km.ForfurtherexplanationpleaseseeFig.4inmaintext.
155(51.0%)
46(15.1%) 92
(30.3%)1
(0.3%)5(1.6%)
2(0.7%)
3(1.0%)
S(n=207) W(n=141)
M(n=11)
18(38.3%)
11(23.4%) 18
(38.3%)0
(0.0%)0(0.0%)
0(0.0%)
0(0.0%)
S(n=29) W(n=29)
M(n=0)
b
d
a
c
IUCN:Leastconcern
IUCN:(Near)Threatened
S+WS+MW+M
n
190(38.3%)
143(28.8%) 100
(20.2%)13
(2.6%)24(4.8%)
9(1.8%)
17(3.4%)
S(n=370) W(n=265)
M(n=63)
21(32.8%)
27(42.2%) 10
(15.6%)3
(4.7%)0(0.0%)
0(0.0%)
3(4.7%)
S(n=51) W(n=40)
M(n=6)
b
d
a
c
IUCN:Leastconcern
IUCN:(Near)Threatened
SWM
n
Singleglobalchangeriskswith10%threshold
Multipleglobalchangeriskswith20%threshold
Page 11
10
FigureS8.PredictionaccuracyofSDMscalibratedinsummerandwinterranges.Modelswereinternally
validatedusinga70-30splitsampleapproachwith3repetitions.AUCistheareaunderthereceiveroperating
characteristiccurve;TSSisthetrueskillstatisticcalculatedasTSS=sensitivity+specificity–1;sensitivityisthetrue
positiverateandspecificitythetruenegativerate.
Summer Winter
0.5
0.7
0.9
AUC
Summer Winter0.5
0.7
0.9
TSS
Summer Winter
0.5
0.7
0.9
Sensitivity
Summer Winter
0.5
0.7
0.9
Specificity
Page 12
11
FigureS9.Effectofdifferentresolutionsandthinningapproachesonresidualspatialautocorrelationandon
estimatesofglobalchangeimpacts.aInthebaselineSDMs,speciesrangemapsandenvironmentaldatawere
griddedat0.5°resolution.WecomparedthesetoSDMsestimatedatcoarserresolutionof1°andtoSDMsestimated
Baseline0.5°(B) Resolution1°(R1) Thinning250km(T) Thinning500km(T5)
èSpecieswith≥40occurrences:
n=715 n=656 n=174 n=13
èIsspatialautocorrelationSACsignificantlyreduced?
ΔSAC:no ΔSAC:ê ΔSAC:ê
èIstheestimatedarealosssignificantlysmaller? Δarea:no Δarea:no Δarea:no
èIstheestimatedmigrationdistanceincreasesignificantlysmaller?
Δdist:no Δdist:no Δdist:no
B B-1000
1000
3000
5000
SA
C d
ista
nce
[km
]
n=715
B R1 B R1-1000
1000
3000
5000
SA
C d
ista
nce
[km
]
n=656
B T B T-1000
1000
3000
5000
SA
C d
ista
nce
[km
]
n=174
B T5 B T5-1000
1000
3000
5000
SA
C d
ista
nce
[km
]
n=13
b c d e
B B
-60
-20
020
40%Δ
are
a
n=715
B R1 B R1
-60
-20
020
40%Δ
are
a
n=656
B T B T
-60
-20
020
40%Δ
are
a
n=174
B T5 B T5
-60
-20
020
40%Δ
are
a
n=13
f g h i
-60
-20
020
40%Δ
dis
tanc
e
n=715
B R1
-60
-20
020
40%Δ
dis
tanc
e
n=656
B T
-60
-20
020
40%Δ
dis
tanc
e
n=174
B T5
-60
-20
020
40%Δ
dis
tanc
e
n=13
j k l m
a
Page 13
12
forspatiallythinneddatawithminimumdistanceof250kmand500kmbetweenpresencepoints(at0.5°resolution).
Becausethenumberofspecieswithatleast40presencesinsummerandwinterrangechangesacrossresolutionand
thinningchoices(i.e.speciesnumbersaredecreasingfromlefttorightpanels),theresultsofthe2nd,3rd,and4th
columnsshowonlytherespectivespeciessub-sets.b-eshowthedistanceclassesatwhichspatialautocorrelationin
modelresidualsbecomesinsignificantforspeciesdistributionmodels(SDMs)calibratedinsummer(red)andwinter
ranges(blue).f-ishowtheprojectedchangeinsummer(red)andwinterarea(blue).j-kshowtheprojectedchangein
migrationdistances.Significantdifferencesweretestedusingtwo-sampleWilcoxontestswithasignificancelevelof
0.05.ProjectionscorrespondtotheensemblemeansfortheRCP4.5-SSP1scenario,withamaximumdispersal
distanceof1000km.
FigureS10.Goodness-of-fitbetweenobservedandpredictedrangepropertiesfordifferentdispersalbuffer
distances.aexemplifieshowmigrationdistancebetweenpredictedsummerandwinterrangecentroids(with
1000kmdispersalbuffer)correspondstomigrationdistanceobtainedfromrangemaps.Thepredictionsfromthe
linearmodelexplain99%oftheobservedvariationinmigrationdistance(asindicatedbytheadjustedr2).b
summarisestheexplainedvariances(adjustedr2)fordifferentrangeproperties.
a b
Dispersal buffer [km]
Adj
. r2
500 1000 2000 full
0.50.60.70.80.91.0
Summer range sizeWinter range sizeSummer latitudeWinter latitudeSummer longitudeWinter longitudeMigration distance
Dispersal 1000 km
Migration distance range maps
Mig
ratio
n di
stan
ce p
redi
ctio
ns
20 40 60 80 100 120
2040
6080
120
adj.r2 = 0.985