1
Vol.:(0123456789)
Scien i ic Repo s | (2020) 10:18705 | h ps://doi.o g/10.1038/s41598-020-74684-8
www.na u e.com/scien i ic epo s
Closely ela ed species show
species‑speci ic en i onmen al
esponses and di e en spa ial
conse a ion needs: P ionailu us
ca s in he Indian subcon inen
And é P. Sil a 1,2,3,4*, Shomi a Mukhe jee 2, Uma Ramak ishnan 3, Ca los Fe nandes
4,5 & Ma s Bjö klund 1
Phylogene ically closely ela ed species a e o en assumed o ha e simila esponses o en i onmen al
condi ions, bu species‑speci ic esponses ha e also been desc ibed. These wo scena ios may
ha e di e en conse a ion implica ions. We es ed hese wo hypo heses o P ionailu us ca s
(P. ubiginosus, P. bengalensis, P. i e inus) in he Indian subcon inen and show i s implica ions
on species cu en p o ec ed a ea co e age and clima ic sui abili y ends h ough ime. We i ed
ecological niche models wi h cu en en i onmen al condi ions and calcula ed niche o e lap. In
addi ion, we de eloped a model o he Jungle Ca Felis chaus o compa e species esponses and niche
o e lap es ima es wi hin P ionailu us wi h hose o a ela ed sympa ic small ca species. Then we
es ima ed he p opo ion o cu en sui able en i onmen co e ed by p o ec ed a ea and p ojec ed
clima ic models om pas (las in e glacial) o u u e (2070; RCP4.5 and RCP8.5) condi ions o show
implica ions on popula ion managemen and conse a ion. The hypo hesis o a simila esponse and
niche o e lap among closely ela ed species is no suppo ed. P o ec ed a ea co e age was lowes
o P. i e inus (mean = 0.071, SD = 0.012) and highes o P. bengalensis (mean = 0.088, SD = 0.006).
In addi ion, he p opo ion o he subcon inen wi h sui able clima e a ied h ough ime and was
species‑speci ic. Fo P. bengalensis, clima ic sui abili y sh unk since a leas he mid‑Holocene, a end
ha can be in ensi ied by human‑induced clima e wa ming. Conce ning P. i e inus, mos p edic ions
show s able u u e clima ic sui abili y, bu a ew indica ed po en ial loss. Clima ic sui abili y o
P. ubiginous was p edic ed o emain s able bu he species exhibi ed a nega i e associa ion wi h
in ensi e ag icul u e. Simila esponses o en i onmen al change by phylogene ically closely ela ed
species should no be assumed and ha e implica ions on p o ec ed a ea co e age and na u al ends
o species clima ic sui abili y o e ime. This should be aken in o accoun du ing conse a ion and
managemen ac ions.
A deep unde s anding o species esponses o he en i onmen is c i ical o guide spa ial conse a ion s a egies
unde ongoing global change1,2. Howe e , pa adoxically, such knowledge is o en lacking o a e and elusi e
species as well as o biodi e si y ich a eas3–5. Phylogene ic niche conse a ism in i s b oade sense p edic s
p ese a ion o ances al ecological ai s among closely ela ed species. Unde his assump ion, occu ence
o common and conspicuous species could po en ially be used o p edic he dis ibu ion and en i onmen al
esponse o closely ela ed species ha a e mo e sec e i e o inhabi emo e egions.
The ecip ocal geog aphic dis ibu ions o sis e axon pai s o bu e lies, bi ds and mammals in allopa y
ha e been shown o ha e high cong uence6. The assump ion o phylogene ic niche conse a ism has also allowed
OPEN
1Depa men o Ecology and Gene ics, Animal Ecology, E olu iona y Biology Cen e, Uppsala Uni e si y,
No by ägen 18D, 752 36 Uppsala, Sweden. 2Sálim Ali Cen e o O ni hology and Na u al His o y, Anaika y
Pos , Coimba o e, Tamil Nadu 641108, India. 3Na ional Cen e o Biological Sciences, TIFR, Bella y Road,
Bangalo e 560065, India. 4cE3c - Cen e o Ecology, E olu ion and En i onmen al Changes, Faculdade de Ciências,
Uni e sidade de Lisboa, 1749-016 Lisboa, Po ugal. 5Faculdade de Psicologia, Uni e sidade de Lisboa, Alameda da
Uni e sidade, 1649-013 Lisboa, Po ugal. *email: [email p o ec ed]
2
Vol:.(1234567890)
Scien i ic Repo s | (2020) 10:18705 | h ps://doi.o g/10.1038/s41598-020-74684-8
www.na u e.com/scien i ic epo s/
he disco e y o new ep iles based on niche p edic ions o closely ela ed species7. Suppo o phylogene ic
niche conse a ism has been ound in se e al o he g oups including amphibians8, eshwa e a h opods9 and
plan s10–12. Howe e , e idence o species-speci ic ecological equi emen s among closely ela ed species has
also been widely epo ed in he li e a u e13, and hence assuming conse a ism in en i onmen al esponses can
lead o e oneous conclusions. Fo example, closely ela ed species o pines14,15, opical amphibians16,17, opical
liza ds18–20 and small mammals21 ha e shown unique niche p ope ies. E idence is conside ably spa se o small
endo he mic species in he opics, hough no comple ely absen 22,23.
He e we e alua e species ecological esponse and spa ial niche conse a ism in he p esence o en i onmen al
change in h ee small ca species (P ionailu us spp.), including wo closely ela ed species, in sou h Asia and
implica ions o p o ec ed a ea co e age. Knowledge abou niche conse a ism may be in o ma i e o con-
se a ion planning24, o example in he spa ial de ini ion o p o ec ed a eas, in egions wi h limi ed baseline
biodi e si y da a, o g oups o species whose ecology is li le known, o when using mo e common species as
p oxies o a e ones25–27. A p e ious s udy showed a lack o da a o wo P ionalu us species (P. planiceps and
P. i e inus)3 and an insu icien le el o knowledge abou species occu ence in he Indian subcon inen 28,
despi e his being a p io i y egion as i is ex emely biodi e se, including h ee biodi e si y ho spo s unde
a ious h ea s29–31. This lack o in o ma ion hinde s es ima ion o species esponses o en i onmen al changes
and p e en s obus planning and e alua ion o spa ial conse a ion s a egies. He e we go a s ep u he han
usual s udies on species dis ibu ion changes ocused on he p esen ime and explo e whe he a po en ial lack
o niche conse a ism can lead o species-speci ic ends in clima ic sui abili y and hei implica ions o spa ial
conse a ion s a egies.
To in es iga e esponses o he s udy species o en i onmen al ac o s, we de eloped ecological niche models,
including a iables ela ed o clima e, opog aphy, land co e , human dis u bance, and p ey occu ence. To es
o spa ial niche conse a ism we es ima ed niche o e lap wi hin he Indian subcon inen o cu en ime, allow-
ing us o include he main en i onmen al ace s ha can cons i u e he species niche. We hen examined how
species ecological a ini ies (habi a equi emen s) can in luence spa ial conse a ion s a egies, namely p o ec ed
a ea co e age. Finally, o explo e he possibili y o di e en clima ic-sui abili y ends o e ime, we ha e also
es ima ed species clima ic sui abili y since he las in e glacial (mo e speci ically, o he pe iod 140–120ka) up
o 2070. We used a species o small ca (Jungle Ca , Felis chaus) ha is sympa ic wi h P ionailu us and belongs
o a lineage closely ela ed o i ( he genus Felis is he ex an sis e clade o he clade con aining P ionailu us;
he wo clades di e ged a 6.18Ma32) as an ou g oup in analyses o es whe he en i onmen al esponse and
niche o e lap a e ela i ely mo e simila wi hin P ionailu us. Wi hin he Indian subcon inen , clima ic niche
conse a ism has been shown o some species wi h sha ed biogeog aphic his o ies bu no o phylogene ically
closely ela ed mammalian species33. Al hough his is he i s b oad-scale compa a i e assessmen o habi a
equi emen s o all P ionailu us species in sou h Asia, exis ing expe knowledge on local habi a equi emen s
poin s o he use o pa icula en i onmen s34–36. We he e o e expec his pa e n o be main ained on a mac-
oscale and isible h ough species-speci ic esponses.
Ma e ials and me hods
S udy species. We used mac oscale compa a i e analyses o look a po en ial conse a ism o niche p ope -
ies wi hin he genus P ionailu us, which is dis ibu ed h ough Asia and is cu en ly ep esen ed by ou species,
h ee o hem occu ing in India (Rus y-spo ed Ca P. ubiginosus, Leopa d Ca P. bengalensis and Fishing Ca
P. i e inus). P. ubiginosus is endemic o India, S i Lanka and Nepal, wi h he la ge pa o i s global popula-
ion occu ing in India37. This lineage o igina ed du ing he Ea ly Pliocene (4.59Ma32), and P. bengalensis and P.
i e inus a e sis e species ha di e ged a he beginning o he Pleis ocene (2.55Ma32). Despi e being closely
ela ed, he meag e in o ma ion a ailable on he ecology o he h ee species sugges s ha hey ha e di e en
ecological niches34–36. P. i e inus, he la ges o he h ee ocal species, is a medium-sized elid weighing up
o 16 kgs38 and s ongly associa ed wi h mang o es, we lands and eed beds39 . The species is mo phologically
adap ed o hun ing in wa e , wi h ish o ming a p ominen pa o i s die along wi h small mammals and
bi ds38. P. bengalensis, al hough dis ibu ed widely in India, is hypo hesized o be es ic ed by high ambien em-
pe a u es exceeding 38°C. I has no been epo ed om he ho cen al and wes e n pa s o he coun y, and
gene ic e idence sugges s ha he Wes e n Gha s popula ion is isola ed om he es o he species’ popula ion40.
The species does no occu in S i Lanka39. I s die includes oden s, o he small mammals, bi ds, amphibians
and ep iles37. Li le is known abou he ecology o P. ubiginosus, bu p elimina y da a sugges i is p ima ily
dis ibu ed h ough deciduous o es s in India and i s die is belie ed o consis o oden s38. All h ee species a e
known o occu in p oximi y o human se lemen s38. F. chaus is a ep esen a i e o he genus Felis ha di e ged
om he P ionailu us lineage a 6.2Ma38. I is a sc ub and open habi a elid, s ongly associa ed wi h wa e and
e y widely dis ibu ed in India41. I has bene i ed om i iga ed ag icul u e ha simula es i s na u al habi a and
hence also occu s in p oximi y o human se lemen s41. I eeds la gely on oden s and bi ds41.
Species occu ences. We collec ed p esence eco ds o each species ac oss he Indian subcon inen
om published sou ces, online da abases, and pe sonal communica ions (TableSM1.2). Following p e ious
ecommenda ions42, eco ds we e p uned o keep only hose con empo a y wi h he ime pe iod o he en i-
onmen al da a used in analyses43. Thus, only eco ds a e 1990 we e included. Mo eo e , duplica e eco ds
wi hin he same cell (10km esolu ion) we e andomly emo ed using he g idSample unc ion in he ‘dismo’
R-package44. Fo he pe iod since 1990, a e emo ing duplica e eco ds (RD) wi hin each cell, we ob ained 75
eco ds o P. bengalensis, 96 o P. i e inus, 54 o P. ubiginosus and 87 o F. chaus. When applying a bal-
anced design (RDbal, see de ails below), he numbe o eco ds we e 57 o P. bengalensis, 43 o P. i e inus,
40 o P. ubiginosus and 62 o F. chaus (see dis ibu ion in Fig.SM1.2; ull da a in TableSM1.2). Al hough ine
3
Vol.:(0123456789)
Scien i ic Repo s | (2020) 10:18705 | h ps://doi.o g/10.1038/s41598-020-74684-8
www.na u e.com/scien i ic epo s/
esolu ion en i onmen al da a is a ailable, analyses we e pe o med a 10km esolu ion based on home- ange
es ima es o he a ge species37,38,45,46, excep o P. ubiginous o which es ima es a e no a ailable. This eso-
lu ion is he e o e likely o minimize po en ial spa ial au oco ela ion while accoun ing o inaccu acies in he
geog aphical posi ioning o eco ds.
En i onmen al da a. En i onmen al p edic o s we e chosen based on he species ecology and knowledge
o hei in luence on ca ni o e occu ence37,40,47–50. Fi e en i onmen al desc ip o s we e conside ed: clima e,
opog aphy, land co e , human dis u bance and p ey occu ence.
Topog aphic (ele a ion) and 19 bioclima ic a iables we e downloaded om he Wo ldClim da a se 51,52
(TableSM1.1). Each cell was also cha ac e ized based on land co e ca ego ies (1km esolu ion) ex ac ed om
he Global Land Co e Map 2000 (including human se lemen s)53 o ma ch he da e o mos occu ence eco ds.
The pe cen age o each land co e ype wi hin each cell was calcula ed by delinea ing each land co e and hen
summing he numbe o pixels wi hin each 10km2 cell using he agg ega e unc ion. Mino land co e classes
we e emo ed o eclassi ied oge he wi h o he simila land co e ypes (TableSM1.1). Human popula ion
densi y es ima es (2.5 a c-minu es esolu ion, ~ 5km) and he Euclidean dis ance o e e y pixel o human dis-
u bance p oxies (e.g., dis ance o oads) we e also included (TableSM1.1). To ob ain he exac esolu ion, all
as e s we e esampled o he Wo ldClim as e s’ esolu ion (30 a c-second) using he bilinea me hod because
he da a is con inuous54. All as e ope a ions we e pe o med in he ‘ as e ’ R-package54. P ey a ailabili y may
also play an impo an ole in small wild ca occu ence because hey a e obliga e ca ni o es and ha e a s ong
p e e ence o oden s, hei s aple p ey37,55,56. Since p ey selec ion has no been epo ed o he ca species
s udied he e (al hough die s udies combined wi h da a on p ey a ailabili y a e a e), we assumed ha p eda ion
may depend mo e on o he ac o s such as ene ge ic cos , p ey abundance and ca chabili y57. We iden i ied he
mos common oden species p eyed upon by ca ni o es in Sou h-Sou heas Asian o es s h ough li e a u e
esea ch (TableSM2.1) and used Maximum En opy Modeling (Maxen )58 o es ima e hei en i onmen al
sui abili y ac oss he Indian subcon inen (see de ails in SM2). We assumed ha oden s o simila body size
p o ide simila ene ge ic bene i s and ha e compa able p obabili y o being caugh , and he e o e we pooled
hem in o h ee body size g oups (‘small’ ≤ 70g, 70g < ‘medium’ < 150g; ‘la ge’ > 150g) acco ding o hei weigh s
(see Fig.SM2.1). We hen used he ou pu o hese models as inpu o he niche models o he s udy species.
Ecological niche models. Fil e ing s a egies and bias iles. Da a clus e ing and bias can in luence en i-
onmen al niche model p edic ion59,60 and spa ial da a il e ing has been ecommended as an e ec i e measu e
o dec ease po en ial bias59. We i s plo ed occu ence eco ds on he Indian subcon inen map and emo ed
ob iously e oneous geo e e enced loca ions based on expe knowledge. Following ecommenda ions om
K ame -Schad e al.61 we p epa ed wo di e en ly il e ed da a se s. In he i s one, all duplica e eco ds wi hin
each cell (10 × 10km) we e emo ed (RD), while in he second one (RDbal) we andomly emo ed eco ds
un il we ob ained simila poin densi ies (balanced design) ac oss coun ies (Bangladesh, Bhu an, Nepal, Paki-
s an, S i Lanka, India) in he s udy a ea (see Fig.SM1.2). All da ase s we e deposi ed in Gi Hub (h ps ://gi hu
b.com/and e psil ade /P ion ailu us). Maxen assumes ha species occu ence da a a e unbiased, independen
samples om he dis ibu ion o he species (see MaxEn u o ial—h ps ://biodi e si yin o ma ics.amnh.o g/
open_sou c e/maxen /), and he e o e does no conside une en sampling e o . As a way o assessing possible
bias in ou da a, we included wo di e en ypes o bias iles (BM01 and BM001) in ou models61. In bo h iles,
a alue o 1 was gi en o cells wi h occu ence eco ds. Fo BM01, cells wi hou samples/ eco ds we e gi en
a alue o 0.1, indica ing 10% o he sampling e o compa ed o cells wi h occu ence eco ds, whe eas o
BM001 he same cells ecei ed a alue o 0.01.
MaxEn modelling and single‑desc ip o models. Small elid occu ence was modelled using a maxi-
mum en opy-based machine lea ning algo i hm ha es ima es he p obabili y dis ibu ion o a species’ occu -
ence based on en i onmen al cons ain s58. This algo i hm was selec ed because i has been demons a ed o
be among he mos obus when using only p esence eco ds62 and o small sample sizes63. Due o he high
numbe o explana o y a iables used in he analyses, in o de o be able o c ea e models con aining only he
mos impo an a iables wi hin each desc ip o (see hyb id model sec ion below), we i s buil sepa a e models
o each en i onmen al desc ip o wi h mo e han one a iable (i.e., clima e, land co e , human dis u bance
and p ey occu ence). Wi hin each desc ip o , o highly co ela ed ( > 0.65) a iables, we omi ed one om
he pai , e aining hose o easie biological in e p e a ion o ep esen ing ex eme en i onmen al condi ions
(e.g. empe a u e o he wa mes mon h) (TableSM1.3). All models we e un in Maxen 3.3.3k58 using 70% o
he eco ds as aining da a ( andom pa i ioning) wi h 10,000 backg ound poin s64,65. The emaining 30% o
he occu ence eco ds we e used o model e alua ion a oiding in la ing he ‘a ea unde he ecei ing ope a -
ing cha ac e is ic cu e’ (AUC) alues by using he same loca ions o aining and es ing. Fo each model, we
an 10 eplica es wi h 5000 i e a ions o allow o model con e gence, and used he subsample s a egy (i.e. he
p esence poin s a e epea edly spli in o andom aining and es ing subse s) as a o m o eplica ion in o de o
maximize he numbe o loca ions used o model es ing. We used Maxen ’s aw ou pu , which exp esses he el-
a i e p obabili y ha a cell con ains a p esence eco d, and consequen ly cells wi h ela i ely low aw alues may
s ill ha e a high absolu e p obabili y o p esence, albei lowe han o he cells. This ou pu a oids de aul assump-
ions on he p obabili y o p esence a loca ions wi h ‘‘a e age’ condi ions o he species, which a e necessa y
o he common logis ic ou pu 66,67. Following p e ious p oposals68, we used species-speci ic model uning o
ind he op imal se ings o Maxen models (see ull de ails in SM1), o alling 24 candida e models pe species.
4
Vol:.(1234567890)
Scien i ic Repo s | (2020) 10:18705 | h ps://doi.o g/10.1038/s41598-020-74684-8
www.na u e.com/scien i ic epo s/
Model selec ion and ela i e a iable impo ance. To ind he bes models in each se we i s selec ed a subse
o models wi h ‘a ea unde a ela i e ope a ing cha ac e is ic cu e’69 o he es da a (AUC es ) > 0.7. Then,
among hose, he ones wi h he lowes es omission a es ( alse nega i es) and AUCdi (AUC ain–AUC es ), a
measu e o o e i ing70, we e selec ed. Th ee h eshold me ics (MTP—minimum aining p esence, P10—10 h
pe cen ile aining p esence, ETS—equal es sensi i i y and speci ici y) we e used o measu e omission a es61.
Finally, o measu e spa ial o e lap be ween bes model p edic ions o each species we calcula ed Schoene ’s
o e lap me ic (D)71 and he modi ied Hellinge me ic (I)72 using he “calc.niche.o e lap” unc ion om he
‘ENMe al’ R-package73. Models wi h AUC es alues be ween 0.7 and 0.9 we e conside ed o ha e use ul accu-
acy, and models wi h AUC es > 0.9 we e conside ed o ha e high accu acy74. Following model selec ion, we
calcula ed a iable pe mu a ion impo ance wi hin each en i onmen al desc ip o ( o de ails on how a iable
impo ance was calcula ed see SM1). To aid ou decision ega ding he mos impo an a iables wi hin each
desc ip o , we c oss-checked hei impo ance o model gain and AUC es o each a iable (see Figs.SM1.2–
SM1.10).
Hyb id model. A species’ en i onmen al niche is likely o be in luenced by mo e han one en i onmen al
desc ip o . In o de o p oduce mo e ealis ic es ima es o he species en i onmen al niches, we buil a hyb id
model o each species. These models included only he mos impo an clima ic, opog aphic, land co e , human
dis u bance and p ey occu ence a iables o each species, as assessed h ough he a iable’s ela i e impo -
ance in he single-desc ip o models. This s a egy educes he numbe o a iables included in he models,
he eby dec easing he po en ial o o e i ing. Maximum co ela ion be ween a iables included in he same
hyb id model was 0.64 (TableSM1.3). To unde s and he ela i e impo ance o clima ic p edic o s compa ed o
o he abio ic and bio ic p edic o s, we also uned models and es ima ed he ela i e impo ance o each a iable
included in he hyb id model as we did o he single-desc ip o models. Finally, o assess i clima e-only models
we e good su oga es o mo e complex models inco po a ing non-clima ic in o ma ion, we calcula ed spa ial
o e lap be ween p edic ions o cu en ime om clima e-only models and om he bes hyb id models. Final
hyb id models used o calcula e niche o e lap and p o ec ed a ea co e age we e also e alua ed using speci ici y
and he symme ic ex emal dependence index (SEDI) based on he P10 h eshold75.
Spa ial o e lap and es s o niche conse a ism. To es po en ial niche conse a ism among he
s udy species we ollowed a ailable quan i a i e app oaches o niche e olu ion72. We used he con inuous aw
p edic ions om he bes hyb id models o calcula e Schoene ’s D and he modi ied Hellinge me ic. In addi ion,
we calcula ed hese me ics in en i onmen al space using he o iginal species occu ence eco ds, o accoun o
possible biases o igina ing om he ex en and dis ibu ion o en i onmen al g adien s in geog aphic space76.
Following he spa ial o e lap analysis, we used niche equi alency and simila i y es s o assess niche di e gence
in en i onmen al space. Following B oennimann e al.76, in he niche equi alency es all occu ences we e
pooled and hen andomly spli in o wo da ase s, each wi h he same o iginal numbe o occu ences. This p o-
cess was epea ed n imes and he dis ibu ion o simula ed alues was compa ed o he obse ed o e lap alue.
The null hypo hesis o niche equi alency canno be ejec ed i he obse ed alue alls wi hin he simula ed dis-
ibu ion. On he o he hand, gi en wo species (X and Y), he niche simila i y es e alua es simila i y be ween
cells whe e species X occu s and a andom se o cells, wi h sample size equal o ha o species Y, wi hin he
s udy a ea o Y72. Thus, in his es , o each pai o species, wo compa isons a e ca ied ou , one be ween spe-
cies X and he null densi y o occu ence o species Y, and ice e sa. Each compa ison is epea ed n imes o
gene a e a dis ibu ion o simula ed alues agains which he obse ed alue is compa ed. All es s we e ca ied
ou in he R-package ‘ecospa ’77 using 1000 eplica ions.
P o ec ed a ea co e age. To e alua e he ag eemen be ween species’ en i onmen al a ini ies and cu -
en spa ial conse a ion s a egies, he e measu ed as p o ec ed a ea co e age, we applied p esence h esholds
(MTP, P10, ETS) o ans o m species en i onmen al sui abili y p edic ions om he hyb id models in o bina y
maps o p edic ed species p esence. Nex , we ex ac ed shape iles o e es ial p o ec ed a eas om he Wo ld
Da abase on P o ec ed A eas78, excluding hose wi h a “P oposed” o “No Repo ed” s a us, and supe imposed
hem o e he bina y p edic ions o species occu ence o calcula e he mean p opo ion o he species’ bina y
p edic ions co e ed by p o ec ed a eas. All analyses and maps we e buil in R 3.4.379.
Species clima ic sui abili y h ough ime. As we de ec ed di e en impo an a iables o each species
unde cu en en i onmen al condi ions (see esul s below), we addi ionally es ima ed species esponse o en i-
onmen al change dynamics h ough ime. We p ojec ed he bes clima e-only models on o p ojec ions o pas
and u u e clima e, and hen calcula ed he p opo ion o he Indian subcon inen wi h sui able clima e o each
species a pa icula ime in e als. We did his only wi h clima ic da a because es ima es o o he en i onmen-
al a iables a e highly unce ain o unknown o pas and u u e condi ions. Mo eo e , du ing ini ial analyses
(da a no shown) we de ec ed ha clima ic p edic o s we e o e all among he mos , i no he mos , impo an
p edic o s o each species a he scale o analysis in his s udy.
To ep esen pas clima e condi ions we used he O o-Bliesne ’s simula ion model80 based on a gene al ci -
cula ion model CCSM2 o he pe iod ~ 140 o 120ka o he las in e glacial (LIG), which p edic s pa icula ly
wa m condi ions du ing he wa mes qua e o he Indian subcon inen (Fig.SM1.1). Fo he emaining ime
pe iods we used clima e p edic ions based on a Communi y Clima e Sys em Model (CCSM) bu we also included
clima e p edic ions om o he Gene al Ci cula ion Models (GCMs) o accoun o unce ain y in clima e model
p ojec ions. Speci ically, o he mid-Holocene (MH) and Las Glacial Maximum (LGM) we used he CCSM4,
MIROC-ESM, and MPI-ESM-P models, ollowing p e ious li e a u e81,82. Fo u u e condi ions (yea 2070), we
5
Vol.:(0123456789)
Scien i ic Repo s | (2020) 10:18705 | h ps://doi.o g/10.1038/s41598-020-74684-8
www.na u e.com/scien i ic epo s/
used models wi h sa is ac o y pe o mance o educed bias83. O hese, he ones downscaled, calib a ed, and
a ailable in he Wo ldClim da abase51, we e selec ed o p ojec ions. The Gene al Ci cula ion Models (GCMs)
used we e GFDL-CM3, MPI-ESM-LR, bcc-csm1-1-m, CCSM4, CNRM-CM5, HadGEM2-ES, and IPSL-CM5A-
LR. Fo p ojec ions o u u e clima e, wo emission scena ios we e conside ed: he Rep esen a i e Concen a ion
Pa hway (RCP) 4.5, an op imis ic scena io whe e emissions peak a ound 2040, and he RCP 8.5, a pessimis ic
scena io whe e emissions con inue o ise h ough he wen y- i s cen u y. Finally, ollowing ecommenda ions
by Nogués-B a o84, we also iden i ied geog aphical a eas in which pas and u u e non-analogous clima es a e
ou o he calib a ion ange (Figs.SM1.14, SM1.15), as his may inc ease he chance o nai e p ojec ions and
he e o e should be aken in o accoun when in e p e ing esul s (Figs.SM1.14, SM1.15).
Resul s
Va iable impo ance. The hyb id models indica ed ha clima e, land co e and opog aphy we e he
main de e minan s o species occu ence (Fig.1). Howe e , a iables wi h he highes pe mu a ion impo ance
wi hin each single-desc ip o model (see ull de ails in supplemen a y ma e ial—SM1), and he e o e included
in he hyb id models, we e di e en among species (Fig.1). P. bengalensis was nega i ely associa ed wi h high
empe a u e in he wa mes qua e (Bio 10), and P. ubiginosus was posi i ely associa ed wi h wa me em-
pe a u es in he coldes qua e (Bio 11) (Fig.SM1.32). P. i e inus was mainly associa ed wi h lowe ele a ion
a eas (Fig.SM1.33), whe eas he e was no single mos impo an a iable o F. chaus. Va iables associa ed wi h
human-modi ied land co e o human p esence also had conside able pe mu a ion impo ance, namely a nega-
i e associa ion wi h i iga ed in ensi e ag icul u e (P. ubiginosus) and posi i e associa ion wi h human- ela ed
landscape ea u es (P. bengalensis and P. ubiginosus) and human popula ion densi y (P. i e inus and F. chaus)
(Fig.SM1.34). The combina ions o en i onmen al a iables included in he hyb id models we e species-speci ic
(Fig.1). O e all, he bes hyb id models had low omission a es ( h eshold; mean ± sd) o P. bengalensis (P10;
0.09 ± 0.02), bu mode a e o P. i e inus (P10; 0.12 ± 0.03), P. ubiginosus (P10; 0.13 ± 0.06) and F. chaus (P10;
0.14 ± 0.04). Model p edic i e accu acy (mean AUC es ± sd) was high o P. bengalensis (0.90 ± 0.02), P. i e -
inus (0.93 ± 0.02) and use ul o P. ubiginosus (0.87 ± 0.02) and F. chaus (0.81 ± 0.01). O e all SEDI con i med
model use ulness o P. bengalensis (0.78 ± 0.02), P. i e inus (0.88 ± 0.01), P. ubiginosus (0.76 ± 0.03) and F.
chaus (0.55 ± 0.06). The bes hyb id models mainly included L, LQ o au o ea u es, bu di e en egula iza ion
Figu e1. Clima e, land co e and opog aphy ended o be, o e all, he mos impo an ac o s explaining
species occu ence. No ably, he mos impo an a iables we e species-speci ic. Va iables (see de ailed
desc ip ion in TableSM1) a e o de ed om highes o lowes pe mu a ion impo ance.

6
Vol:.(1234567890)
Scien i ic Repo s | (2020) 10:18705 | h ps://doi.o g/10.1038/s41598-020-74684-8
www.na u e.com/scien i ic epo s/
mul iplie s (RM1–RM2) and sampling s a egies. Full de ails o hyb id models can be ound in supplemen a y
ma e ial (TablesSM1.6–SM1.8).
Spa ial o e lap and niche conse a ism es s. Hyb id model p edic ions showed ha a eas wi h high
ela i e occu ence a e end o di e be ween P ionailu us species. P. bengalensis had highe occu ence a e
p edic ed o No heas India and along he mid and low ele a ions o he Himalayas and Wes e n Gha s, as well
as o S i Lanka (whe e he species is no known o occu ) (Fig.2a). The highes a e o occu ence o P. i e -
inus was, in u n, p edic ed o low ele a ions along he Himalayas and, in pa icula , Bangladesh, bu also in
S i Lanka (Fig.2a). In con as , he mos sui able a eas o P. ubiginosus we e es ima ed o ex end ac oss sou h
and cen al India. Fo F. chaus he e we e a ew pa icula a eas wi h highe occu ence a es along sou heas e n
India, Eas e n Gha s and he Himalayan lowlands, bu op imal condi ions seem o be widesp ead h oughou
he subcon inen (Fig.2a). Al hough he sis e species P. bengalensis and P. i e inus had a highe es ima e o
o e lap (Schoene ’s D: 0.340 ± 0.077), his was no conside ably g ea e han he es ima es ob ained be ween P.
ubiginosus and P. bengalensis (Schoene ’s D: 0.257 ± 0.024) and be ween P. ubiginosus and P. i e inus (Sch-
oene ’s D: 0.184 ± 0.050) (Fig.2b). A endency o g ea e spa ial o e lap be ween P ionailu us species and he
ou g oup was obse ed, bu s ill wi hin he ange o he highes o e lap es ima es among P ionailu us spe-
cies (Schoene ’s D o F. chaus s. P. i e inus: 0.377 ± 0.049; F. chaus s P. ubiginosus: 0.422 ± 0.021; F. chaus
s P. bengalensis: 0.451 ± 0.041) (Fig.2b). When conside ing ull en i onmen al space (de ined by all a iables
included in hyb id models), niche equi alency could no be ejec ed (P = 0.325) be ween P. i e inus and P.
ubiginosus and be ween P. i e inus and F. chaus (Fig.SM1.38). Niche simila i y be ween species could also no
be ejec ed (P = 0.170) in any o he es s (Figs.SM1.38, SM1.39).
P o ec ed a ea co e age. P o ec ed a ea co e age was o e all low (min = 0.056; max = 0.105) and he -
e ogeneous among species (Fig.2c). Lowe co e age was ound o F. chaus (0.062 ± 0.004) and P. i e inus
(0.071 ± 0.012), and he highes co e age was ound o P. bengalensis (0.088 ± 0.006). P. ubiginosus had in e -
media e co e age (0.080 ± 0.001). Pa e ns o p o ec ed a ea co e age we e simila ac oss he di e en h esholds
es ed, wi h he mos no able di e ences being ha he es ima es we e sligh ly lowe o P. i e inus and highe
o F. chaus when using he MTP h eshold (TableSM1.10, Fig.SM1.40).
Species clima ic sui abili y h ough ime. The impo ance o pa icula clima ic a iables o each spe-
cies led o species-speci ic ends o clima ic sui abili y h ough ime (Fig.3). Fu u e (yea 2070) clima ic sui abil-
i y is es ima ed o dec ease o P. bengalensis and F. chaus, compa ed o cu en ime, unde bo h RCP scena ios.
Clima ic sui abili y o P. bengalensis appea s o ha e been declining conside ably since a leas he mid-Holo-
cene (~ 6ka), bu he nega i e end will accele a e gi en he p edic ed human-induced clima e change. Fo P.
ubiginosus, models es ima ed an inc ease in clima ic sui abili y om he las in e glacial (~ 140 o 120ka) o he
Figu e2. (a) Species ela i e occu ence a es (ROR), displayed as log(ROR), in he Indian Subcon inen as
p edic ed by he bes hyb id models; (b) Species spa ial o e lap (Schoene ’s D). Simila pa e ns we e ound
wi h he modi ied Hellinge me ic I (Fig.SM1.37); (c) P o ec ed a ea co e age o species po en ial occu ence
(calcula ed based on bina y maps—P10 h eshold); Black illed do s co espond o p edic ions o he bes
models. Simila pa e ns we e ound wi h p o ec ed a ea co e age calcula ed using di e en h esholds (MTP,
ETS) (Fig.SM1.40).
7
Vol.:(0123456789)
Scien i ic Repo s | (2020) 10:18705 | h ps://doi.o g/10.1038/s41598-020-74684-8
www.na u e.com/scien i ic epo s/
Las Glacial Maximum (~ 22ka), and s abili y since hen up o 2070. P edic ions o P. i e inus we e less clea
as hey a ied acco ding o he bes models conside ed. While mos models in e ed a sligh inc ease in median
clima ic sui abili y since he LGM, a ew models es ima ed a d op in clima ic sui abili y since he mid-Holocene,
con inuing in he o ecas ed u u e. Es ima ed ends based on MTP and ETS h esholds showed simila pa e ns
o hose using P10 h esholds (Figs.SM1.16, SM1.17). Species-speci ic esponses o clima ic ac o s also explain
he di e en p edic ed clima e e ugia a eas among he s udied species, wi h e ugia only o e lapping po en ially
in he lowlands o Bangladesh and Wes Bengal (Fig.SM1.23). The conside able o e lap be ween he p edic ions
o clima e-only and hyb id models o he p esen ime (Fig.SM1.27), wi h he excep ion o a la ge di e ence
o P. i e inus, sugges s ha clima e-only model p ojec ions o he pas and u u e may closely app oxima e
mo e complex models.
Discussion
We show ha P ionailu us species ha e less o simila spa ial o e lap wi h each o he han in ela ion o an ou -
g oup species. Also, wi hin P ionailu us he e was no g ea e spa ial o e lap be ween sis e species, wi h niche
equi alency ejec ed in en i onmen al space. The e o e, we ound no suppo o phylogene ic niche conse a-
ism wi hin P ionailu us. Ins ead, pa icula en i onmen al niche cha ac e is ics esul in species-speci ic en i-
onmen al esponses, ansla ing in o po en ially he e ogeneous esponses o u u e clima e change. Mo eo e ,
cu en a eas o high en i onmen al sui abili y o he di e en species a e no equally, and adequa ely, co e ed
by he exis ing p o ec ed a ea ne wo k. In his scena io mul ispecies spa ial conse a ion planning becomes
challenging.
Insigh s on species esponses o global en i onmen al change. Ou s udy highligh s ha clima ic
sui abili y o closely ela ed species can be species-speci ic. This inc eases he body o e idence showing lack o
phylogene ic niche conse a ism in mammalian species, and ein o ces he impo ance o unde s anding deep-
ime species his o y and specia ion mechanisms be o e assuming common esponses and conse a ion s a e-
gies delinea ion24,85. In pa icula , he pa e ns ound ma ch wi h a specia ion p ocess ela ed o en i onmen al
di e gence, simila o ha ound in o he opical mammals22,23. Such p ocess would ha e equi ed a po en ial
niche shi om a common ances o , hence sugges ing ha u u e esponse o en i onmen al change may be
de e mined by o he key ecological p ocesses such as dispe sal and bio ic in e ac ions, po en ially wi h species
wi h simila niche bu no necessa ily closely ela ed. This s esses he necessi y o no ocusing only on he
sea ch o common conse a ion a eas bu also on main aining connec i i y and communi y s uc u e.
Fu he , assuming abio ic niche s abili y, ou esul s appa en ly sugges ha endo he mic species wi h a opi-
cally es ic ed geog aphic ange (P. ubiginosus and P. i e inus) may see hei occu ence a eas main ained o
inc eased wi h global wa ming, whe eas mo e he mal gene alis species (P. bengalensis and F. chaus) may no
be able o cope well wi h highe empe a u es in he opical pa s o hei anges (Fig.3). This is in con as o
p edic ions o empe a e egions, whe e global wa ming is p edic ed o ha e a nega i e e ec on species wi h a
Figu e3. Species clima ic sui abili y (P10 h eshold) in he Indian subcon inen since he las in e glacial
(LIG; ~ 140 o 120ka) up o 2070 ( o an op imis ic scena io, RCP 45, and a business-as-usual scena io, RCP
8.5). Colo s indica e he clima e models used o pas and u u e p ojec ions. LGM—Las Glacial Maximum
(~ 22ka); MH—Mid-Holocene (~ 6ka); Cu en ime (1950–2000). Spa ial ep esen a ion o mean aw
p edic ions o each ime pe iod is p o ided in SM1 (Fig.SM1.18–SM1.21).
8
Vol:.(1234567890)
Scien i ic Repo s | (2020) 10:18705 | h ps://doi.o g/10.1038/s41598-020-74684-8
www.na u e.com/scien i ic epo s/
p e e ence o cold clima e, and a posi i e e ec on mo e he mal gene alis species86–89. This is also con a y o
p edic ions o opical ec o he ms, which ha e been sugges ed o be pa icula ly ulne able o clima e wa ming
because o hei physiological sensi i i y o empe a u e changes90,91.
In addi ion we ound indica ions ha occu ence con ac ions may ha e begun be o e an h opogenic impac s
al hough hese may ha e con ibu ed o he end. Fo example, i was es ima ed ha P. bengalensis’ clima ic
sui abili y may ha e been dec easing since he LGM. Al hough he impo ance o na u al dynamics o clima ic
sui abili y h ough ime o explaining cu en biodi e si y pa e ns has been long known84,92, and he e is ecen
wo k acili a ing he unde s anding o such dynamics93,94, ou s udy highligh s he need o ways o disen angle
na u al species’ sui abili y dynamics and human-d i en species dis ibu ion changes, so ha bo h can be aken
in o accoun when p edic ing species ulne abili y o global change.
Implica ions o P ionailu us ca s’ conse a ion. P ionailu us species a e unequally co e ed by he
cu en p o ec ed a ea ne wo k. P o ec ed a eas a e o en loca ed in o es ed egions (gene ally assumed o be
associa ed wi h highe species ichness), in habi a s used by cha isma ic species, such as ige ese es in India,
o in places less sui able o humans due o ough opog aphy o low economic po en ial95–97. This is pe haps
he eason why P. bengalensis, which is associa ed wi h opical semi-e e g een o es s and can inhabi highe
al i udes, has g ea e co e age o i s ange by p o ec ed a eas han he o he species. I is impo an o no e ha
he me ic used o measu e p o ec ed a ea co e age is sensi i e o he ex en o he species’ geog aphic ange,
so ha species wi h a la ge dis ibu ion a ea, such as F. chaus, end o ha e a smalle p opo ion o i co e ed
by p o ec ed a eas, al hough in absolu e e ms hey may ha e mo e geog aphic ange co e ed. O e all, he low
es ima es o p o ec ed a ea co e age (below 10%) indica e impo an gaps in he cu en p o ec ed a ea ne wo k
in he Indian subcon inen . Mo eo e , i is impo an o no e ha he coa se g ain size used in he analyses can
lead o o e p edic ion o p esence a ea ( his o e p edic ion being a iable be ween species). This is pa icula ly
possible in he e ogeneous landscapes98,99 such as he Himalayas o he Wes e n Gha s. The e o e, he low o e all
es ima e o p o ec ed a ea co e age may s ill, in ac , be an o e es ima ion.
This s udy highligh s clima ic change and human-d i en land con e sion as he wo main mac oscale ac o s
nega i ely in luencing he occu ence o P ionailu us ca s in he Indian subcon inen . Clima ic sui abili y sh ink-
age is pa icula ly e iden o P. bengalensis. Fo P. bengalensis, inc eased sui abili y om he LIG o LGM is in
ag eemen o clima ic sui abili y expansion ound o o es oden species du ing he same pe iod in Sou heas
Asia100. Also, a endency o highe , bu o e lapping, sui able clima ic a ea du ing he LGM compa ed o he mid
Holocene esembles he pa e n desc ibed o o he Indochinese mammals101. Mo eo e , conside able gene ic
di e en ia ion has been ound be ween no he n and sou he n Indian Leopa d Ca popula ions40, which may
ha e been exace ba ed by la e Holocene en i onmen al e en s, as p edic ed by he d op in clima ic sui abili y
om he mid Holocene o he p esen ime. While clima ic sui abili y ac oss he Himalayas appea s o ha e been
s able o e ime (Fig.SM1.18), i is much mo e dynamic in sou h and cen al India, wi h an expec ed u u e
decline in he Eas e n and Wes e n Gha s. In ac , he models p edic a comple e loss o clima ic sui abili y in he
Eas e n Gha s unde he mos pessimis ic RCP scena io; on he o he hand, in he Wes e n Gha s he comple e
loss o clima ic sui abili y is no expec ed unde any RCP scena io (Fig.SM1.18). The Wes e n Gha s is ac ually
a clima e e uge (Fig.SM1.23) whe e he popula ions o P. bengalensis ha bo unique gene ic di e si y40,102 and
he species is commonly de ec ed40,49,103. Obse a ions in he Eas e n Gha s a e, in con as , much spa se (bu
see o ins ance87). The e o e, mi iga ing cu en an h opogenic h ea s o he popula ions o P. bengalensis in
he Wes e n Gha s should be a p io i y o a oid a cumula i e e ec o human impac s and na u al oscilla ions o
clima ic sui abili y. This s a egy has he ad an age o main aining a g ea e e olu iona y po en ial in he species
and allows he possibili y o e-expansion du ing pe iods o clima e cooling, as es ima ed o ha e occu ed a e
he las in e glacial (Fig.SM1.18). I is impo an o no e ha F. chaus, o which a demog aphic expansion has
been es ima ed a 271–166ka40, a ime in e al ha includes he in e glacials o Ma ine Iso ope S age (MIS) 7104,
and is now a widesp ead species in he Indian subcon inen , may no be immune o u u e clima ic change, since
disconnec ion be ween clima ic sui abili y pa ches in he Wes e n Gha s and cen al-no h India is p edic ed
(Fig.SM1.21), a pa e n simila o ha cu en ly in e ed o P. bengalensis.
Fo P. i e inus, he consequences o clima e change a e unclea , since he selec ed bes clima e-only models
showed conside able a ia ion in p edic ed u u e clima e sui abili y (Fig.3). Such a ia ion could a ise om a i-
a ion in he clima e models used o ep esen pas and u u e clima e scena ios, bu Maxen models p ojec ed in o
he same clima e scena ios s ill show a ied p edic ions (Fig.3), sugges ing ha p edic ion a ia ion o igina es
om model i ing ins ead. Consequen ly, we canno exclude a po en ial dec ease in he clima ic sui abili y o he
species unde he clima e change scena ios used he e. Howe e , a p elimina y Cy b s udy (Shomi a Mukhe jee,
unpublished da a) in e ed connec i i y be ween No he n and Eas e n India popula ions ha is compa ible
wi h s able clima ic sui abili y o e ime. Ou a e aged p edic ions o e ime a e also b oadly cohe en wi h pas
popula ion dynamics in e ed o hei Indochinese ange om mi ochond ial DNA105. Gi en his unce ain y, we
ecommend conse a ion ocus on a eas wi h s able clima ic sui abili y o e ime, namely he lowlands o Bang-
ladesh and Wes Bengal (Fig.SM1.23). We also no ed a disag eemen be ween p edic ed sui able en i onmen
om hyb id models (i.e., he undamen al niche) and he highly agmen ed dis ibu ion a eas o he species35.
This may be an indica ion ha unconside ed ac o s, such as illegal killing106 and oad kills107, may cu en ly
es ic he occu ence o he species. Un o una ely, P. i e inus is likely bene i ing li le om p o ec ed land
(showed he lowes p o ec ed a ea co e age—Fig.2C). We he e o e sugges inc easing p o ec ed habi a o he
species and echo he need o u gen local conse a ion ac ions35, especially in he lowlands o Bangladesh and
Wes Bengal, he clima e e ugia iden i ied in 90% o he models.
P. ubiginosus was conside ed an endemic species o sou h India and S i Lanka, bu has ecen ly been de ec ed
in no h India and Nepal108–110. Al hough his could be a consequence o he ecen inc ease in came a- ap
9
Vol.:(0123456789)
Scien i ic Repo s | (2020) 10:18705 | h ps://doi.o g/10.1038/s41598-020-74684-8
www.na u e.com/scien i ic epo s/
su eys, a phenomenon ha is likely con ounding es ima ion o popula ion ends o o he species in India111,
i may be he esul o na u al expansion o P. ubiginosus in o p e iously coole a eas due o clima e wa ming.
The Himalayan lowlands we e no iden i ied as clima ic e ugia (Fig.SM1.23) bu may ha e sui able clima e in
he u u e (Fig.SM1.20). We he e o e ecommend ha popula ion iabili y in clima ic e ugia (sou h India and
S i Lanka, Fig.SM1.23) should be ensu ed, and ha explo a o y su eys should be conduc ed h oughou he
Himalayan lowlands o in es iga e a po en ial no hwa d expansion o he species.
Al hough we ha e con idence in ou models, we ad ise cau ion ega ding p edic ions o dese a eas in
Pakis an and India, as well as o cen al India, unde mo e ex eme u u e clima e condi ions (RCP8.5), as he
empe a u e o he wa mes qua e (Bio 10) can exceed he ange used o model calib a ion (Fig.SM1.11). The
same should be aken in o accoun when conside ing he in luence o he empe a u e o he coldes qua e
(Bio 11) in sou h India (Fig.SM1.12).
I should also be s essed ha o he d i e s o en i onmen al change (e.g., land use) can exace ba e o mi iga e
changes in clima ic sui abili y, and possible in e ac i e e ec s ha e ecen ly been desc ibed112,113. Fo example,
he nega i e ela ionship o P. ubiginosus wi h in ensi e i iga ed ag icul u e sugges s ha No h India´s c op-
land bel a ound he Ganga basin is a majo unsui able a ea sepa a ing sui able en i onmen in sou h India and
sui able habi a pa ches in he Himalayan lowlands (Fig.2A). The expec ed inc ease in c opland114 is he e o e
likely o u he nega i ely impac he species. The ela ionship o he species wi h human dis u bance was no ,
howe e , s aigh o wa d, as hey all showed a high p obabili y o occu ence close o human s uc u es o wi h
inc easing human popula ion densi y, al hough human dis u bance a iables ended o be o low impo ance.
This posi i e ela ionship may be an a i ac due o biased da a ega ding human p esence (despi e co ec ions
wi h bias iles) o may co espond o species’ use o human-modi ied en i onmen s, due o ins ance o g ea e
esou ce a ailabili y (e.g., p ey abundance), as we de ec ed associa ion o some oden gene a wi h human se le-
men s and popula ion densi y (Fig.SM2.3). The in ica e link be ween human and oden p ey p esence, oge he
wi h he impo ance o clima ic ac o s o mos o he gene a o oden p ey (Fig.SM2.3) and he species-speci ic
esponses o each small ca species, poin o complex esponses o global change.
Recei ed: 2 Ma ch 2020; Accep ed: 5 Oc obe 2020
Re e ences
1. Powe s, R. P. & Je z, W. Global habi a loss and ex inc ion isk o e es ial e eb a es unde u u e land-use-change scena ios.
Na . Clim. Chang. 9, 323–329 (2019).
2. F anklin, J. Mapping Species Dis ibu ions: Spa ial In e ence and P edic ion (Camb idge Uni e si y P ess, Camb idge, 2010).
3. Zanin, M. & dos Ne es, B. S. Cu en elid (Ca ni o a: Felidae) dis ibu ion, spa ial bias, and occu ence p edic abili y: es ing
he eliabili y o a global da ase o mac oecological s udies. Ac a Oecol. 101, 103–488 (2019).
4. Lomolino, M. V. & Heaney, L. R. F on ie s o Biogeog aphy: New Di ec ions in he Geog aphy o Na u e. (sidalc.ne , 2004).
5. Meye , C., K e , H., Gu alnick, R. & Je z, W. Global p io i ies o an e ec i e in o ma ion basis o biodi e si y dis ibu ions.
Na . Commun. 6, 1–8 (2015).
6. Pe e son, A. T., Sobe on, J. & Sanchez-Co de o, V. Conse a ism o ecological niches in e olu iona y ime. Science 285, 1265–
1267 (1999).
7. Raxwo hy, C. J. e al. P edic ing dis ibu ions o known and unknown ep ile species in Madagasca . Na u e 426, 837–841
(2003).
8. Hu, J. e al. Niche conse a ism in Gynand opaa ogs on he sou heas e n Qinghai-Tibe an Pla eau. Sci. Rep. 6, 32624 (2016).
9. Mo iniè e, J. e al. Phylogene ic niche conse a ism explains an in e se la i udinal di e si y g adien in eshwa e a h opods.
Sci. Rep. 6, 26340 (2016).
10. Liu, H. e al. S ong phylogene ic signals and phylogene ic niche conse a ism in ecophysiological ai s ac oss di e gen lineages
o Magnoliaceae. Sci. Rep. 5, 12246 (2015).
11. Liu, H., Edwa ds, E. J., F eckle on, R. P. & Osbo ne, C. P. Phylogene ic niche conse a ism in C4 g asses. Oecologia 170, 835–845
(2012).
12. C isp, M. D. e al. Phylogene ic biome conse a ism on a global scale. Na u e 458, 754–756 (2009).
13. Lo enzen, E. D. e al. Species-speci ic esponses o La e Qua e na y mega auna o clima e and humans. Na u e 479, 359–364
(2011).
14. Agui e-Gu ié ez, J., Se na-Cha ez, H. M., Villalobos-A ambula, A. R., de la Rosa, J. A. P. & Raes, N. Simila bu no equi alen :
ecological niche compa ison ac oss closely- ela ed Mexican whi e pines. Di . Dis . 21, 245–257 (2014).
15. Pe e , D. L., Leslie, A. B. & Sax, D. F. Na u alized dis ibu ions show ha clima ic disequilib ium is s uc u ed by niche size in
pines (Pinus L.). Glob. Ecol. Biogeog . 28, 429–441 (2018).
16. G aham, C. H., Ron, S. R., San os, J. C., Schneide , C. J. & Mo i z, C. In eg a ing phylogene ics and en i onmen al niche models
o explo e specia ion mechanisms in dend oba id ogs. E olu ion 58, 1781–1793 (2004).
17. Kozak, K. H. & Wiens, J. J. Clima ic zona ion d i es la i udinal a ia ion in specia ion mechanisms. P oc. Biol. Sci. 274, 2995–3003
(2007).
18. Moussalli, A., Mo i z, C., Williams, S. E. & Ca na al, A. C. Va iable esponses o skinks o a common his o y o ain o es
luc ua ion: conco dance be ween phylogeog aphy and palaeo-dis ibu ion models. Mol. Ecol. 18, 483–499 (2009).
19. A onso Sil a, A. C. e al. T opical specialis s. clima e gene alis : Di e si ica ion and demog aphic his o y o sis e species o
Ca lia skinks om no hwes e n Aus alia. Mol. Ecol. 26, 4045–4058 (2017).
20. Logan, M. L., Huynh, R. K., P ecious, R. A. & Calsbeek, R. G. The impac o clima e change measu ed a ele an spa ial scales:
new hope o opical liza ds. Glob. Chang. Biol. 19, 3093–3102 (2013).
21. Mo i z, C. e al. Impac o a cen u y o clima e change on small-mammal communi ies in Yosemi e Na ional Pa k, USA. Science
322, 261–264 (2008).
22. Kamila , J. M. & Muldoon, K. M. The clima ic niche di e si y o malagasy p ima es: a phylogene ic pe spec i e. PLoS ONE 5,
e11073 (2010).
23. B az, A. G., Lo ini, M. L. & Vale, M. M. Clima e change is likely o a ec he dis ibu ion bu no pa apa y o he B azilian
ma mose monkeys (Calli h ix spp.). Di . Dis . 25, 536–550 (2018).
24. Coope , N., F eckle on, R. P. & Je z, W. Phylogene ic conse a ism o en i onmen al niches in mammals. P oc. Biol. Sci. 278,
2384–2391 (2011).