Response o ba s and noc u nal insec s o u ban g een a eas in
Eu ope
Lucía Villa oya-Villalba
a,b,1,
*, Joan Casanelles-Abella
a,c,1
, Ma co Mo e i
a
,
Ped o Pinho
d
, Roeland Samson
e
, Anskje Van Mensel
e
, F an¸cois Chi on
,
Flo ian Zellwege
g
, Ma in K. Ob is
a
a
Biodi e si y and Conse a ion Biology, Swiss Fede al Resea ch Ins i u e WSL, CH-8903 Bi mensdo , Swi ze land
b
Uni e si 
e de Mon pellie , F-34090 Mon pellie , F ance
c
Landscape Ecology, Ins i u e o Te es ial Ecosys ems, ETH Z€
u ich, CH-8049 Zu ich, Swi ze land
d
Cen e o Ecology, E olu ion and En i onmen al Changes, Faculdade de Ci^
encias, Uni e sidade de Lisboa,
P-1749-016 Lisboa, Po ugal
e
Lab o En i onmen al and U ban Ecology, Resea ch G oup En i onmen al Ecology & Mic obiology (ENdEMIC),
Dep . Bioscience Enginee ing, Uni e si y o An we p, B-2020 An we p, Belgium
Uni e si 
e Pa is-Saclay, CNRS, Ag oPa isTech, Ecologie Sys 
ema ique E olu ion, F-91405 O say, F ance
g
Fo es Resou ces and Managemen , Swiss Fede al Resea ch Ins i u e WSL, CH-8903 Bi mensdo , Swi ze land
Recei ed 2 June 2020; accep ed 21 Janua y 2021
A ailable online 22 Janua y 2021
Abs ac
Animal biodi e si y in ci ies is gene ally expec ed o be uni o mly educed, bu ecen s udies show ha his is modula ed by
he composi ion and configu a ion o U ban G een A eas (UGAs). UGAs ep esen a he e ogeneous ne wo k o ege a ed
spaces in u ban se ings ha ha e epea edly shown o suppo a significan pa o na i e diu nal animal biodi e si y. Howe e ,
noc u nal axa ha e so a been unde s udied, cons aining ou unde s anding o he ole o UGAs on main aining ecological
connec i i y and enhancing o e all biodi e si y. We p esen a well- eplica ed mul i-ci y s udy on he ac o s d i ing ba and
noc u nal insec biodi e si y in h ee Eu opean ci ies. To achie e his, we sampled ba s wi h ul asound eco de s and flying
insec s wi h ligh aps du ing he summe o 2018. Resul s showed a g ea e abundance and di e si y o ba s and noc u nal
insec s in he ci y o Zu ich, ollowed by An we p and Pa is. We iden ified a ificial ligh ing in he UGA o lowe ba di e si y
by p obably fil e ing ou ligh -sensi i e species. We also ound a nega i e co ela ion be ween bo h ba ac i i y and di e si y
and insec abundance, sugges ing a op-down con ol. An in-dep h analysis o he Zu ich da a e ealed di e gen esponses o
he noc u nal auna o landscape a iables, while poin ing ou a bo om-up con ol o insec di e si y on ba s. Thus, o e ec-
i ely p ese e biodi e si y in u ban en i onmen s, UGAs managemen decisions should ake in o accoun he combined eco-
logical needs o ba s and noc u nal insec s and conside he specific spa ial opology o UGAs in each ci y.
© 2021 The Au ho s. Published by Else ie GmbH on behal o Gesellscha ü Ökologie. This is an open access a icle unde
he CC BY license (h p://c ea i ecommons.o g/licenses/by/4.0/)
Keywo ds: U banisa ion; U ban biodi e si y; Noc u nal auna; Flying a h opods; Chi op e a; An we p; Pa is; Zu ich
*Co esponding au ho a : Biodi e si y and Conse a ion Biology, Swiss Fede al Resea ch Ins i u e WSL, CH-8903 Bi mensdo , Swi ze land.
E-mail add ess: [email p o ec ed] (L. Villa oya-Villalba).
1
These wo au ho s sha e fi s au ho ship.
h ps://doi.o g/10.1016/j.baae.2021.01.006
1439-1791/© 2021 The Au ho s. Published by Else ie GmbH on behal o Gesellscha ü Ökologie. This is an open access a icle unde he CC BY license
(h p://c ea i ecommons.o g/licenses/by/4.0/)
Basic and Applied Ecology 51 (2021) 5970 www.else ie .com/loca e/baae
In oduc ion
U banisa ion is a he e ogeneous, wo ldwide phenomenon
imposing impo an challenges o wildli e conse a ion.
Rapid u banisa ion in he las decades nega i ely impac s
biodi e si y in complex ways (e.g. Pa is, 2016;
Ri kin e al., 2019). Howe e , he specific e ec s o u bani-
sa ion a e nei he linea no cons an , bu a y amongs and
wi hin ci ies (Beninde, Vei h & Hochki ch, 2015). In
Eu ope, mos ci ies ha e de eloped in o me ag icul u al
lands sha ing simila en i onmen al his o ies. None heless,
hey a e s ill he e ogeneous in hei ci yscape. Al hough his
he e ogenei y has been no iced p e iously (Ramalho &
Hobbs, 2012), mos u ban ecology s udies s ill ollow a sin-
gle-ci y app oach hampe ing he ex apola ion o he esul s
o o he ci ies (Beninde e al., 2015). Finally, despi e u bani-
sa ion educing he amoun o a ailable habi a , ci ies also
con ain a ne wo k o u ban g een a eas (UGAs) ha ha e
been shown o be a key ac o enhancing biodi e si y
(McIn y e, Rango, Fagan & Fae h, 2001;Sa le , Duelli,
Ob is , A le az & Mo e i, 2010).
In mos Eu opean ci ies, densely buil -up dis ic s wi h li -
le o no g een a eas ep esen a small ac ion o he whole
u ban a eas. The majo i y o he u ban dis ic s ha e in e -
media e le els o u banisa ion and con ain a a iable and
some imes dominan p opo ion o UGA. UGAs ep esen a
ne wo k o highly he e ogeneous pa ches usually dis ibu ed
o ming mosaics, such as pa ks, g een oo s o ee pi s
(Lepczyk e al., 2017). Locally, UGAs s ongly a y in
many ea u es such as size, s uc u e, ege a ion composi-
ion, wa e a ailabili y, owne ship o managemen . Fo
ins ance, UGAs may ange om la ge si es wi h a complex,
mul i-laye ed ege a ion s uc u e o linea , highly managed
lawn s ipes. A he landscape scale, he con ibu ion o
UGAs o he habi a amoun is media ed by hei composi-
ion and configu a ion as well as by he pe meabili y o he
su ounding ma ix (Lepczyk e al., 2017). Al oge he , his
hampe s disen angling he ac o s shaping biodi e si y and
ul ima ely making conse a ion assessmen s. Finding ools
o success ully measu e and syn hesise hese complex ela-
ionships is a key s ep owa ds a mo e holis ic u ban biodi-
e si y managemen .
A long-las ing p oblem in u ban ecology is how o eli-
ably in e he e ec s o u ban in ensifica ion (e.g. modifica-
ions on habi a amoun , he e ogenei y, dis u bances,
s esso s o isola ion) on biodi e si y. Remo e sensing ools
such as LiDAR (i.e. Ligh De ec ion and Ranging) o a ifi-
cial ligh a nigh (ALAN) maps a e becoming mo e accessi-
ble, and ha e been p o en o be good p oxies o habi a
amoun and dis u bance, espec i ely (S
anchez De
Miguel e al., 2019). S ill, he po en ial o emo e sensing
ools emains unde u ilised as ew s udies o da e ha e
included his ype o me ics (bu see Hale, Fai b ass, Ma -
hews, Da ies & Sadle , 2015;Zellwege e al., 2016), p ob-
ably because hei po en ial is no ye ully unde s ood.
Habi a , land-co e and land-use maps a e gene ally
a ailable o mos ci ies in Eu ope and used as p oxies o
habi a amoun , he e ogenei y and connec i i y (e.g.
Munzi e al., 2014). Howe e , hey significan ly a y in
g ain and mos impo an ly, in he ecological in o ma ion
used o define he di e en mapping ca ego ies. Finally,
ALAN maps ha e been used o show he esponses o a
wide a ie y o o ganisms o ligh pollu ion (Hale e al.,
2015;Knop e al., 2017). Ne e heless, he e is limi ed
unde s anding o he e ec s o a ificial ligh ning on bio ic
communi ies (Sande s & Gas on, 2018).
Ba s and noc u nal insec s ep esen an unde s udied
assemblage showing ecological ea u es ha make hem a
s iking g oup o moni o he e ec s o u banisa ion. Fi s ,
hey ep esen a p ey-p eda o sys em as Eu opean ba s eed
on noc u nal insec s. P ey-p eda o sys ems migh exhibi
wo ypes o esponses owa ds modifica ions o he habi a
amoun o dis u bances: bo om-up, such as g ea e eeding
ba ac i i y wi h inc easing insec biomass (Th el all, Law
& Banks, 2012a) o op-down, o ins ance insec pes con-
ol by ba s (Puig-Mon se a e al., 2020). Second, noc u nal
animals, such as ba s and noc u nal insec s, migh ha e a di -
e en suscep ibili y owa ds an h opogenic dis u bances (e.
g. ligh pollu ion) han diu nal ones. Fu he , hey s ongly
di e om diu nal o ganisms in e ms o he o ien a ion sys-
em hey use o mo e and o age. These wo ai s egula e
he scale a which he o ganisms pe cei e he en i onmen
and hus he modifica ions and dis u bances ha occu
(Concepci
on, Mo e i, Al e ma , Nobis & Ob is , 2015).
Finally, ba s can be classified in h ee gene al guilds (i.e.
long-, mid- and sho - ange echoloca o s) wi h di e ences
in hei o aging s a egies (see F ey-Eh enbold, Bon adina,
A le az & Ob is , 2013;F oide aux, Zellwege , Bollmann
& Ob is , 2014). The e o e, ba s can be conside ed a po en-
ial bioindica o o land-use changes, pa icula ly u banisa-
ion, bu s udies on hese g oups a e s ill sca ce (bu see
Jones, Jacobs, Kunz, Wilig & Racey, 2009).
P io esea ch es ing he e ec s o u banisa ion on ba s
and noc u nal insec s has yielded mixed esul s. Gene ally,
u banisa ion has been epo ed o simpli y ba communi ies
by fil e ing ou sensi i e species and keeping hose wi h gen-
e alis ic ai s leading o a ce ain deg ee o bio ic homogeni-
sa ion (Russo & Ancillo o, 2015), a p ocess documen ed
also o o he axa (Chong e al., 2014; bu see
Fou nie , F ey & Mo e i, 2020). Fo ins ance, ea lie
esea ch has e ealed a dec ease in ba di e si y ollowing
educ ions o he amoun o a ailable habi a due o u ban
in ensifica ion (e.g. Ku a & Te amino, 1992). None heless,
o he u ban in ensifica ion d i e s ha e unclea e ec s.
Ligh pollu ion has been shown o nega i ely impac many
g oups, pa icula ly noc u nal animals (S one, Ha is &
Jones, 2015). S ill, some s udies p opose ha his dis u -
bance migh no ha e a significan e ec on ba s, as hey
could mo e o mo e sui able pa ches (K auel & LeB-
uhn, 2016); o e en be ad an ageous o hem. Fo ins ance,
s ee ligh s on linea pa hs a ac la ge amoun s o noc u nal
insec s acili a ing o aging o ligh -oppo unis ic ba s
60 L. Villa oya-Villalba e al. / Basic and Applied Ecology 51 (2021) 5970
(Russo & Ancillo o, 2015). Con e sely, ba species wi h
low ole ance o ligh pollu ion (i.e. ligh -a e se) a e fil e ed
ou (S one e al., 2015). On he o he hand, he e ec s o
u banisa ion on noc u nal insec s ha e been li le in es i-
ga ed bu some pa e ns ha e been epo ed. Simila ly o
ba s, a nega i e ela ionship appea s be ween building den-
si y and noc u nal insec di e si y (K auel & LeBuhn, 2016;
Russo & Ancillo o, 2015). Th el all e al. (2012a) sugges
ha he nega i e e ec s on noc u nal insec s’biomass could
be explained by he inc ease in impe ious su aces, leading
o loss and agmen a ion o a ailable habi a and, hus, o
low p ima y p oduc i i y. On he con a y, u ban na i e eg-
e a ion has been p oposed o sus ain noc u nal insec popula-
ions (Russo & Ancillo o, 2015). Thus, i is expec ed ha
insec abundance and i s d i e s g ea ly influence ba dis i-
bu ion in u ban a eas (K auel & LeBuhn, 2016).
In his pape , we s udied he influence o bio ic and abi-
o ic ac o s in shaping he di e si y, ac i i y and abundance
o ba s and noc u nal insec s in u ban g een a eas using wo
con as ing designs in e ms o he numbe o ci ies included
and he ype o he u ban in ensifica ion p oxies used. Fi s ,
we in es iga ed he esponses ac oss h ee Eu opean ci ies
(An we p, Pa is and Zu ich) o es whe he he e ec s o
u banisa ion on he esponse a iables we e consis en . We
used a s anda dised se o u banisa ion p oxies ac oss all ci -
ies including land use and ligh pollu ion maps. Second, we
addi ionally s udied he same se o esponses o ba s and
noc u nal insec s bu adding high- esolu ion p edic o s a ail-
able o he ci y o Zu ich o u he in e he amoun o
a ailable habi a and he ole o ege a ion s uc u e.
Ma e ials and me hods
S udy egion and selec ion o sampling si es
The s udy egion is Wes e n and Cen al Eu ope, in pa ic-
ula Pa is, F ance (48°5102300N, 2°2005800E), Zu ich, Swi -
ze land (47°2204000N, 8°3202300E) and An we p, Belgium
(51°1204800N, 4°2405500E). These ci ies a y in e ms o pop-
ula ion densi y, size and pe cen age o g een a eas
(Uni ed Na ions, 2019).
We ocused on he UGAs mapped and defined in he
Eu opean U ban A las (see EEA, 2012) o selec pa ches.
We used an o hogonal g adien o pa ch size (a ea in m
2
)
and connec i i y. Connec i i y was calcula ed using he
P oximi y Index (PI) which conside s he a ea and he dis-
ance o all nea by pa ches wi h a a ou able habi a , wi hin
a gi en sea ch adius. Thus, he PI measu es he deg ee o
pa ch isola ion, wi h highes alues gi en o less isola ed
pa ches (McGa igal, Cushman & Ene, 2012). We conside ed
as a ou able habi a all pa ches wi h high p obabili y o
ha ing ees ( ha is, UGAs, u ban o es and g ey u ban
land-co e wi h less han 30% impe ious su ace, see
EEA, 2012). The sea ch adius was se o 5 km om each
ocal pa ch, in o de o accommoda e all possible animal
mobili y anges. Lowe bu e alues ( om 500 m onwa ds)
did no g ea ly change he PI alues, because he dis ance is
squa ed, hus g ea ly limi ing he impac o pa ches beyond
a ce ain dis ance. To selec pa ches using he o hogonal
design, all possible pa ches we e classified in six size classes
and six classes o PI (36 possible combina ions). Wi hin
hese combina ions pa ches we e selec ed andomly ( andom
s a ified sampling design). Due o esou ce limi a ions we
only used
1
/
3
o he possible combina ions in Pa is and An -
we p (maximizing he g adien ) and he ull ange o combi-
na ions in Zu ich (32 combina ions, he o he combina ions
we e no a ailable). This esul ed in he final selec ion o 56
si es: 12 in Pa is, 12 in An we p and 32 in Zu ich. Si es
we e selec ed keeping a minimum dis ance o 500 m (excep
o wo si es in Zu ich selec ed by hei posi ion in he u ban
g adien , sepa a ed by 360 m). Median dis ance o he nea -
es si e was 1050 m and 85% o he si es we e sepa a ed by
a leas 800 m.
Field da a collec ion
Ci ies we e isi ed wice om mid-May o mid-July 2018.
Ba eco dings and noc u nal insec collec ions we e con-
duc ed simul aneously a he 12 selec ed UGAs o each ci y,
excep he 32 si es in Zu ich, which we e spli in wo sam-
pling pe iods due o limi ed equipmen . Samplings we e
scheduled o 5 consecu i e nigh s pe si e o each o he
wo sampling pe iods, in o de o a oid insu ficien da a
sizes and eliably sample he species communi y. Samplings
we e ca ied ou unde a ou able wea he condi ions,
ha is, wi hou ain and empe a u es abo e 12 °C
(Hu son, Micklebu gh & Racey, 2001).
Acous ic ba su ey
Echoloca ion sampling was conduc ed wi h acous ic da a
logge s ins alled on he unk o a sui able ee wi h open
canopy (1 m
2
wi h no b anches in on o he de ice),
be ween 34 m abo e g ound. Ba o aging ac i i y was
measu ed wi h he au onomous ul asound eco de s Ba log-
ge M (Elekon AG, 2018). The eco ding sys ems we e sen-
si i e om 10 o 150 kHz (§5 dB) and we e se up o
eco d om 15 min be o e sunse o 15 min a e sun ise 
adap ed o each ci y and da e du ing 5 consecu i e nigh s.
Ba echoloca ion calls we e iden ified using Ba scope 3
(h p://www.ba scope.ch;Ob is & Boesch, 2018). This so -
wa e au oma ically p ocesses he sequences and assigns
each single call o a sui able species wi h a mean a e o co -
ec classifica ions o 95.7% (Ob is & Boesch, 2018). No
all calls could be iden ified o species le el o some c yp ic
calls. In such cases, classifica ion was done manually o he
bes possible axonomic le el (see Appendix A), he eby
a oiding e o s ha can occu in au oma ed species iden ifi-
ca ion (Russo & Voig , 2016;Rydell, Nyman, Ekl€
o , Jones
& Russo, 2017). Subsequen ly, axa we e classified in o
L. Villa oya-Villalba e al. / Basic and Applied Ecology 51 (2021) 5970 61
h ee ecologically meaning ul guilds, acco ding o hei clu -
e esis ance and echoloca ion ange: sho - ange echoloca-
o s (SRE), mid- ange echoloca o s (MRE) and long- ange
echoloca o s (LRE) (F ey-Eh enbold e al., 2013;
F oide aux e al., 2014).
Ba o aging ac i i y was calcula ed on a daily basis,
compu ing he ac i i y o he h ee conside ed guilds in
windows o 5 min, o balance o possible ex ended o -
aging bou s o a single ba a ound a si e. Coun s we e
hen weigh ed by he numbe o possible obse a ional
5-minu e pe iods pe nigh , esul ing in a ela i e ac i -
i y. Ba di e si y was es ima ed pe nigh wi h he Shan-
non Index in wo ways: using he bes possible
axonomic le el (i.e. ba g oup di e si y) and only eli-
ably iden ified species (i.e. ba species di e si y).
Noc u nal insec collec ion
We collec ed noc u nal flying insec s (he ea e
e e ed o as insec s) in pa allel o ba s. Based on he
assump ion ha noc u nal insec s possess posi i e pho o-
axis (P ice & Bake , 2016; an G uns en e al., 2014),
we designed an in e cep LED ligh - ap o sample
insec s ( o de ails see Appendix A). To minimize in e -
e ences, he ain (5 lx) ligh - aps we e ins alled in a
sui able ee 1020 m dis an om he Ba logge , hung
om an open oliage b anch a leas 4.5 m abo e
g ound. T aps we e emp ied daily a e each sampling
nigh . In he labo a o y, insec s we e classified o he
o de le el using an Olympus SZ40 s e eo mic oscope,
en omological guidebooks (e.g. Chine y, 1988)and
expe ad ice. Insec di e si y was es ima ed pe nigh
wi h he Shannon Index and insec abundance as he
coun s o each sample. As o ganism ype o size may
influence i s quali y as ba ood (Hu son e al., 2001),
each o de was assigned a ela i e ac o acco ding o i s
body leng h and a we biomass index calcula ed pe sam-
ple. We assumed insec shapes o be ellipsoids o he
calcula ions. Some o de s we e addi ionally di ided in o
size classes (e.g. Lepidop e a: <5mm,520 mm,
>20 mm; see Appendix A: Table 2). No e ha insec
a iables could no be es ima ed o all he si es, as se -
e al o he ini ial ligh - aps we e andalised in he
cou se o he sampling. F om he ini ial 56 sampling
pa ches, we ob ained unbalanced da a o 12 si es in
Pa is, 11 in An we p and 26 in Zu ich.
Bio ic a iables
We a emp ed o s udy he bo om-up and op-down con-
ols, as ou g oups ep esen a p ey-p eda o sys em. To
in e he bo om-up con ol, we used insec di e si y and
abundance, while op-down con ol was measu ed wi h ba
g oup and species di e si y and bo h o al ba and guild ela-
i e ac i i y (Table 1).
En i onmen al and landscape a iables
We used en i onmen al and landscape a iables o eco-
logical ele ance o he s udied o ganisms (Cusimano,
Massa & Mo gan i, 2016;Th el all, Law & Banks, 2012b).
We ob ained clima ic p edic o s including nigh empe a-
u e, p ecipi a ion le els and wind speed o Zu ich (Me eo
Schweiz da abase), An we p and Pa is (Wea he Unde -
g ound, 2018). Landscape a iables we e calcula ed wi h
di e en bu e adii om he ocal sampling poin : 50, 100,
350 and 500 m, in o de o sui he dispe sal abili ies o
insec s (Ropa s, Dajoz, Fon aine, Mu a e & Geslin, 2019)
and ba s (Hu son e al., 2001). Fo each sampling si e, we
used he Eu opean U ban A las (EEA, 2012) o calcula ing
he size (a ea o each pa ch), he P oximi y Index, he edge-
o-edge dis ance o he nea es UGA and he p opo ion o
impe ious su aces (e.g. u ban ab ic, oads). We es ima ed
he dis ance o he nea es wa e body ollowing K auel and
LeBuhn (2016) and P ice and Bake (2016). The land-use
he e ogenei y (Ma hies, R€
u e , Schaa schmid & P asse,
2017;McIn y e e al., 2001) was calcula ed as he Shannon
Index o habi a s pe si e, using he Eu opean U ban A las
o he h ee ci ies and in Zu ich also using a high- esolu ion
land-co e map (G uen S ad Zue ich, 2010) (see Table 1).
All he connec i i y and landscape measu emen s we e cal-
cula ed in ESRI A cMap 10.4.1.
Mo eo e , o he ci y o Zu ich we used Ai bo ne Lase
Scanning (ALS) me ics o woody ege a ion (>1 m). These
ypes o emo e sensing da a a e su oga es o habi a
amoun and e ical he e ogenei y o ege a ion s uc u e
and ha e been shown o be good p edic o s o biodi e si y
in u ban ecosys ems (F ey e al., 2018;Zellwege e al.,
2016). The habi a amoun was es ima ed as he woody eg-
e a ion co e and he he e ogenei y o s uc u es as he s an-
da d de ia ion o woody ege a ion heigh s a 50, 100, 350
and 500 m adii (Table 1; o de ails see F ey e al., 2018).
Inc easing ALAN has been shown o dis u b ba and insec
biodi e si y (Knop e al., 2017;Lewanzik & Voig , 2017). We
es ima ed ALAN by measu ing illuminance le els in a se o
andomly selec ed poin s in each ci y (Pa is = 50, Zu ich = 40,
An we p = 43) using a lux me e Tes o 540 (h ps://www. es o.
com). We in ended o co e he adiance ange o e e y ci y o
calib a e a nigh image om he In e na ional Space S a ion
(Ea h Science & Remo e Sensing Uni NASA, 2018). Once
he as e was calib a ed, ligh emission in lux a each sampling
si e was ex apola ed. Howe e , he co ela ion o he lux me e
eadings wi h he calib a ion da a was low in Pa is
(R
2
= 0.1189), hus he aw RGB colou alues (o he nigh
image) o he as e we e calcula ed a each ocal sampling poin
(pixel size o 40 £40 m) and used as su oga e o ALAN
(Table 1).
S a is ical analysis
All he analyses and s a is ical figu es we e done wi h he s a-
is ical compu ing so wa e R .3.5.1 (R Co e Team, 2018)and
62 L. Villa oya-Villalba e al. / Basic and Applied Ecology 51 (2021) 5970
packages lme4 .1.1-20, glmmLasso .1.5.1, ca 3.0-3, npa -
comp .3.0 and mul comp .1.4.14.
Va iable selec ion
Due o he high numbe o p edic o s, we fi s selec ed he
ele an ones using an L1-Penalised Es ima ion ia he
glmmLasso unc ion (G oll & Tu z, 2014). In addi ion, we
also included possible in e ac ions (iden ified by explo a o y
analysis) and ecologically meaning ul a iables. Finally, we
checked he co ela ion amongs p edic o s (see Appendix
A) and disca ded hose highly co ela ed ( >0.7) o a oid
collinea i y (Zuu , Ieno & Elphick, 2010).
E ec s o p edic o s on ba and insec esponse a iables
in he h ee ci ies and in Zu ich
We pe o med Gene alised Linea Mixed-E ec s Models
(GLMMs) on insec abundance and ela i e ba ac i i y (i.e.
o al, LRE, MRE and SRE), wi h a nega i e binomial e o
Table 1. Classifica ion o p edic o s in o ca ego ies. The able depic s he se en ca ego ies (i.e. Ci y, Connec i i y, Landscape, Remo e sens-
ing, P ey and P eda o es ima es and S a is ical in e ac ions) used o agg ega e he 44 p edic o s included in he models in o de o acili a e
epo ing o he esul s (see Figs. 13). Each ca ego y is desc ibed and he numbe and ype o p edic o s included and hei a ailabili y in
each ci y o analysis is epo ed (A=An we p, P=Pa is, Z=Zu ich).
Ca ego y Desc ip ion Numbe o
p edic o s
P edic o s A ailabili y
Ci y Con ains a single a iable, he
ci y iden i y
1 Ci y 
Connec i i y Me ics o habi a connec i i y
used as p oxy o he amoun o
a ailable habi a based on he
Eu opean U ban A las
6 Dis ance nea pa ch A, P, Z
Dis ance wa e A, P, Z
A ea A, P, Z
Isola ion A, P, Z
P oximi y Index (500 and 5000 m) A, P, Z
Landscape Landscape me ics measu ing he
di e si y o land-uses (land-use
he e ogenei y) and he amoun o
impe ious su aces (impe ious
co e ) used as p oxy o he
amoun o a ailable habi a and
based on he Eu opean U ban
A las and he Zu ich Habi a Map
12 Land-use he e ogenei y (50, 100, 350,
500 m) o Eu opean U ban A las
A, P, Z
Land-use he e ogenei y (50, 100, 350,
500 m) o Zu ich Habi a Map
Z
Impe ious co e (50, 100, 350, 500 m) A, P, Z
Remo e sensing High- esolu ion emo e sensing
p edic o s used o assess ligh
pollu ion (ALAN and ALAN
RGB) and ege a ion s uc u e
( ege a ion co e , he e ogenei y
and oliage heigh di e si y)
14 ALAN A, P, Z
ALAN RGB A, P, Z
Vege a ion co e (50, 100, 350, 500 m) Z
Vege a ion he e ogenei y (50, 100, 350,
500 m)
Z
Foliage heigh di e si y (50, 100, 350,
500 m)
Z
P ey es ima es Insec biodi e si y me ics 2 Insec di e si y A, P, Z
Insec abundance A, P, Z
P eda o es ima es Ba biodi e si y me ics 9 Ba g oup di e si y A, P, Z
Ba species di e si y A, P, Z
Ba ela i e ac i i y A, P, Z
Long-/Mid-/Sho -Range Echoloca o s
ela i e ac i i y
A, P, Z
Long-/Mid-/Sho -Range Echoloca o s
ela i e ac i i y quad a ic
A, P, Z
S a is ical in e ac ions In e ac ion be ween pai s o
p edic o s
4 Ci y x Dis ance wa e A, P, Z
Ci y x Impe ious co e (100, 500 m) A, P, Z
Ci y x P oximi y Index 5000 m A, P, Z
L. Villa oya-Villalba e al. / Basic and Applied Ecology 51 (2021) 5970 63

s uc u e and log link o he o me , and a binomial
e o s uc u e and logi link o he la e (Zuu , Ieno,
Walke , Sa elie & Smi h, 2009). We an he unc ion
glmmTMB and glme , espec i ely. In each model, he
sampling si e was included as a andom ac o o allow
o epea ed measu es o he same UGA on di e en
nigh s. Me eo ological alues we e also included as an-
dom ac o s o a oid hem masking he e ec s o o he
p edic o s. Once he models we e fi ed, he Va iance
Infla ion Fac o (VIF) was used o iden i y mul icolli-
nea i y ia he unc ion i .WhenVIF>3, we sequen-
ially d opped he p edic o wi h he highes VIF and
ecalcula ed he VIFs (Zuu e al., 2010). We also
checked ha he selec ed model was he one wi h he
lowes co ec ed AIC alue (AICc). The goodness-o -fi
was in es iga ed by calcula ing he condi ional coe fi-
cien o de e mina ion o Gene alised Mixed-E ec s
Models (R
2C
)(Nakagawa & Schielze h, 2013). Finally,
he model assump ions we e alida ed by plo ing he
esiduals agains fi ed alues and he Q-Q plo s o he
andom e ec s (Zuu & Ieno, 2016). We an Linea
Mixed-E ec s Models (LMMs) on ba and insec di e -
si y, wi h a no mal e o (C awley, 2007), using he
unc ion lme . The model ha bes explained he da a
a iabili y was he one wi h no mul icollinea i y (VIF
<3), he lowes AICc alue and a plo o esiduals e -
sus fi ed alues wi hou pa e n.
Models we e un sepa a ely o each esponse a iable:
fi s o he 36 si es in Pa is, Zu ich and An we p (12
each) including he ci y as a fixed ac o in o de o iden-
i y biodi e si y di e ences be ween he ci ies, and sec-
ondly o he 32 si es in Zu ich. O e all, he g aphics o
he esiduals e sus he fi ed alues o he GLMMs and
LMMs poin ed ou o a good fi o he model, as he
poin clouds did no exhibi any pa icula end and
we e well dispe sed o e he axis (see Appendix A). We
also compu ed he Mo an's I es on he model esiduals
o check o spa ial au oco ela ion and no au oco ela-
ion was ound (da a no shown).
Pai wise compa isons be ween he h ee ci ies
To in es iga e he possible di e ences be ween he h ee
ci ies on each esponse a iable, we made he espec i e
mul iple pai wise compa isons ia he unc ions mc p (coun
da a) and glh (con inuous and bina y da a).
Resul s
We sampled a o al o 12,714 insec s and eco ded
283,126 ba passes con aining 5 million echoloca ion calls.
Small Dip e a and T ichop e a ep esen ed 56% o he sam-
pled indi iduals, while medium-sized Lepidop e a and
small-sized Coleop e a oge he ep esen ed 20%. The
emaining g oups accoun ed o less han 10% o o al
abundance each. A ound 91% o he ba calls belonged o
Pipis ellus pipis ellus, while LREs and SREs accoun ed
o 5% and 4% o he ba calls, espec i ely. Thus, MREs
we e significan ly mo e p esen han he emaining guilds.
Noc u nal insec di e si y showed a significan nega i e
e ec wi h inc easing isola ion. Con e sely, insec abun-
dance was educed wi h inc easing quad a ic MRE ela i e
ac i i y and ba g oup di e si y, bu inc eased wi h ba spe-
cies di e si y (Fig. 1). We ound a ificial ligh (i.e. ALAN
RGB) o ha e a ci y-specific e ec on ba g oup di e si y.
Mo eo e , he ange o ALAN RGB and he esponse o ba
di e si y di e ed amongs ci ies (Fig. 2). In An we p and
Pa is ba di e si y was s ongly educed wi h inc easing al-
ues o ALAN RGB, despi e ha ing a di e en ange o illu-
mina ion. Con e sely, Zu ich had he sho es ange o
illuminance alues, which ansla ed in o a small inc ease o
ba di e si y (Fig. 2). In addi ion, Zu ich ha bou ed a mo e
di e se ba communi y han An we p and Pa is (see Appen-
dix B). The o al ba ela i e ac i i y in An we p was signifi-
can ly highe han in he o he ci ies (see Appendix B).
Mo eo e , we ound he ela i e ac i i y o he guild o LRE
o be significan ly dec eased wi h inc easing ALAN RGB in
all ci ies, bu no wi h any o he ca ego y o p edic o s. Simi-
la ly, no significan e ec s we e ound o he connec i i y,
landscape o emo e sensing p edic o s on MRE. The model
only selec ed an inc ease o he MRE ela i e ac i i y wi h
insec di e si y gain. Finally, SRE ac i i y was nega i ely
a ec ed by he p opo ion o impe ious co e in he 100 m
adius (Fig. 1). The pai wise compa ison be ween ci ies
e ealed ha Pa is and An we p we e di e en in e ms
o o al ba , LRE and MRE ela i e ac i i y; while Pa is
and Zu ich di e ed in insec abundance, ba g oup and
species di e si y and LRE and SRE ela i e ac i i y.
Likewise, An we p and Zu ich showed di e ences in
insec abundance and ba g oup and species di e si y
(see Appendix B).
The analyses o ba and insec esponses in an ex ended
se o sampling si es in Zu ich un eiled he influence o
new p edic o s (Fig. 3). Insec esponses we e enhanced
by he size o he UGA and by he ac i i y and di e si y
o ba s. S ikingly, we ound a scale-dependan esponse
wi h inc easing ege a ion he e ogenei y, which was posi-
i e a 100 m and nega i e a 350 m adius. Ba di e si y
and ac i i y esponses depended on a combina ion o p e-
dic o s including connec i i y (dis ance o wa e ), bo om-
up (insec di e si y) and landscape (impe ious su aces
a di e en adii) a iables. O e all, ba di e si y and
ac i i y inc eased oge he wi h insec di e si y gain bu
was cons ained by inc easing dis ance o wa e and he
amoun o impe ious su aces a 100 m. The land-use
he e ogenei y p edic o s, bo h om he U ban A las and
om he ecologically-based land-co e map o Zu ich,
only showed significan e ec s o small landscape scales
(i.e. 50 and 100 m).
De ailed model esul s (e.g. fi ed models, p- alues) can
be ound in Appendix B: Tables 2 and 3.
64 L. Villa oya-Villalba e al. / Basic and Applied Ecology 51 (2021) 5970
Fig. 1. Hea map o he significan esul s o Gene alised Linea Mixed-E ec s Models o he h ee ci ies. Significance h eshold is defined a
a 0.05 le el. The colou g adien eflec s he magni ude o a p edic o ’s es ima e o indi idual esponse a iables. G ey colou indica es lack
o s a is ical significance. SRE: sho - ange echoloca o s, MRE: mid- ange echoloca o s, LRE: long- ange echoloca o s (F ey-
Eh enbold e al., 2013).
Fig. 2. Rela ionship be ween he RGB alues o noc u nal as e s o he ci ies (ALAN RGB) and ba g oup di e si y o An we p, Pa is and
Zu ich. Each ci y shows a specific ange in adiance and a esponse o he ba di e si y. Lines ep esen second o de polynomial GLM mod-
els and bands he 95% confidence le el in e al o model p edic ions. Do s ep esen indi idual ba g oup di e si y measu emen s.
L. Villa oya-Villalba e al. / Basic and Applied Ecology 51 (2021) 5970 65
Discussion
The mul i-ci y app oach e ealed some consis en
esponses o noc u nal biodi e si y o u banisa ion ac oss
ci ies. Isola ion had a nega i e influence on noc u nal insec
di e si y, which migh be ela ed o he limi ed mobili y o
insec s and hei dependence on ege a ion o eaching new
pa ches (Concepci
on e al., 2015). Con as ingly,
Tu ini and Knop (2015) ound ha pa ch isola ion plays a
limi ed ole in u ban ecosys ems and a h opods can be
di e se when su ficien ege a ed space is p o ided. Excep
o dis ance o wa e , connec i i y me ics did no show sig-
nifican e ec s on ba s, likely because hey a e highly
mobile o ganisms and can each dis an pa ches (K auel &
LeBuhn, 2016). Howe e , ou connec i i y me ics we e
mainly a landscape scale, which migh ha e unde ep e-
sen ed he ole o small-sized connec ing elemen s (hedges,
ee lines, e c.) seen in p e ious s udies (F ey-
Eh enbold e al., 2013). Unlike SRE and LRE ba s ha a e
espec i ely clu e - o open-adap ed species, MRE ba s
include a wide ange o in e media e echoloca ion s a egies
ha could explain hei ole ance o agmen a ion and migh
make he mos common guild in ou s udy p one o c oss
less sui able a eas (F ey-Eh enbold e al., 2013).
We iden ified ALAN o be a p ominen ac o lowe ing
ba g oup di e si y. Pa icula ly, Pa is and An we p showed
high adiance le els and low ba di e si y, sugges ing o
he fil e ing ou o ligh -a e se species. ALAN can modi y
he o aging beha iou o ba communi ies (Russo & Ancil-
lo o, 2015), enhancing a subse o ligh -oppo unis ic spe-
cies (e.g. Pipis ellus pipis ellus, he mos common ba in
ou s udy) ha can ake ad an age o insec esou ces a he
expense o ligh -a e se ones. Howe e , noc u nal biodi e -
si y esponses a e species- and con ex -dependan
(Ma hews e al., 2015) as shown by he pai wise compa i-
sons be ween ci ies (see Appendix B). This migh indica e
an ongoing beha iou al change o he u ban ba communi y,
possibly caused by adap a ion o plas ici y owa ds ligh
Fig. 3. Hea map o he significan esul s o Gene alised Linea Mixed-E ec s Models in he ci y o Zu ich. Significance h eshold is defined
a a 0.05 le el. The colou g adien eflec s he magni ude o a p edic o ’s es ima e o indi idual esponse a iables. G ey colou indica es
lack o s a is ical significance. SRE: sho - ange echoloca o s, MRE: mid- ange echoloca o s, LRE: long- ange echoloca o s (F ey-
Eh enbold e al., 2013). GSZ: habi a map o he ci y o Zu ich, EUA: Eu opean U ban A las.
66 L. Villa oya-Villalba e al. / Basic and Applied Ecology 51 (2021) 5970
pollu ion as no ed in mo hs (Al e ma & Ebe , 2016), ye
e idence is lacking. Fu u e s udies del ing in o he eco-e o-
lu iona y dynamics o u ban auna could p o ide impo an
insigh s o be e in o m noc u nal biodi e si y managemen
(Lambe & Donihue, 2020). Fu he , ecen ends o
eplace adi ional s ee ligh ing wi h LED migh help o
ebalance his fil e ing e ec , as LEDs can educe he pho o-
axis o insec s and hei a ailabili y (Wakefield, B oyles,
S one, Jones & Ha is, 2016), especially when combined
wi h app op ia e choice o longe -wa e emission spec a (i.
e. >500 nm, see Bennie, Da ies, C use, Inge & Gas on,
2018;Bollige e al., 2020). Mo eo e , he nega i e e ec o
ba g oup di e si y and ela i e ac i i y on insec abundance
seems o indica e a op-down con ol. S ill, ou esul s ha e
o be aken cau iously due o opposi e e ec o ba species
di e si y. In any case, ou esul s poin ou he impo ance
o conside ing such u ban ophic dynamics
(Shocha , Wa en, Fae h, McIn y e & Hope, 2006).
Ou single-ci y app oach e ealed ha insec di e si y
inc eased bo h ba ela i e ac i i y and di e si y in Zu ich,
simila ly as shown by Lewanzik and Voig (2017). These
esul s emphasise a bo om-up egula ion in he noc u nal
insec -ba ophic sys em in Zu ich. In addi ion, ege a ion
he e ogenei y appea ed o ha e a scale-dependan e ec on
noc u nal insec s, posi i e a 100 m and nega i e a 350 m
adius. Thus, ou esul s a e consis en wi h p e ious wo ks
on a h opods (Tu ini & Knop, 2015) and o he li le mobile
o ganisms, which mo e likely ely on local pa ch cha ac e -
is ics (i.e. 100 m adius) (Concepci
on e al., 2015;
Th el all e al., 2012a). On he o he hand, we we e no able
o assess an e ec o ege a ion s uc u e me ics on ba s
al hough hei significance has been p o en on p io s udies
in u ban en i onmen s (Sua ez-Rubio, Ille & B uckne ,
2018). Fu he mo e, Th el all e al. (2017) ha e poin ed ou
ha na i e plan species in UGAs inc ease ba di e si y,
likely because ba s eed on insec s ha depend on plan s.
Simila ly, land use he e ogenei y de i ed om he de ailed
land-co e map o Zu ich did no show significan e ec s
o mos esponses. In his ega d, high- esolu ion p edic o s
did no con ibu e subs an ially o ou da a unde s anding.
In e es ingly, no model selec ed ALAN o ALAN RGB,
likely because o he low adiance alues wi hin he ci y
(Fig. 2). S udying Eu opean ci ies wi h in e media e adi-
ance le els migh help finding he ipping poin a which
ALAN s a s dec easing noc u nal biodi e si y.
In summa y, ou s udy p o ides e idence o he ole o
UGAs sus aining noc u nal biodi e si y in Eu opean ci ies
and sheds ligh on he impo ance o p oxies used o in e
he e ec s o u ban in ensifica ion. Noc u nal habi a s a e
main ecological niches in ci ies esponsible o se e al eco-
sys em se ices such as insec egula ion and pollina ion,
bu ha e been so a neglec ed in u ban managemen
(Pinho e al., 2021 in his issue). We de eloped a s anda d-
ized mul i-ci y design o consis en ly compa e esponses o
ba s and noc u nal insec s o u ban in ensifica ion. Ou
esul s may se a s a ing poin o planning and managing
noc u nal biodi e si y in UGAs o Wes e n and Cen al
Eu ope. O e all, ou findings showed an u gen need o eg-
ula ing and educing ALAN, bu accoun ing also o he
exis ing ophic con ols. The specific esponses o ou s udy
g oups o u banisa ion highligh he need o mani old man-
agemen s a egies in UGAs ha eflec he di e en ai s o
he noc u nal species. Cu en ly, Eu opean ci ies ace a sce-
na io o u ban densifica ion o a oid he sp awl o o he eco-
sys ems, leading o inc easing le els o s esso s and hus
h ea ening biodi e si y wi hin ci ies. In his con ex , UGAs
a e key elemen s o p ese e and enhance p esen and u u e
u ban biodi e si y.
Decla a ion o Compe ing In e es
None.
Acknowledgmen s
We would like o hank all collabo a o s o he BioVEINS
p ojec (h ps://www.biodi e sa.o g/1012). We hank H. Eggen-
be g o designing he ap nes s. A special hank goes o A.
Zane a o scien ific ad ice, T. Hallikma and K. Kilchho e o
logis ical suppo , Ch. Ginzle o p epa ing LiDAR da a, M.
Nogue a and C. Pe ez-Mon o making fieldwo k and da a
p ocessing mo e pleasan and A. G oll o his p ecious s a is ical
help. We also acknowledge he au ho i ies o An we p, Pa is
andZu ich o hepe missions osample hei UGAand‘G €
un
S ad Z€
u ich’(GSZ) o he pe mission o use hei high- esolu-
ion habi a map o he ci y o Zu ich. This wo k was pa ially
unded by he Eu opean ERA Ne Biodi ERsA p ojec “Bio-
VEINS: Connec i i y o g een and blue in as uc u es: li ing
eins o biodi e se and heal hy ci ies”(H2020 Biodi-
ERsA32015104), he F ench Resea ch Agency (ANR-16-
EBI3-0012) and WSL unds. Finally, we acknowledge unding
o he Swiss Na ional Science Founda ion o F. Zellwege (p oj-
ec 172198 and p ojec 193645) and J. Casanelles-Abella (p oj-
ec 31BD30_172467).
CRediT au ho ship con ibu ion s a emen
LVV: Valida ion, Fo mal analysis, In es iga ion, Resou -
ces, Da a Cu a ion, W i ing - O iginal D a , W i ing -
Re iew & Edi ing, Visualiza ion, P ojec adminis a ion.
JCA: Me hodology, In es iga ion, W i ing - O iginal D a ,
W i ing - Re iew & Edi ing, P ojec adminis a ion. MM:
Concep ualiza ion, Me hodology, W i ing - Re iew & Edi -
ing, Supe ision, P ojec adminis a ion, Funding acquisi-
ion. PP: Concep ualiza ion, Me hodology, Fo mal analysis,
Resou ces, W i ing - Re iew & Edi ing, Funding acquisi-
ion. RS: Concep ualiza ion, W i ing - Re iew & Edi ing,
Supe ision, P ojec adminis a ion, Funding acquisi ion.
AVM: In es iga ion, W i ing - Re iew & Edi ing. FC:
L. Villa oya-Villalba e al. / Basic and Applied Ecology 51 (2021) 5970 67