light pollution at the urban forest edge negatively ... filecontents lists available at...

Contents lists available at ScienceDirect

Biological Conservation

journal homepage: www.elsevier.com/locate/biocon

Light pollution at the urban forest edge negatively impacts insectivorousbats

Joanna K. Haddocka,⁎, Caragh G. Threlfalla,b, Bradley Lawc, Dieter F. Hochulia

a School of Life and Environmental Sciences, Heydon-Laurence Building, Science Road, The University of Sydney, NSW 2006, Australiab School of Ecosystem & Forest Sciences, The University of Melbourne, Richmond, VIC 3121, Australiac Department of Primary Industries, Level 12, 10 Valentine Avenue, Parramatta, NSW 2124, Australia

A B S T R A C T

Connectivity and quality of vegetation in cities, including urban forests, can promote urban biodiversity. However the impact of anthropogenic pressures at theforest-matrix edge, particularly artificial light at night (ALAN), on connectivity has received little attention. We assessed the influence of artificial light at forest edgeson insectivorous bats. We acoustically surveyed 31 forest edges across greater Sydney, Australia, half with mercury vapour streetlights and half in ambient darkness,and compared the bat assemblage and activity levels to urban forest interiors. We also sampled the flying insect community to establish whether changes in insectdensities under lights drive changes in insectivorous bat activity. We recorded 9965 bat passes from 16 species or species groups throughout our acoustic survey. Theactivity of all bats, and bats hypothesised to be sensitive to artificial light, was consistently higher in forest interiors as opposed to edges. We found that slower flyingbats adapted to cluttered vegetation or with a relatively high characteristic echolocation call frequency; Chalinolobus morio,Miniopterus australis, Vespadelus vulturnus,and Nyctophilus spp., were negatively affected by artificial light sources at the forest edge. The emergence time of Vespadelus vulturnus was also significantly delayedby the presence of streetlights at the forest edge. Conversely, generalist faster flying bats; Chalinolobus gouldii, Ozimops ridei, Austronomous australis, Saccolaimusflaviventris, and Miniopterus orianae oceanensis, were unaffected by artificial light at the edge of urban forest, and used light and dark forest edges in a similar way.Insect surveys showed that larger lepidopterans seemed to be attracted to lit areas, but in low numbers. Artificial light sources on the edges of urban forest havediverse effects on bats and insects, and should be considered an anthropogenic edge effect that can reduce available habitat and decrease connectivity for light-sensitive species.

1. Introduction

Urbanisation is one of the leading causes of biodiversity lossworldwide (Czech et al., 2000; McKinney, 2006). Habitat fragmentation(Fischer and Lindenmayer, 2007), along with noise pollution (Ortega,2012), air pollution (Leonard and Hochuli, 2017), reduced waterquality (Blakey et al., 2018), reduced vegetation cover and structure(Threlfall et al., 2016; Threlfall et al., 2017) and artificial light (Hölkeret al., 2010) all contribute to degrading natural habitat for urbanwildlife. Although only taking up a small percentage of the planet'sterrestrial surface, urbanized areas are predicted to grow by 1.2 mil-lion km2 by 2030, impacting many global biodiversity hotspots (Setoet al., 2012). Hence, investment in the conservation of urban biodi-versity is essential for many reasons (Dearborn and Kark, 2010).

Providing “stepping stones and corridors” of forests and native ve-getation is a commonly suggested conservation action for urban bio-diversity (Dearborn and Kark, 2010; Beninde et al., 2015). Connectingforest habitat in cities can mitigate species loss and biotic homo-genisation (Lindenmayer and Nix, 1993; Fischer and Lindenmayer,2007) by facilitating movement (Dearborn and Kark, 2010) and genetic

diversity (Aguilar et al., 2008) of native animals across a degradedurban landscape. Over the last few decades, research has focused onquantifying the characteristics of habitat remnants that maximize theirvalue to native plants and animals, including their size (Evans et al.,2009), shape (Hawrot and Niemi, 1996), and connectedness (Keittet al., 1997; McGarigal et al., 2002; Beninde et al., 2015). Calculatingthe amount of viable habitat existing in urban areas is a more complexprocess. It must incorporate not only structural elements of forest areas,such as their size and shape, but also functional connectivity (Kupfer,2012), and whether structural habitat features are actually used. An-thropogenic pressures, such as ALAN, may impact functional con-nectivity (Hale et al., 2012) by narrowing wildlife corridors and redu-cing functional patch sizes, however, the extent to which this occurs ispoorly understood currently.

ALAN is a growing problem, escalating by 6% each year (Hölkeret al., 2010). It is caused by illumination from anthropogenic lighting,and is most prevalent in urban areas (Falchi et al., 2016). Only rela-tively recently has ALAN been widely discussed as a global threat tobiodiversity (Rich and Longcore, 2013; Gaston et al., 2015). Streetlightsdisrupt migration patterns (La Sorte et al., 2017), breeding cycles

https://doi.org/10.1016/j.biocon.2019.05.016Received 18 September 2018; Received in revised form 5 May 2019; Accepted 14 May 2019

⁎ Corresponding author.E-mail address: [email protected] (J.K. Haddock).

Biological Conservation 236 (2019) 17–28

0006-3207/ © 2019 Elsevier Ltd. All rights reserved.

T

http://www.sciencedirect.com/science/journal/00063207

https://www.elsevier.com/locate/biocon

https://doi.org/10.1016/j.biocon.2019.05.016


mailto:[email protected]


http://crossmark.crossref.org/dialog/?doi=10.1016/j.biocon.2019.05.016&domain=pdf

(Navara and Nelson, 2007) and predator-prey interactions (Gorenzeland Salmon, 1995). Streetlights may also have an effect on the func-tional value and connectivity of proximal urban forests and vegetation.Streetlights positioned along edges of urban wildlife corridors nega-tively affect some species whilst attracting others (Azam et al., 2018).The effects of artificial light at the forest edge for nocturnal wildlifemay be significant; light may penetrate dark vegetation anywhere from50m (Kempenaers et al., 2010) to 380m (Pocock and Lawrence, 2005).Dark habitats are currently at risk from the edge effects of ALAN.

Insectivorous bats are an ecologically diverse group, and respond ina variety of ways to urbanisation (Russo and Ancillotto, 2015). Somefaster flying open-space adapted bat species find roosts in buildings(Kunz, 1982) and can commute across urbanized landscapes (Jung andKalko, 2011). Conversely, other slower flying clutter-adapted bat spe-cies commonly avoid urban areas, they cannot adapt to changes in roostavailability and instead are reliant on networks of urban forest to sur-vive in cities (Basham et al., 2011). Artificial light is one anthropogenicpressure driving these diverse responses to urban areas. Ultravioletradiation attracts high numbers of insects (van Grunsven et al., 2014)and could offer urban feeding grounds for faster flying bat speciesadapted to exploit this resource (Rydell, 1992; Rydell and Racey, 1995).These bats may be able to dive through the light cone when foraging(Blake et al., 1994) and perhaps able to evade any aerial predators thatuse the lit areas as hunting grounds. Conversely, slower flying clutter-adapted species often avoid artificially lit areas (Stone et al., 2012) and

may be constrained by lights at the patch edge, spending a majority oftheir foraging time within dark patches (Threlfall et al., 2013; Haleet al., 2015), and therefore reducing their functional urban habitat.Artificial light also markedly delays some species' emergence timesfrom roosts (Jones and Rydell, 1994; Downs et al., 2003; Boldogh et al.,2007), meaning that their foraging time is reduced and the health of thepopulation may at risk (Boldogh et al., 2007). There is a global patternemerging that ALAN negatively affects species adapted to foraging ineither cluttered vegetation or along habitat edges, like Rhinolophushipposideros (Stone et al., 2012), Eptesicus serotinus (Azam et al., 2018),and Nyctophilus gouldi (Threlfall et al., 2013). The mechanistic driversof this light phobia in bats are not completely understood (Stone et al.,2015a, 2015b, Rowse et al., 2016) but could include predator avoid-ance (Speakman, 1991; Stone et al., 2009; Lima and O'Keefe, 2013),morphology that leaves slower flying bats less able to exploit airborneinsect prey at lights compared with faster flying bats like Pipistrelluspipistrellus (Haffner and Stutz, 1985) sensitivity to ultraviolet radiationemitted by some streetlights (Gorresen et al., 2015), or a combinationof these. Bats with many of these traits are declining in cities (Jung andThrelfall, 2016; Jung and Threlfall, 2018) hence research on the impactof public lighting on this group is urgently required.

We hypothesised that permanent streetlights along the forest edgewould reduce the activity of some insectivorous bats due to a decline inhabitat quality. We predicted that the activity of bats with slow flightspeed would be lower at forest edges with streetlights than edges with

Fig. 1. Map showing location and distribution of forest patch sites (red triangles; n=15 pairs of control interior and edge sites) and connected forest sites (yellowcircles; n=16 pairs of control interior and edge sites) relative to Sydney, Australia (inset). (For interpretation of the references to colour in this figure legend, thereader is referred to the web version of this article.)

J.K. Haddock, et al. Biological Conservation 236 (2019) 17–28

18

no lights. We also predicted that the activity of faster flying, light-ex-ploiting bats would either be unaffected or be higher at edges withstreetlights than edges with no lights. Finally, we predicted that totalbat activity, and the activity of bats with slow flight speed, would bothbe higher in the forest interiors than the forest edges.

2. Method

2.1. Site selection

Our survey was carried out in Greater Sydney, a sub-tropical citywith a population of over 4.5 million, on the east coast of New SouthWales, Australia. Sydney has much remnant native forest abutting theurban matrix, both in continuous national parks and in smaller isolatedpatches surrounded by housing (Benson and Howell, 1995). Two maingeologies in the Sydney region, shale and sandstone, have led to dif-ferent levels of primary productivity and soil fertility across the city,and these have been shown to influence insect prey and bat diversity inthis region (Threlfall et al., 2011, Threlfall et al., 2012a, 2012b). Hence,we only included sites on full or majority sandstone with transitionalsoil type to control for this effect of geology.

2.2. Experimental design

We conducted an acoustic survey along 31 forest edge sites (Fig. 1);16 of these sites were at the edges of connected forest (forest connectedto large natural areas, Fig. 2a), and 15 were on the edges of isolatedforest patches (Fig. 2b),> 30 ha in size, but surrounded on all sides byurban matrix. Of all connected patch edge sites, half the sites (n= 7)had 80W mercury vapour streetlights along the edges (light intensity of10.65 lx ± 1.89) and half of the sites (n=8) were dark (light intensityof 0.52 lx ± 0.34). Of all forest patch edge sites, half the sites (n= 8)had 80W mercury vapour streetlights along the edges (light intensity of10.20 lx ± 1.23) and half of the sites (n=8) were dark (light intensityof 0.97 lx ± 0.52). All edge sites were defined as roads or pathwaysover 4m in width with dense vegetation on only one side. Dark edgeswere defined as 30m of uninterrupted dark conditions (artificial lightlevels comparable to ambient darkness and no external light sources onnearby houses). Artificially lit edges were defined as 30m of 80Wmercury vapour lights along one side of the road, with the streetlightsat a median distance apart of 12.1 m. Usual low-level urban lightingfrom houses in the surrounding matrix was present at both dark andlight edges. We additionally located control sites in the dark interior ofall 31 sites (lux of 0.55 ± 0.20), which were located along an internaledge comprising a pathway or track (2m typical width) between 400mand 1 km perpendicular from the sampled forest edge. This allowed us

to sample the bat assemblage at 31 edge sites, and 31 dark interiorcontrol sites, leading to 62 sites in total. We sampled between No-vember and December of 2016, the maternity season for bats whenresource requirements and activity levels are highest. The light in-tensity (lux) at each site was measured using a lux meter (QM1587;Reduction Revolutions Pty Ltd., Parramatta, Australia). We also mea-sured insect biomass at all forest patches, at both the interior and edgesite, using a black light intercept traps (Australian Entomological Sup-plies, Murwillumbah, Australia, see Appendix A).

2.3. Bat recording

The acoustic survey was carried out using Anabat II detectors (TitleyElectronics, Ballina, Australia) placed on the ground with the highfrequency microphone positioned 1m above the ground and pointingupwards at a 45° angle to record all echolocation calls of passing bats.We used one detector per site, and predicted to still record high flyingbats to be recorded due to louder calls from those species (Jung andKalko, 2011). At artificially lit edges the detector was placed no furtherthan 3m from the base of the streetlight pole, no> 3m away from theedge of the forest. At both the edge and interior sites, the microphonewas pointed parallel down the forest edge and down the interior trackor path, respectively, to optimise the amount of habitat sampled (Lawet al., 1998; Threlfall et al., 2012a, 2012b). Each of the 62 sites weresampled between 3 and 5 consecutive nights (average 4.35 nights± s.e.0.19, due to unexpected equipment failure we did not manage to recordfor 5 nights at each site), totaling 270 recording nights across the entiresurvey. A maximum of eight sites were surveyed on the same night,where the edge and its respective interior control were concurrentlysampled. The detectors passed the data through a Zero-Crossings In-terface Module (ZCAIM; Titley Electronics). The acoustic files collectedduring the survey were processed using Anascheme and a bat callidentification key developed for Sydney bats (Adams et al., 2010). For abat call to be identified to species level, three or more pulses were re-quired and have characteristics that fall within the program's para-meters for that species. A pass was defined as at least three valid pulseswith a minimum of 6 pixels per pulse. Successful species identificationswere made only when a minimum of 50% of pulses within a pass wereidentified as the same species (Adams et al., 2009; Threlfall et al.,2012a, 2012b). The calls identified as Chalinolobus dwyeri, Falsistrellustasmaniensis, Nyctophilus spp., Saccolaimus flaviventris, Scoteanex ruep-pellii and Scotorepens orion are known to be complex or rare and weremanually checked against known parameters and confirmed or re-identified (Adams et al., 2010). In addition, calls were run through aspecies filter in AnaScheme which is specifically designed to identifycalls of Chalinolobus gouldii with alternating frequency. The call

Fig. 2. Images taken from Google Earth of two of the sites in this study demonstrating the distinction between the patch and connected sites, with the green markershowing the position of the detector at the forest interior site and the blue marker showing the position of the detector at the forest edge site, a) Patch site:Cumberland State Forest, Sydney, b) Connected site: St Ives, Garigal National Park. (For interpretation of the references to colour in this figure legend, the reader isreferred to the web version of this article.)


19

characteristics of Nyctophilus gouldi and Nyctophilus geoffroyi are indis-tinguishable using the AnaScheme method and so were pooled as onetaxon; Nyctophilus spp. Only species that were positively identifiedusing the key, filters and manual checking were included for furtheranalysis to eliminate any bias caused by using partially identified spe-cies.

2.4. Assignment to functional groups

Assignment of each species to a functional group was based onmorphological traits and foraging styles (Rhodes, 2002, Haddock et al.in press). Chalinolobus gouldii, Ozimops ridei, Austronomous australis,Saccolaimus flaviventris, and Miniopterus orianae oceanensis were cate-gorised as light-exploiting due to their open-space foraging area pre-ference, or edge-space foraging area preference with low or mediumecholocation call frequency (Rhodes, 2002). Light-exploiting bats maytend to use the well-lit areas to forage around or commute past. Thelight-sensitive group consisted of both maneuverable edge-space fora-ging species with high echolocation call frequency (Chalinolobus morio,Miniopterus australis, Vespadelus vulturnus; Adams et al., 2009) and alsospecies adapted to cluttered vegetation with a slower flight pattern(Rhinolophus megaphyllus and Nyctophilus spp., consisting of two speciesNyctophilus gouldi and Nyctophilus geoffroyi that have echolocation callsindistinguishable from the other) that may avoid lit areas. We basedthese groupings on known response groups found in previous studies,however we acknowledge that there may be species that do not ne-cessarily fit this trend.

2.5. Measurement of environmental variables

The percentage of native vegetation cover (> 3m tall) within aradius of 250m of each site was calculated using Arc Map (ESRI,Redlands, USA, ver. 10.2) and was calculated through intersecting GPSpoints of our sites with the GIS layer ‘The Native Vegetation of theSydney Metropolitan Area - Version 3, VIS_ID 4489’ (NSW Office ofEnvironment and Heritage, Sydney). This area has been found to beindicative of local habitat as shown in previous studies (Lacoeuilheet al., 2014). The same method was used to calculate the percentage ofsandstone and transitional sandstone soil type within the 250m radiusof each site. Average moon illumination was calculated by noting thepercentage of the moon's face visible each night, and then for each site

taking an average of the percentages across all the nights sampled.Although we avoided surveying on full moon, logistical limitationsmeant that we could not control for the moon cycle completely.

2.6. Statistical analysis

All analyses were carried out in SPSS (version 22, SPSS Inc.,Chicago, USA). For each of the 31 sites and their dark interior urbanforest controls, average activity and total species richness was calcu-lated for the bat assemblage, for the light-exploiting functional group,for the light-sensitive functional group and for each individual species.Average activity was calculated by summing the total number ofidentified passes detected at that site during the recording time, andthen dividing by the number of sampling nights at that site. Totalspecies richness was calculated by counting the number of species de-tected at that site over the recording time. To calculate the time untilfirst activity, and therefore potential delay or disturbance to the bats,we located the earliest call for a functional group or species at a par-ticular site and then subtracted that time from the time of sunset thatday (EST), leaving the numbers of minutes after sunset that the call wasrecorded. When discussing the comparison of a response variable acrosslight treatments, we are including light edges, dark edges and darkinteriors.

Generalized linear models (GLMs) were used to establish statisticaldifferences in both lux level and vegetation extent between the threelight treatments, with the habitat treatment (connected or patch) andlight treatment (light edge or dark edge) as fixed effects.

A series of correlations were used to assess whether average moonillumination across recording time, or the percentage of sandstone soilwithin a 250m radius of each site were significant predictors of batactivity or species richness. As none were significant predictors ofspecies richness or total bat activity, they were omitted from furthermodels. A series of GLMs were used to assess if emergence times of C.gouldii, O. ridei, V. vulturnus and Nyctophilus spp. (the two most com-monly detected of each response group) were affected by moon illu-mination, moon illumination has been linked to emergence of bats. Asno response variables were significantly affected by any environmentalor weather variables, all were omitted from further models.

We then used a series of GLMMs to compare bat and insect responsevariables (Table 1) across light treatments. We only included siteswhere both the edge and interior had a record for that species. Othersites were excluded from this analysis. The GLMMs allowed us to in-clude the site (which contained both the edge and dark interior pair) asa random effect, and the habitat treatment (connected or patch) andlight treatment (light edge or dark edge) as fixed effects. We considereda variety of response variables (Table 1). We then also carried outanalyses on the two most commonly recorded light-exploiting species;C. gouldii and O. ridei, and the two most commonly recorded light-sensitive species; V. vulturnus and Nyctophilus spp. as activity of otherspecies was too low or too patchy in distribution to allow for adequateanalysis.

We used Poisson distribution models with a log link function androbust variance estimates for all but two of the GLMMs as the data didnot follow a normal distribution (Yau and Kuk, 2002), the exceptionswere light-sensitive bat activity and Nyctophilus spp. activity. Light-sensitive bat activity was low across all treatments and so a negativebinomial distribution with a log link function with robust variance es-timates was used in the GLMM. Soft-calling Nyctophilus spp. were re-corded in such low numbers that activity of this taxa was converted topresence-absence data, and a GLMM was run using a binomial dis-tribution with a probit link function, again with robust variance esti-mates. All post hoc tests were Fisher's Least Significant Difference.

When light-sensitive or light-exploiting bats were detected at a site,we wanted to know what proportion of the calls were detected alongthe edge compared to the interior for that site, as an indicator ofwhether edges or interior were more preferred by each bat group. For

Table 1All bat and insect response variables included in generalized linear mixedmodels.

Category Measurement Response variable

Bat activity Average activity; All batsThe light-exploiting groupChalinolobus gouldiiOzimops rideiThe light-sensitive groupNyctophilus spp.Vespadelus vulturnus

Bat species richness Average species richness; All batsFor light-exploiting groupFor light-sensitive group

Bat emergence times Average time until firstactivity;

All bats

For the light-exploitinggroupFor the light-sensitive groupChalinolobus gouldiiOzimops rideiNyctophilus spp.Vespadelus vulturnus

Insect abundance Averages; Number of LepidopteracollectedBiomass of Lepidoptera


20

each edge and interior site pair, we therefore calculated the proportionof calls recorded at the forest edge compared that site's interior controlfor both light-sensitive and light-exploiting functional groups. A oneway ANOVA was used to compare the proportion of calls recorded atlight edges and at dark edges for both light-exploiting and light-sensi-tive groups.

3. Results

We recorded 9965 bat passes throughout our acoustic survey, with62.4% identified to species or species group level. We detected 16species or species groups, seven of which have a conservation status ofvulnerable under the NSW Biodiversity Conservation Act 2016;Chalinolobus dwyeri, Falsistrellus tasmaniensis, Miniopterus orianae ocea-nensis, Miniopterus australis, Micronomous norfolkensis, Scoteanex ruep-pellii, Saccolaimus flaviventris.

Artificially lit edges, dark edges and interiors all differed sig-nificantly from each other in the amount of vegetation cover within250m of each site. Interior control sites had greater vegetation extentthan dark edges (p < 0.005), and light edges had greater vegetationextent than dark edges (p < 0.05). Forest patches had significantly lessvegetation cover within a 250m radius of each site than connectedforest (p < 0.001). Artificially lit edges were significantly brighter(F2,59= 157.9, p < 0.005) than dark edges (p < 0.005) and dark in-teriors (p < 0.005).

Echolocation calls of bats were recorded at all sites. Light-exploitingbat species were detected at 95% of sites; with 96% of edge sites and93% of interior sites. Light-sensitive bat species were detected at 51% ofsites, including 41% of edge sites, and 61% at interior sites.

Total bat activity was significantly different among light treatments,with light edges and dark edges having significantly lower activity thandark interior sites (Table 2). Total bat activity was not significantlydifferent between light and dark edges (p > 0.05; Fig. 3a). Speciesrichness was not significantly affected by light treatment (Table 2).

3.1. Light-sensitive group responses

Dark interior sites supported higher light-sensitive species richness(Table 2), with no difference between light edges and dark edges(p > 0.05). Similarly, dark interior sites supported the highest light-sensitive bat activity (Table 2), but then dark edges supported higheractivity that light edges (p=0.037, Table 2, Fig. 3b). A similar patternwas seen for the most commonly recorded light-sensitive species; Ves-padelus vulturnus activity was highest at dark interior sites, lower atdark edges, and lowest at light edges (p=0.02, Table 2; Fig. 3f). Theactivity of Nyctophilus spp. was again highest at the dark interior sites,and significantly lower at both dark and light edges (Table 2). Therewas no significant difference between the Nyctophilus spp. activity atlight and dark edges (p > 0.05).

A higher proportion of the light-sensitive group's echolocation callswere recorded at the dark edges than at the light edges, when comparedwith the proportion recorded at that site's dark interior site(F1,23= 4.96, p= 0.038; Fig. 4). Where light-sensitive species werepresent, they were more likely to be active at the edge if it was darkthan if it was artificially lit.

3.2. Light-exploiting group responses

The species richness of the light-exploiting group was not affectedby light treatment (Table 2). Dark interior sites supported higher light-exploiting bat activity than light and dark edges (Table 2, Fig. 3c), andthere was no difference between dark and light edges (p > 0.05). Theactivity of C. gouldii followed a similar pattern, with activity highest atdark interior sites (Table 2, Fig. 3d), with no difference between lightand dark edges (p > 0.05). However, the activity of O. ridei followed adifferent pattern (Fig. 3e); activity of this species was highest at light

edges (Table 2, Fig. 3e), lower at dark edges, and lowest at dark interiorsites (p= 0.007). There was no significant difference in the proportionof the light-exploiting group's echolocation calls recorded at light edgesand dark edges, when compared with the proportion recorded at that

Table 2Results of GLMM with habitat type and light treatment as fixed effects, and siteas a random effect, with response variables listed on the far left column, and thedark interior control treatment and patch treatment as the reference categories.

Parameter Estimate Standarderror

t value p value

Total bat activityIntercept 2.834 0.3099 9.145 <0.001Connected forest versuspatch

0.079 0.4288 0.185 0.854

Light edge versus interior −0.814 0.0765 −10.642 <0.001Dark edge versus interior −0.604 0.0791 −7.629 <0.001

Average bat species richnessIntercept 4.475 0.5258 8.511 <0.001Connected forest versuspatch

0.330 0.6009 0.550 0.585

Light edge versus interior −0.518 0.7435 −0.696 0.489Dark edge versus interior −0.577 0.7277 −0.793 0.431

Average light-sensitive speciesrichness

Intercept 2.256 0.3569 6.319 <0.001Connected forest versuspatch

−0.044 0.3951 −0.112 0.911


Average light-sensitive speciesactivity

Intercept 1.379 0.5022 2.745 0.009Connected forest versuspatch

−0.262 0.6275 −0.417 0.679

Light edge versus interior −2.790 0.4203 −6.638 <0.001Dark edge versus interior −0.305 0.1991 −1.531 0.134

Average Nyctophilus spp.activity

Intercept 1.872 0.6269 2.985 0.005Connected forest versuspatch

−0.854 0.5699 −1.498 0.144


Average V. vulturnus activityIntercept 1.282 0.6090 2.105 0.043Connected forest versuspatch

−0.533 0.7454 −0.714 0.480

Light edge versus interior −3.100 0.5899 −5.256 <0.001Dark edge versus interior −0.061 0.2235 −0.272 0.787

Average light-exploiting speciesrichness


−0.019 0.2304 −0.084 0.934

Light edge versus interior 0.338 0.2851 1.185 0.241Dark edge versus interior 0.058 0.2790 0.208 0.836

Average light-exploiting speciesactivity


−0.004 0.4623 −0.008 0.994


Average C. gouldii activityIntercept 2.214 0.4017 5.512 <0.001Connected forest versuspatch

−0.130 0.5573 −0.232 0.817


Average O. ridei activityIntercept 0.040 0.4916 0.082 0.935Connected forest versuspatch

−0.182 0.6445 −0.282 0.779

Light edge versus interior 0.644 0.3634 1.772 0.082Dark edge versus interior −0.489 0.1863 −2.626 0.011


21

a a c b

a a b a a b

a b

a b c a b b

g) Species richness of light-sensitive group h) Biomass of lepidopterans

Ave

rage

spe

cies

rich

ness

Ave

rage

lepi

dopt

eran

bio

mas

s (m

g )

a b b b a a

e) Ozimops ridei activity f) Vespadelus vulturnus activity

(caption on next page)


22

site's dark interior site (F1,2= 1.08, p > 0.05; Fig. 4).Habitat type, whether the forest was an isolated patch or connected

to other forested areas, had no significant effect on the activity or thespecies richness of the bat assemblage, the light-sensitive group, or thelight-exploiting group (Table 2).

3.3. Temporal activity patterns

Overall, there was no difference in first recording time betweenlight treatments across species (Table 3), however differences werefound between functional groups.

The light-sensitive functional group was active significantly earlierat interior sites when compared with dark and light edges (Table 3).There was no significant difference between the first recordings at lightand dark edges (p > 0.05). Nyctophilus spp. showed the same pattern(Table 3), again with no difference between first recordings at light anddark edges (p > 0.05). There was no difference between first recordingtimes of Vespadelus vulturnus at both dark interior sites and dark edges(Table 3). However, first recording time was significantly later at lightedges than at dark interior sites (Table 3) and dark edges (p= 0.008).

Overall, the first activity of the light-exploiting functional groupwas not significantly different across light treatments (Table 3), butthere were species-specific responses. First activity of the light-ex-ploiting C. gouldii was not significantly different across light treatments,however it was first recorded significantly earlier in the night in con-nected forest than in patches (Table 3). Emergence time in connectedforest averaged 151.9 ± 9.1min after sunset, versus203.1 ± 16.9 min after sunset in isolated forest patches. The time offirst activity for O. ridei was significantly earlier at dark interior sitesthan at both light and dark edges (Table 3). O. ridei was also recordedsignificantly earlier at light edges than at dark edges (p= 0.001).

Throughout the night, the difference in the activity of the light-sensitive group between light and dark edges was not as pronounced asthe difference between interior sites and pooled edge sites. The activityof the light-sensitive group was higher at interior sites throughout thenight, and was at its highest just after sunset (Fig. 5).

3.4. Insect responses

Lepidoptera biomass was significantly affected by light treatment(Fig. 3h); biomass was higher at light edges than dark interior sites(Table 4) and posthoc tests showed that biomass at light edges washigher than at dark edges also (p < 0.001). The average number ofLepidoptera caught at dark interior sites did not differ from the numbercaught at dark edges (Table 4). However, the number of Lepidopteracaught at light edges was significantly lower than dark interior sites(Table 4), and posthoc tests showed that the number of Lepidoptera atlight edges was lower than at dark edges also (p=0.012).

4. Discussion

Our study demonstrates the disruptive effects of artificial light at theedges of urban forest on light-sensitive and light-exploiting in-sectivorous bats. Our findings are consistent with research that artificial

Fig. 3. Plots showing marginal means (± 1 s.e.) after controlling for site. Panel a. Bat activity compared across light treatments; panel b. Activity of the light-sensitive group compared across light treatments; panel c. Activity of the light-exploiting group compared across light treatments; panel d. Activity of C. gouldiicompared across light treatments; panel e. Activity of O. ridei compared across light treatments; panel f. Activity of V. vulturnus compared across light treatments;panel g. Species richness of the light-sensitive group compared across light treatments; panel h. Lepidopteran biomass (mg) compared across light treatments.Statistically significant differences when comparing variable measurements between treatments are depicted using differing letters (a, b and c).

0

10

20

30

40

50

60

70

Light tolerant Light sensitive

Perc

enta

ge o

f bat

cal

ls re

cord

ed a

t site

edg

es

Functional group

Dark edges

Light edges

Fig. 4. Average percentage (± se) of light-sensitive and light-exploiting func-tional group echolocation calls recorded at light edge sites and at dark edgesites. Calculated by combining calls for each site with calls from associatedinterior control site and then calculating the percentage of the total calls re-corded at the edge.

Table 3Results of GLMM with habitat type and light treatment as fixed effects, and siteas a random effect, with response variables listed on the far left column, and thedark interior control treatment and patch treatment as the reference categories.


t value p value

Minutes until first bat activityIntercept 4.967 0.1635 30.383 <0.001Connected forest versus patch −0.089 0.2312 −0.386 0.700Light edge versus interior −0.010 0.0289 −0.352 0.726Dark edge versus interior 0.021 0.0278 0.744 0.460

Minutes until first light-sensitivegroup activity

Intercept 5.386 0.1347 39.973 <0.001Connected forest versus patch −0.013 0.1718 −0.077 0.939Light edge versus interior 0.265 0.0397 6.687 <0.001Dark edge versus interior 0.232 0.0380 6.092 <0.001

Minutes until first Nyctophilusspp. activity

Intercept 5.500 0.1772 31.047 <0.001Connected forest versus patch 0.147 0.2225 0.659 0.518Light edge versus interior 0.465 0.0615 7.563 <0.001Dark edge versus interior 0.462 0.0591 7.825 <0.001

Minutes until first V. vulturnusactivity

Intercept 5.413 0.1748 30.962 <0.001Connected forest versus patch 0.151 0.2275 0.664 0.514Light edge versus interior 0.252 0.0482 5.228 <0.001Dark edge versus interior 0.023 0.0559 0.411 0.685

Minutes until first light-exploiting group activity

Intercept 5.064 0.0813 62.273 <0.001Connected forest versus patch −0.045 0.1027 −0.442 0.660Light edge versus interior 0.054 0.0974 0.557 0.580Dark edge versus interior 0.050 0.0975 0.517 0.607

Minutes until first C. gouldiiactivity

Intercept 202.150 16.7063 12.100 <0.001Connected forest versus patch −53.183 20.6049 −2.581 0.013Light edge versus interior −15.014 23.1303 −0.649 0.519Dark edge versus interior 26.846 24.3959 1.100 0.276

Minutes until first O. rideiactivity

Intercept 5.272 0.1409 37.419 <0.001Connected forest versus patch −0.129 0.1919 −0.672 0.506Light edge versus interior 0.176 0.0333 5.291 <0.001Dark edge versus interior 0.376 0.0419 8.961 <0.001


23

light has a negative effect on certain species in ecologically sensitivehabitat (Straka et al., 2016) and that artificial light at the urban forestedge contributes to a loss of functional connectivity (LaPoint et al.,2015). Streetlights at the forest edge negatively affected the activity ofthe putative light-sensitive group of bats. Sydney's smallest species, V.vulturnus was particularly negatively affected by light, being less activeand emerging later at artificially lit edges when compared with darkedges. Despite light edges having greater vegetation cover with 250m,bat activity and richness was still lower here than at dark edges.Available functional habitat in urban forests is reduced by artificiallight, and with some species unable to cross lit areas (Hale et al., 2015),urban biodiversity may continue to decline unless the degrading im-pacts of ALAN are addressed.

Species predicted to be light-exploiting were either unaffected bystreetlights along the forest edge, such as C. gouldii, or increased inactivity, such as O. ridei. Ozimops ridei showed the strongest positiveresponse to artificial light at the forest edge, with higher activity andearlier emergence time at artificially lit edges than dark edges, con-sistent with previous findings (Threlfall et al., 2013). The species-spe-cific differences within the a priori fast flying functional group suggestthat responses to ALAN may be more complex than basic guild-relatedresponses. Our data implies that other aspects of the urban environmentway exert more influence on those species.

Whether the forest was connected to other forest or in an isolatedpatch, had little influence on the effects of artificial light at the forestedge. Chalinolobus gouldii were recorded earlier in the night in con-nected forest than in isolated patches, suggesting that as well as C.gouldii being unaffected by artificial lights at the forest edge, this spe-cies may be more likely to roost in connected forest and hence emergeearlier there. Our findings suggest that the bat assemblage appears re-latively similar across all forest habitat in this city, and this needsfurther investigation as it would inform urban conservation strategy.

Recent multi-taxa research concluded that large connected patchesof vegetation are critically important for urban biodiversity (Benindeet al., 2015). Our data supports this, showing that total bat activity, theactivity of the light-sensitive species and C. gouldii were consistentlyhigher on forest interiors when compared with edges. Nyctophilus gouldiwere rarely recorded anywhere but bushland interiors, demonstratingthe importance of forest interiors for bat diversity in cities.

With regards to insect trapping, Lepidoptera numbers were lowestat light edges, suggesting that the biomass was made up of fewer butlarger individuals. We conclude that our findings are consistent withothers (Pintérné and Pödör, 2017), and may have been due to street-lights in the lit conditions outshining our light traps, and attractingmost of Lepidoptera away from the trap. In future studies, suction traps(Shortall et al., 2009), sticky traps, malaise traps (Hosking, 1979;Hallmann et al., 2017) or camera traps (Rydell, 1992) may be betteralternatives for sampling phototactic insects in artificially lit condi-tions.

4.1. Mechanisms driving light sensitivity in insectivorous bats

In line with our predictions, V. vulturnus was light sensitive, usingforest edges less so when they were lit. Thus ALAN may be contributingto edge effects and habitat loss for this species within cities (Gonsalvesand Law, 2017). Of all V. vulturnus calls, 27% were recorded at darkedges, and only 2% were recorded at artificially lit edges. This specieswas active significantly later at lit edges than at dark edges. Vespadelusvulturnus is an edge-space foraging bat, although it is quite maneuver-able and so may not be as dependent on edges and flyways as otheredge-space foraging bats (Law et al., 2011). Vespadelus vulturnus alsoflies closer to the ground in thinned as opposed to unthinned under-storey, perhaps for cover and protection (Adams, 2012). Despite theseflight preferences, our data shows that this bat uses forest edges in

Fig. 5. Activity of light-sensitive bats throughout the night. This is calculated by noting sunset and sunrise each night of the survey and calculating what percentageof the night had elapsed when each call was recorded. The graph shows the number of calls recorded within each 10% increment throughout the night. The bottomgraph shows the combined data from the edge sites compared with the data from the interior sites.

Table 4Results of GLMM with habitat type and light treatment as fixed effects, and siteas a random effect, with response variables listed on the far left column, withpatch and dark edge treatment as reference categories.


t value p value

Biomass of LepidopteraIntercept 5.731 0.0955 60.003 0.000Connected forest versus patch 0.143 0.3798 0.375 0.710Light edge versus interior 0.244 0.0309 7.896 0.000Dark edge versus interior −0.104 0.0262 −3.979 0.000

Average number of Lepidopteracaught

Intercept 3.209 0.1598 20.080 0.000Connected forest versus patch −0.533 0.6323 −0.843 0.407Light edge versus interior −0.546 0.1289 −4.239 0.000Dark edge versus interior 0.126 0.0845 1.495 0.146


24

cities, and is negatively impacted by light sources along those edges. Itsedge-space foraging preference makes V. vulturnus, and other slower-flying edge-space foragers, susceptible to habitat loss and decline fromstreetlights at the forest edge. This is consistent with previous researchthat failed to detect V. vulturnus in small pockets of urban forest inhighly urbanized parts of Sydney (Gonsalves and Law, 2017).

In comparison, 92% of calls identified as Nyctophilus spp., the spe-cies group adapted to foraging in cluttered vegetation, were recorded atinterior sites. Although Nyctophilus spp. have been observed using openand edge habitat in woodland (Brigham et al., 1997), our study con-firms low activity of this species at the urban forest edge, regardless ofwhether streetlights were present or not. Artificial light at the forestedge did not further reduced the low activity of Nyctophilus spp. at theselocations. Our findings are consistent with a previous study that foundthat movements of these taxa were mostly restricted within the boundsof an urban forest remnant (Threlfall et al., 2013). These taxa appear tobe light-sensitive (Threlfall et al., 2013), but the lack of edge habitat usesuggests both habitat fragmentation and disruption in the connectivityof dark vegetation across cities have critically impacted the urbansurvival of this group. Some bats, including Nyctophilus spp., could haveretained a sensitivity to ultraviolet light (Gorresen et al., 2015), and thismay cause these species to avoid streetlights emitting ultraviolet, likethe ones in this study.

4.2. The importance of dark connected forest for urban bats

Our study reveals that total bat activity was higher at dark interiorsites than at either light or dark forest edges, confirming previous re-search that dark forest is crucial for the persistence of some in-sectivorous bats in cities (Threlfall et al., 2013; Hale et al., 2015; Strakaet al., 2016; Azam et al., 2018), Haddock et al., in press). Activity atthese dark forest interior sites peaks early in the night, suggesting thatthis is where most bats roost in hollow trees. Throughout the course ofthe night, the majority of calls were recorded at interior sites, whetherlooking at light-sensitive or light-exploiting functional groups. Ourfindings provide support for the conservation of dark forest in cities,and reconfirm the importance of connectivity between natural areas(Dearborn and Kark, 2010; Goddard et al., 2010; Hale et al., 2012;Beninde et al., 2015), particularly dark corridors for light-sensitivespecies (Bolívar-Cimé et al., 2013; Hale et al., 2015; Straka et al., 2016).

4.3. Artificial light altering the competition for resources

Niche differentiation between bat species can sometimes be subtle(Krüger et al., 2014) as bats exploit slightly different insect prey in si-milar habitat. Clutter-adapted bats will glean moths and beetles fromthe ground and from leaves, whereas open-space adapted bats willpursue prey on the wing (Denzinger and Schnitzler, 2013). Artificiallight sources are known to attract insect prey out of dark habitat anddeplete foraging grounds for light-sensitive species (Rydell, 1992;Rydell and Racey, 1995; Arlettaz et al., 2000). Our data supports thiseffect, where overall moth biomass was significantly higher at lit edgescompared to dark edges and interior sites. Overall moth numbers didnot follow the same pattern, suggesting that larger insects dominatednear streetlights. This potential “vacuum cleaner effect”, where pho-totactic insects migrate out of dark forest and towards artificial light,may be decreasing the amount of available prey to bats restricted todark areas, making it harder for them to persist in cities (Arlettaz et al.,2000). Investigation is needed into the extent of this depletion of darkforaging grounds for light-sensitive bats. Our data shows that light-sensitive bats are less likely to use artificially lit edges, and wouldtherefore be less able to access the increased insect biomass there. Si-milar to previous experiments, artificial light sources could be creatingforaging areas that mainly light-exploiting bats can exploit (Arlettazet al., 2000; Stone et al., 2015a, 2015b). ALAN, particularly artificiallight sources at forest edges in cities, may be disrupting the important

environmental niches that allow bat species to co-exist.

4.4. Light as an edge effect

Artificial light as an edge effect has been studied in combinationwith traffic noise, mostly in the context of road ecology. In this contextit has strong negative effects on many nocturnal taxa (Kempenaerset al., 2010; Hale et al., 2015; Azam et al., 2018). Mammal speciesrichness increases at the point where light levels return to ambientdarkness, and bird diversity increases with distance from the road(Pocock and Lawrence, 2005). Small and narrow urban forest could behighly light-impacted habitat (Azam et al., 2018). In our study, artifi-cial light at the forest edge negatively impacted species that may al-ready be struggling to survive in cities. Functional diversity in im-portant urban forest patches and corridors may be significantlyimpacted, driven by an alteration of the artificial lighting regime. In-vestigation into the zone of influence from permanent artificial lightsources on other taxa is a research priority.

4.5. Lunar phobia and light phobia

Light phobia could be driven by an innate lunar phobia, withchanging moon phase and the brightness of the moon linked to theemergence time and behaviour of some light-sensitive species such as V.vulturnus and others (Law, 1997; Saldaña-Vázquez and Munguía-Rosas,2013). Lunar phobia is linked to morphological traits associated withforaging strategy; slow-flying bats respond more negatively to brightmoonlight (Saldaña-Vázquez and Munguía-Rosas, 2013). These speciesare also more likely to emerge later in the evening (Jones and Rydell,1994), theoretically because they are adapted to hunting prey availablein the darkest part of the night and so avoid crepuscular predator peaks.Our data confirms this trend, that slow flying bat species avoid artificiallight. This morphological group may be responding negatively to arti-ficial light as they have evolved to respond negatively to moonlight.The species-specific differences here warrant further research. Fasterflying bats, those that appear more exploiting of artificial light, are lessaffected by moonlight (Saldaña-Vázquez and Munguía-Rosas, 2013).These species may be pre-adapted, or at least resilient, to artificially litenvironments. However much like the drivers of light phobia in bats,the mechanisms driving lunar phobia are difficult to pinpoint. Periodsof bright moonlight are associated with reduced levels of insect prey(Lang et al., 2006), and may leave bats more vulnerable to predation(Law, 1997), particularly important for slow flying bats. Very few stu-dies have provided evidence that predation risk for bats increases undermoonlit skies, even though it is discussed frequently in the literature(Saldaña-Vázquez and Munguía-Rosas, 2013; Stone et al., 2015a,2015b). Future studies teasing apart the causes of artificial light phobia,changes in food availability or predation risk, would elucidate themechanisms influencing lunar and light phobia in bats.

4.6. Future directions and conclusions

In urbanized Sydney, five species are at particular risk of ALANdegrading viable habitat. Elsewhere the impact may be greater, forexample; locations like tropical Barro Colorado Island in Panama arehome to a much higher proportion of slower flying bat species(Denzinger et al., 2017). If the parameters of light sensitivity discussedhere are applied to other bat communities, many more species could beidentified as at risk of habitat loss due to urban lighting at the edges ofnatural forest. This is an issue for particular consideration in expandingcities and peri-urban areas. Lighting near ecologically important areas,like drinking troughs, is already established as having a disruptive ef-fect on bat drinking behaviours (Russo et al., 2017; Russo et al., 2018).Foraging by many bats species is affected also, and like others(Spoelstra et al., 2015; Straka et al., 2016), we recommend avoiding theinstallation of streetlights near ecologically sensitive areas in cities,


25

such as native forests and wetlands. Other mitigation strategies, such aspart night lighting, may be an alternative to constant lighting (Azamet al., 2015); but need to be turned off early enough in the night tosuccessfully mitigate the impact on vulnerable species. If lighting nearecologically important forest is necessary, red lights (usually between620 and 750 nm) have been shown to have minimal effect on a varietyof mammal taxa (Spoelstra et al., 2015; Spoelstra et al., 2017), althoughtheir suitability as a bat-friendly alternative is controversial (Voigtet al., 2018) and needs further testing. LED technology allows lights tobe manufactured to emit specific spectra, which has recently beenfound to be useful for minimising disruption for bats (Haddock et al., inpress). Research could be used to help manufacturers create lights withthe least disruptive spectra for nocturnal biodiversity. However, con-serving dark forested areas in cities remains crucial for the persistenceof nocturnal biodiversity such as light-sensitive insectivorous bats.

Role of the funding source

This research was financially funded by the Holsworth WildlifeResearch Endowment (grant numbers: HOLSW2015-2-F147 andHWRE2016R2063CONT).

Acknowledgements

I would like to thank Kyle Armstrong at Specialized Zoology for theloan of the Anabat II detectors.

This research was supported by the Australian Government'sNational Environmental Science Program through the Clean Air andUrban Landscapes Hub.

Ethics declaration

This research has been approved by the Animal Ethics Committee atThe University of Sydney, protocol number 2015/916.

Appendix A

i. Insect sampling

Insects were sampled at the edge and interior locations of all patchsites (n=30) for one night. For each patch, the edge and the interiorwere sampled on the same night to control for variation across nights.Sampling took place two weeks after acoustic sampling to avoid in-terfering with naturally occurring bat activity, however temperatureand humidity were not significantly different and so insect samplingnights were deemed comparable to bat sampling nights. We avoidedsurveying on full moon and new moon nights, and for 3 days either sideof both events. Black light intercept traps (Australian EntomologicalSupplies, Murwillumbah, Australia) were placed on the ground, in thesame location used for the Anabat II detectors two weeks prior,< 3mfrom the base of the streetlight pole. 50ml of ethyl acetate was added tothe bottom of the light trap, bulbs were attached to 12 V batteries, andthe trap was turned on between 2000 and 2300. Excalibur digital timers(Merlin Distribution, Roseville, Australia) automatically turned thelights off after 2 h, and traps were collected within an hour of the lightturning off. Individual insects were then separated and identified toorder. Only Lepidoptera were included in the analysis as only a smallnumber of insects of other orders, Dipterans and Coleopterans, werecaught. Body length of each individual insects was measured, andbiomass was calculated using established methods (Sample et al.,1993).

ii. Limitations of insect surveys

Measurements of lepidopteran densities and biomass were counter-intuitive; firstly, biomass was higher at light edges than dark edges or

interior control sites, consistent with previous research (Eisenbeis andEick, 2010; Barghini and Souza de Medeiros, 2012; van Grunsven et al.,2014; Plummer et al., 2016). However secondly, Lepidoptera numberswere lowest at light edges, suggesting that the biomass was made up offewer but larger individuals. For light-exploiting, larger aerial-hawkingbats, moths attracted to artificially lit areas may therefore comprise anenergy-rich food source of larger insect prey. Our result of lowernumbers of moths around lights has been found in another study(Pintérné and Pödör, 2017), which found a higher number of lepi-dopteran in dark, semi-rural areas than in well-lit, urban areas. Ourdata may have been due to streetlights in the lit conditions outshiningour light traps, and attracting most of Lepidoptera away from the trap(Pintérné and Pödör, 2017). Although it seems illogical to sample thedifference between light and dark conditions using an additional lightsource, this method is standard practice in studies of light (Spoelstraet al., 2015; Pintérné and Pödör, 2017).

References

Adams, M.D., 2012. Vertical Stratification of Insectivorous Bats (Microchiroptera) inHarvested Forests: Assessing the Role of Structural Clutter in Shaping Patterns ofFlight Activity. Doctor of Philosophy thesis. School of Biological Sciences, Universityof Wollongong, pp. 2012. http://ro.uow.edu.au/theses/3702.

Adams, M.D., Law, B.S., French, K.O., 2009. Vegetation structure influences the verticalstratification of open-and edge-space aerial-foraging bats in harvested forests. For.Ecol. Manag. 258 (9), 2090–2100.

Adams, M.D., Law, B.S., Gibson, M.S., 2010. Reliable automation of bat call identificationfor eastern New South Wales, Australia, using classification trees and AnaSchemesoftware. Acta Chiropterologica 12 (1), 231–245.

Aguilar, R., Quesada, M., Ashworth, L., Herrerias-Diego, Y., Lobo, J., 2008. Geneticconsequences of habitat fragmentation in plant populations: susceptible signals inplant traits and methodological approaches. Mol. Ecol. 17 (24), 5177–5188.

Arlettaz, R., Godat, S., Meyer, H., 2000. Competition for food by expanding pipistrelle batpopulations (Pipistrellus pipistrellus) might contribute to the decline of lesser horse-shoe bats (Rhinolophus hipposideros). Biol. Conserv. 93 (1), 55–60.

Azam, C., Kerbiriou, C., Vernet, A., Julien, J.F., Bas, Y., Plichard, L., Maratrat, J., Le Viol,I., 2015. Is part-night lighting an effective measure to limit the impacts of artificiallighting on bats? Glob. Chang. Biol. 21 (12), 4333–4341.

Azam, C., Le Viol, I., Bas, Y., Zissis, G., Vernet, A., Julien, J.-F., Kerbiriou, C., 2018.Evidence for distance and illuminance thresholds in the effects of artificial lighting onbat activity. Landsc. Urban Plan. 175, 123–135.

Barghini, A., Souza de Medeiros, B.A., 2012. UV radiation as an attractor for insects.Leukos 9 (1), 47–56.

Basham, R., Law, B., Banks, P., 2011. Microbats in a ‘leafy’ urban landscape: are theypersisting, and what factors influence their presence? Austral Ecology 36 (6),663–678.

Beninde, J., Veith, M., Hochkirch, A., 2015. Biodiversity in cities needs space: a meta-analysis of factors determining intra-urban biodiversity variation. Ecol. Lett. 18 (6),581–592.

Benson, D.H., Howell, J., 1995. Taken for Granted: The Bushland of Sydney and ItsSuburbs. Kangaroo Press in association with the Royal Botanic Gardens Sydney.

Blake, D., Hutson, A., Racey, P., Rydell, J., Speakman, J., 1994. Use of lamplit roads byforaging bats in southern England. J. Zool. 234 (3), 453–462.

Blakey, R., Law, B., Straka, T., Kingsford, R., Milne, D., 2018. Importance of wetlands tobats on a dry continent: a review and meta-analysis. Hystrix, the Italian Journal ofMammalogy. Online first. 29 (1).

Boldogh, S., Dobrosi, D., Samu, P., 2007. The effects of the illumination of buildings onhouse-dwelling bats and its conservation consequences. Acta Chiropterologica 9 (2),527–534.

Bolívar-Cimé, B., Laborde, J., Cristina, M., MacSwiney, G., Muñoz-Robles, C., Tun-Garrido, J., 2013. Response of phytophagous bats to patch quality and landscapeattributes in fragmented tropical semi-deciduous forest. Acta Chiropterologica 15 (2),399–409.

Brigham, R.M., Francis, R.L., Hamdorf, S., 1997. Microhabitat use by two species ofNyctophilus bats: a test of ecomorphology theory. Australian Journal of Zoology 45(6), 553–560.

Czech, B., Krausman, P.R., Devers, P.K., 2000. Economic associations among causes ofspecies endangerment in the United States: associations among causes of speciesendangerment in the United States reflect the integration of economic sectors, sup-porting the theory and evidence that economic growth proceeds at the competitiveexclusion of nonhuman species in the aggregate. BioScience 50 (7), 593–601.

Dearborn, D.C., Kark, S., 2010. Motivations for conserving urban biodiversity. Conserv.Biol. 24 (2), 432–440.

Denzinger, A., Schnitzler, H.U., 2013. Bat guilds, a concept to classify the highly diverseforaging and echolocation behaviors of microchiropteran bats. Front. Physiol. 4.

Denzinger, A., Tschapka, M., Schnitzler, H.U., 2017. The role of echolocation strategiesfor niche differentiation in bats. Can. J. Zool. 96 (3), 171–181.

Downs, N., Beaton, V., Guest, J., Polanski, J., Robinson, S., Racey, P.A., 2003. The effectsof illuminating the roost entrance on the emergence behaviour of Pipistrellus pyg-maeus. Biol. Conserv. 111 (2), 247–252.


26

http://ro.uow.edu.au/theses/3702

http://refhub.elsevier.com/S0006-3207(18)31368-5/rf0010

























































Eisenbeis, G., Eick, K., 2010. Attraction of nocturnal insects to streetlights with specialregard to LEDs. In: Society for Conservation Biology, 24th Annual Meeting, . http://www.conbio.org/activities/meetings/2010/images/abstracts.pdf (checked10.04. 11).

Evans, K.L., Newson, S.E., Gaston, K.J., 2009. Habitat influences on urban avian assem-blages. Ibis 151 (1), 19–39.

Falchi, F., Cinzano, P., Duriscoe, D., Kyba, C.C., Elvidge, C.D., Baugh, K., Portnov, B.A.,Rybnikova, N.A., Furgoni, R., 2016. The new world atlas of artificial night skybrightness. Sci. Adv. 2 (6), e1600377.

Fischer, J., Lindenmayer, D.B., 2007. Landscape modification and habitat fragmentation:a synthesis. Glob. Ecol. Biogeogr. 16 (3), 265–280.

Gaston, K.J., Visser, M.E., Hölker, F., 2015. The biological impacts of artificial light atnight: the research challenge. Philosophical Transactions of the Royal Society B:Biological Sciences 370 (1667), 2014013.

Goddard, M.A., Dougill, A.J., Benton, T.G., 2010. Scaling up from gardens: biodiversityconservation in urban environments. Trends Ecol. Evol. 25 (2), 90–98.

Gonsalves, L., Law, B., 2017. Seasonal activity patterns of bats in North Sydney, NewSouth Wales: implications for urban bat monitoring programs. AustralianMammalogy 40 (2), 220–229.

Gorenzel, W.P., Salmon, T.P., 1995. Characteristics of American crow urban roosts inCalifornia. J. Wildl. Manag. 638–645.

Gorresen, P.M., Cryan, P.M., Dalton, D.C., Wolf, S., Bonaccorso, F.J., 2015. Ultravioletvision may be widespread in bats. Acta Chiropterologica 17 (1), 193–198.

Haffner, M., Stutz, H., 1985. Abundance of Pipistrellus pipistrellus and Pipistrellus kuhliiforaging at street-lamps. Myotis 23 (24), 167–168.

Hale, J.D., Fairbrass, A.J., Matthews, T.J., Sadler, J.P., 2012. Habitat composition andconnectivity predicts bat presence and activity at foraging sites in a large UK con-urbation. PLoS One 7 (3), e33300.

Hale, J.D., Fairbrass, A.J., Matthews, T.J., Davies, G., Sadler, J.P., 2015. The ecologicalimpact of city lighting scenarios: exploring gap crossing thresholds for urban bats.Glob. Chang. Biol. 21 (7), 2467–2478.

Hallmann, C.A., Sorg, M., Jongejans, E., Siepel, H., Hofland, N., Schwan, H., Stenmans,W., Müller, A., Sumser, H., Hörren, T., 2017. More than 75 percent decline over 27years in total flying insect biomass in protected areas. PLoS One 12 (10), e0185809.

Hawrot, R.Y., Niemi, G.J., 1996. Effects of edge type and patch shape on avian com-munities in a mixed conifer-hardwood forest. Auk 586–598.

Hölker, F., Wolter, C., Perkin, E.K., Tockner, K., 2010. Light pollution as a biodiversitythreat. Trends Ecol. Evol. 25, 681–682.

Hosking, G., 1979. Trap comparison in the capture of flying Coleoptera. New ZealandEntomologist 7 (1), 87–92.

Jones, G., Rydell, J., 1994. Foraging strategy and predation risk as factors influencingemergence time in echolocating bats. Philosophical Transactions of the Royal SocietyB: Biological Sciences 346 (1318), 445–455.

Jung, K., Kalko, E.K., 2011. Adaptability and vulnerability of high flying Neotropicalaerial insectivorous bats to urbanization. Divers. Distrib. 17 (2), 262–274.

Jung, K., Threlfall, C.G., 2016. Urbanisation and Its Effects on Bats—A Global Meta-analysis. Bats in the Anthropocene: Conservation of Bats in a Changing World.Springer, pp. 13–33.

Jung, K., Threlfall, C.G., 2018. Trait-dependant tolerance of bats to urbanisation: a globalmeta-analysis. Proc. R. Soc. B Biol. Sci. 285 (1885).

Keitt, T., Urban, D., Milne, B., 1997. Detecting critical scales in fragmented landscapes.Conserv. Ecol. 1 (1), 1–17.

Kempenaers, B., Borgström, P., Loës, P., Schlicht, E., Valcu, M., 2010. Artificial nightlighting affects dawn song, extra-pair siring success, and lay date in songbirds. Curr.Biol. 20 (19), 1735–1739.

Krüger, F., Clare, E., Greif, S., Siemers, B., Symondson, W., Sommer, R., 2014. An in-tegrative approach to detect subtle trophic niche differentiation in the sympatrictrawling bat species Myotis dasycneme and Myotis daubentonii. Mol. Ecol. 23 (15),3657–3671.

Kunz, T.H., 1982. Roosting ecology of bats. In: Ecology of Bats. Springer, pp. 1–55.Kupfer, J.A., 2012. Landscape ecology and biogeography: rethinking landscape metrics in

a post-FRAGSTATS landscape. Prog. Phys. Geogr. 36 (3), 400–420.La Sorte, F.A., Fink, D., Buler, J.J., Farnsworth, A., Cabrera-Cruz, S.A., 2017. Seasonal

associations with urban light pollution for nocturnally migrating bird populations.Glob. Chang. Biol. 23 (11), 4609–4619.

Lacoeuilhe, A., Machon, N., Julien, J.-F., Le Bocq, A., Kerbiriou, C., 2014. The influence oflow intensities of light pollution on bat communities in a semi-natural context. PLoSOne 9 (10), e103042.

Lang, A.B., Kalko, E.K., Römer, H., Bockholdt, C., Dechmann, D.K., 2006. Activity levelsof bats and katydids in relation to the lunar cycle. Oecologia 146 (4), 659–666.

LaPoint, S., Balkenhol, N., Hale, J., Sadler, J., Ree, R., 2015. Ecological connectivity re-search in urban areas. Funct. Ecol. 29 (7), 868–878.

Law, B.S., 1997. The lunar cycle influences time of roost departure in the commonblossom bat, S. australis. Australian Mammalogy 20, 21–24.

Law, B., Anderson, J., Chidel, M., 1998. A bat survey in state forests on the south-westslopes region of New South Wales with suggestions of improvements for future sur-veys. Australian Zoologist 30 (4), 467–479.

Law, B., Chidel, M., Tap, P., 2011. Bat activity in ephemeral stream-beds in the Pilligaforests: clarifying the importance of flyways and buffer widths in open forest andwoodland. In: Law, B., Eby, P., Lunney, D., Lumsden, L. (Eds.), The Biology andConservation of Australasian Bats, pp. 308–321.

Leonard, R.J., Hochuli, D.F., 2017. Exhausting all avenues: why impacts of air pollutionshould be part of road ecology. Front. Ecol. Environ. 15 (8), 443–449.

Lima, S.L., O'Keefe, J.M., 2013. Do predators influence the behaviour of bats? Biol. Rev.88 (3), 626–644.

Lindenmayer, D.B., Nix, H.A., 1993. Ecological principles for the design of wildlife

corridors. Conserv. Biol. 7 (3), 627–630.McGarigal, K., Cushman, S.A., Neel, M.C., Ene, E., 2002. FRAGSTATS: Spatial Pattern

Analysis Program for Categorical Maps.McKinney, M.L., 2006. Urbanization as a major cause of biotic homogenization. Biol.

Conserv. 127 (3), 247–260.Navara, K.J., Nelson, R.J., 2007. The dark side of light at night: physiological, epide-

miological, and ecological consequences. J. Pineal Res. 43 (3), 215–224.Ortega, C.P., 2012. Effects of noise pollution on birds: a brief review of our knowledge.

Ornithol. Monogr. 74 (1), 6.Pintérné, E.N., Pödör, Z., 2017. The effect of artificial lights on nocturnal

Macrolepidoptera (Lepidoptera: Macroheterocera) communities. Acta Silvatica etLignaria Hungarica 13 (1), 41–54.

Plummer, K.E., Hale, J.D., O'Callaghan, M.J., Sadler, J.P., Siriwardena, G.M., 2016.Investigating the impact of streetlighting changes on garden moth communities.Journal of Urban Ecology 2 (1).

Pocock, Z., Lawrence, R.E., 2005. How Far Into a Forest Does the Effect of a Road Extend?Defining Road Edge Effect in Eucalypt Forests of South-eastern Australia. RoadEcology Center.

Rhodes, M.P., 2002. Assessment of sources of variance and patterns of overlap in mi-crochiropteran wing morphology in southeast Queensland, Australia. Can. J. Zool. 80(3), 450–460.

Rich, C., Longcore, T., 2013. Ecological Consequences of Artificial Night Lighting. IslandPress.

Rowse, E., Lewanzik, D., Stone, E., Harris, S., Jones, G., 2016. “Dark matters: the effects ofartificial lighting on bats.” Bats in the Anthropocene: Conservation of Bats in aChanging World. Springer, pp. 187–213.

Russo, D., Ancillotto, L., 2015. Sensitivity of bats to urbanization: a review. Mamm. Biol.80 (3), 205–212.

Russo, D., Cistrone, L., Libralato, N., Korine, C., Jones, G., Ancillotto, L., 2017. Adverseeffects of artificial illumination on bat drinking activity. Anim. Conserv. 20 (6),492–501.

Russo, D., Ancillotto, L., Cistrone, L., Libralato, N., Domer, A., Cohen, S., Korine, C., 2019.Effects of artificial illumination on drinking bats: a field test in forest and deserthabitats. Anim. Conserv. 22 (2), 124–133.

Rydell, J., 1992. Exploitation of insects around streetlamps by bats in Sweden. Funct.Ecol. 744–750.

Rydell, J., Racey, P., 1995. Street lamps and the feeding ecology of insectivorous bats. In:Symposia of the Zoological Society of London. The Society, London, pp. 1960–1999.

Saldaña-Vázquez, R.A., Munguía-Rosas, M.A., 2013. Lunar phobia in bats and its ecolo-gical correlates: a meta-analysis. Mamm. Biol. 78 (3), 216–219.

Sample, B.E., Cooper, R.J., Greer, R.D., Whitmore, R.C., 1993. Estimation of insect bio-mass by length and width. Am. Midl. Nat. 234–240.

Seto, K.C., Güneralp, B., Hutyra, L.R., 2012. Global forecasts of urban expansion to 2030and direct impacts on biodiversity and carbon pools. Proc. Natl. Acad. Sci. 109 (40),16083–16088.

Shortall, C.R., Moore, A., Smith, E., Hall, M.J., Woiwod, I.P., Harrington, R., 2009. Long-term changes in the abundance of flying insects. Insect Conservation and Diversity 2(4), 251–260.

Speakman, J., 1991. The impact of predation by birds on bat populations in the BritishIsles. Mammal Rev. 21 (3), 123–142.

Spoelstra, K., van Grunsven, R.H., Donners, M., Gienapp, P., Huigens, M.E., Slaterus, R.,Berendse, F., Visser, M.E., Veenendaal, E., 2015. Experimental illumination of naturalhabitat—an experimental set-up to assess the direct and indirect ecological con-sequences of artificial light of different spectral composition. Phil. Trans. R. Soc. B370 (1667), 20140129.

Spoelstra, K., van Grunsven, R.H., Ramakers, J.J., Ferguson, K.B., Raap, T., Donners, M.,Veenendaal, E.M., Visser, M.E., 2017. Response of bats to light with different spectra:light-shy and agile bat presence is affected by white and green, but not red light. Proc.R. Soc. B Biol. Sci. 284 (1855).

Stone, E.L., Jones, G., Harris, S., 2009. Streetlighting disturbs commuting bats. Curr. Biol.19 (13), 1123–1127.

Stone, E.L., Jones, G., Harris, S., 2012. Conserving energy at a cost to biodiversity?Impacts of LED lighting on bats. Glob. Chang. Biol. 18 (8), 2458–2465.

Stone, E.L., Harris, S., Jones, G., 2015a. Impacts of artificial lighting on bats: a review ofchallenges and solutions. Mamm. Biol. 80 (3), 213–219.

Stone, E.L., Wakefield, A., Harris, S., Jones, G., 2015b. The impacts of new streetlighttechnologies: experimentally testing the effects on bats of changing from low-pres-sure sodium to white metal halide. Philosophical Transactions of the Royal Society ofLondon B: Biological Sciences 370 (1667), 20140127.

Straka, T.M., Lentini, P.E., Lumsden, L.F., Wintle, B.A., Ree, R., 2016. Urban bat com-munities are affected by wetland size, quality, and pollution levels. Ecology andEvolution 6 (14), 4761–4774.

Threlfall, C., Law, B., Penman, T., Banks, P.B., 2011. Ecological processes in urbanlandscapes: mechanisms influencing the distribution and activity of insectivorousbats. Ecography 34 (5), 814–826.

Threlfall, C.G., Law, B., Banks, P.B., 2012a. Influence of landscape structure and humanmodifications on insect biomass and bat foraging activity in an urban landscape. PLoSOne 7 (6), 10.

Threlfall, C.G., Law, B., Banks, P.B., 2012b. Sensitivity of insectivorous bats to urbani-zation: implications for suburban conservation planning. Biol. Conserv. 146 (1),41–52.

Threlfall, C., Law, B., Banks, P., 2013. The urban matrix and artificial light restricts thenightly ranging behaviour of Gould's long-eared bat (Nyctophilus gouldi). AustralEcology 38 (8), 921–930.

Threlfall, C.G., Ossola, A., Hahs, A.K., Williams, N.S., Wilson, L., Livesley, S.J., 2016.Variation in vegetation structure and composition across urban green space types.


27

http://www.conbio.org/activities/meetings/2010/images/abstracts.pdf

http://www.conbio.org/activities/meetings/2010/images/abstracts.pdf









































































































































































Front. Ecol. Evol. 4, 66.Threlfall, C.G., Mata, L., Mackie, J.A., Hahs, A.K., Stork, N.E., Williams, N.S., Livesley,

S.J., 2017. Increasing biodiversity in urban green spaces through simple vegetationinterventions. J. Appl. Ecol. 54 (6), 1874–1883.

van Grunsven, R.H., Donners, M., Boekee, K., Tichelaar, I., Van Geffen, K., Groenendijk,D., Berendse, F., Veenendaal, E., 2014. Spectral composition of light sources andinsect phototaxis, with an evaluation of existing spectral response models. J. Insect

Conserv. 18 (2), 225–231.Voigt, C.C., Rehnig, K., Lindecke, O., Pētersons, G., 2018. Migratory bats are attracted by

red light but not by warm-white light: implications for the protection of nocturnalmigrants. Ecology and Evolution 8 (18), 9353–9361.

Yau, K.K., Kuk, A.Y., 2002. Robust estimation in generalized linear mixed models. Journalof the Royal Statistical Society: Series B (Statistical Methodology) 64 (1), 101–117.


28














light pollution at the urban forest edge negatively ... filecontents lists available at...

Documents