BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit publishers, academic institutions,research libraries, and research funders in the common goal of maximizing access to critical research.

Do Green Roofs Provide Habitat for Bats in Urban Areas?Author(s): Huma Pearce and Charlotte L. WaltersSource: Acta Chiropterologica, 14(2):469-478. 2012.Published By: Museum and Institute of Zoology, Polish Academy of SciencesDOI: http://dx.doi.org/10.3161/150811012X661774URL: http://www.bioone.org/doi/full/10.3161/150811012X661774

BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the biological, ecological,and environmental sciences. BioOne provides a sustainable online platform for over 170 journals and bookspublished by nonprofit societies, associations, museums, institutions, and presses.

Your use of this PDF, the BioOne Web site, and all posted and associated content indicates your acceptance ofBioOne’s Terms of Use, available at www.bioone.org/page/terms_of_use.

Usage of BioOne content is strictly limited to personal, educational, and non-commercial use. Commercialinquiries or rights and permissions requests should be directed to the individual publisher as copyright holder.

INTRODUCTION

As the world population continues to increase,with more than 50 percent of people now residingwithin cities, remaining natural and semi-naturalhabitats within urban areas play an increasingly im-portant role for people and wildlife (Mortberg andWallentinus, 2000; Benedict and McMahon, 2002;Jim, 2004; Town and Country Planning Association,2004; Bierwagen, 2007; Gill et al., 2007; Tratalos etal., 2007). Urban development has been shown to bea direct threat to biodiversity, by causing and contin-uing to cause habitat destruction and degrada-tion (McNeely and Mainka, 2005). Green roofs area technology that seeks to lessen the impacts of urbanisation on people and wildlife.

A green roof is the roof of a building that hasbeen deliberately (partially or completely) coveredwith vegetation and a growing medium. They can bebroadly categorised into ‘intensive’ and ‘extensive’type roofs according to their depth of substrate andmanagement requirements. Intensive roofs are the

equivalent of parks or gardens on roofs with treesand shrubs included within their design. They have a deep substrate (200–400 mm) and require investments in plant care and irrigation. Convers-ely, extensive roofs have shallower substrates(20–200 mm, but usually about 35–75 mm) and re-quire less maintenance. In general, they comprise‘sedum roofs’ that support low growing succulentplants, or ‘biodiverse roofs’ which are planted withand/or have become colonised by a variety of wild-flowers, herbs, sedums, mosses and grasses.

Green roofs provide ecosystem services such asa reduction in the urban heat island (Kumar andKaushik, 2005; Jim and Hongming, 2010), allevia-tion of the impact of flash floods following summerstorms (Mentens et al., 2006; Carter and Jackson,2007; Stovin, 2010), improved thermal and energyperformance of buildings (Okeil, 2010; Teemuskand Mander, 2010) and improved air quality (Clarket al., 2008), as well as urban green space that offersamenity and wildlife value (Cle mants et al., 2006;English Nature, 2006; Dunnett and Kingsbury, 2008;

Acta Chiropterologica, 14(2): 469–478, 2012PL ISSN 1508-1109 © Museum and Institute of Zoology PAS

doi: 10.3161/150811012X661774

Do green roofs provide habitat for bats in urban areas?

HUMA PEARCE1, 3 and CHARLOTTE L. WALTERS2

1Mostlybats, 59 High Street, Hunsdon, Hertfordshire SG12 8NJ, United Kingdom2Institute of Zoology, Zoological Society of London, Regent’s Park, London NW1 4RY, United Kingdom

3Corresponding author: E-mail: mostlybats@gmail.com

Green roofs, (roofs that are deliberately vegetated), are a technology that seeks to lessen the impacts of urbanisation on people andwildlife. This study investigates their value for UK bat species within the context of the built environment. Green roofs werecategorised as ‘sedum’ or ‘biodiverse’ according to their dominant vegetation type. Bat activity was monitored over 13 biodiverse,nine sedum and 17 conventional un-vegetated roofs located within the Greater London area for seven nights using Anabat SD1detectors. Influence of roof type and environmental variables on bat activity were evaluated using generalised linear models.Pipistrellus pipistrellus were most frequently recorded followed by Pipistrellus pygmaeus, Nyctalus/Eptesicus and Pipistrellusnathusii. The mean number of call sequences per night was 5.2. Feeding events accounted for 16% (217) of call sequences. Batactivity was significantly higher over biodiverse roofs compared to conventional roofs. A greater extent of suitable habitat withinthe surrounding area had a positive influence on bat activity but numbers of call sequences and feeding events were significantlyhigher over biodiverse roofs compared to conventional roofs when suitable habitat within a 100 metre radius of the roof was below33% cover. Other factors affecting bat activity included roof height (negative influence with increased height) and the month ofsurvey. No significant differences were found between sedum and conventional roofs. The findings suggest that biodiverse roofsoffer enhanced habitat for bats within the context of urbanised environments. Further studies are needed to predict more accuratelytheir value as a bat conservation measure.

Key words: green roofs, bat activity, Anabat, generalised linear modelling, urban environments

Gedge and Little, 2008; Surhone et al., 2010). Con -sequently, policy to encourage green roofs withindevelopments has been established for a number ofcities in the UK and Europe, for example The Lon -don Plan (Greater London Authority, 2008).

In recent years, researchers have turned their at-tention to the role that green roofs can play in con-serving biodiversity within towns and cities (e.g.,Grant et al., 2003; Clemants et al., 2006). A growingbody of evidence suggests that green roofs provideliving space for plants and animals including a num-ber of threatened or rare invertebrate (Jones, 2002;Brenneisen, 2005; Kadas, 2006; N. Gong, unpub-lished data) and bird species (Gedge and Kadas,2005; Baumann, 2006; Lee, 2009; H. Burgess, inlitt.). How ever, there is a complete lack of dataavailable on their value to mammals.

The aim of this study was to carry out a prelimi-nary investigation into the value of green roofs tomammals. Bats were chosen as the mammal groupfor investigation because they are mobile flying ani-mals that travel large distances and are thereforemore likely to encounter green roofs compared toother terrestrial mammals. They also occur in a var-i ety of habitats including cities (Mickleburgh, 1988;Briggs et al., 2007). All UK bats have suffered sig -nificant population declines over the last 60 yearsand remain vulnerable to pressures associated withlandscape change, agricultural intensification, de-vel opment and habitat fragmentation (Mickle burgh,1987; Guest et al., 2002). We hypothesised that sinceall UK bat species are insectivorous and that greenroofs can support high insect biomass (Kadas, 2006),they may offer a conservation measure for bats bymitigating against the effects of development andproviding alternative suitable commuting and/orforaging habitats within the built environment.

MATERIALS AND METHODS

Study Area

The study was conducted within the Greater London areabetween 21st May and 26th September 2010. Bat activity wasmonitored over a sample of green and conventional un-vegetat-ed roofs for a period of seven days. Generally, one convention-al roof and one green roof were surveyed at each site but, forsome locations, more than one green roof was present and tomaximise the quantity of data that could be collected over onebat activity season, for five sites, two green roofs and one con-ventional roof were surveyed. To control against variations insur rounding habitat and environmental conditions, and en-sure that the level of bat activity observed over each roof was independent, each roof at a given site was located no less than50 m and no more than 250 m from one other.

Only ‘extensive’ type green roofs were surveyed during thestudy. These were broadly categorised as sedum or biodiverseaccording to their construction and the dominant vegetationtype present. Conventional roofs were predominantly flat orshallow pitched roofs covered with bitumen felt or paving slabs.Sites were chosen according to their accessibility. A total of ninesedum, 13 biodiverse and 17 conventional roofs were surveyed.Roof sites comprised a variety of heights and areas, but everyeffort was made to ensure that these factors were comparablewithin samples. The area of each roof was estimated from aeri-al photographs using the polygon tool of the Google Earth Proprogram. The height of the roof was described according to thenumber of floors of the building surveyed.

Collection of Bat Data

Access to roof sites was not possible during the night be-cause of health and safety concerns. Consequently, roof siteswere surveyed using automated bat detectors, specificallyAnabat SD1 bat detectors (Titley Electronics, Ballina, NSW,Australia). The hardware of an Anabat consists of a frequencydivision detector, a real-time analysis and storage module, andsoftware (AnalookW) for the analysis of recordings. An outputfor the number of bat call sequences per species per night wasgenerated.

Anabat detectors were contained within a waterproof casetogether with a 12 V, 1.3 Ah battery which powered the device.A frequently used method of survey which allows bat calls to bedetected but prevents damage to the bat detector microphone(e.g. from the ingress of rainwater) was used (e.g., Collins andJones, 2009; Stone et al., 2009). This involved creating a circu-lar opening within the water-proof case to expose the micro-phone and covering this with a single layer of cling-film to pro-tect against rainwater. The bat detector was located as near tothe centre of each roof as possible with the microphone posi-tioned upwards.

The detection range of an Anabat SD1 is ca. 25 m in any di-rection. Although the range at which bat calls can be detectedvaries between species, this range was considered to fall withinthe average range for the species likely to be recorded, notablyPipistrellus spp., Nyctalus spp. and Eptesicus serotinus whichoccur in more open habitats than for example Myotis spp. andtypically echolocate at a greater amplitude thus making themmore detectable. The length or breadth of the roofs studied didnot exceed 50 m and therefore the range of one detector was suf-ficient to pick up any bats flying over. The sensitivity of the de-tectors was set at 5 to minimise the likelihood of detecting batsoutside the perimeter of the roof. The detectors were program -med to be operational no less than 30 min before sunset until 30 minutes after sunrise to coincide with the period of batactivity.

Collection of Environmental Data

EL-USB-2 Data Loggers (Lascar Electronics Ltd, Salisbury,UK) were left at each roof site to monitor temperature and rela-tive humidity levels at 30 minute intervals throughout each sev-en day survey period. The data collected was used to equate the mean night temperature and humidity recorded at each roofbetween the hours of sunset and sunrise i.e. the time periodwhen bats would be active. Information on rainfall was estimat-ed using archives on www.weatheronline.co.uk (accessedFebruary 2011) which provides a daily estimate of the quantityof precipitation per day (in mm). These data were collected to

470 H. Pearce and C. L. Walters

investigate ‘weather’ as a factor which may influence differ-ences in bat activity at different sites.

Collection of Habitat Data

To account for the influence of surrounding habitat on batactivity, an estimate of the percentage cover of vegetation andwater habitats within a 100 m radius of the centre of each roofwas calculated from aerial photographs using the Google EarthPro circle and polygon tools. Vegetation within the urban land-scape typically comprises grassland, shrubs, trees and ornamen-tal planting. Water habitats comprise canals and lakes. Thesehabitats typically support good insect biomass and were consid-ered to offer suitable bat habitat within the context of the ur-banised environment.

Identification of Bat Species and Bat Feeding Activity

Ten bat species are known to occur in the Greater Londonarea. These include Pipistrellus pipistrellus (common pip-istrelle), P. pygmaeus (soprano pipistrelle), P. nathusii (Nathu -sius’ pipistrelle), Plecotus auritus (brown long-eared bat), My -otis nattereri (Natterer’s bat), M. daubentonii (Daubenton’sbat), M. brandtii (Brandt’s bat), Nyctalus noctula (noctule), N. leisleri (Leisler’s bat), and Eptesicus serotinus (serotine)(www.londonbats.org.uk — accessed March 2010). The suc-cessful identification of bat species from their echolocation callsis dependent on a number of factors such as acoustic clutter,recording quality, recording methodology and call characteris-tics (Parsons and Jones, 2000). Pipistrellus spp. can usually besuccessfully separated by their characteristic frequency.Conversely, the typical call parameters of N. noctula, N. leisleriand E. serotinus often overlap, making classification to specieslevel difficult (Vaughan et al., 1997). For these reasons, in this study Pipistrellus species were identified to species levelbut N. noctula, N. leisleri and E. serotinus were grouped.

Data Analysis

Only data collected from green and conventional type roofson the same nights at a given site were used in the statisticalanalysis. All statistical analysis was completed using the R-statistics programming language (R Development Core Team,2011). Generalised linear models were used to analyse the influ-ence of roof and environmental variables on bat activity.

Bat activity included all bat call sequences recorded overroof sites. Evidence of bat feeding events was assumed in twocircumstances: Type 1 when ‘feeding buzzes’ were positivelyidentified within a call sequence (Fig. 1A) and Type 2 when several call sequences by the same individual were recordedconsecutively (Fig. 1B) as opposed to a single call sequence(Fig. 1C), i.e. when the time between consecutive same speciescalls was less than one minute — this was assumed to be thesame individual bat.

Factors related to the roof that were included within theanalysis were roof type (biodiverse, sedum or conventional),roof area and roof height. Environmental variables includedtemperature, humidity, rainfall, amount of suitable bat habitatwithin the immediate surrounding area (i.e. the percentage cov-er of vegetation and water habitats within a 100 m radius of theroof) and the month of survey.

A stepwise procedure was used to select the explanatoryvariables to include in the model and to fit the functions relating

each of these variables (additive or dependant) to the proba-bility of recording bat call sequences and bat feeding events.Variable selection procedures stringently guarded against the inclusion of extraneous variables in a model and the removal of significant variables by ANOVA testing each stage of the se-lection process.

RESULTS

Pipistrellus pipistrellus, P. pygmaeus, P. nathusii,N. noctula, N. leisleri and probable E. serotinuswere recorded during the study. The mean numberof bat call sequences recorded per night over roofs(conventional, biodiverse, sedum) was 5.2. Pipi -strel lus pipistrellus were the most frequently record-ed species followed by P. pygmaeus and Nyctalus/Eptesicus spp. (Fig. 2). A high level of variation inthe number of bat call sequences was observed with-in and between sites and bat activity is likely to beinfluenced by factors related to the roof as well asenvironmental factors.

Bat Activity

Bat activity was significantly higher over biodi -verse roofs compared to conventional roofs (t = 3.25,n = 39, P < 0.01). No significant difference wasfound between the number of bat call sequencesrecorded over sedum and conventional roofs. In -creasing roof height had a negative effect on bat activity (t = -3.78, n = 39, P < 0.001 — Table 1) andthere was some variation in the species recorded atdifferent roof heights (Fig. 3). Nyctalus/Eptesicusspp. accounted for less than 10% of all bat call se-quences recorded over lower buildings (≤ 3 storeysin height) but this increased to 32% over four storeybuildings. The contribution of Pipistrel lus speciescall sequences however increased again to 98% forthe sample of 5-storey buildings surveyed. An unex-pectedly high number of Pipistrellus spp. call se-quences were recorded at one 5-storey site whichwas located adjacent to a canal. If data from this onesite is excluded as an outlier, the pattern of a greatercontribution of activity by Nyctalus/Eptesicus atheight continues and they account for 20% of the total number of call sequences recorded (Fig. 3 —bar 5*).

The number of bat call sequences recorded in-creased with suitable bat habitat within the immedi-ate surrounding area (t = 3.10, n = 39, P < 0.01). Aninteraction was also identified between surroundinghabitat and roof type t = -2.65, n = 39, P < 0.05)(Table 1). To explain this interaction, the model wasre-run using habitat classes that described the extent

Bats and green roofs 471

472 H. Pearce and C. L. Walters

FIG

. 1. S

pect

rogr

ams

of b

at c

alls

att

ribu

ted

to t

hree

typ

es o

f ba

t ac

tivi

ty, s

how

ing

A —

Fee

ding

buz

z (F

eedi

ng e

vent

— T

ype

1), B

— C

onse

cuti

ve b

at c

all

sequ

ence

s by

the

sam

e sp

ecie

s,

assu

med

to

be t

he s

ame

indi

vidu

al,

repr

esen

ting

an

assu

med

fee

ding

eve

nt (

Fee

ding

eve

nt —

Typ

e 2)

and

C —

Sin

gle

call

seq

uenc

e w

itho

ut a

fee

ding

buz

z (n

on-f

eedi

ng e

vent

)

Frequency (kHz)

Tim

e (

s)

Tim

e (

s)

Tim

e (

s)

2.2

2.3

2.4

2.5

2.6

2.7

2.8

0.0

1.0

2.0

3.0

4.0

5.0

0.0

1.0

2.0

3.0

4.0

5.0

80

70

60

50

40

35

30

80

70

60

50

40

35

30

80

70

60

50

40

35

30

A B C

Bat species

P. pipistrellus P. pygmaeus P. nathusii Nyctalus/Eptesicus spp.

Mean n

um

ber

of

bat

call

sequences p

er

nig

ht

Biodiverse

Sedum

Conventional

of suitable bat habitat within a 100 m radius of the roof; ‘low’ (<33% cover), ‘medium’ (33% to66% cover) and ‘high’ (>66% cover). Bat activitywas significantly higher over biodiverse roofs compared to conventional roofs when the extent ofsuitable bat habitat in the surrounding area was‘low’ (t = 2.63, n = 39, P < 0.05 — see Fig. 4). The values for ‘high’ habitat extent are based on one data set from a conventional and biodiverse roof and therefore these data could not be analysedstatistically.

Bat activity was affected by the month of surveyand a higher number of call sequences were record-ed in May. The lowest number of bat counts was reported in August. Temperature, humidity, rainfall

and roof area were not found to influence bat activ-ity and were removed from the model.

Bat Foraging Activity

Ninety-four feeding events were recorded duringthe survey. Call sequences that were attributed to batfeeding events accounted for 16% (217) of the totalnumber of call sequences recorded. Pipistrellus pipi -strellus contributed 89.4% of all feeding events. Pi pi strellus pygmaeus and N. noctula accounted for 7.4% and 3.2%, respectively. The mean numberof feeding events recorded per night was greaterover biodiverse roofs (0.56 feeding events/night)compared to conventional (0.38) and sedum type

Bats and green roofs 473

FIG. 2. Mean number of bat call sequences recorded per night over biodiverse, sedum and conventional un-vegetated roofs (error bars show SD of the mean)

TABLE 1. Summary of results from generalised linear modelling of habitat and environmental associations for the number of bat callsequences recorded over sedum, biodiverse and conventional un-vegetated type roofs. [glm (formula = encounter ~ factor(roof) * habitat + roof height + month, family = quasipoisson)]

Bat encounter rates Estimate SE t-value P-level

Biodiverse roof 2.737 0.830 3.249 **Sedum roof 1.399 1.674 0.836 nsHabitat 0.049 0.016 3.100 **Height -0.703 0.186 -3.777 ***Month July -0.466 0.476 -0.978 nsMonth June 0.078 0.528 0.147 nsMonth May 1.591 0.524 3.038 **Month September -1.314 0.674 -1.950 nsBiodiverse roof:habitat -0.042 0.016 -2.649 *Sedum roof:habitat -0.055 0.042 -1.289 ns

* — P < 0.05, ** — P < 0.01, *** — P < 0.001, ns — not significant; ANOVA F-test: P < 0.05

14

12

10

8

6

4

2

0

roofs (0.03). Furthermore, all three species wererecorded feeding over biodiverse roofs whereas,with the exception of one record of a N. noctula, allother feeding events over conventional and sedumroofs were for P. pipistrellus.

The number of bat feeding events was signifi-cantly higher over biodiverse roofs compared toconventional roofs (t = 5.60, n = 39, P < 0.001).Roof height had a negative influence on bat feed-ing activity (Table 2) and with the exception of one biodiverse roof no feeding events were report-ed over buildings where roof height exceeded 2-storeys.

474 H. Pearce and C. L. Walters

FIG. 3. Influence of roof height on Pipistrellus spp. and Nyctalus/Eptesicus spp. activity. (* represents data with theoutlier removed. Error bars show SD of the mean. Light error bars refer to Pipistrellus spp.; dark error bars refer to

Nyctalus/Eptesicus spp.)

Bat feeding events Estimate SE t-value P-level

Biodiverse roof 8.137 1.453 5.600 ***Sedum roof -185.3 0.000 -0.005 nsHabitat 0.122 0.021 5.758 ***Height -1.245 0.274 -4.569 ***Month July -1.783 0.661 -2.696 *Month June -0.432 0.587 -0.737 nsMonth May 3.831 0.710 5.396 ***Month September -2.426 0.777 -3.121 **Biodiverse roof:habitat -0.127 0.244 -5.192 ***Sedum roof:habitat 3.978 895.6 0.004 ns

* — P < 0.05, ** — P < 0.01, *** — P < 0.001, ns — not significant; ANOVA F-test P < 0.05

TABLE 2. Summary of results from generalised linear modelling of habitat and environmental associations for bat foraging activityrecorded over sedum, biodiverse and conventional un-vegetated type roofs. [glm (formula = feed ~ factor(roof) * habitat + height + month, family = quasipoisson)]

Roof height (number of storeys)

1 2 3 4 5 5*

Mean n

um

ber

of

call

sequences p

er

nig

ht

20

16

12

8

4

0

Nyctalus/Eptesicus spp.

Pipistrellus spp.

The extent of suitable habitat within the sur-rounding area had a positive influence on the proba-bility of recording bat feeding activity (t = 5.76, n = 39, P < 0.001). An interaction was also identi-fied between roof type and surrounding habitat (t = -4.57, n = 39, P < 0.001 — Table 2). By plottingthe mean number of bat feeding events per nightagainst roof type and habitat extent (Fig. 5), it wasagain found that more feeding records were obtainedover biodiverse roofs compared to conventionalroofs when the extent of suitable habitat in the sur-rounding area was ‘low’ (< 33% cover).

Bat feeding activity was affected by the month ofsurvey. Feeding activity was recorded during allmonths, but a higher number of feeding events were recorded in May (t = 5.40, P < 0.001) and few-er were reported in July (t = -2.70, P < 0.05) andSeptember (t = -3.12, P < 0.01; in all cases n = 39 —Table 2). Temp e r ature, humidity, rainfall and thearea of the roof were not found to influence feedingactivity and were removed from the model.

DISCUSSION

This study provides some initial evidence in sup-port of the potential for green roofs to provide habi-tat for bats within the context of an urbanised envi-ronment. By modeling the number of call sequencesand feeding events against a variety of environmentalvariables, the results suggest that in urbanised areas,biodiverse roofs generally offered enhanced habitatfor bats when compared to conventional and sedumroofs.

Bat Commuting and Foraging Activity

Overall, the number of call sequences recordedover roof sites was low (mean of 5.2 call sequences

Extent of habitat in the surrounding area

Mean n

um

ber

of

bat

call

sequences p

er

nig

ht

Low Medium High

25

20

15

10

5

0

Biodiverse

Sedum

Conventional

per night) and the majority comprised only singlecall sequences that did not contain a feeding buzzesand were not followed by further activity. Althoughnot conclusive without direct observation of the bats, because of the sporadic nature of these re-cords where by the time between call sequences wasoften hours or even nights and the shortness of thecall se quences obtained (see Fig. 1C), these re-cords are considered likely to convey bat commutingbehaviour.

Pipistrellus pipistrellus was the most frequentlyrecorded species over all roof types and since this isthe most abundant bat species in the London(Mickleburgh, 1987; Briggs et al., 2007), this activ-ity was not unexpected. Records for Nyctalus/Epte -sicus spp., which accounted for 7.6% of the total callsequences, are more notable since N. noctula and E. serotinus populations have declined in Londonover the last 20 years and N. leisleri are uncommon(Briggs et al., 2007). Overall, significantly highernumbers of call sequences were recorded over bio-diverse roofs compared to conventional roofs and we propose that bats may be preferentially in-cluding these vegetated roofs within their flightroutes in order to opportunistically exploit potentialfood resources and/or because these vegetated roofs offer a more navigable feature within the urbanlandscape.

Investigations into the invertebrate fauna ongreen roofs within the Greater London area havebeen undertaken by Jones (2002) and Kadas (2006).Several invertebrates discovered are also bat prey

species, e.g. Coleoptera (beetles), Aranaea (spiders),Hemiptera (bugs), Hymenoptera (bees, wasps, antsetc), Diptera (flies), Lepidoptera (butterflies andmoths) and Orthoptera (grasshoppers) (Vaughan,1997). Pipistrellus spp., whose diet comprises most-ly Diptera, were most frequently recorded for agingover roof sites. Although Jones (2002) dis cover edDiptera only infrequently and considered these to bemerely vagrants, Pipistrellus spp. are generalistfeeders (Dietz et al., 2009) and may be exploitingother small flying insects over green roofs. Con -verse ly, a range of invertebrate species that occur inthe diet of Nyctalus/Eptesicus spp. were found to occur on green roofs. Although only a low numberof feeding events by Ny cta lus/Epte sicus bats werenoted, these species are relatively uncommon inLon don (Briggs et al., 2007) and therefore theserecords are considered noteworthy.

Overall more feeding events by a greater diver-sity of species were recorded over biodiverse roofscompared to conventional and sedum roofs. We pro-pose that although some species of flying insectsmay occur indiscriminately at roof height and not directly benefit from the presence of roof vegeta-tion, others are likely to reside within and emergefrom the vegetation at night. The plant species and structural diversity of the vegetation associat-ed with biodiverse roofs may offer a wider variety of habitats for a greater range of bat prey spe-cies to live in compared to conventional and se-dum roofs and thus offer more feeding opportunities for bats.

Bats and green roofs 475

FIG. 4. Influence of roof type and the extent of suitable bat habitat in the surrounding area on bat activity (error bars show SD of the mean)

Influence of Roof Height

Roof height had a negative influence on bat activity. These findings are consistent with ob-servations made by Collins and Jones (2009) whorecorded fewer bat contacts at 30 m height than at ground level. They also observed differences inspecies encounter rates such that at ground level,90% of bat calls were from Pipistrellus spp. and6.9% were Nyctalus/Eptesicus spp., meanwhile at 30 m height, 65.1% of bat records were Pipi -strellus spp. and 34.1% were from Nyctalus/Eptesicus. Similar results were found during thisstudy such that Nyctalus/Eptesicus spp. only ac-counted for approximately 8% of bat activity over 1- and 2-storey buildings (i.e. 3.5 to 7 m high)but their occurrence increased to 32% and 20% over four (i.e., ca. 14 m) and five (i.e., ca. 17.5 m) storey buildings, respectively (when an outlier wasremoved).

Based on these findings, we propose that the establishment of green roofs on high rise build-ings may be of most benefit to Nyctalus/Eptesicusspp. whereas green roofs constructed on single and two storey buildings are likely to be of greaterhabitat value to Pipistrellus spp. These findingsshould be given due consideration when designingthe planting schemes for green roofs so that they include plant species that offer habitat for the typesof invertebrates that the bat species that are mostlikely to use them can feed on.

476 H. Pearce and C. L. Walters

FIG. 5. Influence of roof type and the extent of suitable bathabitat in the surrounding habitat on bat feeding events per

night (error bars show SD of the mean)

Extent of habitat in the surrounding area

Mean n

um

ber

of

feedin

g e

vents

per

nig

ht

Low Medium High

3

2

1

0

Biodiverse

Sedum

Conventional

Influence of Roof Area

Roof area was not found to influence the numberof call sequences recorded. This is an unexpectedfinding since larger vegetated roof sites are likely to be more prominent features within the urban landscape and they should support greater insectbiomass and offer more bat foraging opportunitiesthan small roofs. Most of the roofs surveyed in thisstudy were small sites with an area of 10 to 400 m²(85% of sites) and therefore any correlations be-tween roof size and bat activity cannot be accurate-ly determined. Further surveys, that encompass awider range of roof sizes, need to be completed be-fore any conclusions can be drawn.

Influence of Surrounding Habitat

The presence of favourable habitat in the surrounding area had a positive influence on bat activity. This is expected since the presence of veg-etation and or water habitats within close proximityto roof sites will enhance overall insect biomasswithin the area of the roof and thus the availabilityof prey. Furthermore linear vegetated features suchas tree lines and canals will provide greater connec-tivity to the roof sites. The interaction between rooftype and habitat within the immediate surroundingarea is considered more notable since significantlymore call sequences and feeding events were record-ed over biodiverse roofs compared to conventionalroofs when the amount of suitable habitat within a 100 m radius of the roof site was less than 33%cover. These findings suggest that biodiverse roofsmay provide enhanced habitat for bats within other-wise low quality urban environments.

Influence of Survey Period and Weather Conditions

Bat activity was influenced by the month of sur-vey. Higher numbers of call sequences were report-ed in May and these observations are likely to becorrelated with increased feeding activity by preg-nant females. Conversely, August is typically a monthwhen bat activity is greatest because both adults andpups are active (Walsh and Harris, 1996), yet thenumber of call sequences recorded was lowest dur-ing this month. Although rainfall was not found to significantly influence bat activity, rainfall washigh est in August and is likely to be contributingfactors to the lower levels of bat activity recordeddur ing this month. Bat activity is typically lower un -der rainy conditions since fewer insects will emergeand be available for bat foraging (Williams, 1961).

Contrary to other studies (e.g., Walsh and Harris,1996; Vaughan et al., 1997; Parsons et al., 2003,Scanlon and Petit, 2008), temperature was not foundto influence bat activity. This is considered to be be cause temperatures remained above 7°C and there was little variation in overall night tempera-tures (me dian and mean temperatures were 15°C)through out the study period. These stable night timetemperatures are likely to be attributed to the urbanheat island effect.

In conclusion, the findings of this study suggestthat buildings that have had biodiverse roofs in-stalled offer enhanced habitat for bats within thecontext of urbanised environments. Conversely, batactivity over sedum roofs was not found to be signif-icantly different from that observed over conven-tional un-vegetated roofs and the overall value ofthese installations, at least in terms of enhancing ur-ban sites for bat species, may therefore be broughtinto question. We propose that the installation ofbiodiverse roofs are likely to offer a more suitablehabitat enhancement measure for inclusion withinnew and existing developments and that a net-work of these roof types should benefit urban batpopulations in the long-term by providing enhancedcommuting habitat between their roost sites and key foraging areas as well as increased feeding opportunities.

Further studies are required to fully understandthe potential benefit of green roofs for bats sincethese findings are based on only a small sample size.Any future studies should encompass a larger sam-ple of roofs that more comprehensively represent a range of roof types, areas and heights. Such inves-tigations would also benefit from the inclusion ofnight time flying insect abundance and biomassstudies to the establish feeding resources availableto bats over different roof types and vegetation surveys to better understand which plant speciescompositions provide enhanced habitat for thesespe cies. The use of alternative bat detecting equip-ment may also allow more accurate identification of bats to species level and enable a more thoroughassessment.

ACKNOWLEDGEMENTS

Many thanks to the London Bat Group for providing fund-ing towards the project, Nick Bonsall of Access Ecology forloaning an Anabat SD1 detector at a reduced price, and all thosenamed below for providing access to the roof sites: GustavoMontes de Oca (Hackney City Farm); Rebecca Scott (WalworthGarden Farm); Jamie Turner (London Zoo); Janice Breslin(Cathedral Primary School); Agnes Knoll (Redcross Gardens);

Stuart Robertson, Paul Richens (Wolff Olins); Charlie Green,Sarah Law (The Office Group), Jane Riddiford (Global Gener -ation); Gary (Maria Fidelis Convent School); Clair Goody, Lyn -da Price (Eversheds); Oliver Burke, Bevan Jones (IslingtonCoun cil); Suzanne Murray (Paradise Park); Liz (Freightliner’sFarm); Ray Edwards (TFL), Daphne and Roland (Oval); Gor -don Lucas (Horniman Museum); Jon Broome, Cheryl Coyne(Shaw’s Cottages); Ed Gooch (St. Martin’s School); JoanneSmith and Tony (Green wich Ecology Park); Ken Greenway(So anes Centre) and Richard Bullock (London Wetland’s Cen -tre). Finally we would like to thank Gary Grant (Ecochemes),Prof. Alan Gange (Royal Holloway University of Lon don) andProf. Gareth Jones (University of Bristol) for comments on earlier versions of the manuscript and their continued supportfor the study.

LITERATURE CITED

BAUMANN, N. 2006. Ground nesting birds on green roofs inSwitzerland: preliminary observations. Urban Habitats, 4:37–50.

BENEDICT, M. A., and E. MCMAHON. 2002. Green infrastructure:Smart conservation for the 21st Century. Renewable Re -sources Journal, 20: 12–17.

BIERWAGEN, B. G. 2007. Connectivity in urbanising landscapes:the importance of habitat configuration, urban area size anddispersal. Urban Ecosystems, 10: 29–42.

BRENNEISEN, S. (ed.) 2005. The natural roof (NADA). Re searchProject Report of the use of extensive green roofs by wildbees. University of Wädenswil, Wädenswil, Switzerland, 23 pp.

BRIGGS, P. A., R. J. BULLOCK, and J. D. TOVEY. 2007. Ten yearsof bat monitoring at the WWT London Wetland Centre — a comparison with the National Bat Monitoring ProgrammeTrends for Greater London. The London Naturalist, 86:47–69.

CARTER, T., and C. R. JACKSON. 2007. Vegetated roofs for storm -water management at multiple special scales. Land scapeand Urban Planning, 80: 84–94.

CLARK, C., P. ADRIAENS, and B. TALBOT. 2008. Green roof eval-uation: a probabilistic economic analysis of environ mentalbenefits. Environmental Science and Technology, 42:2155–2161.

CLEMANTS, S. E., J. MARINELL, G. MOORE, and E. PETERS. 2006.Green roofs and biodiversity. Urban Habitats, 4: 1–101.

COLLINS, J., and G. JONES. 2009. Differences in bat activity in relation to bat detector height: implications for bat surveysat proposed wind farm sites. Acta Chiropterologica, 11:343–350.

DIETZ, C., D. NILL, and O. VON HELVERSEN. 2009. Bats ofBritain, Europe and North-west Africa. A & C Black, Pub -lishers Ltd, London, UK, 400 pp.

DUNNETT, N., and N. KINGSBURY. 2008. Planting green roofsand living walls. Timber Press, Cambridge, UK, 339 pp.

ENGLISH NATURE. 2006. Living roofs. English Nature, Peter -borough, UK, 27 pp.

GEDGE, D., and G. KADAS. 2005. Green roofs and biodiversity.Biologist, 52: 161–169.

GEDGE, D., and J. LITTLE. 2008. The DIY guide to green and living roofs. E-book, 49 pp. Available at http://living roofs.org/about-livingroofs.org-living-roofs/greenroof-diy-guide.html.

GILL, S. E., J. F. HANDLEY, A. R. ENNOS, and S. PAULIET. 2007.

Bats and green roofs 477

478 H. Pearce and C. L. Walters

Adapting cities for climate change — the role of the greeninfrastructure. Built Environment, 33: 115–133.

GRANT, G., L. ENGLEBACK, and B. NICHOLSON. 2003. Greenroofs: their potential for conserving biodiversity in urban ar-eas. English Nature Research Report No. 498, 508 pp.

GREATER LONDON AUTHORITY. 2008. The London Plan — spa -tial development strategy for Greater London. Consolidat edwith alterations since 2004. Greater London Authority, Lon -don, UK, 508 pp.

GUEST, P., K. E. JONES, and J. TOVEY. 2002. Bats in GreaterLondon. Unique evidence of a decline over 15 years. BritishWildlife, 14: 1–5.

JIM, C. Y. 2004. Green-space preservation and allocation for sus -tainable greening of compact cities. Cities, 21: 311–320.

JIM, C. Y., and H. HONGMING. 2010. Coupling heat flux dynam-ics with meteorological conditions in the green roof ecosys-tem. Ecological Engineering, 36: 1052–1063.

JONES, R. A. 2002. Tecticolous invertebrates: a preliminary in-vestigation of the invertebrate fauna on green roofs in urbanLondon. English Nature, Peterborough, UK, 36 pp.

KADAS, G. 2006. Rare invertebrates colonising green roofs inLondon. Urban Habitats, 4: 66–86.

KUMAR, R., and S. C. KAUSHIK. 2005. Performance evaluationof green roof and shading for thermal protection of build-ings. Building and Environment, 40: 1505–1511.

LEE, M. 2009. Case study: installing green roofs. Ecologist,29th January 2009. http://www.theecologist.org/how_to_make_a_difference/wildlife/360290/case_study_installing_green_roofs.html.

MENTENS, J., D. RAES, and M. HERMY. 2006. Green roofs as a tool for resolving the rainwater runoff problem in the urbanised 21st Century? Landscape and Urban Planning,77: 217–226.

MCNEELY, J. A., and S. A. MAINKA. 2005. Conservation for a New Era. IUCN, Gland, Switzerland, 220 pp.

MICKLEBURGH, S. 1987. Distribution and status of bats in theLondon area. The London Naturalist, 66: 41–91.

MICKLEBURGH, S. 1988. Bat records from the London area dur-ing 1987. The London Naturalist, 67: 161–170.

MORTBERG, U., and H. WALLENTINUS. 2000. Red-listed forest birdspe cies in an urban environment — assessment of green spacecorridors. Landscape and Urban Planning, 50: 215–226.

OKEIL, A. 2010. A holistic approach to energy efficient buildingforms. Energy and Buildings, 42: 1437–1444.

PARSONS, K. N., J. JONES, and F. GREENWAY. 2003. Swarming

activity of temperate zone microchiropteran bats: effects ofseason, time of night and weather conditions. Journal ofZool ogy (London), 261: 257–264.

PARSONS, S., and G. JONES. 2000. Acoustic identification oftwelve species of echolocating bat by discriminate functionanalysis and artificial neural networks. Journal of Ex peri -mental Biology, 203: 2641–2656.

R DEVELOPMENT CORE TEAM. 2011 R: a language and environ-ment for statistical computing. R foundation for StatisticalComputing, Vienna, Austria. http://www.r-project.org.

SCANLON, A. T., and S. PETIT. 2008. Effects of site, time, weath-er and light on urban bat activity and richness: considera-tions for survey effort. Wildlife Research, 35: 821–834.

STONE, E. L., G. JONES, and S. HARRIS. 2009. Street lighting dis-turbs commuting bats. Current Biology, 19: 1123–1127.

STOVIN, V. 2010. The potential for green roofs to manage urbanstormwater. Water and Environment Journal, 24: 192–199.

SURHONE, L. M., M. T. TIMPLEDON, and S. MARSEKEN. 2010.Sus tainable landscape architecture: sustainable design, sus-tainable urban drainage systems, fauna, flora, green roof,roof garden, context theory. Betascript Publishing, Mauri -tius, 73 pp.

TEEMUSK, A., and Ü. MANDER. 2010. Temperature regime ofplanted roofs compared with conventional roofing systems.Ecological Engineering, 36: 91–95.

TOWN AND COUNTRY PLANNING ASSOCIATION. 2004. Biodiver si -ty by design. A guide for sustainable communities. Townand Country Planning Association, London, UK, 35 pp.

TRATALOS, J., R. A. FULLER, P. H. WARREN, R. G. DAVIES, andK. J. GASTON. 2007. Urban form, biodiversity potential andecosystems services. Landscape and Urban Planning, 83:308–317.

VAUGHAN, N. 1997. The diet of British bats (Chiroptera). Mam -mal Review, 27: 77–94.

VAUGHAN, N., G. JONES, and S. HARRIS. 1997. Habitat use bybats (Chiroptera) assessed by means of a broad-band acoust -ic method. Journal of Applied Ecology, 34: 716–730.

WALSH, A. L., and S. HARRIS. 1996. Factors determining theabundance of the abundance of vespertilionid bats in Brit -ain: geographical, land class and local habitat relationships.Journal of Applied Ecology, 33: 519–529.

WILLIAMS, C. B. 1961. Studies in the effect of weather condi-tions on the activity and abundance of insect populations.Philosophical Transactions of the Royal Society of London,244B: 331–378.

Received 04 February 2012, accepted 14 September 2012