Effects of Artificial Lighting on Insectivorous Bat Communities in Urban Ecosystems Joanna Kate Haddock

Home , Myotis macropus, Nyctophilus, Ozimops, Ozimops ridei

Effects of artificial lighting on insectivorous bat communities in urban ecosystems

Joanna Kate Haddock

December 2018

A thesis submitted in fulfilment of the requirements for the degree of Doctor of Philosophy in the School of Life and Environmental Sciences, Faculty of Science, at The University of Sydney.

Photo credits: Clockwise from left Taken by M. Law - J. Haddock with blacklight intercept trap Taken by J. Haddock - Scoteanax rueppellii Taken by L. Gonsalves - Nyctophilus gouldi Taken by B. Sloggett - J. Haddock with Scoteanax rueppellii

Declaration

I hereby declare that this work is my own, except where otherwise acknowledged. It has not been submitted in any form for another degree or diploma at any university or other institution. I consent to this thesis being made available for photocopying and loan under the appropriate Australian copyright laws.

Role of the funding source This research was financially funded by the Holsworth Wildlife Research Endowment (grant numbers: HOLSW2015-2-F147 and HWRE2016R2063CONT).

Acknowledgements CGT is supported by the Australian Government’s National Environmental Science Program through the Clean Air and Urban Landscapes Hub.

Joanna Kate Haddock

April 2018

i Table of Contents

General Introduction ...... 1 Light pollution ...... 1 Historical developments in the field of light pollution research ...... 2 Ecological consequences of light pollution ...... 5 Mechanisms of ecological light sensitivity ...... 6 Urban ecological light pollution ...... 7 Urban bats and light pollution ...... 8 Open space adapted bats and ALAN ...... 12 Clutter adapted bat species and ALAN ...... 13 Trawling bats and ALAN ...... 14 Edge adapted species and ALAN...... 15 Aims and chapter structure ...... 15 References ...... 18 Chapter 2: Responses of insectivorous bats and nocturnal insects to local changes in street light technology ...... 28 Abstract ...... 28 Introduction ...... 29 Methods...... 35 Results ...... 44 Discussion ...... 49 References ...... 53 Chapter 3: Light pollution at the urban forest edge and its impact on insectivorous bats 60 Abstract ...... 60 Introduction ...... 61 Method ...... 65 Results ...... 73 Discussion ...... 82 References ...... 89 Appendix ...... 97 Chapter 4: Short term effect of lights on bat activity and feeding: a case study on Myotis macropus ...... 100 Abstract ...... 100 Introduction ...... 101 Methods...... 104 Results ...... 111

ii Discussion ...... 118 References ...... 124 Chapter 5: Experimentally introduced red and white lights at wetlands cause diverse effects on insectivorous bats...... 129 Abstract ...... 129 Introduction ...... 130 Methods...... 133 Results ...... 141 Discussion ...... 152 References ...... 158 Chapter 6: General Discussion ...... 165 Overview of findings ...... 165 Emerging themes...... 169 Management implications ...... 172 Future research ...... 174 Concluding remarks ...... 175 References ...... 176

iii Author attributions

Chapter 2 of this thesis is accepted for publication as “Haddock, J. K., Threlfall, C. G., Law, B. S. and Hochuli, D. F. Responses of insectivorous bats and nocturnal insects to local changes in street light technology. Austral Ecology. Accepted author manuscript: 27 November 2018.” J. K. Haddock, C. G. Threlfall and D.F. Hochuli conceived the project. J. K. Haddock collected the data. All authors analysed the data and contributed to manuscript preparation.

Chapter 3 of this thesis is accepted for publication as “Haddock, J. K., Threlfall, C. G., Law, B. S. and Hochuli, D. F. (2019) Light pollution at the urban forest edge negatively impacts insectivorous bats. Biological Conservation 236, 17-28” All authors conceived the project. J. K. Haddock collected the data. All authors analysed the data and contributed to manuscript preparation.

Chapter 4 of this thesis has been prepared for submission as “Haddock, J. K., Threlfall, C. G., Gonsalves, L., Law, B. S. and Hochuli, D. F. Short term effect of lights on bat activity and feeding: a case study on Myotis macropus.” All authors conceived the project. J. K. Haddock collected the data. All authors analysed the data and contributed to manuscript preparation.

Chapter 5 of this thesis has been prepared for submission as “Haddock, J. K., Threlfall, C. G., Gonsalves, L., Law, B. S. and Hochuli, D. F. Experimentally introduced red and white lights at wetlands cause diverse effects for insectivorous bats.” All authors conceived the project. J. K. Haddock collected the data. All authors analysed the data and contributed to manuscript preparation.

Associate Professor Dr Brad Law Dr Caragh Threlfall Dr Leroy Gonsalves Dieter Hochuli

iv Acknowledgements

There are so many people who have contributed to this thesis in different ways, and have made the last four years some of the most enjoyable years of my life. First and foremost, I would like to thank you my supervisors Dieter Hochuli, Caragh Threlfall and Brad Law. Dieter gave me this amazing opportunity, and his knowledge, endless patience and encouragement has been one of the biggest contributing factors to the completion of this thesis - I am deeply thankful. Caragh inspired my love of bats in the first place, trapping and tracking at Cumberland State Forest with Caragh for many years before my PhD began, and she has inspired and challenged me ever since. Caragh is my role model in so many ways, and I cannot thank her enough for all the support and feedback over the last four years. Brad has taught me so much about research, and I am very grateful to him for giving me support, advice and feedback – and all his amazing bat knowledge – over the course of my PhD.

Thank you also to my contributors Dr Leroy Gonsalves (Department of Primary Industries), Dr Alison Towerton (Greater Sydney Local Land Services), Dr Kyle Armstrong (Specialised Zoology), Dr John Martin (Royal Botanical Gardens, Sydney), and Professor Fritz Geiser, Dr Clare Stawski and Dr Anna Doty (The University of New England). Your time, equipment, knowledge and passion for bats inspired me, and made me feel very welcome in the bat community! I would also like to thank Ben Sloggett for his hours of field work assistance – nocturnal field work is tough, and it’s so much more fun with lovely people.

Thank you to all members of the Integrative Ecology Lab, past and present – our weekly lab meetings allowed me to find my feet in research, and gain confidence discussing papers, practising conferences talks, and drinking way too much coffee (and beer). I am especially grateful to Jayne Hanford, Manuel Lequerica, Henry Lydecker, Ryan Leonard, Ryan Keith, Lizzy Lowe, Fran Van Den Berg, Lucy Taylor, Lily van Eeden, Lisa Harvey, Margot Law, Matthew Crowther, Matthew Hall and Caitlyn Drayton-Taylor– many of you helped with field work, proof reading, and endless encouragement when my imposter syndrome got the better of me.

v My colleagues at the Faculty of Science Outreach and Engagement Team have been such a wonderful support to me throughout my PhD with amazing teaching opportunities, and now in my new role over the last few months – my passion for science engagement has grown in this environment, and with these inspirational people. I am very lucky to be able to learn from them every day.

Now to the people who like me even though they should know better – I don’t know how I have got so lucky, but I know that I would not have been able to complete this PhD without my friends. Distracting me from studying, taking me out for dinner, and listening to my bat stories even when they didn’t want to. Special mention goes to Stefanie, Masaya, Anna, Jeff, Megan, Lily, Anna Doty, Emily, Jules, Sarah, Nikola, Tim, Ash, Pauly, Dan, and Dieter. Thank you for everything.

To my family, near and far – Edd, Carreen, Ben, Rebecca - I appreciate your love and support over the last four years, and to Michelle and Mike for providing a much-needed bed near the field site during field work season.

Finally, this thesis is dedicated to six people. To Liz, your enthusiasm for life and knowledge is infectious, and our all night chats on the Glebe balcony did my thesis the world of good. To Annie, you have been with me through everything and know me in high definition – thank you for celebrating my research and for always believing in me, you are one of the main reasons this thesis got completed. To Nan, your enthusiasm for education, our dinners together and your support during my PhD always encouraged me to do my best. To Bryce, your strength, determination and hard-working nature give me the motivation to follow my dreams, and a safe and happy place to come home to. To Dad, your never- ending curiosity and respect for the natural world has inspired this thesis. To Mum, you keep me happy and teach me new things every day – my passion for learning started with you.

Abstract

One of the greatest defining forces of nature is the light-dark cycle, which is now highly disrupted by anthropogenic light pollution. Artificial light is most prevalent in cities, and negatively affects many nocturnal taxa in urban ecosystems. Research has historically

focused on the impacts of light on biodiversity compared with ambient darkness, and has

only recently begun to address how different spectra of light affect taxa in different ways.

Insectivorous bats are likely to be among the taxa most affected by light pollution, due to

their nocturnality and reliance on an insect prey itself disrupted by artificial light, and there

is increasing evidence that artificial light benefits some bat species whilst negatively impacting others. Street lighting technology is being upgraded for cost and energy efficiency purposes, with little understanding of the impact of changing spectra on biota.

Defining bats’ traits associated with responses to light, and the responses of bats to different lighting technologies, will lead to a better understanding of the impact of artificial light and urbanisation on bat species. I used acoustic surveying methods and radio telemetry techniques to examine the responses of species and guilds to light pollution, in a series of studies testing commuting and foraging behaviour. My research addressed the issues of public lighting in terms of light locations and light spectra, and in different ecologically important habitats; urban forests, waterways and wetlands.

My results showed that a change in street lighting spectra from the common mercury vapour to new light emitting diode (LED) technology causes a decrease in the activity of fast and slow flying bats. Bat diversity was also highest in dark forest remnants, and species with slow flight speed and high echolocation call frequency were the species most reliant on this habitat. This is consistent with the hypothesis that bats with these traits are

vii sensitive to artificial light, indicating that they are most likely to be impacted by the

introduction of light to the urban forest.

In a further experiment investigating the impact of light pollution on functional

connectivity for bats, I found that the activity of this putatively light avoidant group was

significantly lower along artificially lit forest edges than on dark edges. Faster flying bats

adapted to flying in open spaces were unaffected by lights at the edges of urban forests,

using light and dark edges in a similar way. The fact that public lighting reduces functional

habitat connectivity for some species but not others indicates that resilience to artificial

light disruption may be important for a species’ persistence in cities.

To examine the impact of artificial light on a specialist bat, Myotis macropus, a vulnerable

trawling bat species and member of genus established to be light avoidant, I conducted a

light introduction study along a semi-urban riparian corridor. This species is reliant on

riparian areas and sources 100 % of its diet from the aquatic environment, but this habitat is

commonly light polluted in urban areas. By radio-tracking and acoustically surveying this

species I revealed that their commuting and foraging behaviour was significantly disrupted

after introducing artificial light to a waterway.

I lastly examined the effects of different colours of light on bats at wetlands and concluded

that white lights cause a significantly negative community-level response where in

comparison red lights did not. There was also one bat species, Vespadelus vulturnus, that

was attracted to light sources in this peri-urban experiment, but had responded negatively

viii in more urban environments, demonstrating that the response to artificial light may be context-dependent.

My research shows that disrupting the light-dark cycle in urban areas is discouraging specialist bats with slow flight speed and high echolocation call frequency whilst encouraging generalist, faster flying species. These impacts may be context-dependent for some species as well. To increase bat diversity in urban areas, we should avoid using street lights near ecologically sensitive areas such as small urban remnants and forests, narrow

corridors and waterways, and use narrower spectrum streetlights (lights emitting

wavelengths in a narrow part of the visible spectrum) in these areas if public lighting is

absolutely necessary.

ix General Introduction

Light pollution

More people than ever are living in cities (United Nations 2018) and the congregation and

intensification of human population is set to continue (Cohen 2003). As urban settlements

grow, so too does urban light pollution, growing at a rate of 6 % per year (Hölker et al.

2010), and intensifying most in cities (Falchi et al. 2016). Light pollution changes the natural

levels of ambient light at night, reduces the fear of crime (Painter 1996), increases the

accuracy of eyewitness testimony (Rea et al. 2009) and decreases the number of injuries

from traffic accidents (Beyer and Ker 2009). Cities feel vibrant and safe when they are well-

lit, and economic and social benefit has been attributed to public lighting (Doll et al. 2006).

Organisations like the Cities of Light Partnership (LUCI) promote urban artificial light as a way of increasing social cohesion, aesthetics, and quality of life. Humans’ diurnal nature can be seen in our festivals and celebrations of artificial light across the world. The festival of light, Diwali, celebrates the triumph of good over evil and light over darkness and is marked by Hindus, Sikhs and Jains. Pingxi Lantern Festival in Taiwan celebrates the lanterns that were once used as an indicator that the town was safe. Vivid Sydney, Amsterdam Light

Festival and Lumiere London generate millions of tourism dollars every year.

While it is often seen as desirable, the benefits of artificial light may not be straight forward. Artificial light may actually pose risks to human health and wellbeing (Green et al.

2015). Light pollution can cause a disruption in natural circadian rhythms (Pauley 2004), a suppression of melatonin (Falchi et al. 2011), and an increase in chronic diseases (for a

review see Navara and Nelson 2007). The relationship between artificial lighting in cities

1 and road safety is not a simple one (Welsh and Farrington 2008). Artificial lighting may only

reduce the fear of crime and not crime itself (Steinbach et al. 2015); there was no increase

in crime or road accidents when lights were turned off in parts of the UK. Astronomical light

pollution is having immeasurable cultural impacts; sightings of the Milky Way from home

are now impossible for all citizens of Singapore, Kuwait and others (Falchi et al. 2016).

Public lighting cost billions of dollars in public money every year, has a large carbon footprint (OECD / IEA 2006), and disrupts our view of the night sky (Falchi et al. 2016).

Historical developments in the field of light pollution research

The brightness (measured in lux, a unit of illumination on a surface one metre away from a

point source of light equal to one candle) of public lighting is no longer the sole focus of

research. The technologies used to light our streets and cities have evolved, and so have

the spectra that these different lighting technologies emit. Bright sunlight has a broad

emission spectrum and is very bright (around 120,000 lux). Moonlight has the same broad

emission spectra but at much lower lux, approximately 1 lux for a full moon. But ambient

nocturnal light levels are being altered by light pollution and the change and impact is

dependent on the lighting technologies. Low pressure sodium vapour lights were some of the first widely used technology. Introduced in Europe in the 1930s, a narrow spectrum light

source only emits wavelengths in the orange part of the visible spectrum (~590-635 nm).

Fluorescent lamps were introduced widely in the 1940s, emitting high amounts of green,

blue and ultraviolet light (~495-570 nms, ~450-495 nms and ~10-400 nms respectively). The twentieth century was largely dominated by high pressure sodium vapour, metal halide and mercury vapour lights, all of which emit high amount of ultraviolet light. LED streetlights, first introduced in the UK in 2011, can be manufactured to emit almost any colour and

2 spectra but usually do not emit ultraviolet light. As technology has changed, the ecological,

astronomical and anthropocentric fields of light pollution research have developed.

Historically, research concentrated on the impacts of the brightness of light sources (for a

review see Longcore and Rich 2004, Cinzano et al. 2001). However, levels of light pollution

are dependent not only on light lux levels, but also on the spectral characteristics (Falchi et

al. 2011) of the lighting technologies used. For humans, the least polluting light sources are

low pressure sodium lamps, and the most polluting lights are strong blue emission such as

LED (Falchi et al. 2011, Falchi et al. 2016). But other species have different peaks in visual sensitivity (Davies et al. 2013), and therefore may be disrupted by different wavelengths

(Table 1.1). Spectra-dependent responses of biodiversity to light pollution has been the focus of recent research, summarised in Table 1.1.

Table 1.1. Studies looking at spectra-dependent responses in a range of taxa

Response (^ = physiological Light type / Taxa response, * = behavioural Type of study Study spectra response, ^* = unclear) White, green and Songbirds No differing responses to the Experiment field Da Silva, de Jong et red lights lights ^* study in rural forest al. 2017 White, green and Captive blue tits Red and white both delayed the Experimental, lab De Jong, Caro et al. red (Cyanistes onset of morning activity ^ conditions 2017 caeruleus)

White, green and Great tit (Parus Effect of light treatment of lay Experiment field De Jong, Ouyang et red LED lights major) and pied date – green and white lights study in rural forest al. 2015 flycatcher advanced the onset of the lay (Ficedula date, whereas red light had no hypoleuca) effect ^*

Red light with Wheat (Triticum Photosynthesis was strongly Experimental, lab Hogewoning, varying degrees of aestivum L., cv. induced using red lights, but conditions Trouwborst et al. blue light ‘USU-Super leaves did not develop as many 2010 combined Dwarf’) abnormalities if some blue light was included also ^ Red, green or Great tit (Parus Higher corticosterone Experiment field Ouyang, de Jong et white light major) levels in adults that nested near study in rural forest al. 2015 white lights ^ Red, white, green Migrating birds Birds were disrupted by red and Experimental field Poot, Ens et al. and blue lights white light, but not blue and study, 2008 green light ^ observational data White, green and Mice, bats, birds Red lights had less of an effect of Long term study in Spoelstra, van red lights and moths mice than green and white lights; a rural forest Grunsven et al. slow flying bats were not 2015 affected by red lights, no clear pattern for birds and moths ^* White metal halide An evergreen Trees photosynthesised faster Experiment across Takagi and lights versus dark broadleaf tree under lit conditions ^ the urban gradient, Gyokusen 2004 (Ilex rotunda) from rural to urban

White, green and Migrating toads White and green had a negative Experimental study Van Grunsven, red light (Bufo bufo) effect on toads, but red light had along a semi-urban Creemers et al. no effect ^* road 2017 Mercury vapour, Insects Mercury vapour was the most Experimental field van Grunsven, low sodium, metal attractive to all families of insects study in agricultural Donners et al. 2014 halide and (highest by far in UV) ^ farmland phosphorous LED, blue light and red light Six different Moths Every family of lepidoptera Experiment field van Langevelde, wavelengths of collected was most attracted to study in rural forest Ettema et al. 2011 light, one of which ultraviolet light ^ being ultraviolet White, green and Migratory Disoriented when under red light, Experimental, lab Wiltschko and red lights European Robins but no effect for white or green conditions Wiltschko 1995 (Erithacus light ^ rubecula)

Ecological consequences of light pollution

“Our diurnal bias has allowed us to ignore the obvious, that the world is different at night

and that natural patterns of darkness are as important as the light of day to the functioning

of ecosystems.” - Longcore and Rich (pg.1, 2006)

Light pollution has been identified as a key, global biodiversity threat (Hölker et al. 2010)

and affects multiple taxa across terrestrial (Spoelstra et al. 2015), marine (Witherington and

Martin 2000) and aquatic (Perkin et al. 2014) ecosystems (for a multi taxa review see

Longcore and Rich, 2004). Light pollution has been discussed as a main driver of the evolution of urban ecosystems (Hopkins et al. 2018). The paths of migrating birds are interrupted by artificial light sources (Van Doren et al. 2017), the movements of hatchling sea turtles are disrupted by light pollution on the horizon (Thums et al. 2016), disruption in light cues can alter the leafing phenology of trees, and many plant-insect interactions

(Bennie et al. 2016), demonstrating how pervasive artificial light can be throughout trophic levels. Many diurnal communities can also be affected, through changes in circadian or behavioural rhythms (Kempenaers et al. 2010), phenology (Bennie et al. 2016), or increased

competition and competitive exclusion around sunrise and sunset (Russ et al. 2015, Da Silva

et al. 2014). However, most species affected by light pollution are nocturnal. Light changes

the most defining circumstance of nocturnal life; darkness, and many nocturnal species

struggle or fail to adapt.

The ecological impacts of light pollution have been studied in many different ways,

including correlative landscape scale surveys (Gorenzel and Salmon 1995, Straka et al.

5 2016), disorientation responses to large existing light sources (Jaeger and Hailman 1973,

Salmon et al. 1995, Ogden 1996) experiments in rural or agricultural landscapes with naïve

ecosystems (Stone et al. 2009, van Langevelde et al. 2011, van Grunsven et al. 2014) and

controlled lab experiments (Wiltschko and Wiltschko 1995, De Jong, Caro et al. 2017).

Artificial light has also commonly been linked to road ecology (Pocock and Lawrence 2005,

Delgado et al. 2007, Berthinussen and Altringham 2012). The effects of light are difficult to

tease apart from other effects of urbanisation on biodiversity as they co-occur so

commonly (Falchi et al. 2016). Only recently have experiments been carried out to

disentangle light from other contributing factors in urban areas (Kuijper et al. 2008, Brüning

et al. 2015, Stone et al. 2015, Lewanzik and Voigt, 2017, Azam et al. 2018), with studies

concluding that light pollution has negative effects on many taxa beyond the impacts of

urbanisation.

Mechanisms of ecological light sensitivity

Changes in presence, abundance, diversity or community composition of species caused by

artificial light may be driven by either direct or indirect responses, ie, behavioural or

physiological responses to light. Three main possible mechanisms drive a species’ response

to artificial light. The species may be physiologically sensitive to a certain wavelength and

respond positively (phototactically, van Langevelde et al. 2011) or negatively

(photophobically, Beier 1995). The artificial light could indirectly make a prey animal more

at risk of predation from visual predators and so the prey species may avoid lit areas to

avoid predation (Predation Risk Hypothesis sensu Mougeot and Bretagnolle 2000, Navara

and Nelson 2007, Yurk and Trites 2000). Finally, light pollution may cause a change in prey

availability and a predator‘s activity indirectly reflects this (Foraging Efficiency Hypothesis,

6 sensu Imber 1975), either positively by using the lit area as a foraging ground (Rydell 1992)

or negatively by avoiding a lit area with little food (Imber 1975). Current knowledge of these

highly diverse and species-specific responses is growing, but more research is needed on

the community and trophic level responses created by the ‘night light niche’ (Longcore and

Rich 2004, Rowse et al. 2016).

Urban ecological light pollution

Urbanised areas are usually composed of a fragmented mosaic of land-use types (Fischer

and Lindenmayer 2006, Fischer and Lindenmayer 2007) and degraded natural habitat

(Fischer and Lindenmayer 2007). Urbanisation is the leading cause of biodiversity loss

(Czech et al. 2000), with many disruptive processes happening simultaneously. This suite of

factors, like invasive species (Brothers and Spingarn 1992Gibbons et al. 2000, Pocock and

Lawrence 2005), air pollution (Leonard et al. 2017), noise pollution (Bayne et al. 2008,

Mcdonald et al. 2009, Francis et al. 2009), habitat fragmentation (Fahrig et al. 2003, Krauss et al. 2010) and the urban heat island effect (Wong and Yu 2005, Meineke et al. 2013) all act as environmental filters to constrain urban plant and animal communities. Light pollution has been, until recently, an understudied environmental filter, with light location, brightness and spectra potentially having disruptive effects for nocturnal and crepuscular biota.

Connectivity of natural habitat across cities is can mitigate species loss (Beninde et al. 2015,

Lindenmayer and Nix 1993). Reducing that connectivity between natural areas in cities means that endemic species cannot move across the landscape without coming across roads (Coffin 2007, Hobday and Minstrell 2008), predators (Fahrig 1998, Baggio et al. 2011)

7 and humans (Fahrig 2003, Kretser et al. 2008). An important distinction is the difference

between structural and functional connectivity, where high functional connectivity is how

easy organisms can move around and use natural habitat in cities, but structural

connectivity is simply the connectivity of hypothetically appropriate habitat (Kupfer 2012).

Light pollution can act as an edge effect (Hölker et al. 2010, Kempenaers et al. 2010, Azam

et al. 2018), with functional connectivity degraded by lights at the urban forest edge

(Kempenaers et al. 2010, Hale et al. 2015, Azam et al. 2016). However, the International

Organisation for Standardisation (ISO) makes few suggestions to install low-impact lighting near ecologically sensitive areas like green spaces, urban forests and wetlands, in either its papers on public lighting standards or on its environmental sustainability

(ISO93.080.40 and ISO14001:2015 revision respectively). Additionally, few mentions are made in the Australian Standards Public Lighting Code (AS/NZS 1158). Currently, not enough is known about light impacts on urban habitat, with only a few papers looking at how light pollution reduces functional urban habitat connectivity (Pocock and Lawrence

2005, Kempenaers et al. 2010, Azam et al. 2018).

Urban bats and light pollution

Insectivorous bats make up a large percentage of native urban mammal fauna (Van der Ree

and McCarthy 2005, Jung and Kalko 2011) and yet are cryptic and quite often unnoticed by human residents. Insectivorous bats use echolocation to orientate (Denzinger and

Schnitzler 2013), they are all nocturnal and highly diverse, and many species persist in cities

(Jung and Threlfall 2016). Due to their diversity, insectivorous bats species and guilds respond differently to anthropogenic pressures, including artificial light. Bat guilds that are

8 based on morphological and behavioural traits are very valuable when looking at the

impacts of urbanisation (Jung and Threlfall 2016). The activity and species richness of bats

generally declines as the level of urbanisation increases (Jung and Kalko 2010, Threlfall et

al. 2012, Jung and Threlfall 2016), but some guilds persist and even increase in urban areas

(Jung and Kalko 2010). Bat species have complex habitat requirements, where each need

varying amounts and types of natural habitat (Duchamp et al. 2004, Avila-Flores and

Fenton 2005, Duchamp and Swihart 2008), habitat connectivity (Hale et al. 2015), water bodies (Kurta and Teramino 1992, Gehrt and Chelsvig 2003), and roost sites (Neubaum et al. 2007, Threlfall et al. 2013) to survive in cities.

As with many taxonomic groups (McKinney 2006), there are some bat species that adapt and persist, and some that avoid the urban environment (Duchamp and Swihart 2008).

Based on their foraging strategy, wing morphology and echolocation call patterns, insectivorous bat species exploit different habitats (Table 1.2) and can be assigned to guilds

(Denzinger and Schnitzler 2013, Figure 1.1). Certain morphological features of a bat wing, such as high aspect ratio (wingspan2 / wing area) and high wing loading (mass / wing area),

and a low characteristic echolocation call frequency means that the bat can fly fast over

long distances (Norberg and Rayner 1987) and their echolocation does not attenuate as

much (Denzinger and Schnitzler 2013).. Conversely, a species with low aspect ratio, low

wing loading and high characteristic or vertically shaped call frequency are adapted to

cluttered environments with their slow and manoeuvrable flight and highly detailed echoes

(Denzinger and Schnitzler 2013). A species’ guild is closely linked to how well that species

persists in the urban environment (Duchamp and Swihart 2008), and morphological traits

9 such as aspect ratio predict how likely a species is to be found in an urbanised area (Threlfall and Jung, in press).

Table 1.2. Bat guilds, with details of wing morphology and echolocation call patterns, and the habitats they prefer

Characteristic call frequency (based on Wing morphology (based Guild Pennay et al. 2004, Adams et al. 2009, on Denzinger and Habitat preference Schnitzler and Kalko 2001) Schnitzler 2013)

Clutter-space adapted High frequency and linear or steep call Low aspect ratio and low Dense vegetation shape, wing loading, broader wings

Trawling Often high frequency and linear or steep High aspect ratio and Riparian areas, and call shape medium wing loading water bodies

Edge-space adapted, 48 – 54 kHz Diverse Habitat edges high echolocating

Edge-space adapted, 39 – 48 kHz Diverse Habitat edges medium echolocating

Edge-space adapted, 32 – 39 kHz Diverse Habitat edges low echolocating

Open-space adapted 28 – 34 kHz High aspect ratio and high Open spaces, and wing loading, narrower above canopy wings

10 a

b c d e f

Figure 1.1. The use of different habitat types by guilds of insectivorous bats; a) Open-space adapted bats have a faster flight speed and can fly over longer distances (Norberg and Rayner 1987) These bats tend forage out in the open. b) c) d) Edge-space foragers are varied in their wing morphology and in the frequency of their characteristic call meaning that these bats can navigate and hunt in open and cluttered space. e) Cluttered-adapted bats have calls to help them differentiate between prey and close objects (Neuweiler 1984) f) Specialist trawling bats are reliant on riparian corridors and wetlands for foraging (Law and Urquhart 2000) and commuting (Anderson et al. 2006)

Faster flying bats with high aspect ratios seem to survive in urban areas (Threlfall and Jung,

in press), whereas slower flying bats with a lower aspect ratio are not as commonly

recorded in cities. This is probably due to a decrease in the dense vegetation, and habitat

connectivity that these slow flying bats require. The differing effects of urbanisation on bat

guilds may also be heavily confounded with impacts of artificial light, but this has been

rarely studied. But urban communities of insectivorous bats are potentially some of the

most affected by artificial light at night (ALAN, Gaston et al. 2015). Their nocturnality,

complex and varied habitat requirements (Figure 1.1) and a reliance on insect prey (itself

highly disrupted by artificial light, van Langevelde et al. 2011 for example) can result in a

11 highly disrupted landscape for urban bats (Hale et al. 2015). Much like urbanisation,

however, how a species adapts to ALAN seems closely linked to its morphological or

biological traits.

Open space adapted bats and ALAN

Bats that are adapted to fly in open space are generally fast flyers (Norberg and Rayner

1987, Kalko et al. 2008) and most are aerial hawking foragers (Norberg and Rayner 1987,

Rydell 2006), which means they can generally swoop in to the light cone and hunt insects

whilst in flight (Rydell 2006). The global trend emerging for these bats is a phototactic

response, ie. the UV in street lights attract insects (van Langevelde et al. 2011, Owens and

Lewis 2018) and these faster flying bats are able to exploit this resource due to their morphology (Jung and Kalko 2010). If this was an indirect response of the bats to the insect

prey, it would follow that when UV was reduced the bats would show a less significant or no

response. In fact, there have been a few studies looking at the impacts of low UV light types

like sodium vapour and LED and found no difference in open-space adapted fast flying bats

at street lights when compared with ambient darkness (Stone et al. 2009, Stone et al.

2012). The driving mechanism attracting these bats appears to be the UV-attracted insects.

Experiments comparing different coloured lights such as red, white and green (Spoelstra et

al. 2017) have concluded that red lights with no UV do not disrupt members of this guild.

There may be an exception however, as red light does attract some fast flying migrating

bats (Voigt et al. 2018). These migrating bats however, did not increase in foraging activity

once attracted to the red lights. This suggests that migratory bats may rely more on a

visual system when migrating, and so migrating patterns are disrupted by red lights, but

these species do not exploit red lights as foraging grounds once attracted (Voigt et al.

12 2018). Similarly, disruption to the migratory patterns of birds have been found in the

presence of red light, suggesting migratory species may be more affected than others by

night lighting (Wiltschko and Wiltschko 1995). Research urgently needs to clarify which

species of fast flying bats are disrupted by red lights as this mitigation strategy has begun

to be deployed.

Clutter adapted bat species and ALAN

Clutter adapted species are slow flyers and often use a specialised foraging strategy called

gleaning to eat insects from vegetation or the ground, or an aerial hawking strategy in

dense vegetation (Denzinger and Schnitzler 2013). Bats in this guild guild have been

identified as most at risk of extinction due to a loss of this dense vegetation (Safi and Kerth

2004) and tends to be light averse (Rydell 2006, Scanlon and Petit 2008, Stone et al. 2009,

Stone et al. 2012, Threlfall et al. 2013). This aversion to light is possibly driven by indirect mechanisms; the guild’s manouverable but slow flight (Denzinger and Schnitzler 2013) may

render them less able to avoid aerial predators that may use streetlights as foraging

grounds (Speakman 1991, Rydell et al. 1996) or less able to exploit insects at street lights

(Jones and Rydell 1994). If this guild is recorded at all in lit areas, it appears to be near lights that emit ultraviolet radiation and presumably attract more insects (Stone et al. 2015,

Rowse et al. 2016). When UV light decreases, the limited foraging advantage may disappear for these species. Although some work has be done looking at the detrimental effect artificial light has on urban forest habitat for this guild (Threlfall et al. 2013, Azam et al. 2018), more investigation needs to be done on species specific, guild and community level responses (Rowse et al. 2016). Mitigation strategies such as red lights may have a

significantly diminished impact on some members of this guild (Spoelstra et al. 2017), and

13 this may be due to a lack of UV light. However this has not been found for all slow flying species (Zeale et al. 2018). More research is needed here to elucidate which species in this guild respond to red lights as a mitigation strategy.

Trawling bats and ALAN

Trawling bats that trawl along the water surface to forage for insects and tiny fish

(Denzinger and Schnitzler 2013) show light avoidant behaviour (Kuijper et al. 2008, Straka et al. 2016) and are also specifically sensitive to UV light (Gorresen et al. 2015).

Immunohistochemical evidence suggests that UV vision is present in only some bat species that use frequency modulated echolocation, including within the trawling bat guild (Müller et al. 2009, Fujun et al. 2012, Xuan et al. 2012). These species may be responding directly to artificial light due to a retained sensitivity to UV light that other bats may not have; ultraviolet-emitting lights may repel trawling bats (Gorresen et al. 2015). Alternatively, these bats could be avoiding light due to an increase predation risk from large fish.

Experimental evidence of artificial light disturbance on trawling bats is limited, but confirms light avoidance (Kuijper 2008). Landscape scale experiments are urgently needed to investigate how pervasive the impacts of light are on this guild, as wetlands and streams in cities can be commonly light polluted (Straka et al. 2016). Research in to different coloured lights as a possible effective mitigation strategy for this guild has found inconsistent results. At hedgerows and through culverts, red light may have a minimal effect on trawling bats (Spoelstra et al. 2018, Zeale et al. 2018), but this has not yet been tested at the landscape scale and at wetlands, the primary foraging habitat of most trawling bats. Most experimental studies in this area to date have been in rural, not urban, contexts.

14 Edge adapted species and ALAN

Edge space adapted bats are evolved to forage and commute along habitat edges, and

most members of this guild have an echolocation call structure that allows them to switch

habitats quickly (Schnitzler and Kalko 2001, Denzinger and Schnitzler 2013). This guild has

diverse wing morphology traits and habitat edge preferences (Adams et al. 2009), and does

not respond to ALAN uniformly. Some members of this guild exploit lit areas (Haffner and

Stutz 1985, Adams et al. 2005), and yet others seem to avoid them. Even within a species,

different environments can change how bats respond to light (Geggie and Fenton 1985).

Vespadelus vulturnus were attracted to artificial light sources in a forested area (Adams et

al. 2005), but no study has tested to see how this changes in an urban and more light

polluted environment. A further subdivision of this guild demonstrates a clearer pattern

(Adams et al. 2005); bats in this guild with low and medium characteristic call frequencies

were the most positively affected by light sources compared with high frequency

echolocators (Adams et al. 2005). Again, how these sub-guilds respond to artificial light in an urban setting has not been tested. It is still unclear if this guild’s varied responses to

ALAN are driven by direct or indirect mechanisms; Geggie and Fenton (1985) found that bat activity closely mirrored insect availability, however no work has been done to rule out a biological sensitivity to light or certain wavelengths.

Aims and chapter structure

The aims of this thesis are to investigate how guilds and individual species of insectivorous

bats respond to different types of light pollution, and to provide practical strategies for

local and international bat conservation and policy making. My research contributes to

15 research by specifically examining insectivorous bats and artificial light in Australia, and will

assist land managers and councils to make ecologically sensitive lighting decisions.

Chapter 2 investigates how the popular strategy of changing from mercury vapour to LED

street lights for efficiency reasons is impacting urban nocturnal fauna. This is a pertinent

question as governments and land managers in many countries are now changing to LED

lights without an understanding of the ecological impact. Specifically, I tested the

hypothesis that the reduction in UV light would mean a decrease in flying insects, and

therefore a decrease in fast flying open space adapted bats and slow flying clutter adapted.

I discuss the importance of dark urban bushland in supporting high urban bat diversity, and

this natural experiment establishes the impact that changing to LED lights has on the bat

assemblage. This chapter has been accepted by Austral Ecology.

In Chapter 3, I test the impact of public lighting along urban forest edges. This chapter

builds on a key finding of Chapter 2, that dark natural habitat in cities is critical in

supporting urban bat and insect populations. I aimed to test if streetlights along edges of

urban forests reduced the activity of slow flying clutter adapted bats due to a decline in

functional habitat, but increased the number of insects and the activity of faster flying

open-space adapted bats. I discuss my findings in terms of artificial light as an edge effect,

degrading habitat corridors and patches, and an effect that informed public lighting policy

can minimise. This chapter is in review in Biological Conservation.

In Chapter 4, I focus on one threatened species of trawling bat, Myotis macropus, known to be light averse at a landscape scale, but its response to the introduction of artificial light has

16 never been experimentally tested. I hypothesise in this chapter that after the experimental introduction of light, this species would reduce its activity due to its traits which have been associated with light avoidance. I tested this using radio telemetry and acoustic monitoring and discuss how lighting waterways and streams impacts activity and behaviour of this light avoidant species.

In Chapter 5, I build on my findings in Chapters 3 and 4, testing a potential mitigation strategy for lighting urban waterways in a way that minimises disturbance for human and bat residents. This landscape scale experiment tests the different responses of the bat assemblage to red and white lights at wetlands. I hypothesise that fast flying, slow flying and trawling bats would be less disrupted by red lights, due to a lack of UV light, in comparison to white lights where I hypothesised that fast flying bats would be attracted, and slow flying and trawling bats would decline. I discuss my findings in terms of landscape- scale shifts in bat communities when lighting conditions change, and I suggest red lights as a potential strategy for ecological sensitive lighting.

In Chapter 6, I synthesise my research, clarify my contribution to our understanding of the relationship between urban bats and light pollution, make suggestions for future lighting policy and management and outline future opportunities for research.

A note on redundancy

This thesis is written as a series of manuscripts. There is some redundancy and repetition in the introductions and discussion of individual chapters. At the beginning of each chapter I have provided details of the publication where appropriate.

17 References

Adams, M. D., B. S. Law and K. O. French (2005). "Effect of lights on activity levels of forest bats: increasing the efficiency of surveys and species identification." Wildlife Research 32(2): 173-182.

Adams, M. D., B. S. Law and K. O. French (2009). "Vegetation structure influences the vertical stratification of open-and edge-space aerial-foraging bats in harvested forests." Forest Ecology and Management 258(9): 2090-2100.

Anderson, J., B. Law and C. Tidemann (2006). "Stream use by the large-footed myotis Myotis macropus in relation to environmental variables in northern New South Wales." Australian Mammalogy 28(1): 15-26.

Avila-Flores, R. and M. B. Fenton (2005). "Use of spatial features by foraging insectivorous bats in a large urban landscape." Journal of Mammalogy 86(6): 1193-1204.

Azam, C., I. Le Viol, J.-F. Julien, Y. Bas and C. Kerbiriou (2016). "Disentangling the relative effect of light pollution, impervious surfaces and intensive agriculture on bat activity with a national-scale monitoring program." Landscape ecology 31(10): 2471-2483.

Azam, C., I. Le Viol, Y. Bas, G. Zissis, A. Vernet, J.-F. Julien and C. Kerbiriou (2018). "Evidence for distance and illuminance thresholds in the effects of artificial lighting on bat activity." Landscape and Urban Planning 175: 123-135.

Baggio, J. A., K. Salau, M. A. Janssen, M. L. Schoon and Ö. Bodin (2011). "Landscape connectivity and predator–prey population dynamics." Landscape Ecology 26(1): 33-45.

Bayne, E. M., L. Habib and S. Boutin (2008). "Impacts of chronic anthropogenic noise from energy‐sector activity on abundance of songbirds in the boreal forest." Conservation Biology 22(5): 1186-1193.

Beier, P. (1995). "Dispersal of juvenile cougars in fragmented habitat." The Journal of Wildlife Management: 228-237.

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

Bennie, J., T. W. Davies, D. Cruse and K. J. Gaston (2016). "Ecological effects of artificial light at night on wild plants." Journal of Ecology 104(3): 611-620.

Berthinussen, A. and J. Altringham (2012). "The effect of a major road on bat activity and diversity." Journal of Applied Ecology 49(1): 82-89.

Beyer, F. R. and K. Ker (2009). " Street lighting for preventing road traffic injuries." Cochrane Database Syst Rev 1(1): CD004728.

Brothers, T. S. and A. Spingarn (1992). "Forest fragmentation and alien plant invasion of central Indiana old‐growth forests." Conservation Biology 6(1): 91-100.

Brüning, A., F. Hölker, S. Franke, T. Preuer and W. Kloas (2015). "Spotlight on fish: light pollution affects circadian rhythms of European perch but does not cause stress." Science of the Total Environment 511: 516-522.

Cinzano, P., F. Falchi and C. D. Elvidge (2001). "The first world atlas of the artificial night sky brightness." Monthly Notices of the Royal Astronomical Society 328(3): 689-707.

Coffin, A. W. (2007). "From roadkill to road ecology: a review of the ecological effects of roads." Journal of Transport Geography 15(5): 396-406.

Cohen, J. E. (2003). "Human population: the next half century." Science 302(5648): 1172- 1175.

Czech, B., P. R. Krausman and P. K. Devers (2000). "Economic associations among causes of species endangerment in the United States: associations among causes of species endangerment in the United States reflect the integration of economic sectors, supporting the theory and evidence that economic growth proceeds at the competitive exclusion of nonhuman species in the aggregate." AIBS Bulletin 50(7): 593-601.

Da Silva, A., M. de Jong, R. H. van Grunsven, M. E. Visser, B. Kempenaers and K. Spoelstra (2017). "Experimental illumination of a forest: no effects of lights of different colours on the onset of the dawn chorus in songbirds." Royal Society Open Science 4(1): 160638.

Da Silva, A., Samplonius, J. M., Schlicht, E., Valcu, M., and Kempenaers, B. H. A. (2014). “Artificial night lighting rather than traffic noise affects the daily timing of dawn and dusk singing in common European songbirds.” Behav. Ecol. 25, 1037–1047.

Davies, T. W., J. Bennie, R. Inger, N. H. Ibarra and K. J. Gaston (2013). "Artificial light pollution: are shifting spectral signatures changing the balance of species interactions?" Global Change Biology 19(5): 1417-1423.

De Jong, M., J. Q. Ouyang, A. Da Silva, R. H. van Grunsven, B. Kempenaers, M. E. Visser and K. Spoelstra (2015). "Effects of nocturnal illumination on life-history decisions and fitness in two wild songbird species." Philosophical Transactions of the Royal Society B: Biological Sciences, 370(1667).

De Jong, M., S. P. Caro, P. Gienapp, K. Spoelstra and M. E. Visser (2017). "Early birds by light at night: effects of light color and intensity on daily activity patterns in blue tits." Journal of Biological Rhythms 32(4): 323-333.

Delgado, J. D., N. L. Arroyo, J. R. Arévalo and J. M. Fernández-Palacios (2007). "Edge effects of roads on temperature, light, canopy cover, and canopy height in laurel and pine forests (Tenerife, Canary Islands)." Landscape and Urban Planning 81(4): 328-340.

19 Denzinger, A. and H.-U. Schnitzler (2013). "Bat guilds, a concept to classify the highly diverse foraging and echolocation behaviors of microchiropteran bats." Frontiers in Physiology, 4, 164.

Doll, C. N., J.-P. Muller and J. G. Morley (2006). "Mapping regional economic activity from night-time light satellite imagery." Ecological Economics 57(1): 75-92.

Duchamp, J. E. and R. K. Swihart (2008). "Shifts in bat community structure related to evolved traits and features of human-altered landscapes." Landscape Ecology 23(7): 849- 860.

Duchamp, J. E., D. W. Sparks and J. Whitaker, John O (2004). "Foraging-habitat selection by bats at an urban–rural interface: comparison between a successful and a less successful species." Canadian Journal of Zoology 82(7): 1157-1164.

Fahrig, L. (1998). "When does fragmentation of breeding habitat affect population survival?" Ecological Modelling 105(2-3): 273-292.

Fahrig, L. (2003). "Effects of habitat fragmentation on biodiversity." Annual Review of Ecology, Evolution, and Systematics 34(1): 487-515.

Falchi, F., P. Cinzano, C. D. Elvidge, D. M. Keith and A. Haim (2011). "Limiting the impact of light pollution on human health, environment and stellar visibility." Journal of Environmental Management 92(10): 2714-2722.

Falchi, F., P. Cinzano, D. Duriscoe, C. C. Kyba, C. D. Elvidge, K. Baugh, B. A. Portnov, N. A. Rybnikova and R. Furgoni (2016). "The new world atlas of artificial night sky brightness." Science Advances 2(6): e1600377.

Fischer, J. and D. B Lindenmayer (2006). "Beyond fragmentation: the continuum model for fauna research and conservation in human‐modified landscapes." Oikos 112(2): 473-480.

Fischer, J. and D. B. Lindenmayer (2007). "Landscape modification and habitat fragmentation: a synthesis." Global Ecology and Biogeography 16(3): 265-280.

Francis, C. D., C. P. Ortega and A. Cruz (2009). "Noise pollution changes avian communities and species interactions." Current Biology 19(16): 1415-1419.

Fujun, X., H. Kailiang, Z. Tengteng, R. Paul, W. Xuzhong and S. Yi (2012). "Behavioral evidence for cone-based ultraviolet vision in divergent bat species and implications for its evolution." Zoologia (Curitiba) 29(2): 109-114.

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

20 Geggie, J. and M. Fenton (1985). "A comparison of foraging by Eptesicus fuscus (Chiroptera: Vespertilionidae) in urban and rural environments." Canadian Journal of Zoology 63(2): 263- 266.

Gehrt, S. D. and J. E. Chelsvig (2003). "Bat activity in an urban landscape: patterns at the landscape and microhabitat scale." Ecological Applications 13(4): 939-950.

Gibbons, J. W., D. E. Scott, T. J. Ryan, K. A. Buhlmann, T. D. Tuberville, B. S. Metts, J. L. Gorenzel, W. P. and T. P. Salmon (1995). "Characteristics of American crow urban roosts in California." The Journal of wildlife management: 638-645.

Gorresen, P. M., P. M. Cryan, D. C. Dalton, S. Wolf and F. J. Bonaccorso (2015). "Ultraviolet vision may be widespread in bats." Acta Chiropterologica 17(1): 193-198.

Green, J., C. Perkins, R. Steinbach and P. Edwards (2015). "Reduced street lighting at night and health: a rapid appraisal of public views in England and Wales." Health & Place 34: 171- 180.

Greene, T. Mills, Y. Leiden and S. Poppy (2000). "The global decline of reptiles, Déjà Vu Amphibians: reptile species are declining on a global scale. Six significant threats to reptile populations are habitat loss and degradation, introduced invasive species, environmental pollution, disease, unsustainable use, and global climate change." AIBS Bulletin 50(8): 653- 666.

Haffner, M. and H. Stutz (1985). "Abundance of Pipistrellus pipistrellus and Pipistrellus kuhlii foraging at street-lamps." Myotis 23(24): 167-168.

Hale, J. D., A. J. Fairbrass, T. J. Matthews, G. Davies and J. P. Sadler (2015). "The ecological impact of city lighting scenarios: exploring gap crossing thresholds for urban bats." Global Change Biology: 21(7), 2467-2478

Hobday, A. J. and M. L. Minstrell (2008). "Distribution and abundance of roadkill on Tasmanian highways: human management options." Wildlife Research 35(7): 712-726.

Hogewoning, S. W., G. Trouwborst, H. Maljaars, H. Poorter, W. van Ieperen and J. Harbinson (2010). "Blue light dose–responses of leaf photosynthesis, morphology, and chemical composition of Cucumis sativus grown under different combinations of red and blue light." Journal of Experimental Botany 61(11): 3107-3117.

Hölker, F., C. Wolter, E. K. Perkin and K. Tockner (2010). "Light pollution as a biodiversity threat." Trends in Ecology & Evolution(25): 681-682.

Hölker, F., T. Moss, B. Griefahn, W. Kloas, C. C. Voigt, D. Henckel, A. Hänel, P. M. Kappeler, S. Völker and A. Schwope (2010). "The dark side of light: a transdisciplinary research agenda for light pollution policy." Ecology and Society 15(4).

21 Hopkins, G. R., K. J. Gaston, M. E. Visser, M. A. Elgar and T. M. Jones (2018). "Artificial light at night as a driver of evolution across urban–rural landscapes." Frontiers in Ecology and the Environment 16(8): 472-479.

Imber, M. (1975). "Behaviour of Petrels in relation to the moon and artificial lights.” Journal of the Ornithological Society of New Zealand: 302.

Jaeger, R. G. and J. P. Hailman (1973). "Effects of Intensity on the Phototactic Responses of Adult Anuran Amphibians: A Comparative Survey." Zeitschrift für Tierpsychologie 33(3‐4): 352-407.

Jones, G. and J. Rydell (1994). "Foraging strategy and predation risk as factors influencing emergence time in echolocating bats." Philosophical Transactions of the Royal Society of London B: Biological Sciences 346(1318): 445-455.

Jung, K. and C. G. Threlfall (2016). “Urbanisation and its effects on bats—a global meta- analysis. Bats in the Anthropocene: Conservation of Bats in a Changing World” Springer: 13-33.

Jung, K. and E. K. Kalko (2010). "Where forest meets urbanization: foraging plasticity of aerial insectivorous bats in an anthropogenically altered environment." Journal of Mammalogy 91(1): 144-153.

Jung, K. and E. K. Kalko (2011). "Adaptability and vulnerability of high flying Neotropical aerial insectivorous bats to urbanization." Diversity and Distributions 17(2): 262-274.

Kalko, E. K., S. Estrada Villegas, M. Schmidt, M. Wegmann and C. F. Meyer (2008). "Flying high—assessing the use of the aerosphere by bats." Integrative and Comparative Biology 48(1): 60-73.

Kempenaers, B., P. Borgström, P. Loës, E. Schlicht and M. Valcu (2010). "Artificial night lighting affects dawn song, extra-pair siring success, and lay date in songbirds." Current Biology 20(19): 1735-1739.

Krauss, J., R. Bommarco, M. Guardiola, R. K. Heikkinen, A. Helm, M. Kuussaari, R. Lindborg, E. Öckinger, M. Pärtel and J. Pino (2010). "Habitat fragmentation causes immediate and time‐delayed biodiversity loss at different trophic levels." Ecology Letters 13(5): 597-605.

Kretser, H. E., P. J. Sullivan and B. A. Knuth (2008). "Housing density as an indicator of spatial patterns of reported human–wildlife interactions in Northern New York." Landscape and Urban Planning 84(3-4): 282-292.

Kuijper, D. P., J. Schut, D. van Dullemen, H. Toorman, N. Goossens, J. Ouwehand and H. Limpens (2008). "Experimental evidence of light disturbance along the commuting routes of pond bats (Myotis dasycneme)." Lutra 51(1): 37.

22 Kupfer, J. A. (2012). "Landscape ecology and biogeography: rethinking landscape metrics in a post-FRAGSTATS landscape." Progress in Physical Geography 36(3): 400-420.

Kurta, A. and J. A. Teramino (1992). "Bat community structure in an urban park." Ecography 15(3): 257-261.

Law, B. and C. A. Urquhart (2000). "Diet of the Large-footed Myotis Myotis macropus at a forest stream roost in northern New South Wales." Australian Mammalogy 22(2): 121-124.

Leonard, R.J. and Hochuli, D.F. 2017. “Exhausting all avenues: why impacts of air pollution should be part of road ecology.” Frontiers in Ecology and the Environment, 15(8): 443-449.

Lewanzik, D. and C. C. Voigt (2017). "Transition from conventional to light‐emitting diode street lighting changes activity of urban bats." Journal of Applied Ecology 54(1): 264-271.

Lindenmayer, D. B. and H. A. Nix (1993). "Ecological principles for the design of wildlife corridors." Conservation Biology 7(3): 627-630.

Longcore, T. and C. Rich (2004). "Ecological light pollution." Frontiers in Ecology and the Environment 2(4): 191-198.

Longcore, T., & Rich, C. (Eds.). (2006). “Ecological consequences of artificial night lighting.” Island Press.

McDonald, R. I., R. T. Forman, P. Kareiva, R. Neugarten, D. Salzer and J. Fisher (2009). "Urban effects, distance, and protected areas in an urbanizing world." Landscape and Urban Planning 93(1): 63-75.

McKinney, M. L. (2006). "Urbanization as a major cause of biotic homogenization." Biological Conservation 127(3): 247-260.

Mougeot, F. and V. Bretagnolle (2000). "Predation risk and moonlight avoidance in nocturnal seabirds." Journal of Avian Biology 31(3): 376-386.

Müller, B., M. Glösmann, L. Peichl, G. C. Knop, C. Hagemann and J. Ammermüller (2009). "Bat eyes have ultraviolet-sensitive cone photoreceptors." PLoS One 4(7): e6390.

Navara, K. J. and R. J. Nelson (2007). "The dark side of light at night: physiological, epidemiological, and ecological consequences." Journal of Pineal Research 43(3): 215-224.

Neubaum, D. J., K. R. Wilson and T. J. O'Shea (2007). "Urban maternity‐roost selection by big brown bats in Colorado." The Journal of Wildlife Management 71(3): 728-736.

Neuweiler, G. (1984). "Foraging, echolocation and audition in bats." Naturwissenschaften 71(9): 446-455.

Norberg, U. M. and J. M. Rayner (1987). "Ecological morphology and flight in bats (Mammalia; Chiroptera): wing adaptations, flight performance, foraging strategy and

23 echolocation." Philosophical Transactions of the Royal Society B: Biological Sciences 316(1179): 335-427.

Ogden, E. L. J. (1996). "Collision course: the hazards of lighted structures and windows to migrating birds." Fatal Light Awareness Program (FLAP), 3.

Organisation for Economic Co-operation and Development and the Internatioanl Energy Agency. (2006). "Light's labour's lost: policies for energy-efficient lighting." Energy World(343): 14-15.

Ouyang, J. Q., M. de Jong, M. Hau, M. E. Visser, R. H. van Grunsven and K. Spoelstra (2015). "Stressful colours: corticosterone concentrations in a free-living songbird vary with the spectral composition of experimental illumination." Biology Letters 11(8): 20150517.

Owens, A. C. and S. M. Lewis (2018). "The impact of artificial light at night on nocturnal insects: A review and synthesis." Ecology and Evolution 8:11337–11358.

Painter, K. (1996). "The influence of street lighting improvements on crime, fear and pedestrian street use, after dark." Landscape and Urban Planning 35(2): 193-201.

Pauley, S. M. (2004). "Lighting for the human circadian clock: recent research indicates that lighting has become a public health issue." Medical Hypotheses 63(4): 588-596.

Pennay, M., Law, B., & Reinhold, L. (2004). Bat calls of New South Wales: Region based guide to the echolocation calls of Microchiropteran bats. NSW Department of Environment and Conservation.

Perkin, E. K., F. Hölker and K. Tockner (2014). "The effects of artificial lighting on adult aquatic and terrestrial insects." Freshwater Biology 59(2): 368-377.

Pocock, Z. and R. E. Lawrence (2005). "How far into a forest does the effect of a road extend? Defining road edge effect in eucalypt forests of south-eastern Australia." Road Ecology Center.

Poot, H., B. J. Ens, H. de Vries, M. A. Donners, M. R. Wernand and J. M. Marquenie (2008). "Green light for nocturnally migrating birds." Ecology and Society 13(2).

Rea, M. S., J. D. Bullough and Y. Akashi (2009). "Several views of metal halide and high- pressure sodium lighting for outdoor applications." Lighting Research & Technology 41(4): 297-320.

Rowse, E. G., S. Harris and G. Jones (2016). "The Switch from Low-Pressure Sodium to Light Emitting Diodes Does Not Affect Bat Activity at Street Lights." PloS ONE 11(3): e0150884.

Rowse, E., D. Lewanzik, E. Stone, S. Harris and G. Jones (2016). “Dark matters: the effects of artificial lighting on bats. Bats in the Anthropocene: conservation of bats in a changing world” Springer: 187-213.

Russ, A., Rüger, A., and Klenke, R. (2015). “Seize the night: European Blackbirds (Turdus merula) extend their foraging activity under artificial illumination.” J. Ornithol. 156, 123–131.

Rydell, J. (1992). "Exploitation of insects around streetlamps by bats in Sweden." Functional Ecology: 744-750.

Rydell, J. (2006). "Bats and their insect prey at streetlights." Ecological Consequences of Artificial Night Lighting 2: 43-60.

Rydell, J., A. Entwistle and P. A. Racey (1996). "Timing of foraging flights of three species of bats in relation to insect activity and predation risk." Oikos: 243-252.

Safi, K. and G. Kerth (2004). "A comparative analysis of specialization and extinction risk in temperate‐zone bats." Conservation biology 18(5): 1293-1303.

Salmon, M., R. Reiners, C. Lavin and J. Wyneken (1995). "Behavior of loggerhead sea turtles on an urban beach. I. Correlates of nest placement." Journal of Herpetology: 560-567.

Scanlon, A. T. and S. Petit (2008). "Effects of site, time, weather and light on urban bat activity and richness: considerations for survey effort." Wildlife Research 35(8): 821-834.

Schnitzler, H. U. and E. K. Kalko (2001). "Echolocation by Insect-Eating Bats: We define four distinct functional groups of bats and find differences in signal structure that correlate with the typical echolocation tasks faced by each group." AIBS Bulletin 51(7): 557-569.

Speakman, J. (1991). "The impact of predation by birds on bat populations in the British Isles." Mammal Review 21(3): 123-142.

Spoelstra, K., J. J. Ramakers, N. E. van Dis and M. E. Visser (2018). "No effect of artificial light of different colors on commuting Daubenton's bats (Myotis daubentonii) in a choice experiment." Journal of Experimental Zoology Part A: Ecological and Integrative Physiology 329:506–510. Spoelstra, K., R. H. van Grunsven, J. J. Ramakers, K. B. Ferguson, T. Raap, M. Donners, E. M. Veenendaal and M. E. Visser (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, The Royal Society 284(1855).

Spoelstra, K., R. H. van Grunsven, M. Donners, P. Gienapp, M. E. Huigens, R. Slaterus, F. Berendse, M. E. Visser and E. Veenendaal (2015). "Experimental illumination of natural habitat—an experimental set-up to assess the direct and indirect ecological consequences of artificial light of different spectral composition." Phil. Trans. R. Soc. B 370(1667): 20140129.

Steinbach, R., C. Perkins, L. Tompson, S. Johnson, B. Armstrong, J. Green, C. Grundy, P.

25 Stone, E. L., A. Wakefield, S. Harris and G. Jones (2015). "The impacts of new street light technologies: experimentally testing the effects on bats of changing from low-pressure sodium to white metal halide." Philosophical Transactions of the Royal Society of London B: Biological Sciences 370(1667): 20140127.

Stone, E. L., G. Jones and S. Harris (2009). "Street lighting disturbs commuting bats." Current Biology 19(13): 1123-1127.

Stone, E. L., G. Jones and S. Harris (2012). "Conserving energy at a cost to biodiversity? Impacts of LED lighting on bats." Global Change Biology 18(8): 2458-2465.

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

Takagi, M. and K. Gyokusen (2004). "Light and atmospheric pollution affect photosynthesis of street trees in urban environments." Urban Forestry & Urban Greening 2(3): 167-171.

Threlfall, C. G., B. Law and P. B. Banks (2012). "Sensitivity of insectivorous bats to urbanization: implications for suburban conservation planning." Biological Conservation 146(1): 41-52.

Threlfall, C. G., B. Law and P. B. Banks (2013). "Roost selection in suburban bushland by the urban sensitive bat Nyctophilus gouldi." Journal of Mammalogy 94(2): 307-319.

Thums, M., S. D. Whiting, J. Reisser, K. L. Pendoley, C. B. Pattiaratchi, M. Proietti, Y. Hetzel, R. Fisher and M. G. Meekan (2016). "Artificial light on water attracts turtle hatchlings during their near shore transit." Open Science 3(5): 160142.

United Nations. (2018). "2018 Revision of World Urbanization Prospects." Department of Economic and Social Affairs, Population Division, New York.

Van der Ree, R. and M. A. McCarthy (2005). "Inferring persistence of indigenous mammals in response to urbanisation." Animal Conservation 8(03): 309-319.

Van Doren, B. M., K. G. Horton, A. M. Dokter, H. Klinck, S. B. Elbin and A. Farnsworth (2017). "High-intensity urban light installation dramatically alters nocturnal bird migration." Proceedings of the National Academy of Sciences 114(42): 11175-11180.

Van Grunsven, R. H., M. Donners, K. Boekee, I. Tichelaar, K. Van Geffen, D. Groenendijk, F. Berendse and E. Veenendaal (2014). "Spectral composition of light sources and insect phototaxis, with an evaluation of existing spectral response models." Journal of Insect Conservation 18(2): 225-231.

Van Grunsven, R. H., R. Creemers, K. Joosten, M. Donners and E. M. Veenendaal (2017). "Behaviour of migrating toads under artificial lights differs from other phases of their life cycle." Amphibia-Reptilia 38(1): 49-55.

Van Langevelde, F., J. A. Ettema, M. Donners, M. F. WallisDeVries and D. Groenendijk (2011). "Effect of spectral composition of artificial light on the attraction of moths." Biological Conservation 144(9): 2274-2281.

Voigt, C. C., K. Rehnig, O. Lindecke and G. Pētersons (2018). "Migratory bats are attracted by red light but not by warm‐white light: Implications for the protection of nocturnal migrants." Ecology and Evolution 8(18), 9353-9361.

Welsh, B. C. and D. P. Farrington (2008). "Effects of improved street lighting on crime." Campbell Systematic Reviews 13: 1-51.

Wilkinson and P. Edwards (2015). "The effect of reduced street lighting on road casualties and crime in England and Wales: controlled interrupted time series analysis." J Epidemiol Community Health 69(11): 1118-1124.

Wiltschko, W. and R. Wiltschko (1995). "Migratory orientation of European robins is affected by the wavelength of light as well as by a magnetic pulse." Journal of Comparative Physiology A 177(3): 363-369.

Witherington, B. E. and R. E. Martin (2000). "Understanding, assessing, and resolving light- pollution problems on sea turtle nesting beaches." Florida Marine Institute, Technical Report, TR-2

Wong, N. H. and C. Yu (2005). "Study of green areas and urban heat island in a tropical city." Habitat international 29(3): 547-558.

Xuan, F., K. Hu, T. Zhu, P. Racey, X. Wang, S. Zhang and Y. Sun (2012). "Immunohistochemical evidence of cone-based ultraviolet vision in divergent bat species and implications for its evolution." Comparative Biochemistry and Physiology Part B: Biochemistry and Molecular Biology 161(4): 398-403.

Yurk, H. and A. Trites (2000). "Experimental attempts to reduce predation by harbor seals on out‐migrating juvenile salmonids." Transactions of the American Fisheries Society 129(6): 1360-1366.

Zeale, M. R., E. L. Stone, E. Zeale, W. J. Browne, S. Harris and G. Jones (2018). "Experimentally manipulating light spectra reveals the importance of dark corridors for commuting bats." Global Change Biology 24:5909–5918.

27 Chapter 2: Responses of insectivorous bats and nocturnal insects to local changes in street light technology

This chapter has been accepted for publication in Austral Ecology:

Haddock, J. K., Threlfall, C. G., Law, B. S. and Hochuli, D. F. Responses of insectivorous bats and nocturnal insects to local changes in street light technology.” Accepted 27 November

2018.

Abstract

Artificial light at night is a pervasive anthropogenic stressor for biodiversity. Many fast- flying insectivorous bat species feed on insects that are attracted to lights emitting ultraviolet radiation (10 nm to 400 nm). Several countries are currently focused on replacing mercury vapour lamps, which emit ultraviolet light, with more cost-efficient LED lights, which emit less ultraviolet radiation. This reduction in ultraviolet light may in turn reduce insect concentrations in cities, cause declines in predatory fast flying bats, and cause declines in some edge foraging and slow flying bats particularly at risk from extinction.

Capitalizing on a scheme to update streetlights from high ultraviolet mercury vapour to low ultraviolet LED in Sydney, Australia, we measured the activity of individual bat species, the activity of different response groups, and the bat and insect communities, before and after the change in technology. We also surveyed sites already with LED lights, sites with mercury vapour lights, and dark bushland remnants. Species adapted to foraging in cluttered vegetation, and some edge-space foraging species, were more active in dark bushland sites than in all lit sites, and decreased in activity at lit sites after the change to

LED lights. The change to LED streetlights was linked to a decrease in the fast-flying

28 Chalinolobus gouldii but not Miniopterus orianae oceanensis, the latter being more influenced by seasonal and environmental variables. Insect biomass was not affected by changing light types, but instead was negatively correlated with the moon’s percentage illuminance. Changing streetlights to LEDs could affect populations and distributions of insectivorous bats in cities. This study confirms that dark urban bushland remnants are important refuges for high bat diversity, particularly for more clutter-adapted species, and some edge-space foraging species. Preventing light penetration in to dark bushland patches and corridors remains essential to protect the urban bat community.

Introduction

Seventy percent of the world’s human population is predicted to be living in urban

settlements by 2050 (United Nations 2018) and cities are expanding to meet this demand

(Cohen 2003). Anthropogenic threats to biodiversity such as light, noise and air pollution,

and the fragmentation or degradation of natural habitat are at their most intense in cities

(Chan and Yao 2008, Hölker et al. 2010, Ortega 2012, Wolch et al. 2014). Despite this, urban

areas can support more threatened species than non-urban areas (Ives et al. 2016) and therefore it is important to facilitate sustainable management for urban biodiversity.

Species rich urban ecosystems contribute to human wellbeing (Fuller et al. 2007, Taylor and

Hochuli 2015). Conservation of urban biodiversity requires an understanding of how anthropogenic pressures are affecting remnant urban habitat and how to mitigate threats.

Artificial light at night (ALAN) is a rapidly growing global threat to biodiversity (Hölker et al.

2010) and is becoming more intense in cities (Falchi et al. 2016). Artificial light

fundamentally alters the light/dark cycle and can change ecosystem services such as

29 pollination (Moore et al. 2000), predation rates (Gorenzel and Salmon 1995), communication (Kempenaers et al. 2010) and migration (Harewood and Horrocks 2008).

Artificial light, in both presence and spectra, can significantly alter urban biodiversity (see

(Rich and Longcore 2013). Higher radiation of orange and red light can induce some plants to photosynthesise and can alter plant growth and floral communities (Bennie et al. 2016), higher levels of ultraviolet light can attract higher densities of insects (van Langevelde et al.

2011, Pawson and Bader 2014, van Grunsven et al. 2014, Park and Lee 2017), and white light can decrease the utilization of habitat by nocturnal mammals (Spoelstra et al. 2015).

Residential areas in Sydney, Australia rely almost entirely on white mercury vapour lighting, but many streetlights are now being replaced with more cost effective and energy efficient LED technology (Ausgrid 2013), following a global trend. These LED lights are manufactured to emit white light in a similar high colour rendering index as mercury vapour lights, but they emit lower levels of ultraviolet radiation (10 – 400nm), attract lower numbers of insects (van Grunsven et al. 2014), and are generally brighter in lux than mercury vapour lights (Table 2.1). It remains a research priority to understand the implications for biodiversity of these changes in urban lighting regimes.

30 Table 2.1. Sampling intensity at all sites, both before and after the change in light type at sites 1 and 2, along with ambient light measures and streetlight type

Bat sampling intensity Insect sampling intensity Site Light intensity (lux) on Treatment (detector nights) (trapping nights) Number clear, full moon night Before After Before After Site 1 80W Mercury Vapour 13 lux changed to 15 lux 7 8 2 3 changed to 29W LED Site 2 80W Mercury Vapour 8.3 lux changed to 13 lux 6 12 2 2 changed to 29W LED Site 3 80W Mercury Vapour 7.9 lux 8 8 2 3

Site 4 80W Mercury Vapour 12 lux 10 10 2 2

Site 5 80W Mercury Vapour 11 lux 8 9 2 2

Site 6 Bushland 0.47 lux 8 8 2 3

Site 7 Bushland 0.66 lux 10 8 2 2

Site 8 Bushland 0.29 lux 8 9 2 2

Site 9 29W LED 16 lux 8 8 2 3

Site 10 29W LED 12 lux 10 12 2 2

Site 11 29W LED 17 lux 8 9 2 2

Insectivorous bats commonly occur in cities (Van der Ree and McCarthy 2005, Jung and

Threlfall 2016), and provide important ecological roles as well as making up a large percentage of urban mammal fauna (Van der Ree and McCarthy 2005). Morphological and behavioural traits in bats are associated with different use of habitats (Avila-Flores and

Fenton 2005, Denzinger and Schnitzler 2013), with the urban landscape representing a mosaic of different habitats for different bat species. The responses of bat species’ to artificial light may also be linked to their morphology and behaviour (Stone et al. 2015,

Rowse et al. 2016).

31 Faster flying bat species, adapted to foraging and flying in open space (Neuweiler 1984), are most resilient to the altered urban environment (Avila-Flores and Fenton 2005, Jung and Kalko 2011, Jung and Threlfall 2016). Many of these species are able to use street lights as foraging grounds, exploiting the increased abundance of insects that are attracted to the

UV radiation emitted by some street lights (Mathews et al. 2015, Stone et al. 2015, Rowse et al. 2016, Lewanzik and Voigt 2017). Besides which, artificial light interferes with moth predator evasion techniques, giving foraging bats an advantage over insect prey (Wakefield et al. 2015). An emerging global pattern is that ultraviolet-emitting street lights support higher activity of faster flying bats, and if ultraviolet radiation is decreased, the fast flying bat activity follows suit, probably tracking a decrease in ultraviolet-attracted insects

(Lewanzik and Voigt 2017). However, this pattern has not been established within the

Australian bat assemblage.

Not all species are able to exploit the ‘night light niche’ so successfully (Rich and Longcore

2013); many slower flying species adapted to cluttered vegetation (Neuweiler 1984) seem to avoid lit areas (Stone et al. 2012, Bader et al. 2015, Jung and Threlfall 2016, Rowse et al.

2016) and remain in dark bushland remnants (Threlfall et al. 2013). These slow flying species are most at risk of extinction (Jones et al. 2003), and so understanding their response to different lighting types and spectra is a research priority. LED lights set up along a hedgerow immediately excluded a threatened species, Rhinolophus hipposideros

(Stone et al. 2009), and lighting near a roost significantly disrupted colonies of slower flying, clutter-adapted bats (Boldogh et al. 2007). This light avoidance could be due to their slower flight speed which means they may not be efficient at hunting the flying insects that gather at light sources (Neuweiler 1984) or which could theoretically render them incapable

32 of evading predators that use artificially lit areas as hunting grounds (Stone et al.

2015)Myotis spp. have similar wing morphology to slow flying bats but may have retained a sensitivity to ultraviolet light (Gorresen et al. 2015), and so avoid these lights; two major studies recorded an increase in Myotis spp. after a change to low-ultraviolet LED lights

(Rowse et al. 2016, Lewanzik and Voigt 2017). Slower flying bats may be able to exploit insects at street lights, but to a far lesser extent than faster flying species. However, we still do not fully understand the impact that changing from mercury vapour to LED lights will have on clutter-adapted, slow flying bats.

Species that forage along the edges of habitat, edge-space foragers (Neuweiler 1984), are diverse in their echolocation call frequencies and flight patterns (Adams et al. 2005). They are not known to respond in a uniform way to artificial light in an urban setting, although edge-space foraging species that have high characteristic-frequency echolocation calls (the frequency at the end of the flattest part of the call, ESH species, Table 2.2, (Law et al. 2002)

(Adams et al. 2005) are most reliant on denser, cluttered vegetation (Adams et al. 2009), have shown the least positive response to artificial light in a forest setting when compared with other edge-space foragers (Adams et al. 2005), and are known to respond differently to different types of artificial lights (Jung and Kalko 2010). The impact of changing lighting technology on this group has not been tested in Australia, though they share many traits with clutter-adapted species and are sensitive to artificial light.

33 Table 2.2. Functional group assignment for predicted response to artificial light, based on previous classifications, physical attributes and known flight patterns

Call description and Foraging area ** Foraging guild Wing Aspect Putative light Species name characteristic frequency (CF) * & ^^ *** loading^^^ ratio^^^ avoidant group & ^ Curved, alternating, 28 – 31 kHz Edge Space Chalinolobus gouldii Edge 8.20 ± 2.26 6.53 ± 0.41 CF Low (ESL) Curved and steep, 35 – 39.5 kHz Edge Space Falsistrellus tasmaniensis *^ Edge CF Medium (ESM) Curved, down-sweeping tail, 43 Miniopterus orianae oceanensis *^ Edge ESM 9.71 ± 1.59 6.66 ± 0.28 – 47 kHz CF Open Space Ozimops ridei Flat, 28.5 – 31 kHz CF Open (OS) Austronomous australis Flat or curved, 10 – 13 kHz CF Open OS 15.46 ± 1.71 7.99 ± 0.45

Saccolaimus flaviventris *^ Curved, 18 – 21.5 kHz CF OS 15.90 ± 2.47 8.25 ± 0.42

Scoteanax rueppellii *^ Curved, 33 – 36 kHz CF Edge ESL 12.56 ± 0.22 6.36 ± 0.27

Scotorepens orion Curved, 34 – 37 kHz CF Edge ESL 10.43 ± 1.59 5.97 ± 0.54

Vespadelus darlingtoni Curved, 42 – 45 kHz CF Edge ESM Curved, down-sweeping tail, Edge Space Chalinolobus morio Edge 6.29 ± 1.08 6.11 ± 0.31 Light avoidant 47.5 – 53 kHz CF High (ESH) Curved, down-sweeping tail, 57 – Miniopterus australis *^ Edge ESH 5.77 ± 1.32 6.79 ± 0.48 Light avoidant 63 kHz CF Curved, up-sweeping tail, 49 – Vespadelus vulturnus Edge ESH 6.36 ± 0.53 5.19 ± 0.54 Light avoidant 54 kHz CF Near vertical, starts 70/80 kHz Nyctophilus gouldi Clutter CA 6.20 ± 1.25 5.46 ± 0.30 Light avoidant dropping to 35/45 kHz CF Near vertical, starts 70/80 kHz Nyctophilus geoffroyi Clutter CA 5.91 ± 0.97 5.60 ± 0.55 Light avoidant dropping to 35/45 kHz CF

Rhinolophus megaphyllus Flat, 66 – 72 kHz CF Clutter CA 6.77 ± 1.05 4.98 ± 1.05 Light avoidant

* = Reinhold et al. (2001) ^ = Pennay et al. (2004) ** = Adams et al. (2009) *** = Adams et al. (2005) ^^ = Milne et al. (2004) ^^^ = Rhodes (2002) *^ = Conservation status: VULNERABLE (NSW Biodiversity Conservation Act 2016)

34 In this study, we had an opportunity to measure the local-scale, short term responses of the

bat assemblage and insect biomass to a change in light type from high-ultraviolet mercury

vapour lights to brighter LED lights, lower in ultraviolet radiation. We also surveyed the bat

assemblage at different light types and dark bushland to assess how longer term differences in

spectra affected the bat assemblage and individual species.

We predicted the following responses;

1) That the urban bat and insect assemblages would respond to a change in light type from

mercury vapour to LED;

2) that the different light types would support different bat assemblages, with high-ultraviolet mercury vapour lights supporting a higher abundance of insects, higher activity of faster flying and higher activity of slow flying and ESH bats;

3) that dark urban bushland would provide a refuge from artificial light for slow flying and ESH bats.

Methods

Study site

The study was carried out in North Turramurra (Figure 2.1, coordinates 33°44'40.8"S

151°06'45.3"E), a leafy suburb of Sydney, Australia (Benson and Howell 1990), where bat diversity is high (Basham et al. 2010). There are around 15 species found in the study area, with diverse morphology and habitat requirements (Table 2.2). Sydney has high levels of remnant

35 native bushland abutting the urban matrix, however this habitat is rapidly decreasing due to

increased urbanisation (Benson et al. 1995).

Figure 2.1. Map showing all 11 sites in Turramurra, a North West suburb of Sydney, NSW, Australia. Scale bar in kilometres.

Experimental design

We surveyed four light treatments; sites being changed from mercury vapour to LED technology (n = 2), sites with mercury vapour lights (n = 3, Figure 2.2a), sites with LED lights (n

= 3, Figure 2.2b), and dark bushland (n = 3) which acted as reference sites. Sites were an average of 1.55 kms (±s.e. 0.15 kms) apart, and therefore bat foraging activity at each site was

36 considered independent. We surveyed each site for an average of 8.3 nights (s.e. ± 0.38) before

the change in light type, and 9 nights (s.e. ± 0.38) immediately after the change in light type

resulting in 192 detector recording nights (Table 2.1, Fischer et al. 2009).

a) b)

Relative power Relative power

Figure 2.2. Absorption spectrum graphs of absorbance versus wavelength, as I measured using a spectrophotometer; a) the mercury vapour streetlight at site 4 in Turramurra, North Sydney and b) the LED streetlight at site 10 in Pymble, North Sydney.

We sampled between March and April of 2015, with the street lights being upgraded at the

changeover sites on the 10th March. The insectivorous bat population in this region can be

reliably measured at this time, as long bouts of hibernation mean the population can be very

low during winter months, and artificially high after December when the young begin to fly

(Churchill 2008).

Bat sampling

At each site, we used an Anabat II detector (Titley Electronics, Ballina, NSW Australia) with the microphone 1 m above the ground and pointing upwards at a 45° angle to optimise the amount of unobstructed airspace sampled and maximize sampling success (Law et al. 1998, Patriquin et al. 2003, Threlfall et al. 2012). All detectors were calibrated to be equally sensitive and were expected to detect a bat at 30 m, depending on species. At the lit sites the detectors were

37 placed a maximum of 3 m away from the base of the streetlight pointing towards the light and

along the road. At bushland sites, detectors were placed along flyways; defined as pathways

over 2 m in width with a break in the tree canopy. By ensuring that the detectors at all sites were sampling habitat edges along the flyways or roads and habitat interiors where present, we controlled as far as possible for differences in the bat guilds present at different treatments.

All recording nights had comparable weather conditions (nights with rainfall over 3mm were excluded, average minimum daily temperature per sampling period was between 16 and 18°C).

The detectors pass the information through a Zero-Crossings Interface Module (ZCAIM, Titley

Electronics) and divide the frequency of the echolocation recording by a constant factor, a

division ratio, so that the recording is audible to humans but retains all major characteristics of

the call needed for species identification (Corben 2004).

Processing of audio data

Recorded calls were identified using the AnaScheme and AnalookW software (Adams et al.

2010). For a bat call to be identified to species level, three or more pulses were required and

have characteristics that fall within the program’s parameters for that species. A pass was

defined as at least three valid pulses with a minimum of 6 pixels per pulse. Positive species

identifications were made only when a minimum of 50% of pulses within a pass were identified

as the same species (Adams et al. 2009, Threlfall et al. 2012). The identification key used in the

analysis was developed for the Sydney area (Adams et al. 2010). The call characteristics of

Nyctophilus gouldi and Nyctophilus geoffroyi are indistinguishable using the AnaScheme

method and so were pooled as one taxon; Nyctophilus spp. The calls identified as Chalinolobus

38 dwyeri, Falsistrellus tasmaniensis, Nyctophilus spp., Saccolaimus flaviventris, Scoteanex

rueppellii and Scotorepens orion are known to be difficult to identify and were checked

manually to ensure conformance to other guides (Pennay et al. 2004, Adams et al. 2010). In

addition, calls were run through a filter in AnaScheme which is specifically designed to identify

alternating or unusual call characteristics of Chalinolobus gouldii (B. Law, pers. comm., 2015).

Only species that were positively identified using the key, filters and manual checking were

included for further analysis to eliminate any bias caused by using partially identified species.

Assignment to a response group

To assess how traits influence responses to light and light changes we pooled activity data for

both slower flying species known to forage in cluttered environments preferentially, and for

species known to forage along the edges of habitat with higher characteristic-frequency echolocation calls (Table 2.2, Norberg and Rayner 1987, Adams et al. 2005), in to one response

group. This included Nyctophilus spp., Rhinolophus megaphyllus, Chalinolobus morio,

Miniopterus australis, and Vespadelus vulturnus. We predicted a low detection rate for these

species at our artificially lit study sites and so pooled the species data to allow for statistical

analysis (Trindade-Filho et al. 2012). We predicted that edge-space foraging species with high

(above 48 kHz) characteristic-frequency echolocation calls (ESH species, Adams et al. 2009,

Adams et al. 2005) and slow flying, clutter-adapted species may show a similar response to a change in light type, and to artificially light generally (Stone et al. 2009, Threlfall et al. 2013,

Lewanzik and Voigt 2017).

39 Insect sampling

Insects were sampled using 20x25cm white sticky traps (Bugs For Bugs Pty Ltd, Mundubbera,

Australia). Traps were deployed at every site at least twice before the change in light type and

a minimum of three times after the change in light type. Although it would have been ideal to sample all sites on the same night, all trapping nights had comparable weather conditions (no rainfall, minimum daily temperature between 16 and 18°C). Traps were set up on wooden stakes at a height of 1.75m above the ground, opened within 30 minutes of sunset and collected within 40 minutes of sunrise. Sampling was conducted on alternative nights to bat recording to avoid impacting bat behaviour. We identified insects to order, and then measured the length of the insect and used algorithms (Sample et al. 1993) to estimate biomass. Over

91% of all insects caught were in the order Diptera, and so other orders were excluded from further analysis.

Environmental variables

Daily minimum temperature (°C), rainfall (mm) and relative daily humidity (%) for all sites were taken from the closest weather station (Parramatta North weather Station 066124, Bureau of

Meteorology). The percentage vegetation cover within a radius of 200 m was calculated using

Arc Map (ESRI, Redlands, USA, ver. 10.2) to be indicative of local vegetation extent near the detector (Lacoeuilhe et al. 2014). Vegetation cover was calculated by intersecting GPS points of our sites with the GIS layer ‘The Native Vegetation of the Sydney Metropolitan Area -

Version 3, VIS_ID 4489’ (NSW Office of Environment and Heritage, Sydney). This vegetation layer is the most complete vegetation mapping for the area and includes non-native

40 vegetation. Light intensity (lux) and spectral outputs of the different light types were measured using an I1Pro Spectrophotometer (X-rite, Melbourne, Australia), with the spectrophotometer placed 1.5 m above the road surface directly below the streetlight, or in the case of the bushland sites, 1.5 m directly above where the detector was situated. One measurement per site was taken before and after the change in light type. All light measurements were taken on a clear, new moon night between 2200 and 0000 hrs. Daily percentage of the moon illuminated was taken from the United States Naval Oceanography

Portal.

Statistical analysis

All analyses were carried out in SPSS (version 22, SPSS Inc., Chicago, USA). We defined overall bat activity as the number of successfully identified passes recorded each night, the activity of individual species as the number of successfully identified passes of that species recorded each night, and the activity of response group species as the number of successfully identified passes of that response group recorded each night. For each of the 11 sites we had two sampling periods, before and after the change in light type. The number of recording nights during each sampling period ranged from 6 to 12 nights (Table 2.1) for bat recording, and between 2 and 4 nights for insect sampling (Table 2.1). Due to a non-normal distribution, insect biomass was log transformed.

We used a general linear model (GLM) to compare the percentage vegetation cover between light treatments, with a normal distribution, a fourth root transformation and a log link

41 function. All models were compared with null models (without treatment) in log likelihood

ratio tests. If the ΔAICc was less than 2 between models, we chose the model with the fewest

number of parameters.

We wanted to elucidate if there were differences in lux from the different light treatments, or

from the change in light type. We performed a GLM with light treatment (mercury vapour,

LED, bushland and changeover), sampling periods (before and after the light change), the interaction included as fixed effects.

Insect biomass

To reduce factors in our model analyzing insect biomass data, we first used Pearson’s

parametric bivariate correlation to assess the relationship between insect biomass and

weather variables (rainfall, temperature and percentage moon illumination and relative

humidity) measured on the corresponding day. As insect biomass was measured on fewer days

than bat activity, these tests have lower degrees of freedom (Table 2.3). The percentage of the

moon illuminated was the only influential factor (p < 0.05), so it was retained and included in

that GLMM. We aimed to establish if insect biomass differed between light treatments, and

before and after the change in light type by using a generalized linear mixed model. In this

GLMM light treatment (mercury vapour, LED, bushland or changeover), sampling period

(before or after the light change), the interaction between the two, and the percentage of the

moon illuminated were fixed effects, and site was a random effect to account for repeated

measurements. For all GLMMs we used a Poisson distribution with a log link function. Fisher’s

Least Significant Difference was used as a post hoc test.

Bat and insect activity analysis

We used Pearson’s parametric bivariate correlation to assess the relationships between daily bat activity measures (activity of all bats, C. gouldii, M. orianae oceanensis and clutter-

adapted/ESH response group) and daily weather variables (rainfall, temperature and

percentage moon illumination and relative humidity), to avoid overparameterization and

reduce variables in our models. This revealed that relative humidity was the only influential factor (p < 0.05) when considering bat activity, C. gouldii activity and M. orianae oceanensis activity, and so it was included in further generalized linear mixed models (GLMMs) for those response variables. Percentage moon illumination was significantly correlated with the activity of the clutter-adapted/ESH response group and so was included in that GLMM.

We aimed to understand if the change in light type was linked to a change in the bat assemblage, and to understand the relative importance of dark bushland as refuge habitat compared with other light treatments, by conducting a series of GLMMs. Our response variables were overall bat activity, the activity of C. gouldii, the activity of M. orianae

oceanensis, and the pooled activity of the clutter-adapted/ESH response group. For all GLMMs

we used a Poisson distribution with a log link function. Light treatment (mercury vapour, LED,

bushland or changeover), sampling period (before and after the light change) and the

interaction between the two were included as fixed effects. We also included either relative

humidity or percentage moon illuminated as fixed effects, as described above. We included

site as a random effect to account for repeated measures at the same site. For the light

43 treatment fixed effect, we set the reference category as the bushland treatment, for the

before/after fixed effect, we set the reference category as the before treatment. Fisher’s Least

Significant Difference was used as a post hoc test.

To assess compositional differences in the bat communities among light treatments, we used

PRIMER (version 7, Quest Research, New Zealand). We used a fourth root transformation as

our bat activity data was skewed. We then generated a Bray-Curtis similarity matrix from

transformed activity data for all species. From the transformed dataset, we conducted a

similarity percentage analysis (SIMPER) and a permutational multivariate analysis of variance

(PERMANOVA).

Results

We detected 14 different bat species at our study sites. Five of these are listed as vulnerable

(Table 2.2, NSW Biodiversity Conservation Act 2016). We successfully identified 6118 calls to species or species group, and this was 56.5% of all calls detected (typical of studies from

Sydney, Threlfall et al. 2011). These 14 species compose 77% of the bat species known to occur in the area (Basham et al. 2010, Threlfall et al. 2011). Species we did not detect, Myotis macropus, Micronomous norfolkensis and Chalinolobus dwyeri, are all known to be either rare or reliant on caves or water bodies, and so it was deemed that further sampling would not have detected these species.

44 Environmental variables among light treatments

The average percentage vegetation within a 200m radius significantly differed among light

treatments (F (3,10) = 4.37, p = 0.049, Table 2.4), with bushland sites having the highest amount

of vegetation compared with sites with LED lights, sites with mercury vapour lights and changeover sites.

Lux differed among light treatments (F (3,14) = 53.5, p < 0.001), with bushland being the darkest

(n = 2, mean = 0.47), compared with LED lights (n = 2, mean = 15, se = 0.97), mercury vapour

lights (n = 2, mean = 10.3, se = 0.78) or at changeover sites (n = 2, mean = 12.45, se = 1.45). Post hoc tests also showed that lux did not change significantly at the changeover sites after the change in light type (p > 0.05), but lux did differ significantly between mercury vapour and LED treatments (p = 0.002).

Insect biomass, light treatments and the moon

Insect biomass was unaffected by light treatment (F (3,20) = 0.68, p > 0.05) and the interaction

between light treatment and sampling period (before and after the change in light type) was

also not significant (F (3,20) = 2.18, p > 0.05). Insect biomass was significantly affected by the percentage of the moon illuminated (F (3,20) = 8.17, p < 0.001).

Activity of Chalinolobus gouldii among treatments

Of the 6118 calls identified to species or species level, 62.8% were C. gouldii. This species was

significantly affected by relative humidity (F(3,183) = 320.67, p < 0.001) and the interaction effect

45 Table 2.3. Posthoc pairwise contrasts for each generalised linear mixed model, including coefficients and confidence intervals, with ‘after’ as the reference category, where a positive contrast estimate indicates a decrease in the response variable from before to after the change in light type, and a negative contrast estimate indicates an increase in the response variable

Contrast 95% Confidence Interval Model term Light Treatment Pairwise Contrast Estimate Std. Error t df Adj. Sig. Lower Upper Chalinolobus gouldii changeover before versus after 2.598 .107 24.302 183 < 0.001 2.387 2.809 MV before versus after -0.987 .074 -13.369 183 < 0.001 -1.133 -0.842 bushland before versus after 0.907 .074 12.301 183 < 0.001 .761 1.052 led before versus after -0.428 .077 -5.519 183 < 0.001 -0.581 -0.275 Miniopterus orianae oceanensis changeover before versus after -1.874 0.305 -6.150 183 < 0.001 -2.475 -1.273 MV before versus after -0.996 0.133 -7.481 183 < 0.001 -1.258 -0.733 bushland before versus after 1.133 0.197 5.758 183 < 0.001 0.744 1.521 led before versus after -0.125 0.284 -0.441 183 0.660 -0.685 0.435 Light avoidant bats changeover before versus after 1.869 0.691 2.704 182 0.008 0.505 -3.232 MV before versus after -0.342 0.791 -0.432 182 0.666 -1.902 1.218 bushland before versus after -0.903 0.551 -1.640 182 0.103 -1.990 0.184 led before versus after -0.707 0.551 1.283 182 0.201 -1.794 -0.380 Insect biomass changeover before versus after 0.660 1.777 -0.371 20 0.714 -4.367 -3.047 MV before versus after -1.276 0.964 -1.324 20 0.201 -3.287 0.735 bushland before versus after -2.099 1.265 -1.659 20 0.113 -4.738 0.540 led before versus after 0.129 0.645 -0.199 20 0.844 -1.217 1.474

was significant (F (3,183) = 320.67, p < 0.001). Post hoc comparisons showed that C. gouldii

activity at the changeover sites and at bushland sites was higher before when compared after

the change in light (Table 2.3). At mercury vapour and LED lit sites , C. gouldii activity was

lower before the change in light type and then increased after the change (Table 2.3).

Activity of Miniopterus orianae oceanensis among treatments

Identified calls of this species made up 11.8% of all identified calls. This species was

significantly affected by relative humidity (F (1,183) = 49.24, p < 0.001) and the interaction was

significant (F (1,183) = 12.75, p < 0.001). Post hoc comparisons showed that M. orianae oceanensis activity at changeover sites and bushland sites was lower before the change in light type (Table

2.3) and then increased. Miniopterus orianae oceanensis activity at mercury vapour sites was higher before the change in light type, and then decreased after the change (Table 2.3).

Clutter-adapted and ESH bats

Bats categorised as within our response group, slow flying and ESH species, were rarely detected, making up only 2.05 % of all identified calls. The activity of this functional group significantly differed among light treatments (F (2,183) = 14.03, p < 0.001), with dark bushland

supporting higher activity (Table 2.3). The activity of the clutter-adapted/ESH response group

was significantly affected by the percentage moon illuminated (F (1,183) = 9.42, p = 0.002). The

interaction between light treatment and before/after the change in light type was significant

(Figure 2.3, F (2,183) = 5.49, p = 0.024), with activity remaining consistent at all light treatments

except for the changeover sites, where activity significantly decreased (Table 2.3).

adapted

- bat passes or ESH verage number of clutter of number verage A

Before After Before After Before After Before After

Changeover Mercury vapour Bushland LED

Figure 2.3. Boxplot showing average number of bat passes recorded from clutter-adapted or ESH species; with bushland supporting significantly higher activity than other light treatments, activity significantly decreasing at the changeover sites after the change to LED, and activity significantly increasing at the bushland sites after the change in lights at the changeover sites. Crosses depicted mean value for the treatment.

Differences in bat communities among light treatments

Composition of bat communities was not significantly different before and after the change in

light type at the changeover sites (Figure 2.4, pseudo F4,21 = 1.5, p > 0.05). When comparing

light treatments (pooled across the before and after the light change), bat communities in

bushland habitat were significantly different to mercury vapour sites (Figure 2.4, pseudo F4,21 =

2.08, p = 0.036) and different to LED sites (Figure 2.4, pseudo F4,21 = 2.23, p = 0.022). These

differences were driven by both the higher activity of faster flying species, C. gouldii and M.

48 orianae oceanensis, at the lit sites, and the absence of the slower flying, clutter-adapted species group Nyctophilus spp. from any of the lit sites.

Figure 2.4. Non-metric multidimensional scaling plot of bat activity data showing that the composition of the bat assemblage at the bushland sites differed to the composition of the bat assemblage in the lit matrix sites. Points represent the bat assemblage for each site either before or after the change in light type. Changeover B represents the bat assemblage at changeover sites before the change in light type, and Changeover A represents the bat assemblage at changeover sites after the change in light type.

Discussion

Our results show that changing street lights from mercury vapour to LED decreased the

activity of our response group, including species adapted to cluttered vegetation and species

adapted to edges of habitat with relatively high-frequency echolocation calls. This decrease in

clutter-adapted and ESH bat activity was related to the installation of LED street lights and

potentially a decrease in UV radiation or an increase in brightness of lighting.

49 Dark urban bushland as refuge for high urban bat diversity

The bat community in the urban bushland significantly differed from the community in the

urban matrix; fast flying species such as C. gouldii and M. orianae oceanensis dominated the lit areas of the urban matrix, while the clutter-adapted/ESH response group was significantly more active in dark bushland remnants. Our results confirm that changing to LED street lights causes at least a short term decrease in clutter-adapted/ESH species, and that dark bushland is important habitat. The activity of the clutter-adapted/ESH response group (Nyctophilus spp.,

Rhinolophus megaphyllus, Vespadelus vulturnus, Chalinolobus morio and Miniopterus australis) was significantly higher in dark bushland sites. We did not have an opportunity to survey dark streets as a procedural control in this experiment, and so we are unable to state that these bats were avoiding artificial light or whether this effect is due to habitat preference. Edge-space foraging and clutter-adapted species are positively influenced by vegetation density (Threlfall et al. 2013, Suarez‐Rubio et al. 2018). From previous research (Threlfall et al. 2013) and our findings, it is apparent that artificial lighting as well as vegetation density plays a role in habitat preference for bats, but further work is needed to disentangle the two driving factors.

However, dark urban bushland seems to be important for the persistence of both slower flying, clutter-adapted bats and edge-space foraging species with higher characteristic-frequency echolocation calls. Slower flying, clutter-adapted bats may have evolved to emerge in the darkest part of the night (Jones and Rydell 1994) and to specialize in hunting insects in cluttered vegetation, along habitat edges, and in small dark open spaces (Brigham et al. 1997).

We support Rydell’s (1992) suggestion that slow flight and a gleaning foraging strategy may be morphological constraints, rendering these species less able either to exploit insects around

50 street lights (Jones and Rydell 1994, Rydell et al. 1996) or to evade visual predators that may

use lit urban areas as hunting grounds (Speakman 1991), although direct evidence for the latter

is still lacking. Slower flying, clutter-adapted species are most at risk of extinction (Jones et al.

2003, Safi and Kerth 2004) due to rapid habitat fragmentation and destruction, and are at

significant risk from increased artificial light pollution. Our findings also highlight however,

that edge-spaced foraging species with high characteristic-frequency echolocation calls may

also be negatively affected by changing to LED lights and a loss of dark habitat. Further

research is needed to disentangle the importance of habitat type and artificial light levels for

these species and understand species-specific responses, as this group may be (Adams et al.

2005) at risk of increasing light pollution in cities.

Bat responses to different light types and light changes

Our data suggest that different street light types do not support significantly different overall

bat assemblages. Contrary to our predictions and previous research (Lewanzik and Voigt 2017),

mercury vapour street lights high in UV radiation did not support higher overall bat activity, higher activity of faster flying bats or higher insect biomass than low UV LED street lights.

However, changing the light type at a site was linked to some immediate species- and group-

specific responses within the bat assemblage. We measured a decrease in the activity of

clutter-adapted/ESH response group after the installation of low-UV LED lights. This decrease in activity did not mirror the availability of insect prey we recorded however, which did not respond to the change to LED lights. However, insect biomass was negatively correlated with the percentage of the moon illuminated over the course of this experiment (Bishop et al. 2000,

51 Guimarães et al. 2000). Whatever was driving the change in insect biomass, this provides

evidence that neither C. gouldii, M. orianae oceanensis nor the clutter-adapted/ESH group activity were tracking insect availability as none showed the same activity patterns as the insects. The bats may be responding to something other than a change in insect biomass; perhaps instead the increased intensity of the LED lights or a change in insect species composition. Activity of M. orianae oceanensis increased in the second half of the experiment, and particularly at changeover sites and bushland sites. This species is known to migrate out of

Sydney to maternity colonies in the summer months (Gonsalves and Law 2017), and return in autumn. The increased activity that we recorded across all light treatments may be attributed to adults and juveniles returning to the city following the autumn dispersal (Dwyer 1963).

Further, the increases in M. orianae oceanensis activity at both the changeover sites after the lights changed to LED, and the bushland sites, suggest that this species is not influenced by light type but is influenced by wider landscape variables that we did not measure in this experiment, like proximity to an appropriate subterranean roost.

Recommendations for urban planning and future research

This study highlights the importance of protecting bushland remnants as refuges for clutter- adapted and ESH bats. Small remnants (<40 ha) have a positive conservation value for highly light avoidant species (Threlfall et al. 2012, Threlfall et al. 2013), but protecting their low ambient light levels may be important for more bat species to forage and commute. We do not know the impact of artificial lighting on the quality of remaining bushland patches. Penetration of light into bushland edges is known to alter mammal, vegetation and bird diversity (Pocock

52 and Lawrence 2005). Future work should focus on the effect of light penetration into remnant

patches, including narrow corridors, for clutter adapted and ESH species, and devise ways to

maximize the conservation value of this remaining habitat with ecologically sensitive public

lighting policies. Research should assess long term and city-wide surveys after spectra changes

in streetlights. Time constraints and the staggered nature of the installation of LED street

lights across Sydney meant that we could only survey a limited number of sites where lights

were being changed. Future research could focus on longer-term studies that examine many different changeovers at various times. It remains a priority to partner with relevant government stakeholders and installation companies to conduct studies investigating the impact of lighting on nocturnal fauna when entire neighborhoods have their street lights changed. Longer term surveying, both before and after changes in lighting, could also uncover long term patterns that were undetectable in this study. For cities to conserve high nocturnal species diversity, the conservation of dark urban bushland should be considered a research and planning priority, and considered when designing street light upgrade projects.

References

Adams, M. D., B. S. Law and M. S. Gibson (2010). "Reliable automation of bat call identification for eastern New South Wales, Australia, using classification trees and AnaScheme software." Acta Chiropterologica 12(1): 231-245.

53 Arlettaz, R., S. Godat and H. Meyer (2000). "Competition for food by expanding pipistrelle bat populations (Pipistrellus pipistrellus) might contribute to the decline of lesser horseshoe bats (Rhinolophus hipposideros)." Biological Conservation 93(1): 55-60.

Ausgrid (2013). LED streetlight trial; Category P LED Luminarie Trial [White Paper], NSW Electricity Provider.

Avila-Flores, R. and M. B. Fenton (2005). "Use of spatial features by foraging insectivorous bats in a large urban landscape." Journal of Mammalogy 86(6): 1193-1204.

Baagøe, H. (1986). "Summer occurrence of Vespertilio murinus and Eptesicus serotinus on Zealand, Denmark, based on records of roost and registrations with bat detectors." Annalen des Naturhistorischen Museums Wien 88(89): 281-291.

Bader, E., K. Jung, E. K. Kalko, R. A. Page, R. Rodriguez and T. Sattler (2015). "Mobility explains the response of aerial insectivorous bats to anthropogenic habitat change in the Neotropics." Biological Conservation 186: 97-106.

Basham, R., B. Law and P. Banks (2010). "Microbats in a leafy urban landscape: are they persisting, and what factors influence their presence?" Austral Ecology 36(6), 663-678.

Bennie, J., T. W. Davies, D. Cruse and K. J. Gaston (2016). "Ecological effects of artificial light at night on wild plants." Journal of Ecology 104(3): 611-620.

Benson, D. and J. Howell (1990). Sydney’s vegetation 1788–1988: utilization, degradation and rehabilitation. Proceedings of the Ecological Society of Australia (Vol. 16, No. 1, pp. 115-27). Benson, D. H., J. Howell (1995). Taken for granted: the bushland of Sydney and its suburbs, Kangaroo Press in association with the Royal Botanic Gardens Sydney.

Bishop, A., H. McKenzie, I. Barchia and L. Spohr (2000). "Moon phase and other factors affecting light‐trap catches of Culicoides brevitarsis Kieffer (Diptera: Ceratopogonidae)." Austral Entomology 39(1): 29-32.

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

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

Chan, C. K. and X. Yao (2008). "Air pollution in mega cities in China." Atmospheric Environment 42(1): 1-42.

Churchill, S. (2008). "Australian Bats., 2nd edn. (Allen and Unwin: Sydney)." Cohen, J. E. (2003). "Human population: the next half century." Science 302(5648): 1172-1175.

54 Corben, C. (2004). "Zero-crossings analysis for bat identification: an overview." Bat Echolocation Research: Tools, Techniques, and Analysis, RM Brigham, EKV Kalko, G. Jones, S. Parsons, and HJGA Limpens, eds. Austin, TX: Bat Conservation International: 95-107.

Denzinger, A. and H. U. Schnitzler (2013). "Bat guilds, a concept to classify the highly diverse foraging and echolocation behaviors of microchiropteran bats." Frontiers in Physiology 4, 164.

Dwyer, P. (1963). "The breeding biology of Miniopterus schreibersi blepotis (Termminck) (Chiroptera) in north-eastern NSW." Australian Journal of Zoology 11(2): 219-240.

Eisenbeis, G. and K. Eick (2010). Attraction of nocturnal insects to street lights with special regard to LEDs. Society for Conservation Biology, 24th Annual Meeting:< http://www. conbio. org/activities/meetings/2010/images/abstracts. pdf>(checked 10.04. 11).

Fischer, J., J. Stott, B. S. Law, M. D. Adams and R. I. Forrester (2009). "Designing effective habitat studies: quantifying multiple sources of variability in bat activity." Acta Chiropterologica 11(1): 127-137.

Fuller, R. A., K. N. Irvine, P. Devine-Wright, P. H. Warren and K. J. Gaston (2007). "Psychological benefits of greenspace increase with biodiversity." Biology Letters 3(4): 390- 394.

Gaston, K. J. (2013). "Sustainability: A green light for efficiency." Nature 497(7451): 560-561.

Gonsalves, L. and B. Law (2017). "Seasonal activity patterns of bats in North Sydney, New South Wales: implications for urban bat monitoring programs." Australian Mammalogy 40, 220-229.

Gorenzel, W. P. and T. P. Salmon (1995). "Characteristics of American crow urban roosts in California." The Journal of Wildlife Management: 638-645.

Gorresen, P. M., P. M. Cryan, D. C. Dalton, S. Wolf and F. J. Bonaccorso (2015). "Ultraviolet vision may be widespread in bats." Acta Chiropterologica 17(1): 193-198.

Guimarães, A. É., C. Gentile, C. M. Lopes and R. P. d. Mello (2000). "Ecology of mosquitoes (Diptera: Culicidae) in areas of Serra do Mar State Park, State of São Paulo, Brazil. III-daily biting rhythms and lunar cycle influence." Memórias do Instituto Oswaldo Cruz 95(6): 753-760.

Harewood, A. and J. Horrocks (2008). "Impacts of coastal development on hawksbill hatchling survival and swimming success during the initial offshore migration." Biological Conservation 141(2): 394-401.

Hölker, F., C. Wolter, E. K. Perkin and K. Tockner (2010). "Light pollution as a biodiversity threat." Trends in Ecology & Evolution(25): 681-682.

55 Ives, C. D., P. E. Lentini, C. G. Threlfall, K. Ikin, D. F. Shanahan, G. E. Garrard, S. A. Bekessy, R. A. Fuller, L. Mumaw and L. Rayner (2016). "Cities are hotspots for threatened species." Global Ecology and Biogeography 25(1): 117-126.

Jones, K. E., A. Purvis and J. L. Gittleman (2003). "Biological correlates of extinction risk in bats." American Naturalist: 601-614.

Jung, K. and E. K. Kalko (2010). "Where forest meets urbanization: foraging plasticity of aerial insectivorous bats in an anthropogenically altered environment." Journal of Mammalogy 91(1): 144-153.

Jung, K. and E. K. Kalko (2011). "Adaptability and vulnerability of high flying Neotropical aerial insectivorous bats to urbanization." Diversity and Distributions 17(2): 262-274.

Jung, K. and C. G. Threlfall (2016). “Urbanisation and Its Effects on Bats - A Global Meta- Analysis. Bats in the Anthropocene: Conservation of bats in a changing world.” C. Voigt and T. Kingston. Cham Heidelberg, Springer: 13-33.

Lacoeuilhe, A., N. Machon, J.-F. Julien, A. Le Bocq and C. Kerbiriou (2014). "The Influence of Low Intensities of Light Pollution on Bat Communities in a Semi-Natural Context." PloS ONE 9(10): e103042.

Law, B., J. Anderson and M. Chidel (1998). "A bat survey in state forests on the south-west slopes region of New South Wales with suggestions of improvements for future surveys." Australian Zoologist 30(4): 467-479.

Law, B. S., L. Reinhold and M. Pennay (2002). "Geographic variation in the echolocation calls of Vespadelus spp.(Vespertilionidae) from New South Wales and Queensland, Australia." Acta Chiropterologica 4(2): 201-215.

Lewanzik, D. and C. C. Voigt (2017). "Transition from conventional to light‐emitting diode street lighting changes activity of urban bats." Journal of Applied Ecology 54(1): 264-27

Mathews, F., N. Roche, T. Aughney, N. Jones, J. Day, J. Baker and S. Langton (2015). "Barriers and benefits: implications of artificial night-lighting for the distribution of common bats in Britain and Ireland." Philosophical Transactions of the Royal Society of London B: Biological Sciences 370(1667): 20140124.

56 Milne, D., M. Armstrong, A. Fisher, T. Flores and C. Pavey (2004). "A comparison of three survey methods for collecting bat echolocation calls and species-accumulation rates from nightly Anabat recordings." Wildlife Research 31(1): 57-63.

Moore, M. V., S. M. Pierce, H. M. Walsh, S. K. Kvalvik and J. D. Lim (2000). "Urban light pollution alters the diel vertical migration of Daphnia." Internationale Vereinigung für theoretische und angewandte Limnologie: Verhandlungen 27(2): 779-782.

Neuweiler, G. (1984). "Foraging, echolocation and audition in bats." Naturwissenschaften 71(9): 446-455.

Norberg, U. M. and J. M. Rayner (1987). "Ecological morphology and flight in bats (Mammalia; Chiroptera): wing adaptations, flight performance, foraging strategy and echolocation." Philosophical Transactions of the Royal Society B: Biological Sciences 316(1179): 335-427.

Ortega, C. P. (2012). "Effects of noise pollution on birds: a brief review of our knowledge." Ornithological Monographs 74(1): 6-22.

Park, J.-H. and H.-S. Lee (2017). "Phototactic behavioral response of agricultural insects and stored-product insects to light-emitting diodes (LEDs)." Applied Biological Chemistry, 60 (2), 137-144. Patriquin, K. J., L. K. Hogberg, B. J. Chruszcz and R. M. Barclay (2003). "The influence of habitat structure on the ability to detect ultrasound using bat detectors." Wildlife Society Bulletin: 475-481.

Pawson, S. and M.-F. Bader (2014). "LED lighting increases the ecological impact of light pollution irrespective of color temperature." Ecological Applications 24(7): 1561-1568.

Pennay, M., B. Law and L. Reinhold (2004). "Bat Calls of NSW: region based guide to the echolocation calls of microchiropteran bats. NSW Dept." Environment and Conservation, Hurstville.

Rich, C. and T. Longcore (2013). “Ecological consequences of artificial night lighting” Island Press.

Rhodes, M. P. (2002). "Assessment of sources of variance and patterns of overlap in microchiropteran wing morphology in southeast Queensland, Australia." Canadian Journal of Zoology 80(3): 450-460.

Rowse, E., D. Lewanzik, E. Stone, S. Harris and G. Jones (2016). “Dark matters: the effects of artificial lighting on bats.” Bats in the Anthropocene: conservation of bats in a changing world, Springer: 187-213.

Rowse, E. G., S. Harris and G. Jones (2016). "The Switch from Low-Pressure Sodium to Light Emitting Diodes Does Not Affect Bat Activity at Street Lights." PloS ONE 11(3): e0150884.

57 Rydell, J. (1992). "Exploitation of insects around streetlamps by bats in Sweden." Functional Ecology: 744-750.

Rydell, J., A. Entwistle and P. A. Racey (1996). "Timing of foraging flights of three species of bats in relation to insect activity and predation risk." Oikos: 243-252.

Safi, K. and G. Kerth (2004). "A comparative analysis of specialization and extinction risk in temperate‐zone bats." Conservation Biology 18(5): 1293-1303.

Sample, B. E., R. J. Cooper, R. D. Greer and R. C. Whitmore (1993). "Estimation of insect biomass by length and width." American Midland Naturalist: 234-240.

Speakman, J. (1991). "The impact of predation by birds on bat populations in the British Isles." Mammal Review 21(3): 123-142.

Stone, E. L., S. Harris and G. Jones (2015). "Impacts of artificial lighting on bats: a review of challenges and solutions." Mammalian Biology, 80(3), 213-219.

Stone, E. L., G. Jones and S. Harris (2009). "Street lighting disturbs commuting bats." Current Biology 19(13): 1123-1127.

Stone, E. L., G. Jones and S. Harris (2012). "Conserving energy at a cost to biodiversity? Impacts of LED lighting on bats." Global Change Biology 18(8): 2458-2465.

Stone, E. L., A. Wakefield, S. Harris and G. Jones (2015). "The impacts of new street light technologies: experimentally testing the effects on bats of changing from low-pressure sodium to white metal halide." Philosophical Transactions of the Royal Society of London B: Biological Sciences 370(1667): 20140127.

Suarez‐Rubio, M., C. Ille and A. Bruckner (2018). "Insectivorous bats respond to vegetation complexity in urban green spaces." Ecology and Evolution 8(6): 3240-3253.

Taylor, L. and D. F. Hochuli (2015). "Creating better cities: how biodiversity and ecosystem functioning enhance urban residents’ wellbeing." Urban Ecosystems 18(3): 747-762.

58 Threlfall, C., B. Law, T. Penman and P. B. Banks (2011). "Ecological processes in urban landscapes: mechanisms influencing the distribution and activity of insectivorous bats." Ecography 34(5): 814-826.

Threlfall, C. G., B. Law and P. B. Banks (2012). "Influence of landscape structure and human modifications on insect biomass and bat foraging activity in an urban landscape." PloS ONE 7(6): 10.

Threlfall, C. G., B. Law and P. B. Banks (2012). "Sensitivity of insectivorous bats to urbanization: implications for suburban conservation planning." Biological Conservation 146(1): 41-52.

Threlfall, C. G., B. Law and P. B. Banks (2013). "Roost selection in suburban bushland by the urban sensitive bat Nyctophilus gouldi." Journal of Mammalogy 94(2): 307-319.

Trindade-Filho, J., R. A. de Carvalho, D. Brito and R. D. Loyola (2012). "How does the inclusion of Data Deficient species change conservation priorities for amphibians in the Atlantic Forest?" Biodiversity and Conservation 21(10): 2709-2718.

Van der Ree, R. and M. A. McCarthy (2005). "Inferring persistence of indigenous mammals in response to urbanisation." Animal Conservation 8(03): 309-319. van Grunsven, R. H., M. Donners, K. Boekee, I. Tichelaar, K. Van Geffen, D. Groenendijk, F. Berendse and E. Veenendaal (2014). "Spectral composition of light sources and insect phototaxis, with an evaluation of existing spectral response models." Journal of Insect Conservation 18(2): 225-231. van Langevelde, F., J. A. Ettema, M. Donners, M. F. WallisDeVries and D. Groenendijk (2011). "Effect of spectral composition of artificial light on the attraction of moths." Biological Conservation 144(9): 2274-2281.

Wakefield, A., E. L. Stone, G. Jones and S. Harris (2015). "Light-emitting diode street lights reduce last-ditch evasive manoeuvres by moths to bat echolocation calls." Royal Society Open Science 2(8): 150291.

Wolch, J. R., J. Byrne and J. P. Newell (2014). "Urban green space, public health, and environmental justice: The challenge of making cities ‘just green enough’." Landscape and Urban Planning 125: 234-244.

59 Chapter 3: Light pollution at the urban forest edge and its impact on insectivorous bats

This paper is published in Biological Conservation as “Haddock, J. K., Threlfall, C. G., Law, B. S.

and Hochuli, D. F. (2019) Light pollution at the urban forest edge negatively impacts

insectivorous bats. Biological Conservation 236, 17-28”

Abstract

Connectivity and quality of vegetation in cities, including urban forests, can promote urban biodiversity. However the impact of anthropogenic pressures at the forest-matrix edge, particularly artificial light pollution, on connectivity has received little attention. We assessed the influence of artificial light at forest edges on 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 31 urban forest interiors. We also sampled the flying insect community to establish whether changes in insect densities under lights drive changes in insectivorous bat activity. We recorded 9965 bat passes from 16 species or species groups throughout our acoustic survey. The activity 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 flying bats 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

60 sources at the forest edge. The time of first recording of Vespadelus vulturnus was also

significantly delayed by the presence of streetlights at the forest edge. Conversely, generalist faster flying bats; Chalinolobus gouldii, Ozimops ridei, Austronomous australis, Saccolaimus flaviventris, 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 have diverse effects on bats and insects, and should be considered an anthropogenic edge effect that can reduce available functional habitat for light- avoidant species.

Introduction

Urbanisation is one of the leading causes of biodiversity loss worldwide (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 water quality

(Blakey et al. 2018), reduced vegetation cover and structure (Threlfall et al. 2016, Threlfall et al.

2017) and artificial light (Hölker et al. 2010) all contribute to degrading natural habitat for

urban wildlife. Although only taking up a small percentage of the planet’s terrestrial surface,

urbanised areas are predicted to grow by 1.2 million km2 by 2030, impacting many global biodiversity hotspots (Seto et al. 2012). Hence, investment in the conservation of urban biodiversity is essential for many reasons (Dearborn and Kark 2010).

61 Providing “stepping stones and corridors” of forests and native vegetation is a commonly

suggested conservation action for urban biodiversity (Dearborn and Kark 2010, Beninde et al.

2015). Connecting forest habitat in cities can mitigate species loss and biotic homogenisation

(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

degraded urban landscape. Over the last few decades, research has focused on quantifying the

characteristics of habitat remnants that maximize their value to native plants and animals,

including their size (Evans et al. 2009), shape (Hawrot and Niemi 1996), and connectedness

(Keitt et al. 1997, McGarigal et al. 2002, Beninde et al. 2015). Calculating the amount of viable

habitat existing in urban areas is a more complex process. 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.

Anthropogenic pressures, such as artificial light, may impact functional connectivity (Hale et al.

2012) by narrowing wildlife corridors and reducing functional patch sizes, however, the extent

to which this occurs is poorly understood currently.

Light pollution is a growing problem, escalating by 6 % each year (Hölker et al. 2010). It is caused by illumination from anthropogenic lighting, and is most prevelant in urban areas

(Falchi et al. 2016). Only relatively recently has light pollution been widely discussed as a global threat to biodiversity (Rich and Longcore 2013, Gaston et al. 2015). Street lights disrupt migration patterns (La Sorte et al. 2017), breeding cycles (Navara and Nelson 2007) and predator-prey interactions (Gorenzel and Salmon 1995). Street lights may also have an effect

62 on the functional value and connectivity of proximal urban forests and vegetation. Street lights

positioned along edges of urban wildlife corridors negatively affect some species whilst

attracting others (Azam et al. 2018). The effects of artificial light at the forest edge for

nocturnal wildlife may be significant; light may penetrate dark vegetation anywhere from 50 m

(Kempenaers, Borgström et al. 2010) to 380 m (Pocock and Lawrence 2005). Dark habitats are

currently at risk from the edge effects of light pollution.

Insectivorous bats are an ecologically diverse group, and respond in a variety of ways to

urbanisation (Russo and Ancillotto 2015). Some faster flying open-space adapted bat species

find roosts in buildings (Kunz 1982) and can commute across urbanised landscapes (Jung and

Kalko 2011). Conversely, other slower flying clutter-adapted bat species commonly avoid

urban areas, they cannot adapt to changes in roost avaliability and instead are reliant on

networks of urban forest to survive in cities (Basham et al. 2011). Artficial light is one

anthropogenic pressure driving these diverse responses to urban areas. Ultraviolet radiation

attracts high numbers of insects (van Grunsven et al. 2014) and could offer urban feeding

grounds for faster flying bat species adapted 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 that use the lit areas as hunting grounds,

although this is not yet established. This successful exploitation of the lit environment by some

faster flying bats may also lead to competitive exclusion of other species (Arlettaz et al. 2000),

although this too needs more investigation. Conversely, slower flying clutter-adapted species often avoid crossing lits areas on route to commuting grounds (Stone et al. 2012) and may be

63 constrained by lights at the patch edge, spending a majority of their foraging time within dark

patches (Threlfall et al. 2013, Hale et al. 2015), and therefore reducing their functional urban

habitat. Artificial light also markedly delays some species’ emergence times from roosts (Jones

and Rydell 1994, Downs et al. 2003, Boldogh et al. 2007), meaning that their foraging time is

reduced and the health of the population may be at risk (Boldogh et al. 2007), although for

some species there could be benefits to an increased number of insect prey around

streetlights. There is a global pattern emerging that light pollution negatively affects species adapted to foraging in either cluttered vegetation or along habitat edges (Chapter 2, Stone et

al. 2015, Azam et al. 2018, Rowse et al. 2018). The mechanistic drivers of this light phobia in bats are not completely understood (Stone et al. 2015, Rowse et al. 2016) but could include predator avoidance (Speakman 1991, Stone et al. 2009, Lima and O'Keefe 2013), morphology that leaves slower flying bats less able exploit airborne insect prey at lights (Haffner and Stutz

1985, Rydell and Racey 1995, Rydell 2006), sensitivity to ultraviolet radiation emitted by some street lights (Gorresen et al. 2015), or a combination of these. Bats with many of these traits are declining in cities (Jung and Threlfall 2016, Jung and Threlfall in press) hence research on the impact of public lighting on this group is urgently required.

We hypothesised that permanent streetlights along the forest edge would reduce the activity of some insectivorous bats due to a decline in functional habitat . We predicted that the activity of bats with slow flight speed would be lower at forest edges with street lights than edges with no lights. We also predicted that the activity of faster flying, light exploiting bats would either be unaffected or be higher at edges with street lights than edges with no lights.

64 Finally, we predicted that total bat activity, and the activity of bats with slow flight speed,

would both be higher in the forest interiors than the forest edges.

Method

Site selection

Our survey was carried out in Greater Sydney, a sub-tropical city with a population of over 4.5

million, on the east coast of New South Wales, Australia. Sydney has much remnant native forest abutting the urban matrix, both in continuous national parks and in smaller isolated patches surrounded by housing (Benson et al. 1995). Two main geologies in the Sydney region, shale and sandstone, have led to different levels of primary productivity and soil fertility across the city, and these have been shown to influence insect prey and bat diversity in this region

(Threlfall et al. 2011, Threlfall et al. 2012). Hence, we only included sites on full or majority sandstone with transitional soil type to control for this effect of geology.

Experimental design

We conducted an acoustic survey along 31 forest edge sites (Figure 3.1); 16 of these sites were at the edges of connected forest (forest connected to large natural areas, Figure 3.2a), and 15

were on the edges of isolated forest patches (Figure 3.2b), >30 ha in size, but surrounded on all

sides by urban matrix. Of all connected patch edge sites, half the sites (n=8) had 80W mercury

vapour street lights along the edges (light intensity of 10.65 lux ± 1.89) and half of the sites

(n=7) were dark (light intensity of 0.52 lux ± 0.34). Of all forest patch edge sites, half the sites

(n=8) had 80W mercury vapour street lights along the edges (light intensity of 10.20 lux ± 1.23)

65 and half of the sites (n=8) were dark (light intensity of 0.97 lux ± 0.52). All edge sites were defined as roads or pathways over 4m in width with dense vegetation on only one side. Dark edges were defined as 30m of uninterrupted dark conditions (artificial light levels comparable to ambient darkness and no external light sources on nearby houses). Artificially lit edges were defined as 30m of 80W mercury vapour lights along one side of the road, with the street lights at a median distance apart of 12.1m. Usual low-level urban lighting from houses in the

Figure 3.1. Map showing location and distribution of forest patch sites (red triangles; n=15 pairs of control interior and edge sites) and connected forest sites (yellow circles; n=16 pairs of control interior and edge sites) relative to Sydney, Australia (inset). Scale bar is in kilometres.

surrounding matrix was present at both dark and light edges. We additionally located control

sites in the dark interior of all 31 sites (lux of 0.55 ± 0.20), which were located along an internal

edge comprising a pathway or track (2 m typical width) between 400 m and 1 km perpendicular

from the sampled forest edge. This allowed us to sample the bat assemblage at 31 edge sites,

a) b)

0 200

Figure 3.2. 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 marker showing 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. Scale bar is in metres.

and 31 dark interior control sites, leading to 62 sites in total. We sampled between November

and December of 2016, the maternity season for bats when resource requirements and activity

levels are highest. The light intensity (lux) at each site was measured using a lux meter

(QM1587; Reduction Revolutions Pty Ltd, Parramatta, Australia). We also measured insect

biomass at all forest patches, at both the interior and edge site, using a black light intercept

traps (Australian Entomological Supplies, Murwillumbah, Australia, see Appendix).

67 Bat recording

The acoustic survey was carried out using Anabat II detectors (Titley Electronics, Ballina,

Australia) placed on the ground with the high frequency microphone positioned 1 m above the

ground and pointing upwards at a 45° angle to record all echolocation calls of passing bats. We used one detector per site, and predicted to still record high flying bats to be recorded due to louder calls from those species (Jung and Kalko 2011). At artificially lit edges the detector was

placed no further than 3 m from the base of the street light pole, no more than 3 m away from

the edge of the forest. At both the edge and interior sites, the microphone was pointed parallel

down the forest edge and down the interior track or path, respectively, to optimise the amount

of habitat sampled (Law et al. 1998, Threlfall et al. 2012). Each of the 62 sites were sampled

between 3 – 5 consecutive nights (average 4.35 nights ± s.e. 0.19, due to unexpected

equipment failure we did not manage to record for 5 nights at each site), totaling 270 recording

nights across the entire survey. We cycled through the sites in time periods of 3 – 5 consecutive

nights due to limited bat detectors. A maximum of eight sites were surveyed on the same

night, where the edge and its respective interior control were concurrently sampled. The

detectors passed the data through a Zero-Crossings Interface Module (ZCAIM; Titley

Electronics). The acoustic files collected during the survey were processed using Anascheme

and a bat call identification key developed for Sydney bats (Adams et al. 2010). For a bat call to

be identified to species level, three or more pulses were required and have characteristics that

fall within the program’s parameters for that species. A pass was defined as at least three valid

pulses with a minimum of 6 pixels per pulse. Successful species identifications were made only

when a minimum of 50% of pulses within a pass were identified as the same species (Adams et

68 al. 2009, Threlfall et al. 2012). The calls identified as Chalinolobus dwyeri, Falsistrellus tasmaniensis, Nyctophilus spp., Saccolaimus flaviventris, Scoteanex rueppellii and Scotorepens orion are known to be complex or rare and were manually checked against known parameters, and confirmed or re-identified (Adams et al. 2010). In addition, calls were run through a species filter in AnaScheme which is specifically designed to identify calls of Chalinolobus gouldii with alternating frequency. The call characteristics of Nyctophilus gouldi and Nyctophilus geoffroyi are indistinguishable using the AnaScheme method and so were pooled as one taxon;

Nyctophilus spp. Only species that were positively identified using the key, filters and manual checking were included for further analysis to eliminate any bias caused by using partially identified species.

Assignment to functional groups

Assignment of each species to a functional group was based on morphological traits and foraging styles (Chapter 2, Rhodes 2002). Chalinolobus gouldii, Ozimops ridei, Austronomous australis, Saccolaimus flaviventris, and Miniopterus orianae oceanensis were categorised as light exploiting due to their open-space foraging area preference, or edge-space foraging area preference with low or medium echolocation call frequency (Rhodes 2002). Light exploiting bats may tend to use the well-lit areas to forage around or commute past. The light-avoidant group consisted of both maneuverable edge-space foraging species with high echolocation call frequency (Chalinolobus morio, Miniopterus australis, Vespadelus vulturnus; (Adams et al. 2009) and also species adapted to cluttered vegetation with a slower flight pattern (Rhinolophus megaphyllus and Nyctophilus spp., consisting of two species Nyctophilus gouldi and Nyctophilus

69 geoffroyi that have echolocation calls indistinguishable from the other) that may avoid lit

areas. We based these groupings on known response groups found in previous studies,

however we acknowledge that there may be species that do not necessarily fit this trend.

Measurement of environmental variables

The percentage of native vegetation cover (>3 m tall) within a radius of 250m of each site was calculated using Arc Map (ESRI, Redlands, USA, ver. 10.2) and was calculated through intersecting GPS points of our sites with the GIS layer ‘The Native Vegetation of the Sydney

Metropolitan Area - Version 3, VIS_ID 4489’ (NSW Office of Environment and Heritage,

Sydney). This area has been found to be indicative of local habitat as shown in previous studies

(Lacoeuilhe et al. 2014). The same method was used to calculate the percentage of sandstone

and transitional sandstone soil type within the 250m radius of each site. Average moon

illumination was calculated by noting the percentage 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 limitations meant that we could not control for

the moon cycle completely.

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 urban forest controls, average activity and total species richness

was calculated for the bat assemblage, for the light-exploiting functional group, for the light-

avoidant functional group and for each individual species. Average activity was calculated by

70 summing the total number of identified passes detected at that site during the recording time,

and then dividing by the number of sampling nights at that site. Total species richness was

calculated by counting the number of species detected at that site over the recording time. To

calculate the time until first activity, and therefore potential delay or disturbance to the bats,

we located the earliest call for a functional group or species at a particular site and then

subtracted that time from the time of sunset that day (EST), leaving the numbers of minutes

after sunset that the call was recorded. When discussing the comparison of a response variable

across light treatments, we are including light edges, dark edges and dark interiors.

Generalized linear models (GLMs) were used to establish statistical differences in both lux level

and vegetation extent among the three light treatments, with the habitat treatment

(connected or patch) and light treatment (light edge or dark edge) as fixed effects.

A series of correlations were used to assess whether average moon illumination across recording time, or the percentage of sandstone soil within a 250 m radius of each site were significant predictors of bat activity or species richness. As none were significant predictors of species richness or total bat activity, they were omitted from further models. A series of GLMs were used to assess if time till first recording of C. gouldii, O. ridei, V.vulturnus and Nyctophilus spp. (the two most commonly detected of each response group) were affected by moon illumination, moon illumination has been linked to emergence of bats. As no response variables were significantly affected by any environmental or weather variables, all were omitted from further models. We then used a series of GLMMs to compare bat and insect response variables (Table 3.1) among light treatments. We only included sites where both the edge and interior had a record for that species. Other sites were excluded from this analysis.

71 The GLMMs allowed us to include the site (which contained both the edge and dark interior pair) as a random effect, and the habitat treatment (connected or patch) and light treatment

(light edge or dark edge) as fixed effects. All models were compared with null models (without treatment) in log likelihood ratio tests. If the ΔAICc was less than 2 between models, we chose the model with the fewest number of parameters. We considered a variety of response variables (Table 3.1). Interactions terms were initially included, but were dropped from the analysis as they were nonsignificant. We then also carried out analyses on the two most commonly recorded light-exploiting species; C. gouldii and O. ridei, and the two most commonly recorded light-avoidant species; V. vulturnus and Nyctophilus spp. as activity of other species was too low or too patchy in distribution to allow for adequate analysis.

Table 3.1. All bat and insect response variables included in generalized linear mixed models Category Measurement Response variable BAT ACTIVITY Average activity; all bats the light-exploiting group Chalinolobus gouldii Ozimops ridei the light-avoidant group Nyctophilus spp. Vespadelus vulturnus

BAT SPECIES RICHNESS Average species richness; all bats for light-exploiting group for light-avoidant group

BAT RECORDING TIMES Average time until first activity; all bats for the light-exploiting group for the light-avoidant group Chalinolobus gouldii Ozimops ridei Nyctophilus spp. Vespadelus vulturnus INSECT ABUNDANCE Averages; number of Lepidoptera collected biomass of Lepidoptera

72 We used Poisson distribution models with a log link function and robust variance estimates for

all but two of the GLMMs as the data did not follow a normal distribution (Yau and Kuk 2002), the exceptions were light-avoidant bat activity and Nyctophilus spp. activity. Light-avoidant bat activity was low across all treatments and so a negative binomial distribution with a log link function with robust variance estimates was used in the GLMM. Soft-calling Nyctophilus spp. were recorded in such low numbers that activity of this taxa was converted to presence- absence data, and a GLMM was run using a binomial distribution with a probit link function, again with robust variance estimates. All post hoc tests were Fisher’s Least Significant

Difference.

When light-avoidant or light-exploiting bats were detected at a site, we wanted to know what proportion of the calls were detected along the edge compared to the interior for that site, as an indicator of whether edges or interior were more preferred by each bat group. For each edge and interior site pair, we therefore calculated the proportion of calls recorded at the forest edge compared that site’s interior control for both light-avoidant and light-exploiting functional groups. A one way ANOVA was used to compare the proportion of calls recorded at light edges and at dark edges for both light-exploiting and light-avoidant groups.

Results

We recorded 9965 bat passes throughout our acoustic survey, with 62.4% identified to species

or species group level. We detected 16 species or species groups, seven of which have a

conservation status of vulnerable under the NSW Biodiversity Conservation Act 2016;

73 Chalinolobus dwyeri, Falsistrellus tasmaniensis, Miniopterus orianae oceanensis, Miniopterus

australis, Micronomous norfolkensis, Scoteanex rueppellii, Saccolaimus flaviventris.

Artificially lit edges, dark edges and interiors all differed significantly from each other in the

amount of vegetation cover within 250 m of each site. Interior control sites had greater

vegetation extent than dark edges (p = 0.031) but not light edges (p > 0.05). Light edges did not

significantly differ in vegetation extent than dark edges (p > 0.05). Forest patches had

significantly less vegetation cover within a 250 m radius of each site than connected forest (p <

0.001). Light levels were significantly different among treatments (F (2,59) = 32.7, p < 0.001),

with artificially lit edges were significantly brighter than dark edges (p < 0.005) and dark

interiors (p < 0.005), but no difference was present between dark edges and dark interiors (p >

0.05).

Echolocation calls of bats were recorded at all sites. Light-exploiting bat species were detected at 95% of sites; with 96% of edge sites and 93% of interior sites. Light-avoidant bat species were detected at 51% of sites, 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 than dark interior sites (Table 3.2). Total bat activity was not significantly different between light and dark edges (p > 0.05; Figure 3.3a). Species richness was not significantly affected by light treatment (Table 3.2).

Table 3.2. Results of GLMM with habitat type and light treatment as fixed effects, and site as a random effect, with response variables listed on the far left column, and the dark interior control treatment and patch treatment as the reference categories Parameter Estimate Standard error t value p value Total bat activity Intercept 2.834 0.3099 9.145 < 0.001 Connected forest versus patch 0.079 0.4288 0.185 0.854 Light edge versus interior -0.814 0.0765 -10.642 < 0.001 Dark edge versus interior -0.604 0.0791 -7.629 < 0.001 Average bat species richness

Intercept 4.475 0.5258 8.511 < 0.001 Connected forest versus patch 0.330 0.6009 0.550 0.585 Light edge versus interior -0.518 0.7435 -0.696 0.489 Dark edge versus interior -0.577 0.7277 -0.793 0.431 Average light-avoidant species richness

Intercept 2.256 0.3569 6.319 < 0.001 Connected forest versus patch -0.044 0.3951 -0.112 0.911 Light edge versus interior -1.475 0.4515 -3.268 0.002 Dark edge versus interior -1.133 0.3953 -2.865 0.007 Average light-avoidant species activity Intercept 1.379 0.5022 2.745 0.009 Connected forest versus patch -0.262 0.6275 -0.417 0.679 Light edge versus interior -2.790 0.4203 -6.638 < 0.001 Dark edge versus interior -0.305 0.1991 -1.531 0.134 Average Nyctophilus spp. activity

Intercept 1.872 0.6269 2.985 0.005 Connected forest versus patch -0.854 0.5699 -1.498 0.144 Light edge versus interior -2.053 0.6654 -3.086 0.004 Dark edge versus interior -1.540 0.5836 -2.639 0.013 Average V. vulturnus activity

Intercept 1.282 0.6090 2.105 0.043 Connected forest versus patch -0.533 0.7454 -0.714 0.480 Light edge versus interior -3.100 0.5899 -5.256 < 0.001 Dark edge versus interior -0.061 0.2235 -0.272 0.787 Average light-exploiting species richness Intercept 2.139 0.2016 10.610 < 0.001 Connected forest versus patch -0.019 0.2304 -0.084 0.934 Light edge versus interior 0.338 0.2851 1.185 0.241 Dark edge versus interior 0.058 0.2790 0.208 0.836 Average light-exploiting species activity Intercept 2.581 0.3339 7.730 < 0.001 Connected forest versus patch -0.004 0.4623 -0.008 0.994 Light edge versus interior -0.630 0.0824 -7.653 < 0.001 Dark edge versus interior -0.740 0.0902 -8.208 < 0.001 Average C. gouldii activity Intercept 2.214 0.4017 5.512 < 0.001 Connected forest versus patch -0.130 0.5573 -0.232 0.817 Light edge versus interior -0.736 0.0886 -8.308 < 0.001 Dark edge versus interior -0.823 0.1071 -7.680 < 0.001 Average O. ridei activity Intercept 0.040 0.4916 0.082 0.935 Connected forest versus patch -0.182 0.6445 -0.282 0.779 Light edge versus interior 0.644 0.3634 1.772 0.082 Dark edge versus interior -0.489 0.1863 -2.626 0.011

75 Light-avoidant group responses

Dark interior sites supported higher light-avoidant species richness (Table 3.2), with no difference between light edges and dark edges (p > 0.05). Similarly, dark interior sites supported the highest light-avoidant bat activity (Table 3.2), but then dark edges supported higher activity than light edges ( Figure 3.3b). A similar pattern was seen for the most commonly recorded light-avoidant species; Vespadelus vulturnus activity was highest at dark

interior sites, lower at dark edges, and lowest at light edges ( Figure 3.3f). The activity of

Nyctophilus spp. was again highest at the dark interior sites, and significantly lower at both dark and light edges (Table 3.2).

a a b a b c

a a b a a b

76 e) Ozimops ridei activity f) Vespadelus vulturnus activity

a b c a b b

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

ess a a b a b b Average species richn Average lepidopteran biomass (mg)

Figure 3.3. Plots showing marginal means (± 1 s.e.) after controlling for site; a) Bat activity compared among light treatments; b) Activity of the light-avoidant group compared among light treatments; c) Activity of the light- exploiting group compared among light treatments; d) Activity of C. gouldii compared among light treatments; e) Activity of O. ridei compared among light treatments; f) Activity of V. vulturnus compared among light treatments; g) Species richness of the light-avoidant group compared among light treatments; h) Lepidopteran biomass (mg) compared among light treatments. Statistically significant differences, using Fisher’s Least Significant Difference test, when comparing variable measurements among treatments are depicted using differing letters (a, b and c)

A higher proportion of the light-avoidant group’s echolocation calls were recorded at the dark

edges than at the light edges, when compared with the proportion recorded at that site’s dark interior site (F 1,23 = 4.96, p= 0.038; Figure 3.4). Where light-avoidant species were present, they

were more likely to be active at the edge if it was dark than if it was artificially lit.

77 70

40 Dark edges 30 Light edges

10 Percentage of bat calls recorded atsite edges 0 Light exploiting bats Light avoidant bats

Figure 3.4. Average percentage (± se) of light-avoidant and light- exploiting functional group echolocation calls recorded at light edge sites and at dark edge sites. Calculated by combining calls for each site with calls from associated interior control site and then calculating the percentage of the total calls recorded at the edge

Light-exploiting group responses

The species richness of the light-exploiting group was not affected by light treatment (Table

3.2). Dark interior sites supported higher light-exploiting bat activity than light and dark edges

(Table 3.2, Figure 3.3c), and there was no difference between dark and light edges (p > 0.05).

The activity of C. gouldii followed a similar pattern, with activity highest at dark interior sites

(Table 3.2, Figure 3.3d), with no difference between light and dark edges (p > 0.05). However, the activity of O. ridei followed a different pattern (Figure 3.3e); activity of this species was highest at light edges (Table 3.2, Figure 3.3e), lower at dark edges, and lowest at dark interior sites (p = 0.007). There was no significant difference in the proportion of the light-exploiting group’s echolocation calls recorded at light edges and dark edges, when compared with the proportion recorded at that site’s dark interior site (F 1,2 = 1.08, p > 0.05; Figure 3.4).

78 Habitat type, whether the forest was an isolated patch or connected to other forested areas,

had no significant effect on the activity or the species richness of the bat assemblage, the light-

avoidant group, or the light-exploiting group (Table 3.2).

Temporal activity patterns

Overall, there was no difference in first recording time among light treatments across species

(Table 3.3), however differences were found between the functional groups.

The light-avoidant functional group was active significantly earlier at interior sites when compared with dark and light edges (Table 3.3). There was no significant difference between the first recordings at light and dark edges (p > 0.05). Nyctophilus spp. showed the same pattern (Table 3.3), again with no difference between first recordings at light and dark edges (p

> 0.05). There was no difference between first recording times of Vespadelus vulturnus at both dark interior sites and dark edges (Table 3.3). However, first recording time was significantly later at light edges than at dark interior sites (Table 3.3) and dark edges (p = 0.008).

Overall, the first activity of the light-exploiting functional group was not significantly different among light treatments (Table 3.3), but there were species-specific responses. First activity of

the light-exploiting C. gouldii was not significantly different among light treatments, however

it was first recorded significantly earlier in the night in connected forest than in patches (Table

3.3). First recording time in connected forest averaged 151.9 ± 9.1 minutes after sunset, versus

203.1 ± 16.9 minutes after sunset in isolated forest patches.

79 Table 3.3 Results of GLMM with habitat type and light treatment as fixed effects, and site as a random effect, with response variables listed on the far left column, and the dark interior control treatment and patch treatment as the reference categories

Parameter Estimate Standard error t value p value Minutes until first bat activity Intercept 4.967 0.1635 30.383 < 0.001 Connected forest versus patch -0.089 0.2312 -0.386 0.700 Light edge versus interior -0.010 0.0289 -0.352 0.726 Dark edge versus interior 0.021 0.0278 0.744 0.460 Minutes until first light-avoidant group activity Intercept 5.386 0.1347 39.973 < 0.001 Connected forest versus patch -0.013 0.1718 -0.077 0.939 Light edge versus interior 0.265 0.0397 6.687 < 0.001 Dark edge versus interior 0.232 0.0380 6.092 < 0.001 Minutes until first Nyctophilus spp. activity Intercept 5.500 0.1772 31.047 < 0.001 Connected forest versus patch 0.147 0.2225 0.659 0.518 Light edge versus interior 0.465 0.0615 7.563 < 0.001 Dark edge versus interior 0.462 0.0591 7.825 < 0.001 Minutes until first V. vulturnus activity Intercept 5.413 0.1748 30.962 < 0.001 Connected forest versus patch 0.151 0.2275 0.664 0.514 Light edge versus interior 0.252 0.0482 5.228 < 0.001 Dark 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 < 0.001 Connected forest versus patch -0.045 0.1027 -0.442 0.660 Light edge versus interior 0.054 0.0974 0.557 0.580 Dark edge versus interior 0.050 0.0975 0.517 0.607 Minutes until first C. gouldii activity Intercept 202.150 16.7063 12.100 < 0.001 Connected forest versus patch -53.183 20.6049 -2.581 0.013 Light edge versus interior -15.014 23.1303 -0.649 0.519 Dark edge versus interior 26.846 24.3959 1.100 0.276 Minutes until first O. ridei activity Intercept 5.272 0.1409 37.419 < 0.001 Connected forest versus patch -0.129 0.1919 -0.672 0.506 Light edge versus interior 0.176 0.0333 5.291 < 0.001 Dark edge versus interior 0.376 0.0419 8.961 < 0.001

The time of first activity for O. ridei was significantly earlier at dark interior sites than at both light and dark edges (Table 3.3). O. ridei was also recorded significantly earlier at light edges than at dark edges (p = 0.001).

80 Throughout the night, the difference in the activity of the light-avoidant group between light and dark edges was not as pronounced as the difference between interior sites and pooled edge sites. The activity of the light-avoidant group was higher at interior sites throughout the night, and was at its highest just after sunset (Figure 3.5).

Figure 3.5. Activity of light avoidant bats throughout the night. This is calculated by noting sunset and sunrise each night of the survey and calculating what percentage of the night had elapsed when each call was recorded. The graph shows the number of calls recorded within each 10 percent increment throughout the night. The bottom graph shows the combined data from the edge sites compared with the data from the interior sites.

Insect responses

Lepidoptera biomass was significantly affected by light treatment (Figure 3.3h); biomass was

higher at light edges than dark interior sites (see appendix Table 3.A1) and posthoc tests

showed that biomass at light edges was higher than at dark edges also (p < 0.001). The average number of Lepidoptera caught at dark interior sites did not differ from the number caught at

81 dark edges (see appendix Table 3.A1). However, the number of Lepidoptera caught at light

edges was significantly lower than dark interior sites (see appendix Table 3.A1), and posthoc

tests showed that the number of Lepidoptera at light edges was lower than at dark edges also

(p = 0.012).

Discussion

Our study demonstrates the disruptive effects of artificial light at the edges of urban forest on

light-avoidant and light-exploiting insectivorous bats. Our findings are consistent with research that artificial light has a negative effect on species movement across a city (Hale et al.

2015) and on certain species in ecologically sensitive habitat (Straka et al. 2016). Artificial light

at the urban forest edge contributes to a loss of functional connectivity (LaPoint et al. 2015).

Despite light edges having greater vegetation cover with 250 m, bat activity and richness was

still lower here than at dark edges. Further work examining the role in vegetation in mitigating

light impacts is needed. We also found that bats in connected forest areas and in isolated

forest patches surrounded by urban matrix were equally impacted by streetlights along edges,

as there were no differences in activity between the two habitat types, apart from C. gouldii

which was recorded earlier in the night in forested areas. This demonstrates that artificial light

in this case does not have a differentially larger impact at more pristine sites. Recent multi-taxa

research concluded that large connected patches of vegetation are critically important for

urban biodiversity (Beninde et al. 2015). Our data show that total bat activity and the activity of

the light avoidant species and C. gouldii was consistently higher on forest interiors when

compared with edges, consistent with previous research (Basham et al. 2011), and

82 demonstrating the importance of dense vegetation and forested areas to bats in cities. Our data imply that bats who show no response to light may not have their commuting behaviour disrupted as recorded in this study, and that other aspects of the urban environment way exert more influence on those species.

Available habitat in urban forests is reduced by artificial light, and with some species unable to cross lit areas (Hale et al. 2015), urban biodiversity may continue to decline unless the degrading impacts of artificial light are addressed. Consistent with our hypotheses, street lights at the forest edge negatively affected the activity of the putative light-avoidant group of bats. V. vulturnus was particularly negatively affected being less active and recorded later at artificially lit edges when compared with dark edges, assuming that these patterns translated into fitness costs. Species predicted to be light-exploiting were either unaffected by street lights along the forest edge, such as C. gouldii, or increased in activity, such as O. ridei. Ozimops ridei showed the most positive response to artificial light at the forest edge, with higher activity and earlier first recording time at artificially lit edges than dark edges. Whether the forest was connected to other forest or in an isolated patch, had little influence on the effects of artificial light at the forest edge. This suggests that the bat assemblage appears relatively similar across all forest habitat in this city, and would need further investigation, as this would inform conservation policy. Chalinolobus gouldii were recorded earlier in the night in connected forest than in isolated patches, suggesting that as well as C. gouldii being unaffected by artificial lights at the forest edge, this species may be more likely to roost in connected forest and hence emerge earlier there. The species-specific differences within this fast flying functional group

83 suggest that responses to artificial light may be more complex than basic guild-related responses. The complex response within the response groups are shown by the varying responses from O. ridei and C. gouldii, demonstrating species-specific responses to light.

With regards to insect trapping, Lepidoptera numbers were lowest at light edges, suggesting that the biomass was made up of fewer but larger individuals. We conclude that ours and similar findings (Pintérné and Pödör 2017), may have been due to street lights in the lit conditions outshining our light traps, and attracting most 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 better alternatives for sampling phototactic insects in artificially lit conditions.

Mechanisms driving light sensitivity in insectivorous bats

In line with our predictions, the light-avoidant bat species V. vulturnus showed a highly significant negative response to street lights at forest edges. Of all V. vulturnus calls, 27 % were recorded at dark edges, and only 2 % were recorded at artificially lit edges. This species was active significantly later at lit edges than at dark edges. Vespadelus vulturnus is an edge-space foraging bat, although it is quite a maneuverable flyer and so may not be as dependent on edges and flyways as other edge-space foraging bats (Law et al. 2011). Vespadelus vulturnus also flies closer to the ground in thinned as opposed to unthinned understorey, perhaps for cover and protection (Adams 2012). Despite these flight preferences, our data shows that this bat uses forest edges in cities, and is negatively impacted by light sources along those edges.

84 Its edge-space foraging preference makes V. vulturnus, and other slower-flying edge-space foragers, susceptible to habitat loss and decline from street lights at the forest edge. This is consistent with previous research that failed to detect V. vulturnus in small pockets of urban forest in highly urbanised parts of Sydney (Gonsalves and Law 2017).

In comparison, 92 % of calls identified as Nyctophilus spp., the species group adapted to foraging in cluttered vegetation, were recorded at interior sites. Although Nyctophilus spp. have been observed using open and edge habitat in woodland (Brigham et al. 1997), our study confirms low activity of this species at the urban forest edge, regardless of whether street lights were present or not. Artificial light at the forest edge did not further reduced the low activity of Nyctophilus spp at these locations. Our findings are consistent with a previous study that found these taxa were mostly restricted within the bounds of an urban remnant (Threlfall et al. 2013). These taxa appear to be light avoidant (Threlfall et al. 2013), but the lack of edge habitat use suggests both habitat fragmentation and disruption in the connectivity of dark vegetation across cities have critically impacted the urban survival of this group. Some bats, including Nyctophilus spp., may have retained a sensitivity to ultraviolet light (Gorresen et al.

2015), and this may cause these species to avoid street lights emitting ultraviolet, like the ones in this study.

The importance of dark connected forest for urban bats

Our study reveals that total bat activity was higher at dark interior sites than at either light or dark forest edges, confirming previous research that dark forest is crucial for the persistence of

85 some insectivorous bats in cities (Chapter 2, Threlfall et al. 2013, Hale et al. 2015, Straka et al.

2016, Azam et al. 2018). Activity at these dark forest interior sites peaks early in the night, suggesting that this is where most bats roost in hollow trees. Throughout the course of the night, the majority of calls were recorded at interior sites, whether looking at light avoidant or light exploiting functional groups. Our findings 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 avoidant species (Bolívar-Cimé et al. 2013, Hale et al. 2015, Straka et al.

2016).

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 similar habitat. Clutter adapted bats will glean moths and beetles from the ground and from leaves, whereas open-space adapted bats will pursue prey on the wing (Denzinger and Schnitzler 2013). Artificial light sources are known to attract insect prey out of dark habitat and deplete foraging grounds for light-avoidant species (Rydell

1992, Rydell and Racey 1995, Arlettaz et al. 2000). Our data supports this effect, where overall moth biomass was significantly higher at lit edges compared to dark edges and interior sites.

Overall moth numbers did not follow the same pattern, suggesting that larger insects dominated near street lights. This potential “vacuum cleaner effect”, where phototactic insects migrate out of dark forest and towards artificial light, may be decreasing the amount of available prey to bats restricted to dark areas, making it harder for them to persist in cities

86 (Arlettaz et al. 2000). Investigation is needed into the extent of this depletion of dark foraging

grounds for light-avoidant bats. Our data shows that light-avoidant bats are less likely to use artificially lit edges, and would therefore be unable to access the increased insect biomass there. Similar to previous experiments, artificial light could be creating foraging areas that mainly light-exploiting bats can exploit (Arlettaz et al. 2000, Stone et al. 2015). Artificial light, particularly light at forest edges in cities, may be disrupting the important environmental niches that allow bat species to co-exist.

Lunar phobia and light phobia

In V. vulturnus and other light-avoidant, slow flying species, the avoidance of light could be an

instinctive lunar phobic response. Lunar phobia is linked to morphological traits associated

with foraging strategy; slow-flying bats respond more negatively to bright moonlight

(Saldaña-Vázquez and Munguía-Rosas 2013). These species are also more likely to emerge

later in the evening (Jones and Rydell 1994), theoretically because they are adapted to hunting

prey available in the darkest part of the night and so avoid crepuscular predator peaks. Our

data confirms this trend, that slow flying bat species avoid artificial light. This morphological

group may be responding negatively to artificial light as they have evolved to respond

negatively to moonlight. The species-specific differences here warrant further research. Faster

flying bats, those that appear more exploiting of artificial light, are less affected by moonlight

(Saldaña-Vázquez and Munguía-Rosas 2013). These species may be pre-adapted, or at least

resilient, to artificially lit environments. However much like the drivers of light phobia in bats,

the mechanisms driving lunar phobia are difficult to pinpoint. Periods of bright moonlight are

87 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 studies have provided evidence that predation risk for bats increases under moonlight skies, even though it is discussed frequently in the literature (Saldaña-Vázquez and Munguía-Rosas 2013,

Stone et al. 2015). Future studies teasing apart the causes of artificial light phobia, changes in food availability or predation risk, would elucidate the mechanisms influencing lunar and light phobia in bats.

Future directions and conclusions

In urbanised Sydney, five species are at particular risk of light pollution degrading viable habitat. Elsewhere the impact may be greater, for example; locations like tropical Barro

Colorado Island in Panama are home to a much higher proportion of slower flying bat species

(Denzinger et al. 2017). If the parameters of light sensitivity discussed here are applied to other bat communities, many more species could be identified as at risk of habitat loss due to urban lighting at the edges of natural forest. This is an issue for particular consideration in expanding cities and peri-urban areas. Lighting near ecologically important areas, like drinking troughs, is already established as having a disruptive effect 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 the installation of street lights near ecologically sensitive areas in cities, such as native forest and vegetation. Other mitigation strategies, such as part night lighting, may be an alternative to constant lighting (Azam et al.

2015); but need to be turned off early enough in the night to successfully mitigate the impact

88 on vulnerable species. If lighting near ecologically important forest is necessary, red lights

(usually between 620 and 750 nm) have been shown to have minimal effect on a variety of

mammal taxa (Spoelstra et al. 2015, Spoelstra et al. 2017), although their suitability as a bat-

friendly alternative is controversial (Voigt et al. 2018) and needs further testing. LED

technology means that lights can be manufactured to emit specific spectra. Research could be

used to help manufacturers create lights with the least disruptive spectra for nocturnal

biodiversity. However, conserving dark forested areas in cities remains crucial for the

persistence of nocturnal biodiversity such as light-avoidant insectivorous bats.

Acknowledgements

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

References

Adams, M. D. (2012). “Vertical stratification of insectivorous bats (Microchiroptera) in harvested forests: assessing the role of structural clutter in shaping patterns of flight activity” Doctor of Philosophy thesis, School of Biological Sciences, University of Wollongong,2012. http://ro.uow.edu.au/theses/3702

89 Aguilar, R., M. Quesada, L. Ashworth, Y. Herrerias‐Diego and J. Lobo (2008). "Genetic consequences of habitat fragmentation in plant populations: susceptible signals in plant traits and methodological approaches." Molecular Ecology 17(24): 5177-5188.

Arlettaz, R., S. Godat and H. Meyer (2000). "Competition for food by expanding pipistrelle bat populations (Pipistrellus pipistrellus) might contribute to the decline of lesser horseshoe bats (Rhinolophus hipposideros)." Biological Conservation 93(1): 55-60.

Azam, C., C. Kerbiriou, A. Vernet, J. F. Julien, Y. Bas, L. Plichard, J. Maratrat and I. Le Viol (2015). "Is part‐night lighting an effective measure to limit the impacts of artificial lighting on bats?" Global Change Biology 21(12): 4333-4341.

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

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

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

Benson, D. H., J. Howell (1995). “Taken for granted: the bushland of Sydney and its suburbs” Kangaroo Press in association with the Royal Botanic Gardens Sydney.

Blake, D., A. Hutson, P. Racey, J. Rydell and J. Speakman (1994). "Use of lamplit roads by foraging bats in southern England." Journal of Zoology 234(3): 453-462.

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

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

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

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

Dearborn, D. C. and S. Kark (2010). "Motivations for conserving urban biodiversity." Conservation Biology 24(2): 432-440.

Denzinger, A. and H. U. Schnitzler (2013). "Bat guilds, a concept to classify the highly diverse foraging and echolocation behaviors of microchiropteran bats." Frontiers in Physiology 4, 164.

Denzinger, A., M. Tschapka and H. U. Schnitzler (2017). "The role of echolocation strategies for niche differentiation in bats." Canadian Journal of Zoology, 96(3), 171-181.

Downs, N., V. Beaton, J. Guest, J. Polanski, S. Robinson and P. A. Racey (2003). "The effects of illuminating the roost entrance on the emergence behaviour of Pipistrellus pygmaeus." Biological Conservation 111(2): 247-252.

Evans, K. L., S. E. Newson and K. J. Gaston (2009). "Habitat influences on urban avian assemblages." Ibis 151(1): 19-39.

Fischer, J. and D. B. Lindenmayer (2007). "Landscape modification and habitat fragmentation: a synthesis." Global Ecology and Biogeography 16(3): 265-280.

Gaston, K. J., M. E. Visser and F. Hölker (2015). The biological impacts of artificial light at night: the research challenge. Philosophical Transactions of the Royal Society B: Biological Sciences, 370(1667): 20140133

Goddard, M. A., A. J. Dougill and T. G. Benton (2010). "Scaling up from gardens: biodiversity conservation in urban environments." Trends in Ecology & Evolution 25(2): 90-98.

91 Gonsalves, L. and B. Law (2017). "Seasonal activity patterns of bats in North Sydney, New South Wales: implications for urban bat monitoring programs." Australian Mammalogy, 40(2), 220-229.

Gorenzel, W. P. and T. P. Salmon (1995). "Characteristics of American crow urban roosts in California." The Journal of Wildlife Management: 638-645.

Gorresen, P. M., P. M. Cryan, D. C. Dalton, S. Wolf and F. J. Bonaccorso (2015). "Ultraviolet vision may be widespread in bats." Acta Chiropterologica 17(1): 193-198.

Haffner, M. and H. Stutz (1985). "Abundance of Pipistrellus pipistrellus and Pipistrellus kuhlii foraging at street-lamps." Myotis 23(24): 167-168.

Hale, J. D., A. J. Fairbrass, T. J. Matthews and J. P. Sadler (2012). "Habitat composition and connectivity predicts bat presence and activity at foraging sites in a large UK conurbation." PloS one 7(3): e33300.

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

Hawrot, R. Y. and G. J. Niemi (1996). "Effects of edge type and patch shape on avian communities in a mixed conifer-hardwood forest." The Auk: 586-598.

Hölker, F., C. Wolter, E. K. Perkin and K. Tockner (2010). "Light pollution as a biodiversity threat." Trends in Ecology & Evolution(25): 681-682.

Hosking, G. (1979). "Trap comparison in the capture of flying Coleoptera." New Zealand Entomologist 7(1): 87-92.

Jung, K. and E. K. Kalko (2011). "Adaptability and vulnerability of high flying Neotropical aerial insectivorous bats to urbanization." Diversity and Distributions 17(2): 262-274.

Jung, K. and C. G. Threlfall (2016). “Urbanisation and its effects on bats—a global meta- analysis. Bats in the Anthropocene: Conservation of Bats in a Changing World” Springer: 13-33.

92 Jung, K. and C. G. Threlfall (in press). " Trait-dependent tolerance of bats to urbanisation: a global meta-analysis." Proceedings of the Royal Society B: Biological Sciences, 285(1885).

Keitt, T., D. Urban and B. Milne (1997). "Detecting critical scales in fragmented landscapes." Conservation Ecology 1(1): 1-17.

Krüger, F., E. Clare, S. Greif, B. Siemers, W. Symondson and R. Sommer (2014). "An integrative approach to detect subtle trophic niche differentiation in the sympatric trawling bat species Myotis dasycneme and Myotis daubentonii." Molecular Ecology 23(15): 3657-3671.

Kunz, T. H. (1982). “Roosting ecology of bats.” Ecology of bats, Springer: 1-55.

Kupfer, J. A. (2012). "Landscape ecology and biogeography: rethinking landscape metrics in a post-FRAGSTATS landscape." Progress in Physical Geography 36(3): 400-420.

La Sorte, F. A., D. Fink, J. J. Buler, A. Farnsworth and S. A. Cabrera‐Cruz (2017). "Seasonal associations with urban light pollution for nocturnally migrating bird populations." Global Change Biology 23(11): 4609-4619.

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

LaPoint, S., N. Balkenhol, J. Hale, J. Sadler and R. Ree (2015). "Ecological connectivity research in urban areas." Functional Ecology 29(7): 868-878.

Law, B., M. Chidel and P. Tap (2011). "Bat activity in ephemeral stream-beds in the Pilliga forests: clarifying the importance of flyways and buffer widths in open forest and woodland." The Biology and Conservation of Australasian Bats’.(Eds B. Law, P. Eby, D. Lunney and L. Lumsden.) pp: 308-321.

Law, B. S. (1997). "The lunar cycle influences time of roost departure in the common blossom bat, S. australis." Australian Mammalogy 20: 21-24.

Leonard, R. J. and D. F. Hochuli (2017). "Exhausting all avenues: why impacts of air pollution should be part of road ecology." Frontiers in Ecology and the Environment, 15(8), 443-449.

Lima, S. L. and J. M. O'Keefe (2013). "Do predators influence the behaviour of bats?" Biological Reviews 88(3): 626-644.

Lindenmayer, D. B. and H. A. Nix (1993). "Ecological principles for the design of wildlife corridors." Conservation Biology 7(3): 627-630.

McGarigal, K., S. A. Cushman, M. C. Neel and E. Ene (2002). "FRAGSTATS: spatial pattern analysis program for categorical maps."

McKinney, M. L. (2006). "Urbanization as a major cause of biotic homogenization." Biological Conservation 127(3): 247-260.

Navara, K. J. and R. J. Nelson (2007). "The dark side of light at night: physiological, epidemiological, and ecological consequences." Journal of Pineal Research 43(3): 215-224.

Ortega, C. P. (2012). "Effects of noise pollution on birds: a brief review of our knowledge." Ornithological Monographs 74(1): 6.

Pintérné, E. N. and Z. Pödör (2017). "The Effect of Artificial Lights on Nocturnal Macrolepidoptera (Lepidoptera: Macroheterocera) Communities." Acta Silvatica et Lignaria Hungarica 13(1): 41-54.

Plummer, K. E., J. D. Hale, M. J. O’Callaghan, J. P. Sadler and G. M. Siriwardena (2016). "Investigating the impact of street lighting changes on garden moth communities." Journal of Urban Ecology 2(1) juw004.

Pocock, Z. and R. E. Lawrence (2005). "How far into a forest does the effect of a road extend? Defining road edge effect in eucalypt forests of south-eastern Australia." Road Ecology Center.

Rhodes, M. P. (2002). "Assessment of sources of variance and patterns of overlap in microchiropteran wing morphology in southeast Queensland, Australia." Canadian Journal of Zoology 80(3): 450-460.

Rich, C. and T. Longcore (2013). “Ecological consequences of artificial night lighting” Island Press.

94 Rowse, E. G., S. Harris and G. Jones (2018). "Effects of dimming light-emitting diode street lights on light-opportunistic and light-averse bats in suburban habitats." Royal Society Open Science 5(6): 180205.

Russo, D. and L. Ancillotto (2015). "Sensitivity of bats to urbanization: a review." Mammalian Biology 80(3): 205-212.

Russo, D., L. Ancillotto, L. Cistrone, N. Libralato, A. Domer, S. Cohen and C. Korine (2018). "Effects of artificial illumination on drinking bats: a field test in forest and desert habitats." Animal Conservation.

Russo, D., L. Cistrone, N. Libralato, C. Korine, G. Jones and L. Ancillotto (2017). "Adverse effects of artificial illumination on bat drinking activity." Animal Conservation 20(6): 492-501.

Rydell, J. (1992). "Exploitation of insects around streetlamps by bats in Sweden." Functional Ecology: 744-750.

Rydell, J. (2006). "Bats and their insect prey at streetlights." Ecological Consequences of Artificial Night Lighting 2: 43-60.

Rydell, J. and P. Racey (1995). “Street lamps and the feeding ecology of insectivorous bats.” Symposia of the Zoological Society of London, London: The Society, 1960-1999.

Saldaña-Vázquez, R. A. and M. A. Munguía-Rosas (2013). "Lunar phobia in bats and its ecological correlates: A meta-analysis." Mammalian Biology 78(3): 216-219.

Sample, B. E., R. J. Cooper, R. D. Greer and R. C. Whitmore (1993). "Estimation of insect biomass by length and width." American Midland Naturalist: 234-240.

Seto, K. C., B. Güneralp and L. R. Hutyra (2012). "Global forecasts of urban expansion to 2030 and direct impacts on biodiversity and carbon pools." Proceedings of the National Academy of Sciences 109(40): 16083-16088.

Shortall, C. R., A. Moore, E. Smith, M. J. Hall, I. P. Woiwod and R. Harrington (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 British Isles." Mammal Review 21(3): 123-142.

Spoelstra, K., R. H. van Grunsven, J. J. Ramakers, K. B. Ferguson, T. Raap, M. Donners, E. M. Veenendaal and M. E. Visser (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.” Proceedings of the Royal Society B: Biological Sciences, 284(1855).

Stone, E. L., S. Harris and G. Jones (2015). "Impacts of artificial lighting on bats: a review of challenges and solutions." Mammalian Biology, 80(3), 213-219.

Stone, E. L., G. Jones and S. Harris (2009). "Street lighting disturbs commuting bats." Current Biology 19(13): 1123-1127.

Stone, E. L., G. Jones and S. Harris (2012). "Conserving energy at a cost to biodiversity? Impacts of LED lighting on bats." Global Change Biology 18(8): 2458-2465.

Threlfall, C., B. Law, T. Penman and P. B. Banks (2011). "Ecological processes in urban landscapes: mechanisms influencing the distribution and activity of insectivorous bats." Ecography, 34(5), 814-826.

Threlfall, C. G., B. Law and P. B. Banks (2012). "Influence of landscape structure and human modifications on insect biomass and bat foraging activity in an urban landscape." PloS ONE 7(6): 10.

Threlfall, C. G., B. Law and P. B. Banks (2012). "Sensitivity of insectivorous bats to urbanization: implications for suburban conservation planning." Biological Conservation 146(1): 41-52.

Threlfall, C. G., L. Mata, J. A. Mackie, A. K. Hahs, N. E. Stork, N. S. Williams and S. J. Livesley (2017). "Increasing biodiversity in urban green spaces through simple vegetation interventions." Journal of Applied Ecology 54(6): 1874-1883.

96 Threlfall, C. G., A. Ossola, A. K. Hahs, N. S. Williams, L. Wilson and S. J. Livesley (2016). "Variation in vegetation structure and composition across urban green space types." Frontiers in Ecology and Evolution 4: 66. van Grunsven, R. H., M. Donners, K. Boekee, I. Tichelaar, K. Van Geffen, D. Groenendijk, F. Berendse and E. Veenendaal (2014). "Spectral composition of light sources and insect phototaxis, with an evaluation of existing spectral response models." Journal of Insect Conservation 18(2): 225-231.

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

Appendix

Insect sampling

Insects were sampled at the edge and interior locations of all patch sites (n = 30) for one night.

For each patch, the edge and the interior were sampled on the same night to control for variation across nights. Sampling took place two weeks after acoustic sampling to avoid interfering with naturally occurring bat activity, however temperature and humidity were not significantly different and so insect sampling nights were deemed comparable to bat sampling nights. We avoided surveying on full moon and new moon nights, and for 3 days either side of both events. Black light intercept traps (Australian Entomological Supplies, Murwillumbah,

Australia) were placed on the ground, in the same location used for the Anabat II detectors two weeks prior, less than 3 m from the base of the street light pole. 50ml of ethyl acetate was added to the bottom of the light trap, bulbs were attached to 12V batteries, and the trap was turned on between 2000 and 2300. Excalibur digital timers (Merlin Distribution, Roseville,

97 Australia) automatically turned the lights off after 2 hours, and traps were collected within an

hour of the light turning off. Individual insects were then separated and identified to order.

Only Lepidoptera were included in the analysis as only a small number of insects of other

orders, Dipterans and Coleopterans, were caught. Body length of each individual insects was

measured, and biomass (grams) was calculated using established methods (Sample, Cooper et

al. 1993).

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 and Eick 2010, Barghini and Souza de Medeiros 2012, van Grunsven,

Donners et al. 2014, Plummer, Hale et al. 2016). Secondly, Lepidoptera numbers were lowest at light edges, suggesting that the biomass was made up of fewer but larger individuals. It is possible that the short sampling period for insects may have biased the collection, favouring larger moths. For light-exploiting, larger aerial-hawking bats, moths attracted to artificially lit

areas may therefore comprise an energy-rich food source of larger insect prey. Our result of

lower numbers of moths around lights has been found in another study (Pintérné and Pödör

2017), which found a higher number of lepidopteran in dark, semi-rural areas than in well-lit, urban areas. Our data may have been due to street lights in the lit conditions outshining our light traps, and attracting most of Lepidoptera away from the trap (Pintérné and Pödör 2017).

Although it seems illogical to sample the difference between light and dark conditions using an

98 additional light source, this method is standard practice in studies of light (Spoelstra, van

Grunsven et al. 2015, Pintérné and Pödör 2017).

Table 3.A1. Results of GLMM with habitat type and light treatment as fixed effects, and site as a random effect, with response variables listed on the far left column, with patch and dark edge treatment as reference categories

Parameter Estimate Standard error t value p value Biomass (grams) of Lepidoptera Intercept 5.731 0.0955 60.003 <0.001 Connected forest versus patch 0.143 0.3798 0.375 0.710 Light edge versus interior 0.244 0.0309 7.896 <0.001 Dark edge versus interior -0.104 0.0262 -3.979 <0.001 Average number of Lepidoptera caught Intercept 3.209 0.1598 20.080 <0.001 Connected forest versus patch -0.533 0.6323 -0.843 0.407 Light edge versus interior -0.546 0.1289 -4.239 <0.001 Dark edge versus interior 0.126 0.0845 1.495 0.146

99 Chapter 4: Short term effect of lights on bat activity and feeding: a case study on Myotis macropus

This prepared for submission as “Haddock, J. K., Threlfall, C. G., Gonsalves, L., Law, B. S. and

Hochuli, D. F. Short term effect of lights on bat activity and feeding: a case study on Myotis macropus.”

Abstract

Functional and connected habitat is critical for the persistence of many species in cities. Light pollution is a significant global anthropogenic threat to biodiversity which can reduce habitat connectivity for many light avoiding bat species. Myotis macropus is known to avoid artificially lit areas, but is dependent on wetlands and waterways, and these habitats are often subject to significant light pollution. We undertook an experimental trial to investigate the short-term response of M. macropus to the introduction of artificial light (light emitting diode; LED), and to identify the potential to undertake larger scale landscape experiments. We measured responses at the scale of the individual and also at the population level using radio-tracking and acoustic survey methods. LED lights were comparable in spectra, light brightness and height to small lights commonly used to illuminate waterways for aesthetic and safety purposes. There was an immediate and substantial decline in M. macropus activity, echolocation calls, and foraging activity after the introduction of artificial light at our study site. Immediate recovery (to pre-light levels) was observed in both acoustic measures once

100 lights were switched off (post-2330 h.) When ambient darkness was restored on nights 7 - 9,

two of the three activity measures (echolocation calls and radio-tracking fixes) returned to pre- light levels. Myotis macropus feeding activity, as measured by feeding buzzes, did not return to

pre-light levels. Emergence times from the roost were not significantly affected by light

treatment, although the bats emerged later as the moon got fuller and moonrise got earlier in

the evening throughout the study. There was high variability among individual bats for time of

emergence. Our findings provide preliminary data that suggest M. macropus may respond negatively to the introduction of light along urban waterways. Replicated BACI experiments carried out at a landscape scale using similar waterways are required to disentangle the effects of artificial and natural (moon) light on M. macropus. Assessment of the responses of M. macropus to lights of different spectral outputs are also needed to identify which lights have the least impact on this threatened species.

Introduction

Global light pollution is growing at a rate of 6% per year (Hölker et al. 2010). As the human

population continues to urbanise (United Nations, 2007), the intensity and spread of light

pollution in urban areas will continue to increase (Falchi et al. 2016). Light pollution is known to

disrupt many nocturnal taxa, either directly or indirectly (Longcore and Rich 2004), but its

impact on habitat connectivity and movement ecology across the urban landscape is not fully

understood. Artificial light can work as a barrier to the movements of urban fauna, significantly

reducing connectivity across the urban landscape (Hale et al. 2015).

101 Connected natural habitat in cities facilitates the movement of animals to foraging or breeding

grounds, promotes genetic diversity (Brooker, Brooker et al. 1999), and aids dispersal (Angold,

Sadler et al. 2006), whilst providing refuge from abiotic factors associated with the urban matrix (Basham, Law et al. 2010). Water pollution (Mitsch and Gosselink 2000), noise pollution

(Slabbekoorn and Peet 2003), simplification of urban vegetation (Threlfall, Ossola et al. 2016),

and invasive species (Bolger, Suarez et al. 2000) all impact urban bushland patches and corridors, reducing the functional value of natural areas and habitat connectivity for many species. However, little research has been dedicated to the impact that artificial light has on urban habitat connectivity.

Large patches and corridors of natural habitat have the strongest protective effect on urban

biodiversity (Beninde et al. 2015, LaPoint et al. 2015), but light sources nearby can considerably

reduce their functional value (Chapter 3, Kempenaers et al. 2010). To improve connectivity in

highly fragmented habitat, current advice is to minimise light sources near ecologically

sensitive urban bushland and wildlife corridors (Beier and Noss 1998, Beier et al. 2006).

However, much research addressing the relationship between habitat connectivity and

artificial light has focussed on the impacts of roads as barriers (Fensome and Mathews 2016)

where light sources include car headlights (Gaston and Holt 2018), or where light co-varies with

traffic noise (Pocock and Lawrence 2005). The impact of light pollution in a variety of urban

habitats is still poorly understood. Experimentally isolating and testing the impacts of artificial

light on urban biodiversity in these environments remains a research priority.

102 Many insectivorous bat species are bioindicators of good water quality and high habitat

connectivity (Jones et al. 2009, Straka et al. 2016). Bats have specific habitat requirements, are

sensitive to urban landscape changes (Threlfall et al. 2012), and rely on patches and corridors in cities (Basham et al. 2010), making them an ideal group to test the effect of artificial light on urban habitat use. Wing morphology traits are a good predictor of how well a species can tolerate an urbanised landscape (Jung and Threlfall in press), and slow flight and broad wings seem to also be associated with greater light sensitivity (Chapter 2, Stone et al. 2009, Rowse et al. 2016). These slower flying species may be avoiding light because they are less efficient at exploiting airborne insect prey at lights (Rydell 1992) or because they cannot evade aerial predators that use lit areas to hunt (Jones and Rydell 1994). These species presumably require dark habitat for commuting, foraging and roosting (Stone et al. 2009, Threlfall et al. 2013), and may be less likely to use ecological corridors if artificial light sources are within 50 metres

(Azam et al. 2018).

The large-footed myotis, Myotis macropus, is Australia’s only ”trawling bat”, with 100 % of its diet sourced from the aquatic environment (Law and Urquhart 2000). Myotis macropus

primarily and preferentially commute and forage along waterways and streams (Anderson et

al. 2006, Clarke-Wood et al. 2016), although can make some overland movements, and prefer

to forage over wetland sites (Straka et al. 2016). As well as water pollution (Clarke-Wood et al.

2016, Straka et al. 2016), bright lights correlate with low activity of M. macropus at wetlands

(Straka et al. 2016). Artificial light could increase risk of predation, or of hunting success for these bats. Therefore, connectivity of naturally dark riparian areas and wetlands may be

103 beneficial for individual fitness in M. macropus (Campbell et al. 2009). Artificial light has been shown to have a negative effect on the commuting behaviour of a morphologically similar trawling species, Myotis dasycneme (Kuijper et al. 2008), with almost all bats turning around when they approached the light instead of continuing along their normal commuting route.

However no studies have experimentally tested the impact of artificial light on M. macropus.

We aimed to examine the impacts on M. macropus of introducing an LED light to an urban waterway. We focused on measuring the short-term effects of lighting (point source) on the time of first recording , activity and foraging movements of M. macropus, at a single site, using both radio-tracking and ultrasonic sampling methods.

Methods

Site description

A small creek lined with riparian bushland (Figure 4.1), 36 km north west of central Sydney, was selected for the study. The site was centred at a known M. macropus roost supporting 50-100

bats and was set in a semi-rural landscape with low density housing. No artificial lighting

occurred along the creek or within approximately 200 m of the roost. The roost itself was

located in a small concrete culvert over a creek. The water catchment above this site was

approximately 990 hectares, with 83% zoned as rural landscape, and the remainder a mix of

residential, main roads and commercial landscapes (Hornsby Council Annual Water Report

2010). The creek flows into an estuarine section and then in to the Hawkesbury River within

approximately 5 km.

104

Figure 4.1. Location map showing the site of the roost (yellow star), and the sites (red points) where radio-tracking volunteers were stationed, where the anabats were stationed with microphone pointing along the stream away from the roost, and where the lights were introduced pointing along the stream away from the roost. Top left inset shows the relative location of Sydney in Australia, and bottom left inset shows our study site (white point) relative to Sydney CBD (white circle). Scale bar is in kilometres.

The water at the study site was within acceptable ranges for most pollutant levels (Hornsby

Council 2016) except for faecal coliforms, which is typical of rural catchments (Hornsby Council

Annual Water Report 2010). We trapped, radio-tracked, and acoustically recorded bats during

April of 2017. Bat activity (echolocation calls), feeding activity (feeding buzzes) and radio- tracking fixes were monitored from two locations, 200 m north of the roost and 212 m south of the roost, both on the creek line (Figure 4.1). These locations were chosen as access to the creek line was not possible anywhere else along the creek line.

105 Survey design and the introduction of lights

Bat activity (measured as echolocation calls and presence or absence of 10 radio tagged bats) was monitored for nine nights at both locations (three before, three during, and three after the introduction of artificial light, Figure 4.1). For the three nights during light introduction, white

LED RAL Pelican 9460 lights (6000 lumens; Pelican, Torrance, California, USA), standing 2.13 metres above the ground, were introduced at both sites, with the lights facing along the creek and away from the roost. Lights were turned on at dusk (1730 h) and turned off at 2330 h on those three nights. Lights could not be run for longer periods due to battery life. This meant that after 2330 h on all nine nights, the creek experienced ambient darkness, and so post-2330 h acoustic data was used as a pseudo-control comparison. The lights were chosen to be comparable in colour temperature, brightness, and height (4000 kelvin, 9.8 lux at a distance of

10m and emitting radiation in ultraviolet part of the spectrum according to manufacturer’s specifications) to small lights commonly used to illuminate waterways for aesthetic and safety purposes. We continued monitoring bat activity for a further three nights after lights were turned off to assess the post disturbance recovery of the roost. Due to time and budget constraints, we were unable to include a procedural control, with the lights present but switched off, to test the bats response to the novel equipment.

Bat trapping

Harp traps (Faunatech, Victoria, Australia) were set up along the creek, three metres upstream from the roost, just before sunset. Forty-six M. macropus individuals were caught in a 20 minute period, and were placed in cloth bags. Body weight was measured as a proxy for body

106 condition. Once we had 10 adult females with above average body weight for the species (9.7

grams ± 0.4 s.e. (Churchill 2008), we released all other individuals to minimise distress and

foraging disturbance at the roost. Radio transmitters (HOLOHIL LB-2N radio tags, Holohil

Systems, Canada) were 0.31 g, and were 3.73% of the bodyweight of a small adult M. macropus

(Churchill 2008). Tags were glued (3M Vetbond, North Ryde, NSW, Australia) to the selected individuals mid-dorsally, and then each bat was released.

Radio-tracking

Two TR-2 RF Base Stations (Telonics, Arizona, USA) and AY/C Yagi Antennas (Titley

Electronics, Ballina, Australia) were positioned at locations one and two (Figure 4.1), with the antennas pointing parallel to the stream and away from the roost, and were used to monitor the movements of the bats along the creek. Each of these loggers were programmed to automatically scan through each of the ten VHF tag frequencies, with a dwell time on each channel of ten seconds. An observer sat with the base station and recorded the time and strength of the signal each time a signal was detected from any of the transmitters. The strength of the signal was graded between one and five; one to three being classed as weak signals, and four and five being classed as strong signals. VHF radio signal is stronger if line of sight and distance from the tag to the antenna is uninterrupted and shorter (Holohil Systems, pers. comm., 2018). We estimate that we could detect the transmitters at a distance of approximately 400 m away from both sites, with strong fixes approximately under 60 m away.

The bats were monitored using this method from 1730 h to 2330 h for nine nights. For the first two nights, 1.5 hours from 2330 h to 0100 h was also spent searching for signals in the

107 surrounding bushland, up to a radius of 3 km from the roost, along the creekline and overland.

However, the steep banks and low topography of the creek resulted in limited detection of the

tags over long distances. Access to all parts of the creek line was also limited with most of the

creek being private property.

Acoustic monitoring

At each of the two monitoring locations, an Anabat II detector and Zero-Crossings Interface

Module (ZCAIM; Titley Electronics, Ballina, Australia) was placed on the ground with a

microphone 0.8 metre above the ground and pointing along the stream, away from the roost,

so to sample as much of the experimentally lit area as possible. All detectors were calibrated to

be equally sensitive and were expected to detect a bat at 30 m, depending on species. The

detectors recorded echolocations calls and feeding buzzes from passing bats from dusk until

dawn. Recorded calls were identified using AnaScheme (Adams et al. 2010) and AnalookW

software (Corben, 2000). For a bat call to be identified to species level, three or more pulses

must be present and have characteristics that fall within the program’s parameters for that

species. Positive species identifications were made only when a minimum of 50% of pulses

within a pass were identified as the same species. The identification key used in AnaScheme

was developed for the Sydney area (Adams et al. 2010). As M. macropus calls are very similar to

Nyctophilus spp. (broad bandwidth), all steep, linear calls were manually checked and at the same time M. macropus feeding buzzes were also identified using call parameters such as increased pulse repetition (Reinhold et al. 2001).

108 Measurement of environmental variables

The percentage of moon face illuminated each night (moon percentage), the moon’s phase

(moon phase) and daily moonrise times were taken from the United States Naval

Oceanography Portal (http://www.usno.navy.mil). Daily minimum temperature (°C), wind speed (km/hour) and rainfall (mm) were taken from the Parramatta North weather station

(approximately 16 km from the creek, Station 066124; Bureau of Meteorology). We noted the daily moon phase, in order to compare bat activity at different phases. However, the experiment fell such that 77.7 % of recording nights were classified as ‘waning gibbous’, 11.1 % were ‘third quarter’, and 11.1 % were ‘waning crescent’. Daily moon rise times throughout the experiment were all before midnight, ranging from 1826 h on the first night of the experiment, to 2359 on the last night of the experiment. This meant that a bright moon was not high in the sky until 2-3 hours after dusk, by which time bats had left their roost. This also meant that the moonrise got steadily later throughout the three treatments of before, during and after the light introduction.

Statistical analysis

All analyses were carried out in SPSS (version 22, SPSS Inc., Chicago, USA). For each radio tagged bat, we tallied the number of fixes and the number of strong fixes for each location and each night. We then calculated the average number of fixes and the average number of strong fixes for each bat by averaging across the three nights, and two sites, within each light treatment (i.e., pre-light introduction, during lighting, post-light introduction).

109 We calculated the nightly activity of M. macropus both before 2330 h and after 2330 h for each

location. We also calculated the number of feeding buzzes recorded both before 2330 h and

after 2330 h. To calculate the time until first activity, we located the earliest M. macropus call

recorded at either site on every night and then subtracted that time from the time of sunset,

leaving the number of minutes after sunset that the call was recorded. A detector failed on one

of the artificially lit nights at location two, and so only two nights for that treatment were

included in the analysis.

To test if the time of first recording was correlated with any moon variable, we carried out a series of Pearson’s parametric bivariate correlations between average time of first recording time regardless of light treatment and moon variables (moon percentage and moonrise time).

We used generalised linear models (GLMs) with a normal distribution to establish if moon percentage, moonrise time, rainfall or temperature were significantly different across light treatments. Moon percentage and moonrise time were omitted from all further GLMMs as

they were confounded with light treatment; the moon was brightest and rising earliest before

the introduction of artificial light (during the ‘waning gibbous’ moon phase), and was dimmest

and rising latest after ambient darkness was restored (finishing the ‘waning gibbous’ moon

phase and ending with the ‘third quarter’ moon phase).

A series of generalized linear mixed models (GLMMs) were then carried out (for all response

variables see Tables 4.1 and 4.2) which allowed us to include light treatment (before, during or

after the introduction of lights) as a fixed effect, night (date) and site (up or down stream) as

110 random effects. For radio-tracking data we also included individual bat (tag frequency) as a

random effect. All models were compared with null models (without treatment) in log

likelihood ratio tests. If the ΔAICc was less than 2 between models, we chose the model with

the fewest number of parameters.

We used a negative binomial distribution model with a log link function in GLMMs analysing

radio-tracking fixes, and a normal distribution model with a log link function for GLMMs

analysing acoustic call data. A robust variance estimate was used in the GLMM to handle

violations of model assumptions or overdispersion (Yau and Kuk 2002). All post hoc

comparisons were Fisher’s Least Significant Difference tests.

Results

We monitored the ten bats using radio telemetry for a total of 5.5 hours per night (from 1730 to

2330 h) for nine consecutive nights. We recorded a total of 1108 radio-tracking fixes, of which

232 were classed as strong fixes, with high variability among individual bats (Figure 4.2).

Although we recorded more fixes at the site north of the roost than at the south site, the difference among sites was not significant (F 1,178 = 0.116, p > 0.05) due to high variability of

fixes among days. Acoustic monitoring carried out for the nine nights recorded 11 species;

Chalinolobus gouldii, Chalinolobus morio, Falsistrellus tasmaniensis, Miniopterus orianae

oceanensis, Ozimops ridei, Myotis macropus, Nyctophilus spp., Rhinolophus megaphyllus,

Scoteanax rueppellii, Scotorepens orion, Vespadelus darlingtoni and Vespadelus vulturnus. We identified 2707 echolocation calls to species, of which 1733 (65 %) were identified as M.

111 macropus calls, with 366 of these calls being feeding buzzes. The species Miniopterus orianae

oceanensis was the second most commonly detected species throughout the acoustic survey,

with 755 (28 %) calls identified.

Environmental variables

The moon was waning and rising later throughout the experiment, and so nightly moon

percentage significantly reduced across light treatments (Figure 4.4, F 2,15 = 66.6, p<0.001).

Neither daily rainfall (F 2,14 = 1.58, p > 0.05) nor daily minimum temperature (F 2,15 = 3.46, p >

0.05) differed among light treatments.

112 40 a) bat 150.911 15 f) bat 151.431 30 10 20 5 10

0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 10 150 b) bat 150.951 g) bat 151.552 8 100 6 50 4 2 0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9

15 8 c) bat 151.072 h) bat 151.472 6 10 4 5 2 0 0 fixes recorded fixes 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9 15 d) bat 151.193 10 i) bat 151.787 tracking tracking

- 10 5 5 radio 0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9

20 j) bat 151.910 e) bat 151.314 10

Number of total 15 10 5 5 0 0 1 2 3 4 5 6 7 8 9 1 2 3 4 5 6 7 8 9

Night Night

Figure 4.2. Total number of fixes recorded for each individual radio tagged bat across the 9 nights. The broken line shows fixes recorded at location 1; south of the roost and the solid line shows fixes recorded at location 2; north of the roost. Grey areas are to denote dark nights, with white area in the middle over the nights when lights were introduced

113

Figure 4.3. Differences in response variables among light treatments ± 1 s.e; a) Average number of strong radio- tracking fixes recorded per bat; b) Average M. macropus calls recorded before 2330 h; c) Average M. macropus feeding buzzes recorded before 2330 h; d) Average M. macropus calls recorded after 2330 h. The difference between during and after the light introduction was determined using Least Significant Difference post hoc tests as described in the text. Asterisks denote a significant difference (p < 0.05) between the two treatments.

First recording time of Myotis macropus

Nightly emergence of M. macropus, ie. time until the first call was recorded, was positively

correlated with nightly moon percentage (r = 0.811, p < 0.001, Figure 4.4) and negatively

correlated with nightly moon rise (r = -0.922, p < 0.001, Figure 4.4). As the moon percentage decreased and the moon rose later in the evening (meaning the period of time after sunset was darker for longer), the time until first recording decreased (Figure 4.4). The time till first recording of M. macropus was not significantly different among light treatments (F(2,13) = 2.06,

p > 0.05, Table 4.2).

114

Time of moonrise (hhmm)

Figure 4.4. Correlation between average time till first recording (left y axis) and moonrise time (x axis), with the percentage of the moon illuminated (right y axis), illustrating that as the time of the moon rise got later in the evening and the moon had less of its face illuminated, Myotis macropus were recorded earlier. Red circles indicate % moon illuminated and open blue circles indicate time till first recording.

Radio-tracking data

The average number of all nightly radio-tracking fixes did not differ among the three time periods (F(2,177) = 0.53, p > 0.05, Table 4.1). However, the average number of strong fixes recorded differed significantly among the three time periods (F(2,177) = 4.16, p = 0.017), and was highest on days before the lights were turned on, lower after the lights were removed, and lowest during the time the lights were on (Table 4.1, Figure 4.3a). The average number of strong fixes decreased by 95 % when the lights were introduced, and rose by 9 % after the

115 lights were turned off, showing a slight recovery of M. macropus. A pairwise comparison of the average number of strong fixes before and after the lights were turned on showed that strong radio-tracking fixes returned to pre-light levels (p = 0.596). We never observed any bats flying through or foraging near the light cone, although some bats could have been flying over the light cone as the light was only 2.13 m high.

Table 4.1. Results of GLMM for radiotracking data, with light treatment as a fixed effect, and location, night and individual bat as random effects, and the after treatment as the reference category. A positive contrast estimate indicates a decrease in the response variable from first stipulated light condition to second, and a negative contrast estimate indicates an increase in the response variable.

Parameter Contrast Estimate Standard error t value p value df Average number of fixes Intercept 0.706 0.9186 0.768 0.443

Before light introduction versus after lights turned off 0.618 0.3536 1.747 0.082 2, 177

During light introduction versus -0.412 0.3701 -1.114 0.267 2, 177 after lights turned off Average number of strong fixes Intercept -3.668 1.0854 -3.380 0.001

Before light introduction versus after lights turned off 2.668 0.9519 2.803 0.006 2, 177

During light introduction versus -2.294 0.9624 2.384 0.018 2, 177 after lights turned off

Acoustic data across the light treatments pre-2330 h

The average pre-2330 h activity of M. macropus differed significantly among the three time periods (F(2,14) = 6.93, p = 0.008), and was highest before the lights were turned on, lower after the lights were permanently removed, and lowest during the time the lights were on (Table

4.2). Comparing among these three light treatments, activity of M. macropus significantly

116 decreased by 63 % during light introduction (Table 4.2, Figure 4.3b), and showed a significant

11 % recovery after the lights were turned off, back to pre-light levels (p > 0.05). Average feeding buzzes showed a different pattern (F(2,14) = 6.90, p = 0.008, Table 4.2), with feeding

activity greatest before the lights were turned on, significantly decreasing by 61 % during the

time the lights were on, and not showing any recovery after the lights were turned off (Figure

4.3c, Table 4.2).

Acoustic data after 2330 h

Activity of M. macropus was not significantly different among light treatments after 2330

(F(2,14) = 3.55, p > 0.05, Figure 4.3d, Table 4.2), when the site was naturally dark throughout the

study. Numbers of feeding buzzes of M. macropus after 2330 h were in such low numbers in all

light treatments that a GLMM was not possible.

117 Table 4.2. Results of GLMM with light treatment as a fixed effect, and location and night as random effects, with response variables listed on the far left column, and the after treatment as the reference category. A positive contrast estimate indicates a decrease in the response variable from first stipulated light condition to second, and a negative contrast estimate indicates an increase in the response variable.

Parameter Contrast Standard t value p value df Estimate error Average number of calls before 2330 h Intercept 3.196 0.6760 4.727 <0.001

Before light introduction versus after lights turned off 1.229 0.3307 3.715 0.002 2, 14

During light introduction versus after lights turned off -0.890 0.3385 2.630 0.020 2, 14

Average number of calls after 2330 h Intercept 3.454 0.3380 10.218 <0.001

Before light introduction versus after lights turned off -0.091 0.1721 -0.527 0.606 2, 14

During light introduction versus after lights turned off 0.273 0.1473 1.857 0.085 2, 14

Average number of feeding buzzes before 2330 h Intercept 6.964 6.1176 1.138 0.274

Before light introduction versus after lights turned off 11.869 3.1951 3.715 0.002 2, 14

During light introduction versus after lights turned off -6.703 3.1951 2.098 0.055 2, 14

Average number of minutes until first activity Intercept 3.107 0.4299 7.227 <0.001

Before light introduction versus after lights turned off -0.145 0.1185 -1.228 0.241 2, 14

During light introduction versus after lights turned off -0.095 0.1548 -0.613 0.550 2, 14

Discussion

Our data suggest that introducing artificial light to a waterway disrupts the activity and

foraging movements of M. macropus, supporting findings of landscape-scale surveys (Straka et al. 2016). Our study was limited in that no separate control site could be included, nor treatments randomised due to budget constraints, and hence some caution is required in interpreting our results. Nevertheless, as measured by both radio-tracking of individuals and

118 the acoustic recording of echolocation calls and feeding activity, we observed a significant and substantial decrease in M. macropus activity and feeding when artificial lights were introduced.

Recovery of echolocation and feeding activity occurred after 2330 h, when lights were turned off on lit nights. We also measured a subsequent recovery of activity, but not feeding activity, once ambient darkness was restored on nights 7 - 9, but with a high variability among individual bats (Figure 4.2). This reduction in activity during the introduction of light is unlikely to have been driven by moon phase, brightness or rise time, which would have been predicted to have driven a steady increase in activity across the experiment instead.

Interestingly, there was rapid recovery in feeding activity after lights were turned off at 2330 on lit nights, but lack of recovery of feeding activity after ambient darkness was permanently restored on nights 7 - 9. Perhaps nearby bats simply moved back in to the area after the lights turned off at 2330, but over several days, lights illuminating the creek between dusk and 2330 disrupted early evening behaviour and the bats failed to return to normal feeding behaviour in ambient darkness. Potential light sensitivity in M. macropus is consistent with findings for this genus, from mensurative (Shirley et al. 2001, Straka et al. 2016) and manipulative (Kuijper et al.

2008) studies. Our data suggests that artificial light sources along waterways and near wetlands could threaten urban habitat connectivity for the already vulnerable M. macropus. A broader-scale experiment is needed to allow for more firm conclusions, and to understand how responses vary within species.

119 The time till first recording of M. macropus was positively correlated with the percentage of the moon illuminated and negatively correlated with moon rise time, regardless of light treatment.

Light treatment probably had no effect on time till first recording because the lights were over

200 m from, and pointed away from, the roost. Bats were recorded earlier on nights when the

moon rose later at night and was less illuminated, consistent with previous research on bats

and lunar cycles (Lang et al. 2006). Trawling bats such as Myotis daubentonii and Myotis

dasycneme are known to emerge later than gleaning bats (Jones and Rydell 1994) and this may

be driven by light avoidance, although it is unclear if it is the same for M. macropus. This genus

has instinctual lunar phobia that may be driving the artificial light sensitivity, and our data

seem to suggest that at this site, M. macropus were emerging later to avoid bright moonlight.

Artificial lights may decrease functional habitat

Activity and feeding behaviour of M. macropus along the waterway in our study were

negatively affected by the introduction of artificial light, as measured by strong radio-tracking fixes (Holohil Systems, pers. comm., 2018). Tagged individuals stayed close enough for weak signals to still be detected, as demonstrated by a lack of change in total radio-tracking fixes among light treatments. It is difficult to confirm whether the bats avoided the lights and continued commuting through adjacent vegetation, or if the presence of the lights caused the bats to turn back towards the roost, due to a lack of directionality with our data. Using variable signal strength radio telemetry equipment in future would help discern the bats’ exact response. It is possible that bats were using alternative commuting pathways or foraging grounds, as we never observed any individuals flying through the light cone, near the light, or

120 traveling down the creek past the light. We also detected a decrease in activity and foraging

along the creek and this may have negative implications for the urban persistence of M.

macropus. Previously, M. macropus have been found to prefer commuting along streams and

waterways (Barclay et al. 2000, Anderson et al. 2006), and our data suggests that artificial light

could cause a decrease in activity and feeding behaviour in creeks, as has been observed for

wetlands (Straka et al. 2016). The energetic demands caused by a disruption of a habitat could

impact an individual bat’s body condition (Duvergé et al. 2000), the health of juveniles, and

roost survival (Boldogh et al. 2007). The impact of disruptive light pollution on bat population

dynamics has rarely been quantified (Boldogh et al. 2007) and remains a research priority.

Part night lighting

On nights that lights were turned on, lights were only operational until 2330 due to battery life

constraints. The period from 2330 h to sunrise was therefore dark for all nine nights throughout

our study, and acted as a pseudo-control comparison. Our acoustic data showed that post-

2330 h activity did not significantly change throughout our study, despite the pre-2330 h

activity being significantly affected by the light treatments. This suggests that M. macropus

activity may have immediately recovered from light disturbance each night after 2330 h. Part

night lighting could therefore be an effective strategy to reduce disruption for some nocturnal

fauna (Azam et al. 2015). Our data are also consistent with recommendations to turn lights off

before midnight to maximize the ecological benefits of part-night lighting (Azam et al. 2015,

Day et al. 2015). However, caution should be exercised with part night lighting. The first part of the night may be more productive for foraging with many more feeding buzzes recorded

121 during this time frame in our survey, and so lighting until midnight may disproportionately

affect preferential foraging time. A further caution would be that the most advantageous time

for switching off lighting for temperate zone bats would be early evening when insect

abundance is typically highest, however this may not the best time for human residents as it tends to be when human activity is highest. If lighting is absolutely necessary, avoiding areas

that may have a greater impact on habitat connectivity for bats, or leaving unlit sections to

provide foraging habitat is recommended, especially in the early part of the night when

foraging activity is highest.

Recovery in the days after a light disturbance

We know little about the time frame in which bats and other taxa recover from short term

lighting disturbance (Shirley et al. 2001, Kuijper et al. 2008, Stone et al. 2009, Schoeman 2016,

Azam et al. 2018), although instances of such disturbances may be frequent in urban areas,

including sports fields, temporary work sites, and security lighting. Many studies do not

specifically measure the recovery of taxa in the days and weeks after lights have been turned

off. Light and noise over a four day festival was found to delay the emergence time of a colony

of Myotis daubentonii (Shirley et al. 2001). However, recovery rates after the disturbance and

the impact of artificial light specifically were not measured, as is commonly the case with

studies experimentally introducing artificial light. Our data suggest a recovery of M. macropus

after lights were turned off at 2330, and radio-tracking data showed the individual bats

recovered fully after nine days, back to pre-light levels. Conversely, feeding behaviour did not

seem to recover to pre-light levels, illustrating that any assessment of the impacts of light

122 should consider multiple measures of activity. Without control sites and replication, it is

difficult to conclude that these changes in activity were due to the introduction of artificial

light, but this recovery shows that a more comprehensive study on this topic is required.

Mechanisms driving light sensitivity in trawling bats

It appears M. macropus in this part of Sydney is light avoidant, and there is evidence that some species in the genus Myotis are particularly sensitive to UV light (Gorresen et al. 2015).

However, many insects have the opposite response and are well documented to be attracted

to artificial light, and attracted specifically to UV light (van Langevelde et al. 2011, van

Grunsven et al. 2014). With aquatic and flying insect numbers increasing around artificially lit

waterways (Perkin et al. 2014), but feeding buzzes of M. macropus and other trawling bats

decreasing (Kuijper et al. 2008), this suggests that trawling bats cannot exploit insect prey

availability in lit areas like some bat species do (Blake et al. 1994, Rowse et al. 2016). This

suggests, that light avoidant behaviour in these bats may be driven by something else. These

bats could be avoiding light areas to evade aerial predators such as owls or predatory fish;

artificial light can alter predator evasion behaviour in other taxa (Gorenzel and Salmon 1995,

Becker et al. 2013), however few studies have examined these factors.

Mitigation strategies for light pollution along waterways

There are many conservation policies for waterways in Australia (SEPP14, NSW Government,

1985) and globally with international environmental agreements and research (Davis 1994,

Finlayson and Davidson 1999), but none currently make mention of artificial lighting. Artificial

123 light levels are one of the main causes of low bat diversity and activity at wetlands (Straka et al.

2016). Mitigation strategies such as part night lighting (Azam et al. 2015) and shielded light

fittings (Reed et al. 1985) can be effective at reducing the disruption to nocturnal biodiversity,

but research is needed to fully understand their benefit. Furthermore, the spectra that lights

emit may have differing effects on biodiversity along waterways, and understanding this has

been identified as a research priority (Straka et al 2016). Many bats of the Myotis genus are

sensitive to UV light (Gorresen et al. 2015), and so may perceive high-UV mercury vapour and metal halide lights as brighter and more disturbing than low-UV LED lights. Insects are also highly disrupted by high-UV white lights (van Grunsven et al. 2014). The most effective mitigation strategy may be to protect urban riparian areas and corridors from light pollution to conserve nocturnal fauna. As a possible alternative to all types of white light, red light does not have such a negative effect on a variety of mammal taxa, including insectivorous bats

(Spoelstra et al. 2015). Illuminating waterways using red lights may be one strategy to ensure safety for human inhabitants but also to minimize the effect on nocturnal fauna and particularly obligate waterway and wetland foragers, like M. macropus. Research in to the use of different spectra of light as a safety compromise should now be a priority.

References

124 Azam, C., C. Kerbiriou, A. Vernet, J. F. Julien, Y. Bas, L. Plichard, J. Maratrat and I. Le Viol (2015). "Is part‐night lighting an effective measure to limit the impacts of artificial lighting on bats?" Global Change Biology 21(12): 4333-4341.

Barclay, R. M., B. J. Chruszcz and M. Rhodes (2000). "Foraging behaviour of the large-footed myotis, Myotis moluccarum (Chiroptera: Vespertilionidae) in south-eastern Queensland." Australian Journal of Zoology 48(4): 385-392.

Basham, R., B. Law and P. Banks (2010). "Microbats in a leafy urban landscape: are they persisting, and what factors influence their presence?" Austral Ecology, 36(6), 663-678.

Becker, A., A. K. Whitfield, P. D. Cowley, J. Järnegren and T. F. Næsje (2013). "Potential effects of artificial light associated with anthropogenic infrastructure on the abundance and foraging behaviour of estuary‐associated fishes." Journal of applied ecology 50(1): 43-5

Beier, P. and R. F. Noss (1998). "Do habitat corridors provide connectivity?" Conservation Biology 12(6): 1241-1252.

Beier, P., K. Penrod, C. Luke, W. Spencer and C. Cabañero (2006). "South Coast Missing Linkages: restoring connectivity to wildlands in the largest metropolitan area in the United States." Connectivity conservation. Cambridge University Press, Cambridge, United Kingdom: 555-586.

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

Blake, D., A. Hutson, P. Racey, J. Rydell and J. Speakman (1994). "Use of lamplit roads by foraging bats in southern England." Journal of Zoology 234(3): 453-462.

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

Campbell, S., P.-J. Guay, P. J. Mitrovski and R. Mulder (2009). "Genetic differentiation among populations of a specialist fishing bat suggests lack of suitable habitat connectivity." Biological Conservation 142(11): 2657-2664.

Churchill, S. (2008). "Australian Bats.” 2nd edn. (Allen and Unwin: Sydney.)

Clarke-Wood, B. K., K. M. Jenkins, B. S. Law and R. V. Blakey (2016). "The ecological response of insectivorous bats to coastal lagoon degradation." Biological Conservation 202: 10-19.

125 Davis, T. J. (1994). The Ramsar Convention manual: a guide to the convention on wetlands of international importance especially as waterfowl habitat, Gland (Switzerland) Ramsar Convention Bureau.

Day, J., J. Baker, H. Schofield, F. Mathews and K. Gaston (2015). "Part‐night lighting: implications for bat conservation." Animal Conservation, 18(6), 512-516.

Duvergé, P. L., G. Jones, J. Rydell and R. D. Ransome (2000). "Functional significance of emergence timing in bats." Ecography 23(1): 32-40.

Fensome, A. G. and F. Mathews (2016). "Roads and bats: a meta‐analysis and review of the evidence on vehicle collisions and barrier effects." Mammal Review 46(4): 311-323.

Finlayson, C. and N. Davidson (1999). "Global review of wetland resources and priorities for wetland inventory." Preface iv Summary Report, 15.

Gaston, K. J. and L. A. Holt (2018). "Nature, extent and ecological implications of night‐time light from road vehicles." Journal of Applied Ecology.

Gorenzel, W. P. and T. P. Salmon (1995). "Characteristics of American crow urban roosts in California." The Journal of wildlife management: 638-645.

Gorresen, P. M., P. M. Cryan, D. C. Dalton, S. Wolf and F. J. Bonaccorso (2015). "Ultraviolet vision may be widespread in bats." Acta Chiropterologica 17(1): 193-198.

Government, N. S. W. (1985). "State Enviornmental Planning Policy No. 14 - Coastal Wetlands." Department of Urban Affairs and Planning, Sydney.

Hölker, F., C. Wolter, E. K. Perkin and K. Tockner (2010). "Light pollution as a biodiversity threat." Trends in Ecology & Evolution(25): 681-682.

Jones, G., D. S. Jacobs, T. H. Kunz, M. R. Willig and P. A. Racey (2009). "Carpe noctem: the importance of bats as bioindicators." Endangered Species Research 8(1-2): 93-115.

Jung, K. and C. G. Threlfall (in press). " Trait-dependent tolerance of bats to urbanisation: a global meta-analysis." Proceedings of the Royal Society B.

126 Kempenaers, B., P. Borgström, P. Loës, E. Schlicht and M. Valcu (2010). "Artificial night lighting affects dawn song, extra-pair siring success, and lay date in songbirds." Current biology 20(19): 1735-1739.

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

LaPoint, S., N. Balkenhol, J. Hale, J. Sadler and R. Ree (2015). "Ecological connectivity research in urban areas." Functional Ecology 29(7): 868-878.

Law, B. and C. A. Urquhart (2000). "Diet of the Large-footed Myotis Myotis macropus at A Forest Stream Roost in Northern New South Wales." Australian Mammalogy 22(2): 121-124.

Perkin, E. K., F. Hölker and K. Tockner (2014). "The effects of artificial lighting on adult aquatic and terrestrial insects." Freshwater Biology 59(2): 368-377.

Pocock, Z. and R. E. Lawrence (2005). "How far into a forest does the effect of a road extend? Defining road edge effect in eucalypt forests of south-eastern Australia." Road Ecology Center.

Reed, J. R., J. L. Sincock and J. P. Hailman (1985). "Light attraction in endangered procellariiform birds: reduction by shielding upward radiation." The Auk: 377-383.

Reinhold, L., B. Law, G. Ford and M. Pennay (2001). "Key to the bat calls of south-east Queensland and north-east New South Wales." Forest Ecosystem Research and Assessment Technical Paper 7.

Rowse, E. G., S. Harris and G. Jones (2016). "The Switch from Low-Pressure Sodium to Light Emitting Diodes Does Not Affect Bat Activity at Street Lights." PloS one 11(3): e0150884.

Rydell, J. (1992). "Exploitation of insects around streetlamps by bats in Sweden." Functional Ecology: 744-750.

Schoeman, M. (2016). "Light pollution at stadiums favors urban exploiter bats." Animal Conservation 19(2): 120-130.

Shirley, M., V. Armitage, T. Barden, M. Gough, P. Lurz, D. Oatway, A. South and S. Rushton (2001). "Assessing the impact of a music festival on the emergence behaviour of a breeding colony of Daubenton's bats (Myotis daubentonii)." Journal of Zoology 254(03): 367-373.

127 Stone, E. L., G. Jones and S. Harris (2009). "Street lighting disturbs commuting bats." Current Biology 19(13): 1123-1127.

Threlfall, C. G., B. Law and P. B. Banks (2012). "Sensitivity of insectivorous bats to urbanization: implications for suburban conservation planning." Biological Conservation 146(1): 41-52.

United Nations Fund for Population Activities. (2007). State of the World’s Population 2007: Unleashing the Potential of Urban Growth. United Nations Population Fund, New York.

van Grunsven, R. H., M. Donners, K. Boekee, I. Tichelaar, K. Van Geffen, D. Groenendijk, F. Berendse and E. Veenendaal (2014). "Spectral composition of light sources and insect phototaxis, with an evaluation of existing spectral response models." Journal of Insect Conservation 18(2): 225-231.

van Langevelde, F., J. A. Ettema, M. Donners, M. F. Wallis, DeVries and D. Groenendijk (2011). "Effect of spectral composition of artificial light on the attraction of moths." Biological Conservation 144(9): 2274-2281.

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

128 Chapter 5: Experimentally introduced red and white lights at wetlands cause diverse effects on insectivorous bats.

This paper has been prepared for submission as “Haddock, J. K., Threlfall, C. G., Gonsalves, L.,

Law, B. S. and Hochuli, D. F. Experimentally introduced red and white lights at wetlands cause

diverse effects for insectivorous bats.”

Abstract

Artificial light is rapidly intensifying, and its disruptive effects on nocturnal fauna are widely

documented. Because street lights are deemed necessary for human residents, it is essential to

find mitigation strategies that effectively reduce impacts to fauna, whilst meeting the lighting

requirements for human society. We assessed the short-term response of insectivorous bats to experimentally introduced red and white halogen lights at replicated peri-urban wetlands by acoustically surveying 12 wetlands. The composition of the bat assemblage differed significantly among sites, but also changed significantly upon the introduction of white lights, but not red lights. Specifically, the activity of the vulnerable specialist trawling bat species

Myotis macropus significantly decreased upon the introduction of white lights but not red lights. Conversely, the vulnerable fast flying bat Micronomous norfolkensis was more active at wetlands with white lights, whereas the edge-space adapted bat Vespadelus vulturnus increased in activity around both red and white lights compared to controls. We found that, irrespective of artificial light, bat diversity was positively associated with wetland size and the amount of vegetation in the surrounding 500 m radius, and negatively associated with the

129 distance of a wetland to a major waterway. Wetlands in cities support high bat diversity, and

red lights with lower colour temperature and ultraviolet light, may be a possible mitigation

strategy for reducing lighting impacts on bats at urban wetlands.

Introduction

Light pollution disrupts the natural processes that rely on the light-dark cycle. Anthropogenic light pollution is increasing (Hölker et al. 2010) and is impacting many different taxa (Rich and

Longcore 2013). The ‘night light niche’ (Longcore and Rich 2004) is a highly altered ecological niche where some species benefit, and others do not. For example, insects attracted to street lights (van Langevelde et al. 2011) are preyed upon by spiders that exploit lit areas as foraging grounds (Heiling 1999). However, this anthropogenically driven niche is not uniform - there are many different types of lighting technology that emit varying spectra, to which biodiversity can respond differently. Narrow spectrum lighting emits radiation in only a narrow part of the visible spectrum, as opposed to broad-spectrum, which is theorised to be disruptive to more biodiversity (Gaston et al. 2012). Spectra-dependent responses have been observed in plants

(Raven and Cockell 2006, Bennie et al. 2016), insects (van Langevelde et al. 2011, van Grunsven et al. 2014), reptiles (Van Grunsven et al. 2017), birds (Poot et al. 2008) and mammals (Chapter

2, Lewanzik and Voigt 2017, Spoelstra et al. 2017). Narrow spectrum red light (620 – 700 nm) is associated with earlier flowering in some plant species, which may increase the amount of visiting pollinators (Heithaus 1974, Potts et al. 2003). Some wavelengths of light, emitted by street lights, including ultraviolet radiation (10 – 400 nm), attract many insect taxa (van

Grunsven et al. 2014), which in turn increases predation by some bats that are able to exploit

130 this food resource (Rydell 1992). Different spectra of artificial lights are causing complex

ecosystem-wide responses, and to best mitigate the impacts we require a better

understanding of the wavelengths that vulnerable species are responding to.

Urbanised areas have the highest levels of light pollution (Falchi et al. 2016), with different

types of light creating a city-wide mosaic. For the persistence of nocturnal fauna in cities,

research is revealing that the conservation of dark habitat is key (Chapter 4, Gaston et al. 2012,

Rich and Longcore 2013). However, owing to public safety lighting is deemed crucial in many

urban areas, and so it remains essential to design lighting that meets both ecological and

human needs. Wetlands are key habitats for biodiversity in many different environments,

including in urban areas (Smith and Chow-Fraser 2010). Urban wetlands support high bird,

mammal, reptile and insect diversity (Walsh et al. 2001, Hamer and McDonnell 2008, Smith

and Chow-Fraser 2010, Stokeld et al. 2014, Straka et al. 2016). However, urbanisation is a

leading cause of wetland degradation (Ehrenfeld 2000, Clarke-Wood et al. 2016). Wetlands in

human dominated environments are often highly polluted by chemical run-off or highly altered

to suit the needs of human residents (Brabec et al. 2002). Light pollution at urban wetlands has

been shown to negatively affect biodiversity (Kuijper et al. 2008, Becker et al. 2013, Straka et

al. 2016). Public lighting at urban wetlands and waterways is very common, but there are few

ecological standards to minimise biodiversity impacts.

Insectivorous bats are highly reliant on wetlands and riparian habitat (Clarke-Wood et al. 2016,

Straka et al. 2016, Blakey et al. 2018); bat diversity and activity in cities is often highest at

131 wetlands and along waterways. Light pollution at wetlands is linked to lower bat diversity and

is known to negatively impact vulnerable species that rely on wetland habitat (Straka et al.

2016). However, the relationship between insectivorous bats and light is complex; some faster

flying species forage at lights for the increased insect prey attracted to the ultraviolet radiation

(Lewanzik and Voigt 2017). But slower flying species that are reliant on cluttered vegetation,

and specialist foragers like trawling bats, are more likely to avoid lights (Kuijper et al. 2008,

Stone et al. 2009) possibly because of an increased risk of predation (Jones and Rydell 1994,

Lima and O'Keefe 2013), innate light avoidance behaviour (Jones and Rydell 1994), or being

less able to exploit the insect prey attracted to lights (Rydell 1992). Furthermore, presence of

narrow-spectrum red lights had no effect on slower flying bats in rural forested areas

(Spoelstra et al. 2017).

This study aimed to assess the differing short-term responses of insectivorous bats and insects to the experimental addition of white and red halogen lights at multiple wetlands in an

urbanising area. We predicted that the introduction of low-ultraviolet narrow-spectrum red lights to wetlands would have no effect on slower flying, specialist bats or faster flying bats.

We also predicted that the introduction of ultraviolet-emitting broad-spectrum white lights would cause an increase in ultraviolet-attracted insects, an increase in activity of faster flying bats, and a decrease in activity of slower flying and specialist bats.

132 Methods

Study site

The study was carried out in the south west region of the Sydney Basin (Figure 5.1) around the

Menangle area, from the indigenous word ‘Manhangle’ meaning ‘of swamps and lagoons’

(Bayley 1966). This area historically consisted of Alluvial Woodland, Shale Plains Woodland, and Shale Hills Woodland (Burton 2005), but has been largely cleared for agricultural purposes

(Benson and Howell 1990). More recently, the area has seen rapid growth and urban development (Benson et al. 1995).

Figure 5.1. Location map showing the locations of the 12 wetlands where acoustic detectors were positioned. Top left inset shows the relative location of Sydney in Australia, and bottom left inset shows our study site (yellow star) relative to Sydney CBD (white circle). Scale bars are in kilometres.

133 We surveyed 12 permanent wetland sites, with varying degrees of surrounding vegetation.

Some sites had red lights introduced (n = 3), some sites had white lights introduced (n = 3), and

sites remained in ambient darkness each night during the experiment and acted as controls (n

= 6). All sites were surveyed simultaneously, for 3 – 4 nights (average of 3.6 nights, s.e. ± 0.2,

average is reported here due to some equipment failures which did not affect the overall

design of the experiment) before differing light treatments were introduced, then surveyed

again for 1 – 3 nights (average of 2.6 nights, s.e. ± 0.2) while the light treatments were applied.

All 12 wetlands were further surveyed for another 2 – 3 nights (average of 2.9 nights, s.e. ± 0.1)

after all sites had been returned to ambient darkness, resulting in 109 detector nights. Sites

were all within a 10.4 km radius (average of 3.6 km apart, s.e. ± 0.45) and wetlands were

between 0.26 and 2.99 ha (average of 1.38 ha, s.e. ± 0.37) in size. Surveys were carried out in

late austral summer (February 2018) a time of year when the bat population is reliably

sampled. The experiment was timed to be centred around the dark new moon to minimise the

effect of moonlight. A small waning crescent moon was present before the lights were

introduced, the new moon was present during light introduction, and a small waxing crescent

moon was present after the lights were turned off.

Sources of introduced light

At the experimentally lit sites, a single artificial light source was introduced to each wetland.

The lights were 100W white halogen Striker spotlights (SL1702 Lightforce; Hindmarsh, South

Australia) attached to 2.13 m high poles, positioned within a metre of the water’s edge and

angled at 45˚ over and across the wetland. These lights emit 2.12 lux at a distance of 50m

134 (approximately the maximum sampling range of the acoustic bat detector) and are

approximately 4500 K in colour temperature. Red light filters (Lightforce; Hindmarsh, South

Australia) that block all other wavelengths (including ultraviolet radiation; Figure 5.2a and

5.2b) were used on the light sources at red light sites. Although dimmer than streetlights (by

approximately 60 %, streetlights range between 9 and 18 lux), the white halogen lights were

chosen to be comparable in brightness and spectra to small pedestrian light sources commonly

used to light urban wetlands.

a) b)

Relative power Relative power

Wavelengths (nm) Wavelengths (nm)

Figure 5.2. Spectrographs of relative absorbance versus wavelength, as I measured using a spectrophotometer 1 m away from the lamp, and 1.5 m from the ground; a) the white halogen lamp on a new moon night whilst in place at site 1 and b) the same lamp with a red filter fitted, in place at site 8 on a new moon night

Acoustic monitoring and call analysis

At each site, an Anabat II detector (Titley Electronics, Ballina, NSW Australia) was placed on

the ground within 1.5 metres of the water’s edge and with the microphone 1 m above the

wetland’s water level. Each microphone was directed towards the centre of the wetland at a

45° angle above the ground to minimize bias in the detectability of different guilds among sites

(Law et al. 1998, Patriquin et al. 2003, Threlfall et al. 2012). Microphones were also centred on lights for wetlands that were experimentally lit. The acoustic files collected during the survey were processed using AnalookW and Anascheme software (Adams et al. 2010). A bat call was

135 defined as three or more pulses, with a minimum of 6 dots per pulse. The number of bat calls

was then used as a measure synonymous with activity. Positive species identifications were

made only when a minimum of 50 % of pulses within a pass were identified as the same species

(Adams et al. 2009, Threlfall et al. 2012). The identification key used in the analysis was

developed for the Sydney area (Adams et al. 2010). Calls identified as Chalinolobus dwyeri,

Falsistrellus tasmaniensis, Nyctophilus spp., Saccolaimus flaviventris, Scoteanex rueppellii and

Scotorepens orion are known to be difficult to identify or are threatened species and were

checked manually to ensure conformance to other guides (Pennay et al. 2004, Adams et al.

2010). In addition, calls were run through a filter which is specifically designed to identify

alternating call characteristics of Chalinolobus gouldii (B. Law, pers. comm., 2015). Only species

that were positively identified using the key, filters and manual checking were included for

further analysis to eliminate any bias caused by using partially identified species. The linear calls of Nyctophilus gouldi, N. geoffroyi and M. macropus cannot be reliably distinguished using the AnaScheme method and so were grouped as ‘linear calls’. All linear calls were manually checked in AnalookW to identify M. macropus calls. Calls with a long sequence length

(Reinhold et al. 2001), pulse interval time <75 ms, and an initial slope > 400 octaves per second were classified as M. macropus (Pennay et al. 2004). All other linear calls were grouped as

Nyctophilus spp. since calls of N. gouldi and N. geoffroyi are indistinguishable.

Measurement of environmental variables

The percentage of the moon’s face illuminated each night (referred to as moon illumination) was taken from the United States Naval Oceanography Portal (www.usno.navy.mil/). The

136 nightly moon phase was taken from MoonGiant.com. Daily humidity (%) and daily rainfall

(mm) were taken from the Campbelltown weather station (Station 068257; Bureau of

Meteorology). The nearest distance (m) to the closest major river (Nepean River) was calculated using the measurement tool in Google Earth (Google Inc., 2017). The size of each wetland (m2) was calculated by drawing a polygon around the wetland edge in Google Maps using the area measurement tool. The percentage vegetation cover (>3 m) within a radius of

500m of each site, excluding the wetland’s water surface, was calculated using Arc Map (ESRI,

Redlands, USA, ver. 10.2) by intersecting GPS points of our sites with the GIS layer

‘CumberlandPlainWest_2013_E_4207’ (NSW Office of Environment and Heritage, Sydney).

The percentage water within a radius of 500m of each site was calculated the same way but instead using the GIS layer ‘NSW Hydro Area Dataset’ (NSW Spatial Data Catalogue). Light brightness (lux) was measured using a lux meter (Digitech QM1587) held at hip height perpendicular to the ground and facing upwards, whilst standing 1 m in front of the light and detector. In the case of the dark control sites, the lux meter was held 1 m in front of the detector at the same height. The percentage of the wetland edge with trees (>3 m) was estimated whilst at the site at the time of the experiment and each wetland was given a number between 0 and 1; 0 being no trees and 1 being completely surrounded by trees, with

0.2 being 20 %, 0.5 being 50 % and so on (Straka et al. 2016).

Statistical analysis

All analyses were carried out in SPSS (version 22, SPSS Inc., Chicago, USA) and PRIMER

(version 7, Quest Research Ltd, New Zealand). One site recorded for just one night in a time

137 period due to equipment failure, but because site was our unit of replication we proceeded with the analysis.

Analysis of environmental variables

To test if bat activity was correlated with any weather variable, we carried out Pearson’s

parametric bivariate correlations between measures of bat activity across all sites regardless of

light treatment (average nightly bat activity and the average nightly activity of the most

commonly recorded species; M. macropus, M. norfolkensis, V. vulturnus, and Nyctophilus spp.)

and weather variables (daily humidity and nightly moon illumination). We then used non-

parametric Kruskal-Wallis rank tests to check if moon illumination, moon phase or daily

humidity differed among time periods (before, during and after the introduction of the lights).

Nightly moon illumination, and daily humidity were omitted from all generalized linear mixed

models as they were confounded with time period (χ2 (2) = 72.3, p < 0.001 and χ2 (2) = 17.19, p <

0.001 respectively). Nightly moon illumination was highest before the introduction of artificial

lights (average 11.6 % ± s.e. 4.14), lowest during the introduction of artificial lights (average

0.67 % ± s.e. 0.67) and marginally increased after the lights were turned off (average 7.67 % ±

s.e. 2.9). Daily humidity showed a similar pattern, and was highest before the introduction of

artificial lights (average 73.9 %, ± s.e. 1.37), lowest during the introduction of artificial lights

(average 53.03 % ± s.e. 4.24) and increased after the lights were turned off (average 68.5 % ±

s.e. 3.71).

138 Daily rainfall was consistently low throughout the experiment so we did not include it in any analysis (77.8 % of nights were dry, and two nights had 2.4 mm during the afternoon). A student’s t-test was used to compare the difference in brightness of the white lights and red lights.

Principal coordinates analyses for compositional changes

To assess differences in the bat assemblages among time periods (before, during, after) within each light treatment, and to understand what species and environmental factors were contributing most to any changes, we separated the data in to three datasets; sites where red lights were introduced, sites where white lights were introduced, and dark control sites.

Within-site differences were so great that this comparison made it easier to see changes at each site among time periods. For each of the three datasets, we used a fourth root transformation as our data were skewed (Clarke and Green 1988), and then generated Bray-

Curtis similarity matrices from transformed activity data for all species. We conducted similarity percentage analyses (SIMPER) and a two factor permutational multivariate analysis of variance (PERMANOVA, with site and time being the two factors) to assess differences in the bat assemblages among time periods and among sites. We then generated two principal coordinate analyses for each light treatment, one with overlaid vectors describing environmental variables with a significant Pearson’s correlation of over 0.6, and one with overlaid vectors describing species dispersions with significant Pearson’s correlation coefficient >0.6.

139 Generalised linear mixed models for species-specific responses

To understand if the introduction of lighting regimes specifically affected the activity of key species, we used generalized linear mixed models (GLMMs). Using GLMMs allowed us to include light treatment (red light, white light or ambient darkness), time period (before, during and after the introduction of lights) and the interaction between the two as fixed effects in our models, and site as a random effect. Site was added as a random effect to control for repeated measurements at the wetlands. Our response variables were overall bat activity, the activity of the three most commonly recorded species Micronomus norfolkensis, Vespadelus vulturnus,

Myotis macropus, and the activity of the slower flying species; Nyctophilus spp. For the light treatment and time period fixed effects, we set the reference category as the before treatment. The environmental variables that explained most of the differences among sites were the nearest distance between the wetland and the Nepean River, wetland size and the percentage of the wetland edge with vegetation, and so we included these as random effects in all our GLMMs by binning the data and factorising it for use in the model. For all GLMMs we used a Poisson distribution with a log link function, except with the pooled data for slower flying species where we used a negative binomial distribution. All models were compared with null models (without treatment) in log likelihood ratio tests. If the ΔAICc was less than 2 between models, we chose the model with the fewest number of parameters. Data were checked for under and over dispersion and then a robust variance estimate was used in the

GLMM to handle violations of model assumptions or overdispersion (Yau and Kuk 2002).

140 Results

We identified 35,662 echolocation calls to 16 species or species group level (see appendix), of

which eight are listed as vulnerable under the NSW Biodiversity Conservation Act 2016. The

most commonly recorded species were M. norfolkensis making up 32.7 % of recorded calls, M.

macropus (19.5%) and V. vulturnus (19.5 %).

Compositional changes and site differences

The bat assemblages at white light sites were significantly different among time periods (Table

5.1, pseudo F (2,8) = 2.24, p = 0.047, Figure 5.3a) and different among sites (pseudo F (2,8) = 6.18, p

< 0.001). The overlaid vectors (Figure 5.3a, Table 5.1) indicate the direction in which species

were changing with the PCO axes. The significant change in assemblage during the white light

introduction was driven by an increase in Ozimops ridei, Falsistrellus tasmaniensis, Micronomous

norfolkensis, Chalinolobus gouldi and Chalinolobus morio and a decrease in Myotis macropus and

Scotorepens orion (Figure 5.3a). In comparison, bat assemblages at the red light sites did not

significantly change during or after the introduction of red lights (Figure 5.3b, Table 5.1, pseudo F (2,8) = 0.84, p = 0.59), but still differed among sites (pseudo F(2,8) = 5.3, p = 0.001).

Finally, the bat assemblage did not change across time periods at the control sites (Table 5.1, pseudo F (2,17) = 1.04, p > 0.05, Figure 5.3c), but again differed among sites (pseudo F (5,17) = 7.97,

p = 0.001).

141 5.3a) white light sites

5.3b) red light sites

142 5.3c) dark control sites

Figure 5.3. Principal components ordination showing the composition of the bat communities before (B), during (D) and after (A) lights were introduced at each of the different light treatments, with overlaid vectors representing species that contribute the most to changes in the assemblage (vectors show variables with Pearson’s correlation coefficient higher than 0.6); a) showing sites that had white lights introduced and the species that contribute the most to changes in the assemblage; b) showing sites that had red lights introduced and the species that contribute the most to changes in the assemblage; c) showing sites that remained dark control sites and the species that contribute the most to changes in the assemblage

Table 5.1. Permutational MANOVA pairwise comparisons of the composition of the bat assemblage before, during and after the introduction of lights, for each light treatment

Light treatment Comparison T value p (perm) Unique permutations

Before – During 1.8029 0.045 60 White Before – After 1.9528 0.031 60 During – After 1.1038 0.36 60

Before – During 0.64077 0.771 996 Control Before – After 1.2864 0.214 998 During – After 1.2076 0.242 996

Before – During 1.1571 0.387 60 Red Before – After 0.70718 0.691 60 During - After 0.83488 0.699 60

143 Environmental factors

There was high variability among sites for the bat assemblages that we recorded, with amount

of surrounding vegetation cover within 550 m and the distance of the wetland to a major

waterway contributing the most to site differences within the white light sites (Figure 5.4a).

The environmental factors significantly explaining the differences among red light sites were wetland size, percentage of the wetland edge covered in vegetation, and the percentage vegetation within 500 m (Figure 5.4b). Environmental factors significantly explaining the differences among dark control sites were the distance from a major waterway, and the percentage vegetation within 500 m. Overall, bat diversity was positively associated with the size of the wetland, the distance to a major waterway, and the percentage vegetation 500m

(Figure 5.4a, 5.4b and 5.4c).

5.4a) white light sites

144

5.4b) red light sites

5.4c) dark control sites

Figure 5.4. Principal components ordination of sites showing the composition of the bat communities before (B), during (D) and after (A) lights were introduced, with overlaid vectors of the environmental variables that contribute the most to changes in the assemblage (vectors show variables with Pearson’s correlation coefficient higher than 0.6); a) showing sites that had white lights introduced and the environmental variables that contribute the most to changes in the assemblage; b) showing sites that had red lights introduced and the environmental variables that contribute the most to changes in the assemblage; c) showing sites that remained dark control sites and the environmental variables that contribute the most to changes in the assemblage.

145 Species-specific responses to different lights

The effect of time period on total bat activity (F(4,27) = 58.60, p < 0.001), the activity of M. macropus, V. vulturnus and M. norfolkensis differed by light treatment (F(4,27) = 44.68, p < 0.001;

F(4,27) = 44.36, p < 0.001; F(4,27) = 122.47, p< 0.001 respectively); the interaction model between

the two factors was significant for all of these response variables. Total bat activity increased in

all light treatments during the period of light introduction, and then declined at red light and

white light sites after the lights were turned off. After the lights were turned off, control sites

maintained higher total bat activity than before the introduction of lights (Figure 5.5a, Table

5.3), whereas total activity at white light sites returned to pre-disturbance levels, and red light

sites fell below pre-disturbance levels (Table 5.3). Myotis macropus activity was negatively

affected by the introduction of white light (Table 5.3), showed no response to red light and

increased in activity at the control sites (Figure 5.5b, Table 5.3). After lights were turned off, M.

macropus activity recovered significantly at white light sites, but not to pre-disturbance levels,

and decreased at control sites but remained higher than pre-disturbance levels (Table 5.3).

Activity of M. macropus remained higher at red light sites after lights were turned off than

before lights were turned on. Vespadelus vulturnus increased in activity at both the white light

and red light sites but not the control sites during the period when the lights were on (Figure

5.5c, Table 5.3), and then decreased in activity at all sites after the lights were turned off, falling

below pre-disturbance levels at red light and control sites, but remaining higher at white light

sites (Table 5.4). Micronomus norfolkensis activity was positively affected by the introduction of

white lights, did not respond to the introduction of red lights, and decreased at control sites

(Figure 5.4d, Table 5.3). Once all lights were turned off, M. norfolkensis activity decreased at

146 white lights sites, but remained higher than pre-disturbance levels (Table 5.3). Activity of M. norfolkensis recovered to pre-disturbance levels at control sites, but was much lower than pre- disturbance levels at red light sites. The activity of slow flying Nyctophilus spp. did not differ among light treatments or time periods (F(4,27) = 0.25, p > 0.05, Table 5.3), but the data on this species were sparse.

147

a) b)

* * * * * * *

* * * *

c) d)

* * * * * * *

* * *

Figure 5.4. Plot showing estimated marginal means (± 1 s.e.) after controlling for wetland size, the distance of the wetland from the Nepean River and the percentage of the wetland edge with vegetation. Sites where white lights were introduced are depicted in blue, sites where red lights were introduced are depicted in red, and dark controls are in black. The coloured asterisks show which pairwise comparisons are significant and therefore if there is a change over time (p < 0.05). A blue asterisk denotes a significant change at the white light sites, red asterisks denote a significant change at the red light sites, and black asterisks denote a significant change at the control sites) with the asterisks below the bracket showing significant differences between before and after the lights were introduced; a) average number of bat passes compared among time periods and among light treatments; b) average number of Myotis macropus calls compared among time periods and among light treatments; c) average number Vespadelus vulturnus calls compared among time periods and among light treatments; d) average number of Micronomous norfolkensis calls compared among time periods and among light treatments.

148 Table 5.3. Results of the Least Significant Difference pairwise comparisons, relating to figure 5.4, performed after generalized linear mixed models, comparing the response variable among light treatments across time periods

Temporal 95% 95% Response Light Treatment Contrast Std. Confidence Confidence t df Adj. Sig. Variable treatment Pairwise Estimate Error Interval Interval Contrasts Lower Upper

white before - after -0.03 0.055 -0.548 27 0.588 -0.143 0.082 white before - during -0.598 0.049 -12.291 27 < 0.001 -0.698 -0.498

white during - after 0.568 0.048 11.786 27 < 0.001 0.469 0.667 red before - after 0.544 0.067 8.114 27 < 0.001 0.407 0.682 red before - during -0.339 0.053 -6.368 27 < 0.001 -0.448 -0.23 red during - after 0.883 0.063 13.919 27 < 0.001 0.753 1.013

Average activity Average control before - after -0.084 0.03 -2.852 27 0.008 -0.145 -0.024 control before - during -0.11 0.029 -3.742 27 0.001 -0.17 -0.05 control during - after 0.026 0.029 0.892 27 0.38 -0.033 0.085

white before - after 0.3 0.093 3.239 27 0.003 0.11 0.49 white before - during 1.829 0.162 11.261 27 < 0.001 1.496 2.162 activity white during - after -1.529 0.166 -9.195 27 < 0.001 -1.87 -1.188 red before - after -0.285 0.127 -2.251 27 0.033 -0.545 -0.025 red before - during -0.184 0.13 -1.418 27 0.168 -0.45 0.082 red during - after -0.102 0.121 -0.842 27 0.407 -0.349 0.146

Myotis macropus macropus Myotis control before - after -0.339 0.067 -5.026 27 < 0.001 -0.477 -0.2 control before - during -0.481 0.066 -7.345 27 < 0.001 -0.616 -0.347

Average control during - after 0.142 0.059 2.398 27 0.024 0.021 0.264 white before - after -0.713 0.14 -5.089 27 < 0.001 -1 -0.425

white before - during -1.569 0.126 -12.445 27 < 0.001 -1.828 -1.31 white during - after 0.856 0.096 8.934 27 < 0.001 0.66 1.053

red before - after 0.984 0.185 5.309 27 < 0.001 0.604 1.364 red before - during -0.962 0.114 -8.464 27 < 0.001 -1.195 -0.729 activity red during - after 1.946 0.169 11.512 27 < 0.001 1.599 2.293 Vespadelus vulturnus control before - after 0.231 0.075 3.084 27 0.005 0.077 0.385 control before - during -0.079 0.069 -1.14 27 0.264 -0.221 0.063 Average control during - after 0.31 0.074 4.208 27 < 0.001 0.159 0.461 white before - after -1.474 0.254 -5.797 27 < 0.001 -1.996 -0.953 white before - during -3.334 0.233 -14.281 27 < 0.001 -3.813 -2.855 white during - after 1.86 0.118 15.76 27 < 0.001 1.618 2.102

red before - after 1.585 0.171 9.244 27 < 0.001 1.233 1.937 red before - during 0.128 0.103 1.237 27 0.227 -0.084 0.34

activity red during - after 1.457 0.173 8.401 27 < 0.001 1.101 1.813

Micronmous norfolkensis norfolkensis Micronmous control before - after 0.036 0.048 0.764 27 0.452 -0.061 0.134 control before - during 0.189 0.05 3.804 27 0.001 0.087 0.291

Average control during - after -0.153 0.05 -3.045 27 0.005 -0.255 -0.05

149 white before - after -0.101 0.701 -0.144 27 0.887 -1.539 1.337 white before - during -0.78 0.7 -1.115 27 0.275 -2.216 0.655 white during - after 0.679 0.699 0.971 27 0.34 -0.756 2.114

red before - after 0.193 0.703 0.275 27 0.786 -1.25 1.636 red before - during -0.581 0.701 -0.829 27 0.415 -2.019 0.858 species red during - after 0.774 0.702 1.103 27 0.28 -0.666 2.214 control before - after -0.277 0.494 -0.56 27 0.58 -1.291 0.738 control before - during -0.333 0.494 -0.674 27 0.506 -1.347 0.681

Average activity of slower flying control during - after 0.056 0.494 0.114 27 0.91 -0.957 1.07

Weather variables

The average nightly activity of M. norfolkensis (r (10) = -0.81, p = 0.004) and V. vulturnus (r (10) =

-0.72, p = 0.019) was negatively correlated with daily humidity. Total bat activity was

negatively correlated with daily humidity (r (10) = -0.68, p = 0.03) and percentage moon (r (10)

= -0.68, p = 0.032). Light brightness (lux) was not significantly different between white and red

light treatments (F(1,4) = 0.14, p > 0.05) with the white lights (average 6.3 lux, ± s.e. 0.1) and the

red lights (average 6.2 lux, ± s.e. 0.25) the same brightness.

150 Table 5.4. Average activity for each species detected, with ± 1 standard error, by light treatment and time period

Control Red lights White lights Before During After Before During After Before During After Chalinolobus dwyeri mean 0.17 0.28 0.28 0.17 0.00 0.22 0.25 0.11 0.11 s.e. 0.08 0.28 0.13 0.08 0.00 0.22 0.14 0.11 0.11

Chalinolobus morio mean 5.08 11.17 5.72 9.17 28.56 2.83 1.83 8.89 2.89 s.e. 2.35 7.05 1.84 8.29 27.06 0.83 0.94 5.30 1.16

Chalinolobus gouldii mean 12.04 15.72 11.22 18.75 17.33 12.22 29.08 26.22 19.89 s.e. 2.65 4.86 2.06 7.57 6.77 8.24 16.14 12.01 14.24

Falsistrellus tasmaniensis mean 0.96 0.83 1.72 1.03 1.44 2.78 0.83 3.00 1.44 s.e. 0.81 0.65 0.71 0.74 51.20 2.12 0.44 2.36 0.56

Micronmous norfolkensis mean 149.29 123.39 143.72 66.75 58.67 13.94 6.33 177.94 27.33 s.e. 98.09 70.99 79.08 63.89 50.02 6.83 3.04 96.19 12.39

Miniopterus australis mean 10.13 11.00 11.50 5.28 3.78 1.61 15.67 3.39 7.89 s.e. 5.78 6.58 5.56 5.11 3.13 1.03 14.55 3.14 6.58

Miniopterus orianae mean 0.08 0.39 0.17 0.36 0.33 0.61 0.00 0.28 0.00 oceanensis s.e. 0.05 0.33 0.11 0.22 0.19 0.45 0.00 0.15 0.00

Myotis macropus mean 63.25 101.67 88.28 36.56 43.67 48.22 91.67 14.72 67.78 s.e. 25.79 61.83 20.32 29.59 38.87 39.03 55.16 5.03 45.50

Nyctophilus spp. mean 12.50 13.94 31.22 6.19 12.44 7.83 5.50 5.17 7.89 s.e. 4.81 5.20 18.53 2.70 8.96 1.92 2.27 1.17 0.56

Ozimops ridei mean 36.33 45.39 25.00 8.33 14.33 6.56 22.42 21.11 25.33 s.e. 22.47 29.93 14.21 4.48 1.17 3.09 15.04 14.22 19.40

Saccolaimus flaviventris mean 0.00 0.06 0.00 0.00 0.22 0.00 0.00 0.00 0.11 s.e. 0.00 0.06 0.00 0.00 0.11 0.00 0.00 0.00 0.11

Austronomous australis mean 1.13 2.83 2.11 1.78 1.44 1.33 1.50 2.83 2.67 s.e. 0.32 1.44 1.16 0.91 0.48 0.88 0.29 1.36 1.20

Scotorepens orion mean 2.17 3.28 19.00 6.25 1.67 0.89 6.08 4.17 6.00 s.e. 1.57 2.75 16.18 5.00 0.84 0.59 1.92 1.73 4.68

Scoteanax rueppellii mean 3.38 1.06 3.06 3.22 3.11 1.72 11.58 5.22 4.11 s.e. 2.45 0.69 1.60 1.63 1.79 0.84 11.33 3.08 3.12

Vespadelus darlingtonia mean 1.63 4.44 1.50 2.08 2.67 2.83 0.25 2.00 0.00 s.e. 0.83 4.38 1.06 2.08 2.50 2.83 0.14 2.00 0.00

Vespadelus vulturnus mean 67.17 72.44 53.00 35.75 93.33 13.44 25.58 121.83 51.67

s.e. 36.01 44.27 16.72 19.92 78.01 6.95 14.09 79.45 19.70

151

Discussion

Our results show that, at least for some species, red lights do not have as disruptive an effect on bat communities at wetlands as white lights, consistent with another forest-based study

(Spoelstra et al. 2017). The introduction of red lights to wetlands, when compared with white

lights, caused no decrease in activity of the vulnerable trawling bat M. macropus and no

increase in the faster flying M. norfolkensis. Vespadelus vulturnus increased in activity around

both red and white lights, and Nyctophilus spp. responded to neither but was rarely recorded.

Myotis macropus is reliant on wetlands and waterways for much of its life cycle (Law and

Urquhart 2000, Anderson et al. 2006, Campbell 2009), and in urban areas these wetlands are

commonly light polluted (Straka et al. 2016). There was high variability among sites, meaning

that we analysed the different light treatments separately to highlight responses to introduced

lights. Our study supports that where public lighting at wetlands is deemed necessary, red

lights could be an effective mitigation strategy to minimize disruption for this vulnerable

species and other bat species.

Hypothetical mechanisms for light sensitivity in bats

The numerous hypotheses outlining mechanisms that drive bat responses to artificial light are

still not well understood. An increase in ultraviolet-attracted insects provides a dependable

food source for some bat species, with insects and bats decreasing when the UV light

decreases (Lewanzik and Voigt 2017). The species that most exploit this ‘night light niche’ are

faster flying bats (Lewanzik and Voigt 2017) that are able to dive in to the light cone and forage

152 aerially (Rydell 1992), but this does not offer an explanation for why activity of other bat species, such as M. macropus in this study, decreased in white light areas. Four possible mechanisms could explain this light phobia in some bat species. Artificial light could offer a feeding ground for nocturnal visual predators such as owls (Speakman 1991) or predatory fish for trawling bats (Becker et al. 2013), causing some slower flying bat species to avoid lit areas due to an increased predation risk (Jones and Rydell 1994), although this has not been experimentally tested. Another hypothesis based on immunohistochemical evidence predicts that some species are more sensitive to blue light than others (Müller et al. 2009), theoretically leaving longer wavelengths of red light indistinguishable from ambient darkness (Spoelstra et al. 2017). No bat species in Australia, including M. macropus, have been tested using the immunohistochemical technique, but these differences in sensitivities to light may help to explain species-specific responses to artificial light. A third hypothesis is that the activity of slower flying and trawling bats may decline at white lights due to a change in insect prey distributions, which can be significantly disrupted by artificial light in both terrestrial and riparian environments (Perkin et al. 2014). Finally, artificial light may be causing interspecific interference competition, attracting species that are able to exploit increased insect or aquatic invertebrate abundances, and competitively excluding other species that might be forced to forage elsewhere (Baagøe 1986, Arlettaz et al. 2000), although this may not apply to trawling bats as other species cannot adopt that specialised foraging style. However, our study did not measure competition, and so cannot conclude whether some bats excluded others at lit areas.

153 Species-specific responses to red and white lights

Consistent with our predictions, the specialist trawling bat, M. macropus, responded negatively to white lights at wetlands. Species in this genus are known to emerge later even than gleaning bats and forage in the darkest part of the night (Jones and Rydell 1994), and do not rely on insect concentrations at street lights (Furlonger et al. 1987). Myotis macropus also has broader wings and relatively slow flight speed for trawling (Bullen et al. 2016), which means they may be less able to evade aerial predators in lit conditions. The instinctual light phobia of this genus coupled with a potential increased risk of predation may be driving artificial white light avoidance in this species. Conversely, this species showed no response to red lights at wetlands. This lack of a response may be driven by the bats simply not perceiving the red light

(Spoelstra et al. 2017). Whatever the mechanisms, our study suggests that using red lights in place of white lights around M. macropus roosting and foraging habitat in urban areas may reduce the negative effects of urban lighting on this species.

Our study shows that M. norfolkensis, a faster flying open-space adapted species, responded

positively to white lights. This species, although listed as vulnerable on the IUCN Redlist and

under the NSW Biodiversity Conservation Act 2016, was one of the most commonly recorded in

our study. Micronomus norfolkensis prefers riparian habitat with little vegetation (McConville et

al. 2014), which our study supports. This species is moderately sensitive to urbanisation

(Threlfall et al. 2012, McConville et al. 2014), and artificial light in cities has been implicated as a potential driver for urban avoidance (McConville et al. 2014). Our data do not support this

154 hypothesis and instead demonstrate that this species is tolerant of light, at least point sources

of light, and so is presumably avoiding urban landscapes for another reason.

Vespadelus vulturnus is an edge-space adapted species that is light avoidant in urban areas

(Chapters 2 and 3), and is also considered urban sensitive (Scanlon and Petit 2008, Basham et

al. 2010, Threlfall et al. 2012). However, our data show that in a peri-urban setting this species may be light tolerant, and even attracted to lights (see also (Adams et al. 2005), both white and red. Insects are attracted to lights in both urban and rural settings (Eisenbeis et al. 2006,

Eisenbeis and Eick 2010), and so are presumed equally beneficial to insectivorous bats.

However, light traps have been found to attract more insects in rural settings than in urban areas (Pintérné and Pödör 2017). Rural light sources could be of higher foraging value than urban lights to foraging insectivorous bats (Scanlon and Petit 2008), but the impact of light sources in different environments is poorly understood. It is difficult to conclude if this species increased in activity at red light sites due to an increase in available prey or if this species can perceive red lights and move towards them for another reason. Future studies should elucidate the exact physiological spectral sensitivity and wavelength discrimination of species using opsin immunohistochemical analyses (Müller et al. 2009).

Landscape-scale movement of species

The introduction of lights caused local-scale changes in bat activity and assemblage, but also may have caused landscape-scale movements. Not only did the activity of M. norfolkensis increase after the introduction of white lights, activity decreased at dark control sites.

155 Conversely, the activity of M. macropus decreased after the introduction of white lights, and

increased at the dark control sites. We suggest that M. norfolkensis may have moved out of

dark areas and into white light sites, and M. macropus moved away from white lights towards

dark control sites. In fact, the disruption of these species is potentially likely to be greater than

the site separation that we measured. Bats are known to respond at a landscape-scale to anthropogenic disturbances in their habitat (Fenton et al. 1992, Grindal and Brigham 1998,

Law et al. 2018), but bat responses to short term artificial light disruption has been rarely studied (Chapter 3).

The importance of wetland traits

Wetlands are essential habitat for biodiversity in all different landscapes, including urban areas

(Smith and Chow-Fraser 2010). However, wetlands are often highly altered, and surrounding vegetation is simplified and disconnected from other natural habitat (Ehrenfeld 2000, Clarke-

Wood et al. 2016). Our results show that, independent of artificial light, landscape traits of a wetland influence the bat assemblage, and meant that the bat assemblage significantly varied among sites. Wetland size, vegetation cover and proximity to a major waterway influenced the bat assemblage recorded, consistent with previous research (Straka et al. 2016). Bat diversity was positively associated with the wetland size and the amount of vegetation in the surrounding 500 m radius, and negatively associated with the distance of a wetland to a major waterway. Wetland vegetation and trees can provide roosting habitat for bats (Campbell

2009), shelter for bats from extreme weather and predators (Rydell et al. 1996, Verboom and

Spoelstra 1999), and can even improve water quality (Busnardo et al. 1992). Our data also

156 confirm that rivers are essential riparian corridors that maintain biodiversity across a region

(Naiman et al. 1993). Although we measured the extent of other water bodies surrounding each wetland, the more important factor for the bat assemblage was the distance to a major river, confirming that the conservation of natural rivers is essential (Lintott et al. 2015).

Ecologically sensitive lighting

Artificial lighting at urban wetlands is directly and indirectly affecting many different taxa of biodiversity (Kuijper et al. 2008, Spoelstra et al. 2015, Straka et al. 2016). Red lights instead of white lights may reduce the disruptive short-term effect on some species in the urban insectivorous bat assemblage at wetlands. This effect may not be limited to bats; some urban populations of amphibians and reptiles reliant on urban wetlands also respond more to white than red light (Downs et al. 2003, Stokeld et al. 2014, Spoelstra et al. 2017, Hamer et al. 2018).

The use of red lights may need to be approached carefully however, as some species may still respond to red wavelengths, for example V. vulturnus in our study, and other bat species in recent research (Voigt et al. 2018, Zeale et al. 2018). This study focused on the effect of one single point of light, rather than many light sources causing more diffuse light pollution. A study replicating a real-life scenario where lighting entirely surrounds a wetland would be beneficial. Also, our study only measured short-term responses, and longer term experiments are urgently needed to understand if nocturnal fauna can become habituated to artificial light disruption, or whether the responses we measured are more permanent.

157 Many countries are updating their public lighting to LED technology for energy and cost

efficiency (OECD 2006, Ausgrid 2013) and this provides an exciting opportunity to install

lighting with custom spectral output, and colour temperature, sensitive to the local ecosystem.

For nocturnal fauna there is probably no substitute for dark habitat (Hölker et al. 2010).

However, changing spectral outputs of street lights to narrow spectrum, low ultraviolet

emittance and warmer colour temperature could help mitigate the impact of artificial lighting

on nocturnal fauna.

Acknowledgements

We would like to thank Alison Towerton from Greater Sydney Land Services for her local knowledge and help choosing study sites.

References

158 Arlettaz, R., S. Godat and H. Meyer (2000). "Competition for food by expanding pipistrelle bat populations (Pipistrellus pipistrellus) might contribute to the decline of lesser horseshoe bats (Rhinolophus hipposideros)." Biological Conservation 93(1): 55-60.

Ausgrid (2013). LED streetlight trial; Category P LED Luminarie Trial [White Paper], NSW Electricity Provider.

Basham, R., B. Law and P. Banks (2010). "Microbats in a leafy urban landscape: are they persisting, and what factors influence their presence?" Austral Ecology, 36(6), 663-678.

Bayley, W. A. (1966). “History of Campbelltown: New South Wales” Campbelltown Municipal Council.

Bennie, J., T. W. Davies, D. Cruse and K. J. Gaston (2016). "Ecological effects of artificial light at night on wild plants." Journal of Ecology 104(3): 611-620.

Benson, D. and J. Howell (1990). Sydney’s vegetation 1788–1988: utilization, degradation and rehabilitation. Proceedings of the Ecological Society of Australia (Vol. 16, No. 1, pp. 115-27).

Benson, D. H., J. Howell (1995). Taken for granted: the bushland of Sydney and its suburbs, Kangaroo Press in association with the Royal Botanic Gardens Sydney.

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

Brabec, E., S. Schulte and P. L. Richards (2002). "Impervious surfaces and water quality: a review of current literature and its implications for watershed planning." Journal of Planning Literature 16(4): 499-514.

Bullen, R., N. McKenzie and A. Cruz-Neto (2016). "Characteristic flight speeds in bats." CEAS Aeronautical Journal 7(4): 621-643.

Burton, R. (2005). "Recovering bushland on the Cumberland Plain: Best practice guidelines for the management and restoration of bushland." NSW Department of Environment and Conservation, Sydney, Australia.

159

Busnardo, M. J., R. M. Gersberg, R. Langis, T. L. Sinicrope and J. B. Zedler (1992). "Nitrogen and phosphorus removal by wetland mesocosms subjected to different hydroperiods." Ecological Engineering 1(4): 287-307.

Campbell, S. (2009). "So long as it’s near water: variable roosting behaviour of the large-footed myotis (Myotis macropus)." Australian Journal of Zoology 57(2): 89-98.

Clarke, K. and R. Green (1988). "Statistical design and analysis for a 'biological effects' study." Marine Ecology Progress Series: 213-226.

Clarke-Wood, B. K., K. M. Jenkins, B. S. Law and R. V. Blakey (2016). "The ecological response of insectivorous bats to coastal lagoon degradation." Biological Conservation 202: 10-19.

Ehrenfeld, J. G. (2000). "Evaluating wetlands within an urban context." Urban Ecosystems 4(1): 69-85.

Eisenbeis, G., C. Rich and T. Longcore (2006). "Artificial night lighting and insects: attraction of insects to streetlamps in a rural setting in Germany." Ecological Consequences of Artificial Night Lighting 2: 191-198.

Fenton, M. B., L. Acharya, D. Audet, M. Hickey, C. Merriman, M. Obrist, D. Syme and B. Adkins (1992). "Phyllostomid bats (Chiroptera: Phyllostomidae) as indicators of habitat disruption in the Neotropics." Biotropica: 440-446.

Furlonger, C., H. Dewar and M. Fenton (1987). "Habitat use by foraging insectivorous bats." Canadian Journal of Zoology 65(2): 284-288.

Gaston, K. J., T. W. Davies, J. Bennie and J. Hopkins (2012). "REVIEW: Reducing the ecological consequences of night‐time light pollution: options and developments." Journal of Applied Ecology 49(6): 1256-1266.

160 Grindal, S. D. and R. M. Brigham (1998). "Short-term effects of small-scale habitat disturbance on activity by insectivorous bats." The Journal of Wildlife Management: 996-1003.

Hamer, A. J., L. J. Harrison and D. Stokeld (2018). "Terrestrial habitat and individual fitness increase survival of a freshwater turtle in an urban landscape." Urban Ecosystems 21(1): 71-83.

Hamer, A. J. and M. J. McDonnell (2008). "Amphibian ecology and conservation in the urbanising world: a review." Biological Conservation 141(10): 2432-2449.

Heiling, A. M. (1999). "Why do nocturnal orb-web spiders (Araneidae) search for light?" Behavioral Ecology and Sociobiology 46(1): 43-49.

Heithaus, E. R. (1974). "The role of plant-pollinator interactions in determining community structure." Annals of the Missouri Botanical Garden 61(3): 675-691.

Hölker, F., C. Wolter, E. K. Perkin and K. Tockner (2010). "Light pollution as a biodiversity threat." Trends in Ecology & Evolution (25): 681-682.

Law, B., A. Doty, M. Chidel and T. Brassil (2018). "Bat activity before and after a severe wildfire in Pilliga forests: Resilience influenced by fire extent and landscape mobility?" Austral Ecology 43, 706–718.

Law, B. and C. A. Urquhart (2000). "Diet of the Large-footed Myotis Myotis macropus at A Forest Stream Roost in Northern New South Wales." Australian Mammalogy 22(2): 121-124.

Lewanzik, D. and C. C. Voigt (2017). "Transition from conventional to light‐emitting diode street lighting changes activity of urban bats." Journal of Applied Ecology 54(1): 264-271.

161 Lima, S. L. and J. M. O'Keefe (2013). "Do predators influence the behaviour of bats?" Biological Reviews 88(3): 626-644.

Lintott, P. R., N. Bunnefeld and K. J. Park (2015). "Opportunities for improving the foraging potential of urban waterways for bats." Biological Conservation 191: 224-233.

Longcore, T. and C. Rich (2004). "Ecological light pollution." Frontiers in Ecology and the Environment 2(4): 191-198.

McConville, A., B. Law, T. Penman and M. Mahony (2014). "Contrasting habitat use of morphologically similar bat species with differing conservation status in south‐eastern Australia." Austral Ecology 39(1): 83-94.

Müller, B., M. Glösmann, L. Peichl, G. C. Knop, C. Hagemann and J. Ammermüller (2009). "Bat eyes have ultraviolet-sensitive cone photoreceptors." PloS ONE 4(7): e6390.

Naiman, R. J., H. Decamps and M. Pollock (1993). "The role of riparian corridors in maintaining regional biodiversity." Ecological Applications 3(2): 209-212.

Organisation for Economic Co-operation and Development and International Energy Agency (2006). "Light's labour's lost: policies for energy-efficient lighting." Energy World(343): 14-15.

Patriquin, K. J., L. K. Hogberg, B. J. Chruszcz and R. M. Barclay (2003). "The influence of habitat structure on the ability to detect ultrasound using bat detectors." Wildlife Society Bulletin: 475- 481.

Pennay, M., B. Law and L. Reinhold (2004). "Bat Calls of NSW: region based guide to the echolocation calls of Microchiropteran bats. NSW Dept." Environment and Conservation, Hurstville.

Perkin, E. K., F. Hölker and K. Tockner (2014). "The effects of artificial lighting on adult aquatic and terrestrial insects." Freshwater Biology 59(2): 368-377.

Pintérné, E. N. and Z. Pödör (2017). "The Effect of Artificial Lights on Nocturnal Macrolepidoptera (Lepidoptera: Macroheterocera) Communities." Acta Silvatica et Lignaria Hungarica 13(1): 41-54.

Poot, H., B. J. Ens, H. de Vries, M. A. Donners, M. R. Wernand and J. M. Marquenie (2008). "Green light for nocturnally migrating birds." Ecology and Society 13(2).

Potts, S. G., B. Vulliamy, A. Dafni, G. Ne'eman and P. Willmer (2003). "Linking bees and flowers: how do floral communities structure pollinator communities?" Ecology 84(10): 2628- 2642.

162 Raven, J. and C. Cockell (2006). "Influence on photosynthesis of starlight, moonlight, planetlight, and light pollution (reflections on photosynthetically active radiation in the universe)." Astrobiology 6(4): 668-675.

Reinhold, L., B. Law, G. Ford and M. Pennay (2001). "Key to the bat calls of south-east Queensland and north-east New South Wales." Forest Ecosystem Research and Assessment Technical Paper 7.

Rich, C. and T. Longcore (2013). “Ecological consequences of artificial night lighting” Island Press.

Rydell, J. (1992). "Exploitation of insects around streetlamps by bats in Sweden." Functional Ecology: 744-750.

Rydell, J., A. Entwistle and P. A. Racey (1996). "Timing of foraging flights of three species of bats in relation to insect activity and predation risk." Oikos: 243-252.

Scanlon, A. T. and S. Petit (2008). "Effects of site, time, weather and light on urban bat activity and richness: considerations for survey effort." Wildlife Research 35(8): 821-834.

Smith, L. A. and P. Chow-Fraser (2010). "Impacts of adjacent land use and isolation on marsh bird communities." Environmental Management 45(5): 1040-1051.

Speakman, J. (1991). "The impact of predation by birds on bat populations in the British Isles." Mammal Review 21(3): 123-142.

Stokeld, D., A. J. Hamer, R. van der Ree, V. Pettigrove and G. Gillespie (2014). "Factors influencing occurrence of a freshwater turtle in an urban landscape: a resilient species?" Wildlife Research 41(2): 163-171.

Stone, E. L., G. Jones and S. Harris (2009). "Street lighting disturbs commuting bats." Current Biology 19(13): 1123-1127.

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

Threlfall, C. G., B. Law and P. B. Banks (2012). "Influence of landscape structure and human modifications on insect biomass and bat foraging activity in an urban landscape." PloS ONE 7(6): 10.

Threlfall, C. G., B. Law and P. B. Banks (2012). "Sensitivity of insectivorous bats to urbanization: implications for suburban conservation planning." Biological Conservation 146(1): 41-52.

van Langevelde, F., J. A. Ettema, M. Donners, M. F. WallisDeVries and D. Groenendijk (2011). "Effect of spectral composition of artificial light on the attraction of moths." Biological Conservation 144(9): 2274-2281.

Verboom, B. and K. Spoelstra (1999). "Effects of food abundance and wind on the use of tree lines by an insectivorous bat, Pipistrellus pipistrellus." Canadian Journal of Zoology 77(9): 1393- 1401.

Walsh, C. J., A. K. Sharpe, P. F. Breen and J. A. Sonneman (2001). "Effects of urbanization on streams of the Melbourne region, Victoria, Australia. I. Benthic macroinvertebrate communities." Freshwater Biology 46(4): 535-551.

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

164 Chapter 6: General Discussion

Overview of findings

Many factors that contribute to how a bat species responds to light. The overarching aim of

this thesis was to investigate the impact of artificial light on insectivorous bats in an urban landscape, assessing this in a number of contexts from multiple perspectives. My research fills gaps in knowledge on the effects of light source location and spectral emission on bat foraging and commuting behaviour. The findings also confirm that dark urban forests, dark waterways and dark wetlands in cities are critical for the persistence of high urban bat diversity. However, the way in which different bats respond to artificial light are highly contextual, and simple statements of the impacts of lights on bats are unlikely to be satisfying. The responses to artificial light differ significantly across the bat assemblage and are consistently well predicted by a species’ guild, although responses could be context-dependent for some species. In

Figure 6.1 I outline the ways in which understanding habitat preferences, light types and regimes, and bat traits may contribute to a deeper understanding of how bats respond to the introduction of acritical light in urban ecosystems.

Bats that preferentially forage in cluttered vegetation are light avoidant, but Chapters 2 and 3 demonstrate that the response by this guild to light is probably more driven by preference for cluttered vegetation than by light avoidance, shown in their high diversity in forest interiors.

Species that forage and commute across open spaces seem either unaffected by ALAN and are driven by other factors such as humidity (Chapter 2) or increase in lit areas, probably to hunt

165 insects attracted to the ultraviolet light (Chapters 2, 3 and 5). Species such as Micronomous norfolkensis increased in activity around a single light at a wetland (Chapter 5) and Ozimops ridei utilised artificially lit habitat along the urban forest edge (Chapter 3). A finer scale approach to to how individual bats responded to light highlighted the highly variable nature of these responses and the challenges of assessing individual responses to anthropogenic stressors at landscape scales (Chapter 4).

Figure 6.1. A framework demonstrating the mechanisms that predict a bat’s response to artificial light, with some of these mechanisms tested in this thesis. Underlined words are the research topics focused on in this thesis, asterisks mark the fields where appropriate management of public lighting can have the greatest effect, and italics mark some suggestions for future research topics. The white arrow shows the broader to finer scale mechanisms that drive a bat’s response to light, and the darker line demonstrates that the morphology of a species is linked to its habitat preference.

My findings have illustrated that edge-space adapted bats, evolved to forage and commute along habitat edges, have diverse responses to artificial light and also seem to show context-

166 dependent responses. In Chapter 3 Vespadelus vulturnus an edge-space adapted bat with a high characteristic call frequency (between 48 and 54 kHz, Adams et al. 2009) decreased in activity at well-lit urban forest edges, but in Chapter 5 increased in activity around peri-urban white and red lights at wetlands. This may be due to the highly urbanised and light polluted nature of the urban forest study site, when compared with the point source nature of the wetland lights. Light traps attracted more insects in a rural setting than in an urban setting

(Pintérné and Pödör 2017), probably due to a low level of surrounding light pollution and therefore a more conspicuous light source. Vespadelus vulturnus therefore could have been exploiting this concentrated rural food source by flying around red and white lights at wetlands

(Chapter 5), but that benefit was not as great at the urban forests. Indeed, this species was attracted to artificial lights in a rural context in a previous study (Adams et al. 2005). This highlights how important context may be in driving a bats’ response to artificial light and demonstrates that a fully replicated and nested study is required to disentangle the context- dependent responses of these bats to light.

Street lights in many countries are being updated from older technologies to LED without full understanding of the ecological impact. I found that slow flying clutter-adapted bats, edge- space adapted bats with high characteristic echolocation call frequency (ESH bats), and some fast flying bats all decreased activity after a change to LED streetlights (Chapter 2). I also found that the bat assemblage was significantly different in small patches of dark urban forest compared with well-lit streets only a few hundred metres away (Chapter 2), confirming that dark urban forests are important in the persistence of urban bat populations.

167 These findings led me to test the hypothesis that street lights surrounding these dark urban

forests could be decreasing the amount of functional habitat for sensitive bat species. I found

evidence that supported this hypothesis, and that lights along edges of urban forests

negatively affected the activity of slow flying clutter-adapted and ESH bats, and also delayed their emergence time (Chapter 3). The activity and emergence time of faster flying bats were unaffected by lights along the edges of urban forests (Chapter 3). My data support the emerging global pattern that slow flying clutter-adapted species are light avoidant (Threlfall et al. 2013, Stone et al. 2015), but I also provide new evidence that ESH bats may also be at particular risk from light pollution.

Guild-specific responses to the introduction of light are also likely to interact with habitat preference. It has been experimentally established that at least one species of trawling bat in the Netherlands avoids artificial light (Kuijper et al. 2008), and this light avoidant pattern has been observed but not experimentally tested in Australia (Straka et al. 2016). This thesis provides pilot data on individual behaviour (Chapter 4) and a fully replicated study (Chapter 5) on the effect of introducing lights to waterways and wetlands. I introduced artificial lights to a waterway and found that an urban trawling bat highly reliant on riparian habitat, Myotis macropus, may be have their habitat severely impacted by light pollution (Chapter 4). This species had a substantial negative response to the introduction of LED lights, and did not fully recover in the period immediately after ambient darkness was restored. Red lights had been found to be an effective mitigation strategy, having no effect on slow flying forest bats in a rural setting (Spoelstra et al. 2017), however their mitigative effect needs further testing.

168 Hence, I tested the effect of red versus white lights at peri urban wetlands. I compared the impact of experimentally introducing red or white lights at wetlands, and demonstrated that the bat assemblage as a whole was disrupted by white lights but showed no response to red lights (Chapter 5), although there were diverse species-specific responses. My findings illustrate the importance of ecologically sensitive lighting choices in cities, in terms of both the spectra and the location of light sources. The popularity of red lights with human residents has yet to be tested.

Emerging themes

Habitat preference: the importance of dark connected habitat

Connected urban forests and natural areas can protect against species loss in cities

(Lindenmayer and Nix 1993, Fischer and Lindenmayer 2007, Dearborn and Kark 2010, Beninde et al. 2015). Corridors and patches of functional habitat allow for movement and genetic diversity across populations (Aguilar et al. 2008), with many anthropogenic processes threatening this functional habitat. My research demonstrates that dark urban forests are imperative for high urban bat diversity (Figure 6.1, and Chapters 2 and 3), and that urban forests should have dark edges to maximise their conservation value to bats and other nocturnal fauna. Wetlands and waterways are also important functional natural habitat, however are commonly light polluted in urban areas (Straka et al. 2016). Much like urban forests, the habitat value of urban waterways is significantly impacted by artificial light

(Chapter 4 and 5). Arguably the best way to protect the functionality of forest edges, wetlands and waterways may be to leave dark areas or to select a coloured light that has minimal impact

169 on the local ecosystem. This reinforces the fundamental role of conserving high quality habitat

in urban ecosystems as central to conserving biodiversity in these systems.

The importance of guilds

Some bat species are able to exploit urban areas, whilst others decline (Duchamp and Swihart

2008, Jung and Threlfall in press). Certain morphological traits associated with adaptation to a

cluttered environment, like low aspect ratio (wingspan2 / wing area) which is linked to low flight speed but higher manouverability, seem to be indicators of urban sensitivity (Duchamp and

Swihart 2008, Jung and Threlfall in press). A global consensus is emerging that slow flying, clutter-adapted bats are most sensitive to artificial light, while faster flying, open-space adapted bats are able to exploit lit areas (Kuijper et al. 2008, Threlfall et al. 2013, Lewanzik and

Voigt 2017, Spoelstra et al. 2017, Rowse et al. 2018, Russo et al. 2018). The findings presented in this thesis provide support for this consensus, with slow flying clutter-adapted species are less likely to be found in lit areas or along lit forest edges (Chapters 2 and 3). My thesis goes further and finds diverse responses to light from edge-space adapted bats. Edge-space adapted species with high echolocation call frequency were negatively impacted by artificial light in an urban setting (Chapters 2 and 3), but that effect seemed context-dependent

(Chapter 5).

Lighting regime and light type

Different lighting technologies, and now the implementation of LED street lights globally, have sparked investigation into the various ecological effects of different wavelengths of light.

170 Many different technologies emit varying spectral signatures with peaks in different parts of

the electromagnetic spectrum. Responses of taxa to different parts of the spectrum, spectra-

dependent responses, are highly variable across taxa and species (Table 1.1 in the general introduction), and research is urgently needed to understand how species and communities respond to different spectra. If properly understood, spectra-dependent responses provide an

opportunity to design light sources that emit less disruptive wavelengths for the local

ecosystem. Along with other studies (Stone et al. 2012, Wakefield et al. 2015, Lewanzik and

Voigt 2017, Rowse et al. 2018), my research has shown that bats exhibit spectra dependant

responses (Chapter 2 and Chapter 5) and adds to existing knowledge on this subject by

demonstrating community-level responses to different coloured lights at wetlands.

In many cases, LED street lights can be manufactured to emit almost any spectral signature,

and much research has looked at the effects of different colour temperatures and spectra

(Pawson and Bader 2014, van Grunsven et al. 2014) with warmer LEDs (lower colour

temperature and kelvins) seemingly the least impactful for some bats (Voigt et al. 2018). LEDs

with many different colour temperatures are already being used in public lighting. This is a very

relevant issue in Sydney, Australia, as LED street lights are being implemented over the next

few years (Ausgrid 2013) and research may be able to inform technology choices. These

possibilities highlight the potential for integrative approaches to light pollution work, including

sensory biology for bats.

171 Morphological traits may predict which species can exploit the ‘night light niche’ and which ones cannot. Slow flying bats and trawling bats are some of the most negatively impacted by artificial light, but interestingly do not always show a negative response to red light (Spoelstra et al. 2017 and Chapter 5). This lack of response to red light may not always be consistent. Red light did significantly reduce the commuting activity of one species, Rhinolophus hipposideros,

(Zeale et al. 2018) but the results may have been exacerbated as the lights created an obstacle on a linear commuting route. The effect was also not recorded for all species, including for

Myotis spp. that did not show a significant response to red light (Zeale et al. 2018). The exact drivers of these differing results, ie. the comparison of foraging and commuting or the study of highly light avoidant species like Rhinolophus hipposideros, is worth considering when interpreting results in this area. It is particularly pertinent as some instances of red street lighting are already in place to protect bats. This thesis presents evidence confirming that at least for some bat communities and species like Myotis macropus, red lights are an effective mitigation tool for urban wetlands, when public lighting is absolutely necessary (Chapter 5).

Management implications

An intended outcome of this thesis was to provide practical data and recommendations for policy makers and land managers, so that public lighting decisions can be informed by ecological understanding. This is no trivial issue, as many street lights in Australia and elsewhere are being changed to LED lights, and having ecological consequences that we do not fully understand. Aspects that may be considered in future planning decisions include;

172 • Changing the spectra of street lights can affect the bat community. Lights create an

artificial feeding ground for fast and sometimes slow flying bats and exclude other

species entirely. This has potential to increase competition for food and roosts in cities.

To mitigate the disruptive effect of light, narrow spectrum lighting (lights that emit

wavelengths in only a narrow part of the visible spectrum) like red lights may be the most

appropriate.

• Edges of urban forests that are high in bat diversity should be protected from excessive

light pollution. When lighting is necessary, the use of appropriate spectra and wide gaps

between light poles may help to mitigate the effects of light pollution.

• Urban waterways and wetlands should be protected from excessive lighting, as it

disrupts the local ecosystem. These habitats are important foraging grounds and

commuting routes across fragmented habitat in cities, but currently wetland

conservation guidelines do not mention public lighting (SEPP14, NSW Government

1985).

Further research is also urgently needed on the impact of different light mitigation strategies.

Motion-censored street lighting systems are being suggested as a way to reduce carbon emissions and cost of urban lighting (OECD 2006). Motion-sensored street lighting will also have ecological impacts, either positive by reducing overall artificial lighting and therefore reducing ecological disruption, or negative by causing startle responses for fauna when the lights turn on unexpectedly. There are many emerging opportunities to test ecological responses to motion-sensored lights using natural experiments that arise, and a properly replicated before-after-control-impact experiment could elucidate whether this technology is a

173 successful lighting strategy. Data on the effectiveness of these lighting systems are needed, as

they could be an efficient way to reduce urban light pollution.

Future research

Trophic interactions and bat biological traits

Future research should also focus on how artificial light can cause different types of

competition between bat species. In figure 6.1, I highlight areas that need further investigation

in order to understand the impacts artificial light is having on nocturnal urban ecosystems.

Faster flying bats are more successful at foraging around streetlights (Rydell 1992) and may

then have a larger population in the surrounding urban and forest area which could have

implications for roost availability for slower flying species that reside exclusively in the forested

areas. There may be increased competition for a finite insect food source (Arlettaz et al. 2000,

Kuijper et al. 2008), or light sources may cause a “vacuum cleaner effect” (Arlettaz et al. 2000)

where insects are attracted out of dark urban forests and towards well-lit foraging grounds that slower flying bats do not utilise as much as faster flying bats. However the extent to which this occurs is unknown.

My research demonstrates the diverse species’ responses to artificial light, and there are many direct and indirect mechanisms that are driving these differences. One possible indirect driver of light phobia is predator avoidance. There is an assumption in the literature that slower flying bats are more at risk or fearful of predation in lit areas (Jones and Rydell 1994, Rydell et al.

1996, Stone et al. 2009), however this has never been experimentally tested. A fully nested

174 experimental design, with predator noises, model visual predators and artificial lights, would help to understand whether predator presence changes bat behaviour, and if behaviour changes more so in lit conditions.

Another possible direct driver of light phobia in some species is a sensitivity to ultraviolet light and other wavelengths (Müller et al. 2009, Xuan et al. 2012, Xuan et al. 2012, Gorresen et al.

2015). Research on this has never been conducted on any species in Australia but may be the key to different species responses to artificial light. Immunohistochemical evidence from species and laboratory-based behavioural responses to different wavelengths would help to test the theory of ultraviolet sensitivity in Australian bats.

Concluding remarks

The field of ecological light pollution research has grown tremendously in the last two decades,

and the impacts on astronomical viewing and ecosystems are beginning to be understood

(Falchi et al. 2011, Falchi et al. 2016). Research is now looking at spectra-dependent responses

for multiple taxa (Ouyang et al. 2015, Da Silva et al. 2017, De Jong et al. 2017), with fitness

implications and costs still relatively little understood. Regarding insectivorous bats, some

species are able to exploit the ‘night light niche’, while some are being excluded (Stone et al.

2015), and this can be mitigated by choosing appropriate locations and spectra for public

lighting. Yet in order to enable effective urban planning on city-wide scales, it is imperative to understand how public lighting decisions affect bat populations at the landscape scale. My research (Chapter 5) has touched on this, by suggesting that Myotis macropus probably moved

175 out of artificially lit wetlands and towards wetlands left in ambient darkness. Micronomous

norfolkensis moved out of dark control sites and towards white light sites during the

experiment. This attractive effect may differ under lighting regimes, where for example an

entire wetland is floodlit, but this needs further investigation. A comparison of point source

lighting and floodlighting at wetlands could be used to investigate whether there is a threshold

level of lighting, over which species are unable to utilise the wetland. On a broader scale, these

movements across the landscape in response to introduced lights could be causing species to

move away from and decline in light polluted urban areas over time. Light could be a major

contributing factor in facilitating the homogenization of the urban bat populations (McKinney

2006).

References

Aguilar, R., M. Quesada, L. Ashworth, Y. Herrerias‐Diego and J. Lobo (2008). "Genetic consequences of habitat fragmentation in plant populations: susceptible signals in plant traits and methodological approaches." Molecular Ecology 17(24): 5177-5188.

Ausgrid (2013). LED streetlight trial; Category P LED Luminarie Trial [White Paper], NSW Electricity Provider.

176 Beninde, J., M. Veith and A. Hochkirch (2015). "Biodiversity in cities needs space: a meta‐ analysis of factors determining intra‐urban biodiversity variation." Ecology Letters 18(6): 581- 592.

Dearborn, D. C. and S. Kark (2010). "Motivations for conserving urban biodiversity." Conservation Biology 24(2): 432-440.

Duchamp, J. E. and R. K. Swihart (2008). "Shifts in bat community structure related to evolved traits and features of human-altered landscapes." Landscape Ecology 23(7): 849-860.

Fischer, J. and D. B. Lindenmayer (2007). "Landscape modification and habitat fragmentation: a synthesis." Global Ecology and Biogeography 16(3): 265-280.

Gorresen, P. M., P. M. Cryan, D. C. Dalton, S. Wolf and F. J. Bonaccorso (2015). "Ultraviolet vision may be widespread in bats." Acta Chiropterologica 17(1): 193-198.

Jung, K. and C. G. Threlfall (in press). " Trait-dependant tolerance of bats to urbanisation: a global meta-analysis." Proceedings of the Royal Society B.

Lewanzik, D. and C. C. Voigt (2017). "Transition from conventional to light‐emitting diode street lighting changes activity of urban bats." Journal of Applied Ecology 54(1): 264-271.

Lindenmayer, D. B. and H. A. Nix (1993). "Ecological principles for the design of wildlife corridors." Conservation Biology 7(3): 627-630.

177 McKinney, M. L. (2006). "Urbanization as a major cause of biotic homogenization." Biological Conservation 127(3): 247-260.

Müller, B., M. Glösmann, L. Peichl, G. C. Knop, C. Hagemann and J. Ammermüller (2009). "Bat eyes have ultraviolet-sensitive cone photoreceptors." PloS one 4(7): e6390.

N. S. W. Government (1985). "State Enviornmental Planning Policy No. 14 - Coastal Wetlands." Department of Urban Affairs and Planning, Sydney.

Organisation for Economic Co-operation and Development and International Energy Agency (2006). "Light's labour's lost: policies for energy-efficient lighting." Energy World(343): 14-15.

Pawson, S. and M.-F. Bader (2014). "LED lighting increases the ecological impact of light pollution irrespective of color temperature." Ecological Applications 24(7): 1561-1568.

Pintérné, E. N. and Z. Pödör (2017). "The Effect of Artificial Lights on Nocturnal Macrolepidoptera (Lepidoptera: Macroheterocera) Communities." Acta Silvatica et Lignaria Hungarica 13(1): 41-54.

Rowse, E. G., S. Harris and G. Jones (2018). "Effects of dimming light-emitting diode street lights on light-opportunistic and light-averse bats in suburban habitats." Royal Society Open Science 5(6): 180205.

Russo, D., L. Cistrone, N. Libralato, C. Korine, G. Jones and L. Ancillotto (2017). "Adverse effects of artificial illumination on bat drinking activity." Animal Conservation 20(6): 492-501.

Rydell, J. (1992). "Exploitation of insects around streetlamps by bats in Sweden." Functional Ecology: 744-750.

Rydell, J., A. Entwistle and P. A. Racey (1996). "Timing of foraging flights of three species of bats in relation to insect activity and predation risk." Oikos: 243-252.

Spoelstra, K., R. H. van Grunsven, J. J. Ramakers, K. B. Ferguson, T. Raap, M. Donners, E. M. Veenendaal and M. E. Visser (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, The Royal Society 284(1855)..

Stone, E. L., G. Jones and S. Harris (2009). "Street lighting disturbs commuting bats." Current Biology 19(13): 1123-1127.

178 Stone, E. L., G. Jones and S. Harris (2012). "Conserving energy at a cost to biodiversity? Impacts of LED lighting on bats." Global Change Biology 18(8): 2458-2465.

Threlfall, C., B. Law and P. Banks (2013). "The urban matrix and artificial light restricts the nightly ranging behaviour of Gould's long‐eared bat (Nyctophilus gouldi)." Austral Ecology 38(8): 921-930. van Grunsven, R. H., M. Donners, K. Boekee, I. Tichelaar, K. Van Geffen, D. Groenendijk, F. Berendse and E. Veenendaal (2014). "Spectral composition of light sources and insect phototaxis, with an evaluation of existing spectral response models." Journal of Insect Conservation 18(2): 225-231.

Xuan, F., K. Hu, T. Zhu, P. Racey, X. Wang and Y. Sun (2012). "Behavioral evidence for cone- based ultraviolet vision in divergent bat species and implications for its evolution." Zoologia 29(2).

179