The environmental impacts of artificial light at night have been a rapidly growing field of global change science in recent years. Yet, light pollution has not achieved parity with other global change phenomena in the level of concern and interest it receives from the scientific community, government and nongovernmental organizations. This is despite the globally widespread, expanding and changing nature of night‐time lighting and the immediacy, severity and phylogenetic breath of its impacts. In this opinion piece, we evidence 10 reasons why artificial light at night should be a focus for global change research in the 21st century. Our reasons extend beyond those concerned principally with the environment, to also include impacts on human health, culture and biodiversity conservation more generally. We conclude that the growing use of night‐time lighting will continue to raise numerous ecological, human health and cultural issues, but that opportunities exist to mitigate its impacts by combining novel technologies with sound scientific evidence. The potential gains from appropriate management extend far beyond those for the environment, indeed it may play a key role in transitioning towards a more sustainable society.

INTRODUCTION While artificial light at night (ALAN) has been a long established man‐made disturbance (Longcore & Rich, 2004), the number of studies documenting its ecological and human health impacts has grown dramatically in the last decade (Figure 1). Collectively, this body of research now highlights the pervasiveness of ALAN's impacts across a broad array of biomes, ecosystems, species and behaviours. The measured biological responses occur at intensities and spectra of artificial light that are currently encountered in the environment, and the global distribution of night‐time lighting means that it is likely already having widespread impacts in marine, freshwater and terrestrial habitats around the world. Figure 1 Open in figure viewerPowerPoint The trend in research outputs associated with light pollution and climate change since the year 2000. Bar heights represent the cumulative number of articles expressed as a percentage of the total number of articles published by the end of 2016; numbers are the cumulative number of articles published by the end of each year. Note that the total number of articles does not reflect the total number published in the research area, only the number returned from the search. The data were collected from a Web of Science search for phrases in article titles. The search phrases used for light pollution research outputs were “Light pollution” OR “Artificial Light at Night” OR “Nighttime lighting” OR “Night‐time lighting” OR “Night time lighting” OR “Street Lighting” OR “LED lighting” OR “Light‐emitting diode lighting.” The search phrase for climate change was “Climate change” and results were not refined by research area. The search for articles on light pollution was refined by research areas: (Plant Sciences or Ornithology or Psychology Multidisciplinary or Environmental Sciences or Evolutionary Biology or Physics Applied or Entomology or Engineering Environmental or Ecology or Urban Studies or Fisheries or Biodiversity Conservation or Biology or Physics Multidisciplinary or Zoology or Oceanography or Geography Physical or Geography or Remote Sensing or Physiology or Marine Freshwater Biology or Public Environmental Occupational Health) While ALAN research has gained notable momentum in recent years, it is yet to achieve notoriety among environmental scientists as a driver of global change. Here, we argue that ALAN should be a focus for global change research in the 21st century. Our argument is broken down into 10 points that highlight the global extent of ALAN, the geographic scale of its influence, the potential to reverse its environmental impacts, the rise of new human–environment conflicts with emerging lighting technologies, its evolutionary novelty, the diverse array of species now known to be affected, the extreme sensitivity of organisms to light, impacts on human health, cultural impacts on human–environment interrelationships, and the feasibility of solutions. While we do not assert that ALAN is more important than other global change phenomena, we draw comparisons where they help highlight the need for greater parity of concern.

1 GLOBALLY WIDESPREAD As with greenhouse gas emissions, ALAN is a globally widespread environmental pollutant. It is estimated that 23% of the land surface between 75°N and 60°S (Falchi et al., 2016) is exposed to artificial skyglow (artificial light that is scattered in the atmosphere and reflected back to the ground). This is comparable to the area of global ice‐free land converted to either pasture or cropland, estimated to be 35% in the year 2000 (Klein Goldewijk, Beusen, van Drecht, & De Vos, 2011). The degree of exposure increased in all global terrestrial ecosystems between 2008 and 2012, with those important for biodiversity conservation often most affected (Bennie, Duffy, Davies, Correa‐Cano, & Gaston, 2015). Exposure to ALAN is not limited to terrestrial environments, with current best estimates indicating that 22% of the worlds’ coastal regions (Davies, Duffy, Bennie, & Gaston, 2014) are experiencing some degree of artificial illumination and 20% of marine‐protected areas are exposed across their entire range (Davies, Duffy, Bennie, & Gaston, 2016). The amount of artificial light is also increasing in 13,061 terrestrial protected areas across Europe, Asia and South and Central America (Gaston, Duffy, & Bennie, 2015) and 1,687 (14.7%) of the world's marine‐protected areas (Davies et al., 2016). Given that more than 95% of global population increases are projected to occur in the cities of economically developing countries over the next 50 years (Grimm et al., 2008), and levels of light pollution are closely associated with population density and economic activity (Gallaway, Olsen, & Mitchell, 2010); ALAN will continue to expand both in spatial extent and intensity throughout the 21st century without intervention.

2 SPHERE OF INFLUENCE Artificial light arises from point sources (municipal, industrial, commercial and residential), giving the impression that its impacts on the environment are highly localized. Indeed, the majority of studies into the ecological impacts of ALAN quantify responses to direct lighting (Gaston, Visser, & Hölker, 2015). Artificial skyglow increases the sphere of ALAN's potential influence far beyond a patch of habitat in the vicinity of a street light (Falchi et al., 2016; Kyba & Hölker, 2013). Numerous taxa are adapted to make use of spatial and temporal patterns of natural sky brightness at intensities equivalent to or less than those created by artificial skyglow (Dacke, Baird, Byrne, Scholtz, & Warrant, 2013; Last, Hobbs, Berge, Brierley, & Cottier, 2016; Moore, Pierce, Walsh, Kvalvik, & Lim, 2000; Naylor, 1999; Warrant & Dacke, 2016), suggesting that lights in urban centres will have impacts on environments tens to hundreds of kilometres away. A dung beetle navigating its landscape using the Milky Way could, for example, become disorientated by artificial skyglow from a city tens or perhaps even hundreds of kilometres away (Kyba & Hölker,2013), an effect comparable to a moth becoming disorientated by a street light hundreds of metres away (van Grunsven, Lham, van Geffen, & Veenendaal, 2014). While ALAN can be misconstrued as being a highly localized anthropogenic stressor, climate warming is likewise misrepresented as globally widespread in its occurrence. Like ALAN, ecologically relevant warming occurs at more localized spatial scales (Hannah et al., 2014) (Figure 2) and is influenced by variable topographical features such as slope and aspect that create refuges where rates of warming are reduced (Bennie, Huntley, Wiltshire, Hill, & Baxter, 2008; Maclean, Suggitt, Wilson, Duffy, & Bennie, 2016). The ecological impacts of climate change — like light pollution — are therefore likely to be spatially heterogeneous for organisms with low mobility, but more widespread for taxa that depend on large‐scale movements for their survival and reproduction. In the case of both stressors, population impacts on the former species are manifest foremost through the loss and fragmentation of suitable habitat (Hannah et al., 2014), while impacts on the latter species are manifest via direct effects on population demography (Gaston & Bennie, 2014). Figure 2 Open in figure viewerPowerPoint 2016 A comparison of fine‐scale spatial variability in environmental warming and artificial light at night on the Lizard peninsula, Cornwall, UK. (a) The increase in the number of growing degree days ( a measure of change in growing season length expressed in °C Days) between 1977 and 2014 (100 m resolution). Adapted with permission from Maclean et al. (). (b) The distribution of artificial light across the same area (750 m resolution) recorded from the VIIRS sensor on board the Suomi NPP satellite

3 LAG EFFECTS Abating future rises in global temperatures constitutes one of the most significant challenges facing humanity in the 21st century. Yet even if all fossil fuel combustion ceased with immediate effect, the recovery of atmospheric CO 2 concentrations, global temperatures, ocean pH and oxygen concentrations to preindustrial levels would take hundreds to thousands of years (Frölicher & Paynter, 2015; Frolicher, Winton, & Sarmiento, 2014; Mathesius, Hofmann, Caldeira, & Schellnhuber, 2015), and there is the very real possibility that temperatures would continue to rise in the medium term (Frolicher et al., 2014). In contrast, globally widespread artificial light can be “switched off” instantaneously. There would be no lag effect on the physical environment following such an event, allowing the biological environment to immediately begin the recovery process. While such a scenario would likely prove controversial, recent technological advances present tangible ways of mitigating the ecological impacts of artificial light at night (see reason 10). Failure to abate the environmental consequences of a man‐made disturbance using available viable solutions would not inspire confidence in our ability to solve the apparently insurmountable challenges posed by global climate change phenomena.

4 THE RISE OF LEDS Light‐Emitting Diodes (LEDs) have grown from a 9% share of the lighting market in 2011 to 45% in 2014 and are forecast to reach 69% by 2020 (Zissis & Bertoldi, 2014). Their rising popularity stems from the variety of colours that LEDs can be tailored to produce, their improved energy efficiency over alternative electric light sources, and ability to produce “white” light that is aesthetically pleasing and provides enhanced visual performance (Pimputkar, Speck, Denbaars, & Nakamura, 2009; Schubert & Kim, 2005). While LEDs are often advocated for their potential to reduce global CO 2 emissions and the ability to tailor their spectra to avoid unwanted environmental impacts (see “Feasibility of solutions”), environmental scientists and human health experts have raised concerns about the broad‐spectrum light (Davies, Bennie, Inger, De Ibarra, & Gaston, 2013; MacGregor, Pocock, Fox, & Evans, 2014) and prominent short wavelength peak (Haim & Portnov, 2013; Haim & Zubidat, 2015) that the commonly used white models emit (Figure 3). Figure 3 Open in figure viewerPowerPoint 2013 The potential ecological impacts of white Light‐Emitting Diode lighting compared to other light sources. Spectral power distributions are given for white Light‐Emitting Diode (LED), Low‐Pressure Sodium (LPS), High‐Pressure Sodium (HPS) and Metal Halide (MH) lights recorded using a MAYA 200 pro spectrometer from street lighting in Cornwall. The amount of light at each wavelength is standardized to relative intensity (radiant energy divided by the maximum radiant energy recorded at any wavelength for each light source) so that the relative distribution of radiant energy across the light spectrum can be compared for each light source. Grey arrows represent the wavelength range over which different types of biological response are expected/recorded. Dashed lines represent the range of wavelengths over which mammal, bird, reptile, insect and arachnid visual systems can detect light [adapted from Davies et al. ()] Firstly, the broad range of wavelengths emitted by white LEDs likely enables organisms to perform colour‐guided behaviours at night that were previously only possible during the day (Davies et al., 2013). A range of intra‐ and interspecific interactions could be affected including foraging (e.g. seeking nectar‐rich flowers), predation (ability to locate and successfully capture prey), sexual communication (ability to locate, identify and assess the fitness of conspecifics through visual displays) and camouflage (ability to avoid detection by predators). Nocturnal species may find themselves competing for resources with diurnal species where such interactions had previously not existed (MacGregor et al., 2014), and differences in the sensitivity of animal visual systems to white LED light spectra could change the balance of species interactions (Davies et al., 2013). Some alternative lighting technologies also emit light across a broad range of wavelengths (e.g. Metal Halide and Mercury Vapour lighting, Figure 3); however, the energy efficiency of LEDs makes them the lighting of choice in the 21st century, and as such research should focus on how any unforeseen deleterious impacts can best be mitigated. Secondly, the short wavelength peak emitted by white LEDs coincides with the wavelengths to which many biological responses are known to be sensitive. Many invertebrate behaviours (Cohen & Forward, 2009; Gorbunov & Falkowski, 2002; Haddock, Moline, & Case, 2010; van Langevelde, Ettema, Donners, Wallisdevries, & Groenendijk, 2011) and the melatonin response (West et al., 2011) are sensitive to short wavelengths of light (between 350 and 500 nm), and some studies have demonstrated that white LED lighting has a greater impact on short wavelength sensitive responses compared to alternative lighting technologies (Pawson & Bader, 2014). Thirdly, because LEDs illuminate a broad range of wavelengths, they have the potential to affect a greater variety of biological responses that are sensitive to specific wavelengths of light. To give one example, while many invertebrate behaviours and the melatonin response are most sensitive to short wavelength light, the phytochrome system in plants—which is associated with the timing of flowering—is sensitive to red/far red light (660 and 720 nm) (Bennie, Davies, Cruse, & Gaston, 2016). Using broad wavelength light sources, such as white LEDs, therefore risks affecting more biological responses across a greater variety of taxa than using narrow wavelength light sources, such as low‐pressure sodium lighting (Gaston, Davies, Bennie, & Hopkins, 2012). Fourthly, the improved energy efficiency offered by LEDs may encourage growth in the amount of artificial light produced around the world. This “rebound effect” can be observed in historical lighting trends (see Kyba, Hänel, & Hölker, 2014) and partly explains why aesthetic and decorative lighting installations are now increasingly seen in municipal centres, on monuments, bridges and waterfront developments. Finally, improvements in the energy efficiency of LED lighting coupled with the production efficiency of solar cells have resulted in a rapid growth in off‐grid lighting installations, typically in remote regions containing previously artificial light naive ecosystems (Adkins, Eapen, Kaluwile, Nair, & Modi, 2010; Dalberg Global Development Advisors 2013; Mills & Jacobson, 2007). The greatest ecological impacts of ALAN over the next 50 years will likely occur in these previously artificial light‐naive regions, with an ecology not previously shaped by night‐time lighting.

5 EVOLUTIONARY NOVELTY Organisms have evolved with large‐scale fluctuations in atmospheric CO 2 , climate temperatures and ocean pH throughout history while sudden changes to natural light regimes are unprecedented over evolutionary time scales. The harmonic movements of the earth, moon and sun provide reliable cues to which many biological events are now highly attuned (Kronfeld‐Schor et al., 2013). The ability of organisms to rapidly adapt to the introduction of ALAN through behavioural, genetic or epigenetic changes is likely to be far more limited than for climate warming due to the unprecedented nature of this change (Swaddle et al., 2015). Furthermore, the scattered growth of artificial lighting around the world is a significant barrier to predicting where organisms will be able to seek out suitably dark habitats in the future and identifying where to allocate dark corridors that enable such migrations to happen. Although challenging, identifying where species need to go to survive climate warming, and how they get there, is made simpler by the predictability of regional climatic shifts (e.g. poleward migrations by land and sea and upward migrations in high‐altitude regions) (Hannah et al., 2007).

6 DIVERSITY OF BIOLOGICAL RESPONSES Artificial light at night is now known to cause a plethora of environmental impacts from altering organism physiology to changing the structure of ecological communities. The diversity of taxa affected continues to grow and now includes birds (Dominoni, 2015; Kempenaers, Borgstrӧm, Loës, Schlicht, & Valcu, 2010), bats (Rydell, 1992; Stone, Jones, & Harris, 2009), sea turtles (Kamrowski, Limpus, Moloney, & Hamann, 2012; Witherington, 1992), marsupials (Robert, Lesku, Partecke, & Chambers, 2015), rodents (Bird, Branch, & Miller, 2004), anurans (Hall, 2016), freshwater and marine fish (Becker, Whitfield, Cowley, Järnegren, & Næsje, 2012; Brüning, Hölker, Franke, Preuer, & Kloas, 2015; Riley, Davison, Maxwell, & Bendall, 2013), moths (Frank, 1988; Wakefield, Stone, Jones, & Harris, 2015), beetles, spiders, harvestmen, woodlice and ants (Davies, Bennie, & Gaston, 2012; Davies et al., 2017), branchiopod (Moore et al., 2000), amphipod (Davies, Coleman, Griffith, & Jenkins, 2015; Davies et al., 2012; Navarro‐Barranco & Hughes, 2015) and copepod (Davies et al., 2015) crustaceans, polychaete worms, colonial ascidians and hydrozoans (Davies et al., 2015), corals (Kaniewska, Alon, Karako‐Lampert, Hoegh‐Guldberg, & Levy, 2015), and terrestrial plants (Bennie, Davies, Cruse, Inger, & Gaston, 2015; Bennie et al., 2016; Ffrench‐Constant et al., 2016). The documented impacts include those on animal communication (van Geffen et al., 2015; Kempenaers et al., 2010), reproductive development (Dominoni, Quetting, & Partecke, 2013; Hansen, Stefansson, & Taranger, 1992), the timing of reproduction (Kaniewska et al., 2015; Robert et al., 2015), orientation (Frank, 1988; Witherington, 1992), habitat selection (Davies et al., 2012, 2015), predator avoidance (Wakefield et al., 2015), predation pressure (Becker et al., 2012; Bolton et al., 2017; Rydell, 1992), circadian disruption (Brüning et al., 2015; Raap, Pinxten, & Eens, 2015, 2016), plant phenology (Bennie et al., 2016; Bennie, Davies, et al., 2015; Ffrench‐Constant et al., 2016) and ecosystem services (Knop et al., 2017; Lewanzik & Voigt, 2014). While those impacts on survival and reproductive success highlight that ALAN is likely causing widespread population losses for a variety of taxa, no population‐level effects have so far been reliably demonstrated. This is in part because satellite images of night‐time lights are not available in sufficiently high spatial resolution for inferences to be drawn regarding impacts on species populations that can be variable on the scale of tens to hundreds of metres (Elvidge et al., 2007). Disentangling the effects of street and residential lighting from those of urbanization and land use change is challenging since all of these explanatory variables likely contribute to population declines but all covary. Analyses using higher resolution images from the international space station (capable of identifying individual roads) may yield further insights, but tend to be focused on cities, preventing comparisons from being drawn across sufficiently large spatial scales. Recent developments in hemispherical photography allow “biologically relevant” artificial skyglow to be mapped from ground level across thousands of square kilometres (Luginbuhl et al., 2009; Zoltan, 2010), better enabling ecologists to quantify its impacts on populations of organisms that utilize celestial patterns of sky brightness, but perhaps not the population effects of direct lighting. Techniques to model the distribution of artificial light across towns and cities have also been developed (Bennie, Davies, Inger, & Gaston, 2014); however, such models can be computationally expensive and have not yet been applied to the question of whether direct lighting has an impact on organism populations. Before After Control Impact (BACI) experiments have the potential to provide insights into the long‐term responses of sessile species populations and those mobile taxa with <1 km home ranges; however, the finances and time required to implement them at appropriate spatial and temporal scales make this approach less feasible in a limited funding environment. For now, quantifying the population‐level impacts of ALAN remains one of the most important and challenging problems facing ecologists working in this area.

7 SENSITIVITY OF BIOLOGICAL RESPONSES Many organisms are extremely sensitive to natural light, utilizing light cues as dim as the moon and the Milky Way to orientate themselves, navigate landscapes and identify conspecifics and resources at night (Dacke et al., 2013; Last et al., 2016; Ugolini, Boddi, Mercatelli, & Castellini, 2005; Warrant & Dacke, 2016). Perhaps, most striking is the growing number of documented responses to white LEDs in marine systems (Bolton et al., 2017; Davies et al., 2015; Gorbunov & Falkowski, 2002; Navarro‐Barranco & Hughes, 2015), where species are both adapted to utilize short wavelengths that penetrate deeper in seawater and are incredibly sensitive to natural light. Examples of this extreme sensitivity include copepods (Calanus sp.) that undergo diel vertical migration to depths of 50 m guided only by variations in moonlight intensity during the arctic winter (Båtnes, Miljeteig, Berge, Greenacre, & Johnsen, 2013; Last et al., 2016); sessile invertebrate larvae that move and identify suitable settlement locations guided by light levels equivalent to moonless overcast nights (Crisp & Ritz, 1973; Thorson, 1964) and polychaete worms, corals and echinoderms that synchronize broadcast spawning events using monthly and annual variations in lunar light intensity (Naylor, 1999). Many of these responses are clearly sensitive enough to be affected both by direct lighting and artificial skyglow (Figure 4), and indeed such impacts have been demonstrated for zooplankton diel vertical migration in freshwater ecosystems (Moore et al., 2000). Given the spatial extent of artificial skyglow in coastal regions (Davies et al., 2014; Falchi et al., 2016), the disproportionate importance of these regions for global biogeochemical cycles [coastal zones account for 30% of global ocean primary production but only 10% of global ocean surface area (Wollast, 1998)], and the role of diel vertical migration in maintaining these cycles (Hays, 2003), it is not unreasonable to surmise that ALAN could have detectable effects on ocean carbon and nutrient budgets in the near future. Figure 4 Open in figure viewerPowerPoint 3 uniform profile, wind = 5 m/s) and spring (b & d; Chlorophyll = 5 mg m3 uniform profile, wind = 5 m/s) water column properties. Models of scalar irradiance with depth are derived from spectral power distribution recorded from the spring high tide mark under a white LED street light on the Barbican in Plymouth (a & b), and artificial skyglow from predominantly white Metal Halide spectra recorded above Falmouth Harbour (c & d). Grey dashed lines indicate the maximum depth at which sufficient artificial light is available to perform species behaviours. SSS = Settlement Site Selection; PR = Polyp Retraction; LP = Larval Phototaxis; DVM = Diel Vertical Migration. Sensitivities to white light were calculated from experimentally derived values in existing literature (Båtnes et al., 2013 1973 1984 2002 1995 1995 1982 The sensitivity of marine invertebrates to direct artificial light and artificial skyglow. Solid lines represent the attenuation of scalar irradiance (between 400 and 700 nm) with depth estimated using radiative transfer models under winter (a & c; Chlorophyll = 0.3 mg muniform profile, wind = 5 m/s) and spring (b & d; Chlorophyll = 5 mg muniform profile, wind = 5 m/s) water column properties. Models of scalar irradiance with depth are derived from spectral power distribution recorded from the spring high tide mark under a white LED street light on the Barbican in Plymouth (a & b), and artificial skyglow from predominantly white Metal Halide spectra recorded above Falmouth Harbour (c & d). Grey dashed lines indicate the maximum depth at which sufficient artificial light is available to perform species behaviours. SSS = Settlement Site Selection; PR = Polyp Retraction; LP = Larval Phototaxis; DVM = Diel Vertical Migration. Sensitivities to white light were calculated from experimentally derived values in existing literature (Båtnes et al.,; Crisp & Ritz,; Forward, Cronin, & Stearns,; Gorbunov & Falkowski,; Svane & Dolmer,; Tankersley, Mckelvey, & Forward,; Young & Chia,

8 IMPACTS ON HUMAN HEALTH In 2007, the World Health Organisation classified shift work that disrupted human circadian rhythms as a probable human carcinogen (International Agency for Research on Cancer, 2007). While this classification is primarily associated with shift work, exposure to ALAN has been linked to a variety of health disorders in people through the same circadian disruption mechanism. These include sleep disorders, depression, obesity and the progression of some cancers (Cajochen et al., 2011; Chang, Aeschbach, Duffy, & Czeisler, 2014; Haim & Portnov, 2013; Keshet‐Sitton, Or‐Chen, Yitzhak, Tzabary, & Haim, 2015). The prominent peak of blue wavelength light emitted by LEDs is of increasing concern since it occurs at the most effective frequency for suppressing the production of melatonin (Haim & Zubidat, 2015; West et al., 2011), a hormone released by the pineal gland that regulates sleep wake cycles and acts as an antioxidant. Over the last decade, LEDs have become a ubiquitous feature of human life and can be found in street, residential, commercial and aesthetic lighting installations, laptops, televisions, e‐readers, smart phones and tablets. Late evening exposure to LED light from handheld devices has been linked to circadian disruption of sleep wake cycles and alertness and cognitive performance during the day (Cajochen et al., 2011; Chang et al., 2014). The extent to which outdoor lighting impacts human health is yet to be reliably determined. While epidemiological studies have found correlations between the amount of outdoor lighting and some health effects (Kloog, Haim, Stevens, Barchana, & Portnov, 2008; Koo et al., 2016), as with ecological patterns, they are limited by the inferences that can be drawn from satellite images (Defence Meteorological Satellite Programme Operational Line Scan) with insufficient spatial resolution (5 km) to differentiate exposure to ALAN from other factors that covary across city districts at fine spatial scales (Elvidge et al., 2007; Kyba, 2016). The need for higher resolution images or novel approaches that can disentangle the effects on both ecology and human health of multiple urban pollutants that covary is clear, although individual‐level sensors can also reveal important impacts of daily light exposure on circadian disruption and stress (Figueiro et al., 2017). A more recent analysis using higher resolution (0.75 km) images from the Visible Infrared Imaging Radiometer Suite (VIIRS) on board the Suomi National Polar‐orbiting Partnership satellite has revealed a significant association between ALAN and breast cancer incidence in the Greater Haifa Metropolitan Area in Israel (Rybnikova & Portnov, 2016). This analysis accounted for several potential covarying explanatory factors, but not noise pollution, and atmospheric pollution explicitly.

9 HUMAN–ENVIRONMENT INTERRELATIONSHIPS In a recent analysis that combined high‐resolution night‐time satellite images with atmospheric dispersion models of artificial skyglow, Falchi et al. (2016) estimated that more than 80% of the worlds’ population currently live under light‐polluted skies, such that the Milky Way is hidden from one‐third of people alive today. This extraordinary change in our night‐time environment escalated in the developed world during the mid to late 20th century and is now rapidly transforming the cultures of billions in the developing world. The trend is concurrent with urbanization [66% of the worlds’ population will reside in urban areas by 2050 (United Nations, 2014)], and it contributes to the growing disconnect between people and nature that has become known as “the extinction of experience” (Miller, 2005). This growing disconnect undermines public support for conservation issues by preventing individuals from connecting with, understanding and forming attachments to the natural world (Miller, 2005). The extinction of experience is another of the great challenges facing humanity in the 21st century. Miller (2005) argues that it can be addressed by designing urban landscapes to facilitate “meaningful interactions with the natural world.” There is perhaps no more profound way in which people can reconnect with nature than giving them access to the Milky Way and allowing them to experience the natural rhythms of moonlight and sunlight that they are evolutionarily preadapted to synchronize their physiology and behaviour with (Cajochen et al., 2013; Wright et al., 2013). Like biodiversity conservation, however, pristine skies have become tourist attractions restricted to regions awarded special status for their value to dark sky conservation (Collison & Poe, 2013; Pritchard, 2017; Rodrigues, Rodrigues, & Peroff, 2014) where many in the developed world can no longer afford to reside or visit. Pritchard (2017) argues that dark sky protection programmes also risk suppressing the economic and cultural development of poorer nations in a way analogous to biodiversity conservation in the 20th century. In her appraisal of NASA's “City Lights,” composite satellite image of the world's lights at night (http://earthobservatory.nasa.gov/Features/IntotheBlack/) Pritchard (2017) warns against “neo‐colonial approaches to the conservation of natural night‐sky brightness.” While it is clear that the continued growth in artificial lighting risks perpetuating the disconnect between people and the environment—and this will inevitably contribute to the concomitant shifting baseline in conservation objectives (Papworth, Rist, Coad, & Milner‐Gulland, 2009; Pauly, 1995)—any intervention should seek to support the modernization of societies while retaining their connections with the natural world. Pritchard (2017) describes achieving this balance as a “new frontier in 21st century conservation.”

10 FEASIBILITY OF SOLUTIONS While the recent growth in LED lighting has raised concerns among environmental scientists and human health experts, this technology offers lighting managers greater flexibility when it comes to tailoring the timing, intensity and spectral power distribution of municipal lighting systems (Davies et al., 2017; Gaston, 2013). Of the local authorities in England, 23% are engaged in permanent part‐night lighting schemes where street lights are turned off between midnight and 04:00 to 05:00 a.m., while 39% are engaged in permanent dimming schemes where lights are dimmed for at least some period of the night (Campaign to Protect Rural England, 2014). Increasing constraints on local authority budgets have incentivized the adoption of these lighting strategies in the wake of the 2008 global financial crash; however, more often the reasons given for their implementation are improved energy savings and reduced CO 2 emissions. Both dimming and part‐night lighting are better enabled by switching to LED and introducing central management systems that use wireless communication technology to programme individual street lights remotely. The ecological benefits of dimming and part‐night lighting are not yet well explored (although see Azam et al., 2015; Day, Baker, Schofield, Mathews, & Gaston, 2015; Davies et al., 2017). A recent emphasis in the ecological literature has instead been on tailoring spectral power distributions to reduce known ecological impacts (Brüning, Hölker, Franke, Kleiner, & Kloas, 2016; Davies et al., 2017; van Geffen et al., 2015; Longcore et al., 2015; Pawson & Bader, 2014; Rivas, Tomillo, Uribeondo, & Marco, 2015; Spoelstra et al., 2015), despite this approach being less popular among lighting managers and engineers who often focus on the improved visual performance offered by broad‐spectrum lighting as a key selling point. These studies collectively present an inconsistent picture of whether spectral manipulation can be used to effectively mitigate the ecological impacts of ALAN. This is partly because some studies compare narrow spectrum (e.g. red, green and blue) light with broad‐spectrum light sources, while others either decrease the amount of light occurring at wavelengths known to manifest certain ecological responses (usually shorter wavelengths in the visible spectrum), or increase the amount of light occurring at wavelengths that do not give rise to these responses (longer wavelengths in the visible spectrum). Even if a unified approach was adopted in spectral manipulation experiments, it seems unlikely that a publically acceptable light spectrum that does not give rise to any ecological impacts can be developed, because different species responses are evolutionarily adapted to utilize different wavelengths of light. Examples of this are abundant in the emerging literature on the ecological impacts of artificial light. The number of beetle taxa aggregating under white LED lighting can be reduced by switching to amber, but this has no discernible effect on the number of spider taxa that aggregate (Davies et al., 2017). Many animal responses are sensitive to short‐wavelength light (van Langevelde et al., 2011; Rivas et al., 2015; Spoelstra et al., 2017), while phenological responses in plants are most sensitive to the longer wavelengths recommended to avoid such effects (Bennie et al., 2016; Bennie, Davies, et al., 2015). Male caterpillars of the moth Mamestra brassicae reared under green artificial light reached a lower maximum mass, pupated earlier and obtained a lower pupal mass than those reared under red light (van Geffen, van Grunsven, van Ruijven, Berendse, & Veenendaal, 2014), while red light inhibited the attractiveness of a female pheromone lure to more adult males of the winter moth Operophtera brumata than did green light (van Geffen et al., 2015). Studies investigating the ecological benefits of part‐night lighting have also highlighted that different taxa respond in different ways (Azam et al., 2015; Davies et al., 2017; Day et al., 2015), and the adoption of part‐night lighting schemes is often inhibited by a perception among political actors that they lack popular support. There are both perceived and realized benefits of artificial light for society, including in the areas of road safety, crime and the economy (Gaston, Gaston, Bennie, & Hopkins, 2015). The night‐time economy in the United Kingdom, for example, was worth £67bn in 2016 (MAKE Associates, personal communication) and accounted for up to 27% of town and city centre turnover and 10% of most locations overall employment figure in 2009 (VisitEngland, 2012). While modern lighting technologies offer the potential to reduce the ecological impacts of ALAN, identifying how this is best achieved is clearly complex. Studies are needed across a wide variety of taxonomic groups and lighting approaches, to develop options that are both socially and ecologically acceptable.

CONCLUSION Research into the ecological, human health and societal consequences of ALAN is now growing rapidly. Here, we have highlighted 10 reasons why ALAN should and likely will be a focus for global change research in the 21st century. Most important to consider is the notion that while ALAN is having widespread and profound impacts on people and the environment, strategies for abating them are already being explored. Solving the challenges posed by ALAN would not only improve environmental and human health outcomes but also enhance the human experience of nature and change perceptions of the natural world in a way that facilitates the necessary transition towards a more environmentally orientated and hence sustainable society. It would also inspire greater confidence in our ability to tackle the problems posed by other global change phenomena. The challenge now is identifying how best to address to the complex array of ecological, human health and cultural problems presented by society's propensity for illuminating the night.

ACKNOWLEDGEMENTS The authors thank Dr Ilya Mclean (University of Exeter) for granting permissions to use images in Figure 2. The work leading to this publication was supported by a European Regional Development Fund funded Sêr Cymru II Fellowship awarded to TWD.