Nature, extent and ecological implications of night‐time light from road vehicles

Abstract The erosion of night‐time by the introduction of artificial lighting constitutes a profound pressure on the natural environment. It has altered what had for millennia been reliable signals from natural light cycles used for regulating a host of biological processes, with impacts ranging from changes in gene expression to ecosystem processes. Studies of these impacts have focused almost exclusively on those resulting from stationary sources of light emissions, and particularly streetlights. However, mobile sources, especially road vehicle headlights, contribute substantial additional emissions. The ecological impacts of light emissions from vehicle headlights are likely to be especially high because these are (1) focused so as to light roadsides at higher intensities than commonly experienced from other sources, and well above activation thresholds for many biological processes; (2) projected largely in a horizontal plane and thus can carry over long distances; (3) introduced into much larger areas of the landscape than experience street lighting; (4) typically broad “white” spectrum, which substantially overlaps the action spectra of many biological processes and (5) often experienced at roadsides as series of pulses of light (produced by passage of vehicles), a dynamic known to have major biological impacts. The ecological impacts of road vehicle headlights will markedly increase with projected global growth in numbers of vehicles and the road network, increasing the local severity of emissions (because vehicle numbers are increasing faster than growth in the road network) and introducing emissions into areas from which they were previously absent. The effects will be further exacerbated by technological developments that are increasing the intensity of headlight emissions and the amounts of blue light in emission spectra. Synthesis and applications. Emissions from vehicle headlights need to be considered as a major, and growing, source of ecological impacts of artificial night‐time lighting. It will be a significant challenge to minimise these impacts whilst balancing drivers' needs at night and avoiding risk and discomfort for other road users. Nonetheless, there is potential to identify solutions to these conflicts, both through the design of headlights and that of roads.

A few studies have examined ecological impacts of some other stationary sources (e.g. communication towers, lighthouses; Jones & Francis, 2003;Longcore et al., 2012), but the ecological impacts of mobile sources of lighting have remained virtually ignored.
The predominant mobile source of artificial night-time light is the emissions from vehicle, and particularly road vehicle, headlights. The ecological impacts that might arise from these have received almost no attention, or only passing reference, either within the literature on impacts of artificial night-time lighting (e.g. see reviews by Longcore & Rich, 2004;Gaston et al., 2013;Gaston, Duffy, Gaston, Bennie, & Davies, 2014), or on the ecological impacts of roads (e.g. see reviews by Coffin, 2007;Spellerberg, 1998;Trombulak & Frissell, 2000;van der Ree, Smith, & Grilo, 2015). Where they have been considered, the focus has been on the dazzling of vertebrates and the resultant potential for these causing collisions with vehicles (e.g. Outen, 2002).
Notwithstanding, there are good reasons to predict that headlight emissions have profound ecological impacts, both because of their general contribution to artificial night-time lighting, and because of the particular challenges posed by the high intensity of their emissions and the pulse-like nature of illuminance caused by passing vehicles.
In this paper, we review the nature, extent and ecological implications of artificial light from vehicle headlights. We do so by exploring in turn each of four key issues that shape the ecological impacts of artificial night-time lighting, namely light intensity, spectrum, spatial extent and temporal pattern. Essentially, we work from the level of individual vehicles to that of the landscape, and explore the ways in which recent and potential developments in vehicle ownership and technology may influence these effects.

| Background
Typical intensities of light emissions measured directly from headlights are around 2,000-8,000 lx for newer cars, but can be higher ( Figure 1); lux (lx) is a measure of luminous flux per unit area based on human photopic vision, and so does not necessarily capture the relative effects of light influencing biological processes with different spectral responses, but its use ensures a direct link to illuminance as commonly measured in the environment and employed in the design and mitigation of artificial lighting systems. This level of luminance is broadly comparable to that from emissions measured directly from streetlights, but vehicle headlights have a much more focused beam, which travels K E Y W O R D S artificial light, light cycles, light pollution, night-time, skyglow, spectra, urban ecology, vehicles further at higher intensities. Therefore, whilst the downward directed emissions from streetlights tend to result in ground-level illuminance of around 10-20 lx directly below the source, which usually declines to <1 lx a few metres away, those from vehicles reach much higher levels over much greater distances, both horizontally and vertically.
For example, emissions for a family car that approached 10,000 lx at source remained at 25 lx at 50 m distance, and exceeded 1 lx at 100 m; moonlight is c. 0.1 lx (for a full moon) . As a result, roadside vegetation and the surrounding area is frequently illuminated at night by emissions at levels of the order of 300 lx, and, depending on the angle to the oncoming traffic and the likelihood of vehicles using full beam (which will tend to be higher on rural roads, given lower levels of traffic), this may on occasion approach levels of around 1,000 lx or more, equivalent to daylight on a heavily overcast day ( Figure 2).
The artificial night-time lighting emitted by streetlights has been shown regularly to exceed the thresholds, which are often low (<1 lx; Gaston et al., 2013Gaston et al., , 2014, that trigger a wide variety of biological effects (e.g. physiological, behavioural, and other responses); this includes attraction and repulsion behaviours of animals, which may or may not influence risks of vehicle collision.
Nonetheless, dose-response relations-how effects change with increasing intensity of emissions-are poorly understood for most of these effects, and research establishing them is regarded as a high priority . The yet greater levels of illuminance at distance from vehicle headlights mean that the upper intensity levels that require exploration will need to be substantially higher than those from streetlights, and than the intensities which have been used in empirical studies thus far (e.g. Davies et al., 2017;de Jong et al., 2016;Sanders et al., 2015).

| Developments
The history of vehicle headlights has largely been one in which the intensity of emissions has progressively increased with technological improvements and innovations. The maximum intensity allowed along the axis of a single headlamp on full-beam (or high-beam) is presently 112,500 cd in Europe (under ECE Regulation 48) and Japan (under Japanese Safety Regulation Article 32), and 75,000 cd in the U.S. (under Federal Vehicle Motor Standard 108; Rumar, 2000); where E is intensity of emissions in lux (lx), I the intensity in candelas (cd), and d is distance in metres. In general, regulations have tended to increase to keep track with vehicle headlight strength, F I G U R E 1 Variation in (a) intensity and (b) correlated colour temperature (CCT) of emissions measured from headlights on full beam for different makes and models of cars, of a variety of ages (year) (n = 35). CCT is the absolute temperature of a blackbody whose chromaticity most nearly resembles that of the light source, and is frequently used to describe the aesthetic appearance of white light, from "warm" orange to "cool" blue light. Symbols represent light type. Data were collected using a UPRtek MK350N PLUS spectrometer, held in a cushioned frame that was placed in a standardised way directly on car headlights and surrounded by blackout fabric that eliminated external ambient light in the visible spectrum. These figures represent forward emissions and not the peak emissions achieved by the angling and reflection of the light. Some of the variation in figures is likely to be due to the shape and configuration of headlight assemblies whilst keeping below a level that creates too much glare for drivers of oncoming vehicles.
Recent technological developments are seeing the replacement of halogen bulbs with high intensity discharge (HID) xenon, light-emitting diode (LED) and, in as yet a very limited way, laser light sources, all of which can reach greater visible outputs. Laser headlights can produce an exceptionally bright white light that is significantly more intense than conventional light sources. Yet, in tacit acknowledgement of the potential for unprecedented levels of vehicular light pollution, laser headlights are not currently authorised in urban areas. In time the current "higher-end" technologies of LED and laser will become more affordable, and will be incorporated into low and medium cost vehicles (LED bulbs are already widely available for retrofitting into vehicle headlight assemblies).
Unless road vehicle technology changes in such a way as to make headlights redundant (see below), there is little evidence that emissions will not continue to increase with further innovations in headlight technology. This said, one potential brake on increasing intensity of emissions from new types of headlights may arise from concerns that these exacerbate effects of glare from oncoming vehicles, particularly for older drivers and in ageing populations. HID lights are especially problematic in this regard. The effect is worse for older drivers due to increased intraocular light scattering, glare sensitivity, and photostress recovery time (Mainster & Timberlake, 2003).

| Background
Streetlight emissions are very different in their spectra from sunlight, moonlight or starlight . Some types emit over very narrow bandwidths (e.g. low-pressure sodium lighting), others do so over a wide range of wavelengths (e.g. high-pressure sodium lighting, "white" LED lighting; Elvidge, Keith, Tuttle, & Baugh, 2010). Current vehicle headlight types tend to be of the latter form not currently widely commercially available and the details of the spectra remain unclear, but they provide focused, high-contrast white light intended to mimic sunlight, and are adapted from bluelaser diode technology (Wierer, Tsao, & Sizov, 2013).
White light is typical for car headlights for superior illumination at night, to avoid causing unnecessary fatigue, and to avoid inhibiting driver's colour vision. However, it is widely held that broad "white" lighting is environmentally especially problematic because of the headlights are close to the international regulated limit of 6,000 K ( Figure 1).

| Developments
It seems likely that there will be increasing use of headlight technologies with greater emissions particularly in the biologically significant blue part of the spectrum. A similar shift has been seen in streetlight technology, and has led to much public discussion over the implications for human health and wellbeing, for aesthetics, and for wider environmental impacts. In particular, there has been public opposition in some areas to the use of LED street lighting with higher CCT values. It would seem sensible to bring the desirability of developments in headlight technology into these same debates. The breadth of concerns here is similarly wide, embracing not just potential environmental impacts, but those on the behaviour and wellbeing of oncoming drivers and pedestrians, on occupants of roadside properties, and on the night-time aesthetics of roads.

| Background
Vehicle headlights illuminate vastly greater areas of habitat than this coverage is such that influences from roads are arguably the norm for areas rather than the exception. Of the coterminous United States, 20% of the total land area has been estimated to lie within 127 m of a road and 83% within 1,061 m (Riitters & Wickham, 2003).
Headlight emissions are not captured well by the satellite imagery that is used widely to analyse spatial patterns of artificial nighttime lighting (Figure 4). This is both because the emissions occur predominantly in the horizontal plane, and because imagery is often processed to represent static/persistent lighting and remove ephemeral lighting (thereby avoiding contamination of images of artificial night-time lighting with the location of fires etc.). In consequence, satellite imagery will tend markedly to underestimate the extent of artificial night-time lighting. This is important, because such imagery has been used to determine the levels to which the night-time environment has been eroded in different ecosystem types (Bennie, Duffy, Davies, Correa-Cano, & Gaston, 2015;de Freitas, Bennie, Mantovani, & Gaston, 2017), across areas protected for conservation , in areas with different species richness (Bennie, Duffy, Inger, & Gaston, 2014), and across the geographic ranges of different species (Duffy, Bennie, Durán, & Gaston, 2015).
Almost invariably, concerns have been expressed as to the levels of F I G U R E 3 Measured spectral irradiances (relative intensity) of three contrasting headlight types: (a) halogen; (b) high intensity discharge xenon; and (c) "white" light-emitting diode. Data were collected using a UPRtek MK350N PLUS spectrometer, held in a cushioned frame that was placed directly on car headlights and surrounded by blackout fabric light pollution being experienced, and the consequent changes in habitat suitability for organisms. However, in ignoring vehicle headlight emissions, these will tend to be substantial underestimates.
As well as causing direct illuminance, upwardly directed or reflected emissions from headlights will also contribute to skyglow, but these emissions are not presently incorporated into the prevailing models of this phenomenon, again underestimating its extent.

| Developments
The spatial extent of influence of vehicle headlights is likely to be growing rapidly alongside that of the road network. Globally, this network increased by 35% in the decade 2000-2009, and it has been estimated that there will be a need for an additional 25 million km of paved roads by 2050 (Dulac, 2013). Inevitably, this will introduce emissions from vehicle headlights into substantial areas in which they have not previously occurred. Of particular concern is that much of this growth in roads is likely to be in regions with rapidly emerging economies (e.g. China, India), with non-OECD countries expected to account for nearly 90% of the global growth in roadway infrastructure (Dulac, 2013). These regions include ones of high global importance for biodiversity and ecosystem services (Laurance et al., 2014).

Some change in the spatial extent of influence of headlights from
individual vehicles may result from increased used of adaptive technologies that, for example, cause these lights to swivel to better illuminate bends in the road and that extend beams on straighter roads.
However, at least in the immediate term, these effects seem likely to be small compared with the overall growth in length of roads and numbers of road vehicles. This may place a primacy on careful planning of where new roads are built so as, alongside other concerns, to limit the propagation of headlight emissions across landscapes, and to incorporate into their design landscape or habitat changes that block or reduce this spread of light. It seems likely that, cognisant of safety issues, landscape profiling and careful planting of appropriate vegetation (akin to sound barriers) could serve markedly to limit the propagation of emissions from headlights both along existing and new roads.

| Background
Street lighting and other static forms of lighting give rise to nighttime-long continuous or reasonably continuous periods of illumination. By contrast, at any one point along a roadside (and its surroundings) illuminance by emissions from headlights is typically  (2015) early hours of the day (see Figure 5). The time of year may also have a significant effect; in winter organisms will experience more traffic-related pulses due not only to a longer period of night-time, but also because these dark or twilight hours are more likely to coincide with peak "rush-hour" traffic. Furthermore, the level and pattern of traffic-related light pulsing is likely to vary both regionally and globally, according to latitude (influencing seasonal variation in length of night-time), level of economic development, and cultural conventions such as typical working and non-working days of the week. This introduces a wholly unnatural regime of light exposure to organisms, unrelated to genuine seasonal or biological cues.
The vast majority of studies of the biological impacts of artificial pulses (commonly of 30 min or 1 hr duration, but sometimes much less) to produce phase response curves to understand the circadian rhythms of a variety of organisms (e.g. Daan & Pittendrigh, 1976;Flari & Lazaridou-Dimitriadou, 1995;Ford & Cook, 1988;Gronfier, Wright, Kronauer, Jewett, & Czeisler, 2004;Kennedy & Hudson, 2016;Kumar & Singaravel, 2014). It also excludes phenomena such as the recovery times of night vision ("dark adaptation") after exposure to artificial lighting (and associated "bleaching" of photopigments), which may in insects and vertebrates take 30 min or more (Martin, 2017;Post & Goldsmith, 1965), with profound consequences for resource acquisition and predator avoidance. Pulsed lighting has regularly been found to act as a repellent to organisms, with limited evidence for adaptive responses (e.g. Hamel, Brown, & Chipps, 2008;Linhart, 1984;Nemeth & Anderson, 1992;Patrick, Sheehan, & Sim, 1982;Sullivan et al., 2016), and to be less of an attractant than continuous lighting (Gehring, Kerlinger, & Manville, 2009). Areas experiencing pulsed lighting may thus as a consequence be avoided and may contribute to the fragmentation of habitats.

| Developments
Globally ownership is growing rapidly, and this trajectory is likely to continue.
Since the growth in vehicle numbers is increasing at a greater rate than that of most countries' road networks, traffic density on the current roads will increase, which is likely to increase headlight pulse frequency by default. This said, the probable future trajectory of road transport and therefore volume is much debated. Whilst shortterm increases in car numbers are inevitable, the overall trends seem likely to be dependent on the type, rate and level of uptake of automated vehicles. Innovations could vary from advanced driver assist functions to full automation of personal cars and haulage vehicles.

Full automation may perhaps appear in combination with a system
where personal car ownership has all but ceased in urban areas, with rentable cars or taxis held in depots. In that case, the volume of traffic on roads could decrease, or become more evenly spread throughout the night-time hours as passengers make long journeys in the (currently unpopular) small hours, by automated vehicle. Increases in night-time traffic would obviously be a major concern for ecological impacts of headlights.

| D ISCUSS I ON
Widespread recognition of the, arguably pervasive, ecological impacts of artificial night-time lighting has only emerged quite recently.
Indeed, while spurred by key earlier contributions, the now rich literature of modeling, observational and experimental studies that documents these impacts has largely developed in the space of just the last decade. These insights have, however, focused almost exclusively on the consequences of emissions from static lighting sources.
The argument that mobile sources, and especially those from road vehicle headlights, are both contributing substantially to overall levels of artificial night-time lighting and to the ecological impacts seems compelling. Moreover, emissions from headlights give rise to particular concerns because of their intensity, predominantly horizontal and long trajectory, prevailing broad "white" spectrum, and the pulsed nature of the illuminance of habitat and organisms that they cause.
This said, it will be important to determine the details of the actual ecological impacts of emissions from vehicle headlights. In particular, it would be helpful to conduct field and mesocosm experiments with suitable study systems (e.g. see Sanders et al., 2015), to measure the effects on individual organisms, populations and communities of pulsed lighting of different intensity, frequency and spectrum. Perhaps more so than with static lighting, a key challenge will be to determine the relative importance   Datta, Samanta, Sinha, and Chakrabarti (2016) of drivers and the avoidance of risks and discomfort of other road users. By contrast, streetlights serve a wide range of purposes, including safety, security, social benefit, and aesthetics, although their general importance for some of these (including impacts on levels of vehicle accidents and crime) is hotly disputed (Gaston, Gaston, Bennie, & Hopkins, 2015). Second, recent developments in headlight technology have not been strongly driven by concerns to further reduce energy demands (albeit there are clearly limits to what can be supplied) and carbon dioxide emissions, or further prolonging the life span of lamps. These factors have, however, been critical considerations in the development of street lighting schemes, particularly at a time when public finances are widely under great pressure following the global financial crisis (Gaston, 2013). Third, headlight technology has predominantly been focused on broad "white" spectrum lamps for a long time, on grounds of safety, and there seems little likelihood of changing this. By contrast, different parts of the world have employed different streetlight technologies, with different spectral characteristics, and the rapidity and extent, benefits and costs, of a switch to broad "white" spectrum lamps is a topic of much debate.
This is not to say that headlight systems could not be redesigned so as to better limit light emissions into places and in forms (e.g. intensities, spectra) that they are not needed. Recognition and understanding of the environmental consequences of these emissions is obviously a key to pressure for such changes. Substantial reduction in these environmental impacts will also require accompanying landscape or habitat changes that reduce the spread and influence of the light emissions from headlights, and the incorporation of such concerns into the planning and design of new roads.

ACK N OWLED G EM ENTS
We