Is part-night lighting an effective measure to limit the impacts of artificial lighting on bats?
Abstract
As light pollution is currently considered to be a major threat to biodiversity, different lighting management options are being explored to mitigate the impact of artificial lighting on wildlife. Although part-night lighting schemes have been adopted by many local authorities across Europe to reduce the carbon footprint and save energy, their effects on biodiversity are unknown. Through a paired, in situ experiment, we compared the activity levels of 8 bat species under unlit, part-night, and full-night lighting treatments in a rural area located 60 km south of Paris, France. We selected 36 study locations composed of 1 lit site and a paired unlit control site; 24 of these sites were located in areas subject to part-night lighting schemes, and 12 sites were in areas under standard, full-night lighting. There was significantly more activity on part-night lighting sites compared to full-night lighting sites for the late-emerging, light-sensitive Plecotus spp., and a similar pattern was observable for Myotis spp., although not significant. In contrast, part-night lighting did not influence the activity of early emerging bat species around streetlights, except for Pipistrellus pipistrellus for which there was significantly less activity on part-night lighting sites than on full-night lighting sites. Overall, no significant difference in activity between part- and full-night lighting sites were observed in 5 of the 8 species studied, suggesting that current part-night lighting schemes fail to encompass the range of activity of most bat species. We recommend that such schemes start earlier at night to effectively mitigate the adverse effects of artificial lighting on light-sensitive species, particularly along ecological corridors that are especially important to the persistence of biodiversity in urban landscapes.
Introduction
Given the current degree of urbanization (United Nations, 2014) and its severe impacts on biodiversity (McKinney, 2006; McDonald et al., 2008), characterizing its effects on biological communities is of major importance (Luniak, 2004; Jung & Kalko, 2010; Hale et al., 2012; Penone et al., 2013; Bates et al., 2014).
Artificial lighting is intrinsically associated with urban sprawl; it has been deployed at a massive scale over the last century and continues to spread at an annual rate of increase of 6% worldwide (Cinzano et al., 2001; Hölker et al., 2010a). As a result, major concerns have been raised about the hidden impacts of artificial lighting and the associated light pollution on biodiversity and ecosystem functioning (Rich & Longcore, 2006; Navara & Nelson, 2007; Hölker et al., 2010b).
Ecological light pollution alters natural light regimes (Longcore & Rich, 2004; Hölker et al., 2010b), and it affects the rhythms of activity of populations of both diurnal and nocturnal species with important implications for individual fitness, sexual selection, and reproductive success (Miller, 2006; Boldogh et al., 2007; Kempenaers et al., 2010; Titulaer et al., 2012; Le Tallec et al., 2013; Nordt & Klenke, 2013). Furthermore, the responses of species to artificial lighting are driven by attraction/repulsion behaviors, so the movements and distribution of species can be altered at multiple spatial scales (Longcore & Rich, 2004). Taken together, the effects of artificial lighting can dramatically affect biological communities (Davies et al., 2012, 2013; Gaston et al., 2013, 2014).
In this context, designing outdoor lighting regulations that save energy and reduce CO2 emissions while limiting the ecological impacts of artificial lighting is a major challenge in land-use planning (Hölker et al., 2010a). Different lighting parameters, such as streetlamp spectrum, intensity, directionality, and duration of lighting, can be managed to limit the negative effects of artificial lighting on biodiversity (Gaston et al., 2012; Kyba et al., 2014). In an attempt at mitigation, many local authorities in Europe have started switching off public streetlights in the middle of the night, primarily to reduce local electricity costs and to save energy (Bennie et al., 2014).
It has been suggested that this measure, so-called part-night lighting, is unlikely to limit the impacts of artificial lighting on biodiversity because it does not coincide with the activity peaks of most nocturnal organisms, which occur at dusk when public demand for outdoor lighting is very high (Gaston et al., 2012). Consistent with this hypothesis, subjecting the light-sensitive bat R. ferrumequinum to simulated part-night lighting scenarios revealed that such schemes are unlikely to be compatible with its peak activity (Day et al., 2015). However, no study has tested the effect of this measure at the community level by simultaneously comparing the responses of multiple species to part-night lighting schemes.
As they are nocturnal and directly exposed to light pollution, microchiropteran bats are good candidates to test the effects of part-night lighting schemes on biodiversity. Increasingly threatened worldwide (Mickleburgh et al., 2002), bats are considered to be indicators of the response of biodiversity to anthropogenic pressure (Jones et al., 2009). Microchiropterans are long-lived insectivorous species, and it has been suggested that their population trends reflect those of lower trophic level species (Jones et al., 2009; Stahlschmidt & Brühl, 2012). Furthermore, several studies have pointed to their value in terms of providing ecosystem services, such as pest control (Cleveland et al., 2006).
Artificial lighting can affect bats in different ways, both in time and space. For many species, including bats, the natural light regime is a cue that synchronizes their window of activity with their environment (Gaston et al., 2013, 2014). The artificial illumination of maternity roosts can delay the emergence of female bats (Downs et al., 2003; Boldogh et al., 2007), which has important fitness costs for reproductive females as they miss the peak abundance of insects at dusk (Rydell, 1992; Jones & Rydell, 1994).
Artificial lighting can also modify resource availability and change species foraging patterns. Slow-flying species adapted to prey on insects in cluttered vegetation, such as Rhinolophus spp., Myotis spp., and Plecotus spp., appear to completely avoid illuminated areas (Rydell, 1992; Stone et al., 2009, 2012; Kuijper et al., 2008) due to increased predation risk from owls and other raptors (Jones & Rydell, 1994; Rydell et al., 1996). In contrast, fast-flying species adapted to hunt insects in the open air, such as Pipistrellus pipistrellus, appear to benefit from new and predictable foraging opportunities provided by streetlights (Rydell, 1992; Blake et al., 1994), which attract a large portion of the surrounding insect biomass (Eisenbeis, 2006; Perkin et al., 2014). However, even the movements and gap-crossing behaviors of light-attracted species in urban landscapes can be altered by artificial lighting (Hale et al., 2015). Therefore, these differences in the response of species to artificial lighting likely induce important changes in the structure of bat communities (Arlettaz et al., 2000; Polak et al., 2011) and raise concerns about ecosystem function.
In this study, we intended to determine whether current part-night lighting schemes effectively limit the impacts of artificial lighting on bats in an inhabited rural region of France. We compared the level of activity of eight species of bats as measured by ultrasound recordings under dark (unlit), part-night, and full-night lighting conditions. We expected to find no difference in activity between part- and full-night lighting treatments for fast-flying species, whose peak activity is known to occur at dusk to exploit the evening peak in insect abundance (Jones & Rydell, 1994; Rydell et al., 1996). However, we expected a potentially positive response from slow-flying species, with more activity on part-night lighting sites than on full-night lighting sites as they are the most light-sensitive species, and they are known to be active later at night than fast-flying species (Jones & Rydell, 1994; Rydell et al., 1996). Furthermore, slow-flying species may also take advantage of the insect biomass attracted by streetlights, once they are turned off.
Materials and methods
Study area
The field experiment was set up in a protected, 849 km² regional park established to promote the sustainable use of natural resources and ecosystems (IUCN Protected Area Category VI), which is located 60 km south of Paris, France. Arable lands represent 58 % of the area, and forests comprise 31 %. Currently, urban areas make up 8 % of the park and are mostly composed of small towns and villages (Fig. 1), but the entire region is subject to pressures from urbanization due to its vicinity to the capital. The park is comprised of 69 municipalities that average 12 km² in size, and 56 % have employed part-night lighting schemes for at least 2 years. These schemes are designed to turn off all public streetlights from midnight (±1 h) to 5:00 hours, representing approximately 65 % of the duration of the night.
Sampling design
We compared bat activity levels under unlit, part-night, and full-night lighting treatments through a paired in situ experiment. We selected 36 study locations composed of 1 lit site and a paired, unlit control site (n = 72 sites); 24 pairs were located in administrations practicing part-night lighting schemes, and 12 were located in municipalities with full-night lighting (Fig. 1). Lit sites were illuminated by 1 high-pressure sodium (HPS) vapor streetlamp (average intensity = 32 lux; range = 10–99 lux) and located away from the town cores to limit any correlation between light treatment and urbanization. Unlit sites were separated from their paired lit site by approximately 250 m, but pairs were located in similar habitats and set along the same types of bat commuting routes, such as forest edges and hedgerows (Walsh & Harris, 1996; Downs & Racey, 2006). The 2 sites of each pair were also located at similar distances from linear elements, such as roads and streams.
We ensured that light treatment was not correlated with the surrounding land uses, especially the proportion of impervious areas, in our study area. We created circular buffers with radii ranging from 50 to 2000 m around each sampled site (n = 72) using GIS (ARCGIS 10/ESRI; http://www.esri.com/) and calculated the proportion of forested, arable, impervious (urban and roads), and open (meadows and gardens) areas using a detailed regional geo-referenced land-use database with a resolution of 25 m (IAURIF, 2008). We then tested the correlation between light treatment and these land-use variables at 6 different spatial scales using a Kruskal–Wallis one-way analysis of variance. We detected significant differences between the 3 light treatments at only small, from 50 to 250 m, spatial scales (Table S1), but these differences did not result in any multicollinearity problems (Variance Inflation Factor (VIF) < 3; Fox & Monette, 1992).
Bat monitoring
Fieldwork was carried out (i) from the first of May to the 28th of August, which corresponds to the seasonal peaks in the activities of the bat species as recommended by the French national bat-monitoring program ‘Vigie-Chiro’ (http://vigienature.mnhn.fr/); (ii) when weather conditions were favorable, that is, no rain, low wind speed (< 7 m s−1), and temperatures higher than 12 °C; (iii) between the third and the first quarter moon to limit the interaction between natural and artificial lighting (Saldaña-Vázquez & Munguía-Rosas, 2013).
Both sites of each pair (1 lit/1 unlit) were sampled on the same night from 30 min before sunset to 30 min after sunrise. Standardized echolocation calls were simultaneously recorded using 1 stationary SM2BAT (http://www.wildlifeacoustics.com/) detector per site, which allowed for the direct comparison of bat activity between the 2 sites of each pair. The detectors automatically recorded all ultrasound (> 12 KHz) while maintaining the characteristics of the original signals. Ambient temperature was also recorded every 30 min for each pair with an EL-USB-1 temperature data logger (Lascar Electronics, Salisbury, UK).
We used the software SonoChiro© (Bas et al., 2013) to automatically classify the echolocation calls to the most accurate taxonomic level possible. We then checked the software classification by screening all ambiguous calls with Syrinx software version 2.6 (Burt, 2006). Identification was possible to the species level in all but two genera, Plecotus spp. and Myotis spp., due to the very low occurrence of the individual species and uncertainties in the acoustic identification to the species level (Obrist et al., 2004; Barataud, 2012). Note that we expected similar responses to the light treatments from the species in these two genera as they appear to have similar foraging behaviors (Arlettaz et al., 2001). As it is impossible to distinguish individual bats from their echolocation calls, we calculated an index of relative bat activity for each sample site, which was defined as the mean number of bat passes per species. A bat pass is defined as the occurrence of a single or several bat calls during a 5-s interval (Millon et al., 2015).
Statistical analysis
For each species, we created a general linear mixed model using the total number of bat passes per night per site as a response variable with a Poisson or a negative binomial error distribution (Zuur et al., 2009). Light treatment (composed of 3 factors: unlit, part-night lighting, and full-night lighting) and the land-use covariables (VIF < 3) were included in the models as fixed effects; the covariables were the proportion of forested, impervious, and open areas within a 250-m radius buffer around each sample site. We selected this scale as it represented the limit beyond which there were no differences in land-use composition between the 3 light treatments (Table S1). We also tested the effect of land-use within a 50-m radius buffer around each sample site, which corresponds to the average detection distance of each species (Barataud, 2012), but none of the land-use covariables were significant at this spatial scale. The average night temperature (°C), the moon phase (composed of 3 factors: ascending, descending or absent), Julian date, and the time of sunset were also included as fixed effects in the models. As random effects, we included the identification number of the pair for each site, which was nested into the identification number of the municipality in which each pair was located. The former random effect allowed the pair-wise comparison of bat activity among light treatments, whereas the purpose of the latter was to take the similarities in environmental management between nearby pairs into account. For each species, we selected the best model by removing each fixed effect one by one and comparing the residual deviance of the subsequent models with a type II anova associated with a chi-squared test (Table S2; Zuur et al., 2009). Model validation was carried out by visual inspection of the patterns of the model residuals (Zuur et al., 2009). All analyses were performed in R 3.15 with the ‘MASS’ package and the ‘glmmPQL’ function.
Results
Bat monitoring
A total of 57 341 bat passes belonging to 6 species and 2 species groups were recorded in the 72 study sites, and the most abundant species was the common pipistrelle bat, P. pipistrellus, representing 83 % of the observations (Table 1). The least abundant species were Plecotus spp. (481 bat passes) and Pipistrellus nathusius (595 bat passes) although they were present in 50 % and 30 % of the 72 sites, respectively.
| Flight behavior | Species | No. of bat passes | Final model formula | Distribution |
|---|---|---|---|---|
| FF | Pipistrellus pipistrellus | 46 314 | Light treatment | Poisson |
| Eptesicus serotinus | 3305 | Light treatment +% open areas | Poisson | |
| Pipistrellus kuhlii | 1156 | Light treatment | Poisson | |
| Nyctalus leislerii | 976 | Light treatment +% open areas | Poisson | |
| Nyctalus noctula | 844 | Light treatment +% forest +% open areas | NB (θ = 6.1) | |
| Pipistrellus nathusius | 595 | Light treatment +% forest +% impervious areas +mean T °C | NB (θ = 5) | |
| SF | Myotis sp. | 3670 | Light treatment +% forest +% open areas +% impervious areas | Poisson |
| Plecotus sp. | 481 | Light treatment +mean T °C | NB (θ = 5.1) |
Effect of light treatment on bat activity
In comparison with the unlit control treatment, full-night lighting had a significant negative effect on Myotis spp. (P < 0.05; Table 2; Fig. 2A) and a nearly significant negative effect on Plecotus spp. (P = 0.08; Table 2; Fig. 2B). In contrast, full-night lighting had a significant positive effect on P. pipistrellus (P < 0.001; Table 2; Fig. 2C), Pipistrellus kuhlii (P < 0.001, Table 2; Fig. 2D), and Nyctalus leislerii (P < 0.01; Table 2) and a nearly significant positive effect on P. nathusius (P = 0.08; Table 2). No effect of full-night lighting on Eptesicus serotinus and Nyctalus noctula was found (Table 2).
| Pipistrellus pipistrellus | Pipistrellus kuhlii | Pipistrellus nathusius | Nyctalus leislerii | Nyctalus noctula | Eptesicus serotinus | Myotis sp. | Plecotus sp. | ||
|---|---|---|---|---|---|---|---|---|---|
| Light treatment | |||||||||
| (A) | Part-night vs. Unlit | 0.15 (0.17) | 1.27 (0.33)*** | 1.80 (0.51)** | 3.10 (1.18)* | 1.18 (0.25)*** | −0.68 (0.40)**** | −0.91 (0.28)** | 0.67 (0.24)** |
| (B) | Full-night vs. Unlit | 0.93 (0.19)*** | 1.25 (0.27)*** | 0.86 (0.70)**** | 2.48 (0.59)** | 0.03 (0.50) | −0.23 (0.40) | −2.40 (1.22)* | −0.74 (0.40)**** |
| (C) | Part- vs. Full-night | −0.78 (0.22)** | 0.01 (0.41) | 0.94 (0.58) | 0.73 (0.93) | 1.14 (0.50)* | −0.44 (0.52) | 1.49 (1.22) | 1.46 (0.47)** |
Similar to the full-night lighting treatment, part-night lighting sites also had significantly less Myotis spp. activity (P < 0.01) and significantly more activity by P. kuhlii (P < 0.001), P. nathusius (P < 0.01), and N. leislerii (P < 0.05; Table 2) than the unlit control sites (Table 2; Fig. 2). For these 4 species, there were no differences between the part-night and full-night lighting sites (Table 2), but for the 3 remaining species, the effect of part-night lighting differed from the effect of full-night lighting. The activity of P. pipistrellus on the part-night lighting sites was half of that under full-night lighting (P < 0.001; Fig. 2C), and there was no significant difference in bat activity between the unlit and part-night lighting sites (Table 2; Fig. 2C). In contrast, there was significantly more Plecotus spp. (P < 0.01; Table 2; Fig. 2B) and N. noctula (P < 0.001; Table 2) activity on part-night lighting sites than on the unlit or full-night lighting sites.
Discussion
To our knowledge, this study is the first to empirically test the effectiveness of part-night lighting schemes on the mitigation of the impacts of artificial lighting on the activity of a bat assemblage. Given that part-night lighting schemes have been in place for several years in our study site, our in situ experiment provided a unique opportunity to characterize how bat species have adapted their foraging and commuting behaviors to this mitigation measure.
Our study examined the effects of only high-pressure sodium (HPS) vapor lamps, which are the most commonly used type of lamp in European public lighting (Eisenbeis, 2006). Due to the low emissions of short wavelength UV, HPS vapor lamps attract fewer insects and, therefore, fewer bats than mercury vapor lamps (Blake et al., 1994; Eisenbeis, 2006). Nevertheless, insect traps illuminated by HPS lamps still catch 27 times the insects caught by traps under dark conditions, and the attraction effect extends to 40 m from a light source (Perkin et al., 2014). Consistent with other studies, our experiment showed that the streetlights were creating additional foraging opportunities for the fast-flying Pipistrellus spp. and Nyctalus spp. (Rydell, 1992; Rydell 2006) while reducing the availability of foraging patches for the slow-flying Myotis spp. and Plecotus spp. (Kuijper et al., 2008; Stone et al., 2009). Similar bat response patterns to artificial lighting have been observed even under low light intensities (Lacoeuilhe et al., 2014), including energy-efficient light-emitting diode (Stone et al., 2012).
It is important to note that the increased foraging opportunities induced by artificial lighting for fast-flying bats may not be stable because the massive attraction of insect species to streetlights is likely to have significant impacts on their long-term demography (Eisenbeis, 2006; Moore et al., 2006). Common macromoths have experienced major declines in the UK in recent decades (Conrad et al., 2006), and it has been hypothesized that urban areas and their associated sky glow may act as ecological sinks, depleting the surrounding landscapes of moth species (Bates et al., 2014). Cascading effects of these declines may be expected in the long term (Van Langevelde et al., 2011).
In our study, the effect of part-night lighting differed among species and highlights the importance of addressing the efficacy of a mitigation measure at the community level. For the 3 fast-flying species, P. kuhlii, P. nathusius and N. leislerii, and the slow-flying Myotis spp., part-night lighting schemes did not drastically change the overall level of activity around streetlights. This suggests that current part-night lighting schemes do not coincide with the activity window of these species (Gaston et al., 2012). This is particularly important for the light-sensitive Myotis spp., which were significantly less active under both light treatments than on the unlit control sites. Slow-flying species, such as Myotis spp., are important conservation targets as they are particularly sensitive to habitat loss and fragmentation (Safi & Kerth, 2004; Frey-Ehrenbold et al., 2013). Disturbances to bat commuting routes induced by artificial lighting can significantly impact the fitness and reproductive success of light-sensitive species by increasing the distance between roosts and suitable foraging sites (Stone et al., 2009). Our results suggest that current part-night lighting schemes do not limit nightscape fragmentation for this genus.
Consistent with our hypothesis, there was no difference in overall bat activity between part- and full-night lighting sites for the 3 fast-flying species, P. kuhlii, P. nathusius, and N. leislerii, which are known to emerge at dusk (Jones & Rydell, 1994; Rydel, 2006). However, P. pipistrellus did not respond similarly to the other fast-flying species; it exploited the part-night lighting sites 2 times less than the full-night lighting sites even before the streetlights were extinguished (Fig. 3). This is surprising because this species appears to have foraging behaviors similar to other fast-flying species (Rydell, 2006). Once they have identified suitable foraging areas, individual bats show strong site-fidelity over time (Bonaccorso et al., 2005; Hillen et al., 2009). As part-night lighting schemes have been in place for several years in the study area, P. pipistrellus individuals may have identified the streetlights used in part-night lighting as less suitable foraging sites than the full-night lighting sites. This would be especially likely during the reproduction period (as in our study) when the energetic costs of reproduction influence female foraging strategies (Racey & Swift, 1985; Barclay, 1989; Rydell, 1989; Duvergé et al., 2000).
In contrast, part-night lighting had a significant positive effect on the slow-flying Plecotus genus with more activity on part-night lighting sites compared to both unlit and full-night lighting sites. This suggests that current part-night lighting schemes overlap with part of its activity range. This finding is consistent with the literature as Plecotus spp. is one of the late-emerging bat taxa (Jones & Rydell, 1994; Rydell et al., 1996). Furthermore, Plecotus spp. forages on prey, such as moths, by gleaning the vegetation or substrate surface (Rydell, 1992; Rydell et al., 1996; Jones & Rydell, 1994). The attraction of insects to streetlights is often coupled with a ‘fixation effect’, meaning the insects stop flying and land on the ground or the surrounding vertical surfaces (Eisenbeis, 2006). They can even remain stationary within the illuminated area for several hours (Frank, 2006), so Plecotus spp. may be taking advantage of the stationary insects around streetlights once the lights have been turned off. A similar response was observed in N. noctula, but this species is a long-range echolocator, and its range of detection can extend up to a 100 m (Obrist et al., 2004; Frey-Ehrenbold et al., 2013). However, due to (i) the limited number of bat pass for this species and (ii) the level of uncertainty associated with the position of each bat pass, we must be cautious with these results.
Our study demonstrates that current part-night lighting schemes fail to overlap with the range of activity of 5 of the 8 bat species studied. This suggests that even if this mitigation measure limits CO2 emissions and enhances energy savings, it is currently not an effective mitigation measure for biodiversity. However, further studies may confirm our results, especially for species such as Plecotus spp., Myotis spp., and Nyctalus noctula as the estimates from the full-night lighting treatment were associated with relatively large standard errors because of the low number of sites sampled (n = 12). For example, the fact that twice as many Myotis spp. were recorded on part-night lighting sites as on full-night lighting sites could suggest a slightly beneficial effect of part-night lighting for this genus, but this difference was not significant due to the highly variable activity of Myotis spp. on the full-night lighting sites. Simulated part-night lighting scenarios have shown that streetlights must be switched off before 23:00 hours to coincide with a significant portion of the activity range of the light-sensitive Rhinolophus ferrumequinum (Day et al., 2015). Therefore, part-night lighting schemes may become an efficient mitigation measure for Myotis spp. if implemented earlier at night. Starting part-night lighting schemes before 23:00 hours at the scale of an entire city or region would likely face resistance from the local inhabitants (Gaston et al., 2012). However, this could be a valuable strategy along ecological corridors, such as urban parks and river banks, that would allow light-sensitive species to persist in urban and peri-urban environments (Jung & Kalko, 2010; Threlfall et al., 2013).
Acknowledgements
We thank the ‘Réseau francilien de recherche et de developpement soutenable’ of the Ile-De-France region for funding and Thomas Bédot, Mathieu Deperrois, Alexandre Emerit, and Emmanuelle Guilmault-Fonchini of the French Gâtinais Regional Park for their assistance and support. We also thank all of the local authorities that allowed us to do our experiment, Pr. Georges Zissis for his insightful comments on lighting measurements and outdoor lighting regulations, Clémentine Reneville for his assistance during the field work, and the Vigie-Nature platform for loaning us the acoustic equipment. We also thank the anonymous reviewer for all his (her) insightful comments.
