Ecosystem-specific moth attraction to warm-coloured LED simulating road lighting conditions

Abstract

Light pollution poses a significant threat to nocturnal insects, yet our understanding of how insects are affected by lighting across ecosystems remains limited. We investigated differences in light-induced attraction regarding abundance, richness and diversity of moths between forest and grassland ecosystems. This novel study presents the first multi-site comparison of moth attraction across ecosystems using warm-coloured LED lighting (3000 K) that simulates typical road lighting conditions in both spectral composition and intensity. Using identical light traps with characterised lighting properties across 32 sites, we found significantly higher moth abundance and diversity in forests compared to open grasslands, with stronger environmental effects on moth attraction in open grassland ecosystems. Even at the low light intensities used in our experiment, which is comparable to spill light from road lighting in natural environments, we found significant impacts on moth activity, which is concerning given their limited activity periods. Moth families exhibited varying attractions across ecosystems, suggesting potential sampling biases in light attraction studies. Our comprehensive light characterisation using real-world lighting conditions provides a standardised framework for future research on light pollution impacts. The findings demonstrate the need for ecosystem-specific approaches in light pollution mitigation strategies and have important implications for conservation across diverse landscapes.

1. Introduction

Light pollution is recognised as a pervasive environmental problem with serious consequences worldwide, including reduced ability for astronomical observations of celestial objects ¹ and far-reaching implications for most species and their ecosystems, ² impacting the majority of economically developed regions across the globe. ³ The problem of nighttime lighting is escalating in almost all countries worldwide, with an annual rise of 2.2% in upward light emissions in already lit areas. ⁴ Light pollution can be broadly categorised into astronomical, ecological and human impacts, ⁵ with this study focusing on the ecological effects.

A group of organisms significantly affected by nocturnal light are insects, whose responses to anthropogenic lighting have been extensively reviewed.^6–9 Insects, the most species-rich animal group on Earth, play essential roles in ecosystems by providing services such as pollination, decomposition, soil formation, food and pest control. ¹⁰ Numerous studies have documented substantial declines in insect diversity and biomass.^8,11–13 These declines are linked to various factors, including habitat destruction and degradation, climate change, land use changes and habitat fragmentation, ¹³ as well as the potential impacts of anthropogenic light at night.^9,14 Many nocturnal insects are often attracted to light, a phenomenon known as positive phototaxis.^15,16 Such attraction to light may cause mortality through circling, exhaustion, injury and increased predation risk when exposed in the light. ¹⁷ Thus, anthropogenic light may potentially affect ecosystem services such as pollination and biodiversity. ¹⁸ Insect attraction to light sources is influenced by several factors, including light intensity, ¹⁹ spectral power distribution (as might be characterised by correlated colour temperature (CCT)), ²⁰ flickering, ²¹ light distribution and optics of the luminaire. ²²

However, very little is known about how the attraction of nocturnal insects to light sources is affected by the spatial context, such as the surrounding ecosystems. For instance, studies of moth attraction to anthropogenic light have been conducted in different types of ecosystems such as grasslands,^23,24 urban environments, ²⁵ prairies, ²⁶ woodland edges and management regimes^27,28 and aquatic ecosystems. ²⁹ In a rare direct comparison between ecosystems, Merckx and Slade ³⁰ found significantly lower moth recapture rates in an open field compared to woodland, which they attributed to lower temperatures and higher background illumination in the open habitat. However, their study had limitations as it was restricted to a single site per habitat type and used a 6 W actinic light trap with a spectral power distribution very different from real-world light installations.

Grasslands, covering up to 40% of Earth’s terrestrial surface, and forests, covering 30% and accounting for most of the terrestrial primary production and biomass, are both crucial biomes providing essential ecosystem goods and services globally.^31,32 Due to differences in topography and elevation of physical features, these ecosystems exhibit varying light distributions and exposures for species. Forest ecosystems typically experience dim light conditions as vegetation filters and absorbs light, allowing limited light to reach the forest floor. ³³ In contrast, open grasslands lack these filtering effects, allowing unobstructed light to dominate the landscape. These differences between closed forests and open grasslands may also result in varied exposure of insects to environmental factors such as weather conditions and moonlight. For instance, lower temperatures may significantly reduce moth catches in light traps, while increased cloud cover can increase them. ³⁴ It has also been shown that trap catches of moths can be considerably higher near the period of the new moon compared to full moon. ³⁵ Consequently, environmental factors can affect insect activity differently across ecosystems with potential interactive effects from anthropogenic light. Therefore, it is essential to consider confounding factors like weather conditions and moonlight exposure in studies of the effects of anthropogenic light attraction in different ecosystems.

The aim of our study was to compare insect attraction to light in two different ecosystems, open grasslands and forests, simulating real-world road lighting rather than specialised insect-attracting lights. Our objective was to investigate differences in light-induced attraction regarding abundance, richness and diversity of moths between these ecosystems. For this purpose, we sampled moths in grassland and forest ecosystems with light traps using identical light exposure (i.e. light sources, intensities and distributions).

2. Method

2.1 Study area

The study area is situated south of Stockholm, in Huddinge, Haninge and Botkyrka municipalities and encompasses a representative boreo-nemoral mixed heterogeneous landscape, including forests with mixed evergreen and deciduous tree species, lakes, agricultural areas and open field grasslands (Figure 1). The forest ecosystem was dominated by, for example, Picea abies, Populus tremula, Betula spp. and Corylus avellana. The understory vegetation and forest floor maintained typical boreal forest characteristics with mosses, ferns and shade-tolerant herbs. The grassland ecosystems consisted of semi-natural meadows with varying management intensities, representative of the region’s traditional agricultural landscape. These sites shared a common vegetation structure dominated by Phleum pratense, Festuca spp., Hypericum spp., Trifolium spp. and Achillea millefolium, forming a diverse herb-rich grassland community. While vegetation composition showed natural variation between sites, this heterogeneity reflects the actual conditions under which moths encounter anthropogenic lighting in real-world settings. Our multi-site approach aligns with landscape ecology principles, where understanding species responses across heterogeneous landscapes is fundamental for capturing ecological processes. ³⁶

Figure 1

Overview of the study sites from the field experiment. The background map is the Topographical map from Lantmäteriet, free to use according to the Creative Commons licence CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/)

The brightest field site experienced sky brightness of 2.64 mcd·m⁻² (Vårby), medium-bright sites had a mean value of 1.10 mcd·m⁻² (five sites) and the darkest situated sites had 0.60 mcd·m⁻² according to the New World Atlas of Artificial Night Sky Brightness which is based on satellite composite data from 2014.^3,37 In the three municipalities, more than 1500 species of Lepidoptera have been recorded in accordance with the Species Observation System ³⁸ out of approximately 2700 resident species in Sweden. ³⁹

2.2 Experimental set-up

2.2.1 Experimental design and light distribution

We performed a field experiment on nocturnal moths using light traps. To represent realistic road lighting conditions, we implemented LED lighting with a CCT of 3000 K, following Swedish road lighting guidelines. ⁴⁰ This aligns with the upper limits recommended for urban environments by international guidelines. ⁴¹ Warm-coloured lighting with minimal blue content is advocated by both EU environmental policy ⁴² and wildlife conservation recommendations from the Convention on the Conservation of Migratory Species of Wild Animals. ⁴³ The preference for warm-coloured lighting is supported by evidence that nocturnal insects exhibit strong phototactic responses to blue wavelengths. ⁴⁴ The lighting intensity used in our study was deliberately low, consistent with realistic outdoor light levels found in the natural environment outside the areas directly lit by road lighting. For example, this could correspond to illuminance levels typically encountered at the periphery of urban or rural roads, where light encroaches on the surrounding natural environment. The illuminance variation in horizontal illuminance on the ground used in this study is comparable with maximum illuminance values for human uses in sensitive areas. ⁴⁵ We used directional (spot) lamps and shielding to restrict spill light, creating a light cone confined to the nearby ground. This set-up allowed us to implement identical attraction radii, enabling a comparison of abundance, richness and diversity between ecosystems. This field-based approach comparing ecosystems with realistic outdoor lighting is novel in studying insect attraction to light at night and facilitates comparisons of light pollution impacts between ecosystems. In our field experiment, we used funnel traps with an attached net bag originally designed for the LepiLED light source ⁴⁶ (Figure 2(a) and (b)). We mounted an LED lamp (3000 K 3.8 W 350 lm 12 V 36° Osram, Germany) to each funnel trap with the light source at 133 cm heights from the ground, to simulate road lighting conditions. The LED source was selected for its lower UV and blue light emissions compared to the LepiLed system. We characterised the lamp’s spectral power distribution using a JeTi Spectro-Radiometer (specbos 1201; see Supplemental Material A for a full spectrum). Due to the physical constraints of the trap design, measurements were taken at 2 cm from the lamp centre at a 10° angle, registering a radiance of 693 W·sr⁻¹·m⁻² and luminance of 2.28 × 10⁵ cd·m⁻² (JeTi Spectro-Radiometer). The measured CCT was 3091 K, confirming the nominal specifications. Photosynthetically active radiation measurements taken directly beneath the lamp at 6.5 cm distance registered 143 µmol·s⁻¹·m⁻² (LI-COR LI-250 Light Meter). To avoid vertical spread of light from the lamp, the upper surface on the lid of the funnel trap was covered with duct tape. The trap was mounted on a stand consisting of metal frames usually used for portable greenhouses (Figure 2(a) and (b)). We used portable power banks as the power supply for the LED lights.

Figure 2

(a) Close-up photo of insect trap. (b) Luminance photo of the insect light trap in the forest. The luminance photo(b) uses false colour to represent luminance values for clearer visualisation of the distribution. (c) Measurements of vertical illuminance at different heights (0.5 m, 1 m, 1.5 m and 2 m) and at different distances from the light trap. The threshold value of 0.3 lx refers to the maximum illuminance of the full moon

We documented luminance of the light trap and its surroundings (Figure 2(b)), and vertical illuminance at four heights and at different distances 0.5 m to 5 m from the light trap (Figure 2(c)), using a calibrated Hagner™ Universal Photometer S5 with a measurement range of 0.01 lx to 199 900 lx (or cd·m⁻²) and ±3% accuracy. Luminance was captured using the LMK Mobile Advanced imaging luminance photometer (based on a Canon EOS 550D), and the associated computer software LMK LabSoft ver. 12.7.23 (TechnoTeam Bildverarbeitung GmbH, Ilmenau, Germany), enabling direct image-to-luminance value conversion. Detailed measurements of the illuminance at various distances and heights from the trap are presented in Supplemental Material B. This comprehensive light characterisation, rarely conducted in moth attraction studies, confirmed that light influence was localised and did not spread extensively through the surrounding terrain, with vertical illuminance dropping below 0.3 lx (which is equivalent to the maximum possible horizontal illuminance of the moon ⁴⁷ ) within a short distance (2 m) from the trap, and horizontal illuminance showing even lower values. Illuminance measurements at 0.5 m from the insect trap and further away confirmed that light levels were consistent with ambient illumination typical of natural areas adjacent to road lighting installations.

2.2.2 Experimental procedure

Light traps were mounted in late afternoons-early evenings (1 h to 2 h before sunset) at each site (Figure 1). We sampled two sites per night, one site situated in an open grassland and one site situated under the canopy cover inside a forest. All grassland sites were visually assessed to ensure intact vegetation with no signs of recent mowing or other major disturbances during sampling periods though we did not systematically document management history. The mean distance between traps in open grassland and the nearest forest was 41 m (range: 15 m to 160 m). At each site, we placed a control trap identical to the light trap but without a light source. We identified potential sites beforehand through aerial photograph analysis. The control traps were positioned near the light traps but maintained a minimum distance of 25 m from them. Throughout the study period, these control traps caught no moths, confirming that our low-intensity shielded light sources only influenced moth behaviour in the immediate vicinity. Additionally, light distribution measurements showed that our modified trap design effectively prevented interference between sampling locations. This stands in contrast to attraction distances documented in previous studies using higher-intensity unshielded light sources with different spectral characteristics. ³⁰

The traps were revisited the next morning and all moth individuals captured were photographed. All moths were released after documentation. This procedure was performed for a total of 16 nights between 26 July 2022 and 5 September 2022, resulting in sampling of moths at a total of 32 sites. The sampling period captured phenological turnover in moth assemblages. This timing accommodated species-specific variations in both phenology and light responsiveness, enabling assessment of light attraction patterns across diverse moth taxa. The multi-week sampling window incorporated natural temporal variation in species composition and abundance, enhancing the robustness of results across ecosystem types. All moths captured in the traps were identified to species level using adequate literature and online databases.^48–50

At each sampling location, alongside with the installation of the light traps, we took notes on the prevailing weather conditions using the weather app provided by the Swedish Meteorological and Hydrological Institute. We recorded the temperature and wind speed from the weather app. In addition, we also noted if the sky was cloudy or clear and also obtained information about the lunar phase during the night from the app Moon Phases and Lunar Calendar (Kinetic stars).

2.3 Data and statistical analysis

All statistical analyses were conducted using R 4.1.0. ⁵¹ We analysed the response variables moth abundance and moth richness at the sites with generalised linear mixed-effects models (in the package lme4 ⁵² ). We used a negative binomial error distribution for these models, as initial analyses using Poisson error distributions indicated overdispersion. In addition, we analysed species diversity using linear mixed-effects models. For all response variables, we included the fixed effects of the ecosystem (forest vs. grassland), temperature (in Celsius), cloudiness (cloudy vs. not cloudy), lunar phase (in %) and wind speed (m·s⁻¹). The sampling date was included as a random variable in all analyses. Since there is a possibility that environmental variables may affect trap catches differently in the ecosystems, we included the two-way interactions between the ecosystem and the environmental variables (i.e. temperature, cloudiness, wind velocity and lunar phase). We used a backward stepwise selection procedure, removing non-significant variables from the full models. The resulting models were then compared using Akaike’s Information Criterion (AIC) to determine the best-fitting model, continuing until no further improvement in AIC was possible. Species diversity (Hill’s effective number of species) of insects in forest and open ecosystems was assessed using Hill’s diversity index, which provides different ways to measure biodiversity. ⁵³ We calculated three complementary diversity measures: q = 0 which counts the total number of species (species richness), q = 1 which accounts for both the number of species and how evenly they are distributed (the exponential of Shannon’s entropy index) and q = 2 which gives more weight to common species (the inverse of Simpson’s concentration index). All taxa are listed in Supplemental Material C.

3. Results

The field experiment yielded 174 moth individuals from 54 species. Species richness at the investigated sites ranged from 1 to 8 species, with abundance ranging from 1 to 21 individuals. Of the 32 sampled sites, six sites captured no moths. Geometridae was the most species-rich family with 13 species (43 individuals), followed by Noctuidae with 11 species (46 individuals) and Crambidae with 10 species (47 individuals). None of the control traps (with no light source) captured any moths.

The total abundance and species richness of the captured moths was significantly higher in the forest compared to the open grassland (abundance: z = −2.38, p = 0.017, species richness: z = −2.36, p = 0.018, Table D1, Supplemental Material D, Figure 3). Abundance and species richness increased with increased cloudiness (abundance: z = 2.42, p = 0.015, species richness: z = 3.33, p < 0.001) while abundance also increased with higher temperatures (z = 2.10, p = 0.035, Table SD1, Supplementary material D, Figure 4). Furthermore, trap catches in the open grassland increased at nights with higher temperatures and slow wind, as indicated by significant ecosystem by temperature (abundance: z = 2.82, p < 0.01, species richness: z = 2.80, p< 0.01) and ecosystem by wind interactions (abundance: z = −2.21, p = 0.027, species richness: z = −2.42, p = 0.016, Table SD1, Supplemental Material D).

Figure 3

Abundance (a) and species richness (b) of the three most dominant families, divided between forest and open grassland ecosystems

Figure 4

3D graphs of abundance (a) and number of species (b) for all moths in forest and open grassland ecosystems with temperature (°C)

In Crambidae, abundance was significantly higher in the forest (z = −5.41, p < 0.001, Table SD1, Supplemental Material D, Figure 5(a)) and increased with temperature (z =2 .14, p = 0.032) and cloudiness (z = 2.27, p = 0.023), and there were significant two-way interactions between ecosystem and temperature (z = 5.36, p < 0.001) and wind speed (z = −2.85, p < 0.01), where trap catches in open grasslands increased with temperature and decreased with wind (Table SD1, Supplemental Material D). However, species richness of Crambidae were only affected by increased cloudiness (z = 2.04, p = 0.042, Table SD1, Supplemental Material D, Figure 5(b)).

Figure 5

(a) Abundance of Crambidae, Geometridae and Noctuidae in relation to temperature and cloudiness. (b) Species richness of Crambidae, Geometridae and Noctuidae in relation to temperature and cloudiness. Error bars represent standard error

In Geometridae, we found no significant main effect of the ecosystem for abundance or species richness (Table SD1, Supplemental Material D, Figure 5), but abundance increased significantly with increased cloudiness (z = 2.20, p = 0.028) and decreased moon phase (z = −2.09, p = 0.037). Furthermore, there was a significant ecosystem by cloudiness interaction effect on Geometridae species richness (z = −2.02, p = 0.044), where the effect of cloudiness varied between ecosystems. Specifically, species richness decreased with increased cloudiness in open grasslands but not in forest (Table SD1, Supplemental Material D, Figure 5(b)).

In Noctuidae, we found no main significant effect of the ecosystem on abundance or species richness (Table SD1, Supplemental Material D, Figure 5). However, for abundance we found significant interaction effects between ecosystem and temperature (z = 2.04, p = 0.042), cloudiness (z = 2.04, p = 0.041) and moon phase (z = −2.01, p = 0.045), where abundance increased in open grasslands with temperature and cloudiness and decreased with moon phase (Table SD1, Supplemental Material D). Species richness of Noctuidae was only significantly affected by temperature (z = 2.41, p = 0.016, Table SD1, Supplemental Material D).

Species diversity, q = 1 (exponential of Shannon’s entropy index, effective number of species) increased significantly with cloudiness (z = 3.81, p = 0.0025), and there was a negative interaction effect between ecosystem and wind speed (z = −2.56, p = 0.025, Supplemental Material D: Table SD2, Figure 6). However, in species diversity q = 2 (inverse of Simpson’s concentration index), we found no significant effects (Supplemental Material D: Table SD2, Figure 6).

Figure 6

Species diversity (Hill’s effective number of species) of insects in forest and open grassland ecosystems.q = 1 (exponential of Shannon’s entropy index, effective number of species) and q = 2 (inverse of Simpson’s concentration index). For q = 0 (species richness), see Figure 5(b). ENS represents the effective number of species. Error bars indicate95% confidence intervals

4. Discussion

To the best of our knowledge, this study provides the first multi-site comparison of moth attraction across ecosystems using warm-coloured LED lighting that simulates typical road lighting conditions, in terms of spectra and intensities. The use of warm-coloured LEDs (3000 K) rather than traditional UV-rich light sources allows for the first realistic assessment of how modern road lighting affects moth populations across different ecosystems, which strengthens the real-world applicability of our findings. Our key findings demonstrate significantly higher moth abundance and species richness in forest than in grassland ecosystems. We also found stronger effects of environmental factors on moth attraction in grasslands.

Forests generally yielded higher catch rates because they act as thermal buffers, moderating temperatures ⁵⁴ and providing shelter from winds, thus creating a more stable and favourable microclimate for insects. This stable environment facilitates extended periods of insect activity, which may explain our findings. In contrast, moth abundance and species richness in open grasslands increased with higher temperatures and increased cloudiness, approaching levels comparable to those observed in forest ecosystems. Our findings align with previous studies demonstrating reduced catches at lower temperatures and decreased cloudiness. ³⁴ Importantly, our study further reveals that forest environments exhibit less variation in catch rates across different weather conditions.

A majority of the research on moth responses to anthropogenic light at night has focused on behaviour and attraction.^{20,23,55–59} Other research has examined environmental differences on moths. Merckx et al. ²⁷ studied different management regimes using light traps in forest areas, finding that sheltered and dark forest had highest moth abundance and species richness. While some species occurred exclusively in dense forest, others were only found in more open forest areas created by coppicing and ride widening, which provided complementary conservation value by supporting additional species with affinity for more open biotopes. However, their study focused on comparing woodland treatments rather than contrasting forests with grassland habitats. A field study by Niermann and Brehm ⁶⁰ demonstrated differences in microhabitats, with higher abundance and species richness in catch traps at moderately sheltered sites (closer to bushes) compared to exposed sites (i.e. more than 10 m from bushes). Additionally, Straka et al. ²⁵ used UV-light traps to catch moths in dry grassland ecosystems with differing tree cover and impervious surface amounts along an urbanisation gradient in the presence of outdoor lighting. They found a positive effect of tree cover density on species abundance and richness, although this effect was primarily driven by results from a single site. These findings, along with our results, underscore the importance of considering ecosystem variability when studying the impacts of light pollution on moth populations.

Previous studies comparing ecosystems ³⁰ used specialised insect-attracting lights with intensities and spectral distributions very different from typical outdoor lighting and were restricted to single sites per habitat type. Our study captured fewer moths (mean 5 individuals per trap night) compared to studies using specialised UV-light traps (e.g. 63 individuals per trap night ⁶⁰ ). Direct comparisons with other moth attraction studies in terrestrial ecosystems are challenging as many do not report catch rates per night for Lepidoptera. However, our lower capture rates are in line with trap catches in a study by Bolliger et al., ⁶¹ who caught 94 Lepidoptera individuals in 44 nights in traps placed directly below LED road lights of 2700 K and 6500 K (2 moths per trap night). Furthermore, Justice and Justice ⁶² showed significantly lower capture rates of moths to LED lights (3000 K: 4 moths per trap night, 5000 K: 3 moths per trap night) compared to conventional light sources (e.g. halogen: 23 moths per trap night). Similarly, Wakefield et al. ²⁴ captured significantly fewer Lepidoptera with LED lamps (2700 K and 5000 K) compared to conventional lighting types (compact fluorescent and tungsten filament). In a follow-up study, Wakefield et al. ²⁸ found that 4250 K LEDs attracted fewer Lepidoptera than metal halide lamps, with high-pressure sodium lamps showing the lowest attraction rates. However, these studies used other experimental designs and lamp wattage and cannot directly be compared with our results. Importantly, the consistently lower moth attraction to warm-coloured LEDs across studies indicates that the use of warm-coloured LEDs is a significant improvement for reducing ecological impact. Crucially, despite using warm-coloured LEDs at low intensities comparable to spill light from road lighting in natural environments, we documented significant moth attraction, with zero captures in control traps. This finding suggests that current road lighting practices may have more extensive ecological impacts than previously recognised.

In our study, we observed distinct differences in catches among various taxonomic groups across forest and open grassland ecosystems. In forests, we caught higher numbers of Geometridae (31) and Crambidae (26) compared to open grasslands (12 and 21, respectively), while Noctuidae showed similar abundance in both ecosystems (22 in forest, 24 in open). Species richness also varied, with forests hosting more Geometridae (18) and Crambidae (14) species than open grasslands (8 for both), while Noctuidae showed higher species richness in open grasslands (14) compared to forests (6). These findings align with Merckx and Slade, ³⁰ who demonstrated family-specific sampling areas (attraction radii) and efficiencies for light traps. They found that Erebidae were attracted from up to 27 m, Geometridae from up to 23 m and Noctuidae from up to 10 m, with varying capture rates among families. While the specific mechanisms for the family-specific differences in attraction to light traps in different ecosystems in our study cannot simply be explained, it may depend on various ecological traits among taxonomic groups. The varied responses across families and ecosystems could potentially bias results if not accounted for in light attraction studies. Therefore, it is important to evaluate trap efficiency differences among families when interpreting results. ³⁰

We measured light distribution at different heights and distances from the light trap, providing detailed knowledge about light exposure around it and ensuring repeatability. This aspect is rarely addressed in other studies investigating insect attraction to light. It has been suggested from a controlled lab experiment that a threshold for impacts of light at night in the Greater wax moth (Galleria mellonella) is approximately 0.3 lx. ¹⁹ Storms et al. ⁶³ demonstrated impacts on moth mating behaviour at their lowest tested intensity of 150 lx, suggesting thresholds for behavioural disruption occur at lower intensities. However, their experimental intensities represented conditions near luminaires, whereas ecological impacts in natural environments occur at greater distances where insects experience lower light exposure. We used a light trap with an intensity exceeding 0.3 lx (bright full moonlight), which illuminated a circular area with a 1.5 m to 2 m radius at different ground heights (see also Supplemental Material B). This resulted in a theoretically estimated insect attraction area of approximately 12.6 m². This approach allows for a more precise quantification of the affected area and light-insect relationships, addressing a significant gap in the existing literature by establishing a defined area of influence. Therefore, we can better estimate the attraction radius, with the potential to extrapolate the results to larger-scale lighting scenarios. The current absence of standardised protocols for lighting, light exposure and biological response metrics in ecological research, combined with heterogeneity in experimental design, significantly limits comparability between studies, making it difficult to determine which factors drive observed differences in results. Our methodological approach offers a reproducible framework for future research on the impacts of anthropogenic light on biodiversity, enabling other researchers to compare their results across ecosystems and light sources. Future studies would also benefit from real-time cloud cover measurements or higher temporal resolution meteorological data aligned with insect sampling periods.

Our findings have important implications for road lighting design guidelines and standards, which typically focus on human visual needs and safety requirements while treating all natural environments uniformly.^64,65 While international frameworks such as IUCN ⁴¹ establish foundational principles for mitigating light pollution in natural areas, the transition from generalised mitigation strategies to evidence-based, ecosystem-specific interventions remains crucial for effective ecological conservation. Since adult moths are active for short periods, further limited by unfavourable weather conditions, ⁶⁶ the added disruption from outdoor lighting in open ecosystems significantly shortens this critical activity window, ultimately reducing opportunities for mating, foraging and reproduction, which could impact population survival and ecosystem functions.

Grassland ecosystems may require more stringent light control measures, such as enhanced luminaire shielding, while forest canopy structure provides natural light attenuation that may partially mitigate anthropogenic light impacts. For example, adaptive lighting systems could be programmed to dim during key periods of insect activity, helping to mitigate impacts on critical behaviours. Further research across ecosystems with varied microclimates could reveal greater differences and lead to more effective mitigation strategies.

Conclusions

This study provides the first multi-site comparison of moth attraction across ecosystems using light sources representing typical road lighting conditions, revealing significant ecological differences and demonstrating the need for ecosystem-specific lighting guidelines. Our findings demonstrate higher catch rates in forests compared to grasslands, with environmental factors such as temperature and cloudiness exerting a stronger influence on moth attraction in open grassland ecosystems. Even at the low light intensities used in our experiment, which is comparable to spill light from road lighting in natural environments, we found significant impacts on moth activity, which is particularly concerning given their limited nocturnal activity periods. The family-specific variations in light trap catches between ecosystems, highlights potential sampling biases that should be considered in future research. Additional research is needed using standardised experimental designs comparing moth attraction across different light sources that represent road lighting conditions to enable inter-study comparisons and ensure real-world applicability. Our method, with comprehensive light characterisation, enables reliable ecosystem comparisons and provides a standardised framework for quantifying light pollution effects in future field studies. These results demonstrate the need for ecosystem-specific approaches and tailored light pollution mitigation strategies. This work calls for further research into the mechanisms driving these patterns and expanded studies across diverse ecosystems, regions and seasons.

Supplemental Material

sj-docx-1-lrt-10.1177_14771535251400296 – Supplemental material for Ecosystem-specific moth attraction to warm-coloured LED simulating road lighting conditions

Supplemental material, sj-docx-1-lrt-10.1177_14771535251400296 for Ecosystem-specific moth attraction to warm-coloured LED simulating road lighting conditions by P Andersson and AK Jägerbrand in Lighting Research & Technology