Introduction
Artificial lights at night (ALAN) is a phenomenon characterised by light pollution in which natural darkness is artificially brightened (Gaston et al. 2013, Guetté et al. 2018). ALAN can be categorised into two forms: direct illumination from a short optical path and scattered light, resulting from the large-scale diffusion of upwelling light, which creates a bright haze known as skyglow (Aubé 2015). Due to skyglow, the effects of ALAN can extend hundreds of kilometres from the source of origin (Duriscoe et al. 2013). As light often serves as a source of information, it can provide information for organisms of their location, time and the characteristics of their surroundings. Consequently, alterations in natural light can provide organisms with misinformation, thereby affecting their behaviour (Gaston et al. 2013).
For fireflies, bioluminescence is crucial for communication and natural darkness is generally considered essential for their successful signalling (Owens and Lewis 2022). Artificial lights are known to disrupt this process, reducing mating success (Elgert et al. 2020, Owens and Lewis 2022) and, consequently, leading to declines in population abundance (Hagen et al. 2015, McNeil et al. 2024). Despite these negative effects, fireflies are still observed in brightly lit cities like Kuala Lumpur, Morelia and Turin (Picchi et al. 2013, Mbugua et al. 2019, Pérez-Hernández et al. 2023). This paradox highlights a major gap in our understanding: while it is generally accepted that artificial light negatively impacts fireflies, there is uncertainty about the extent to which these small, limited-mobility insects can exploit their environment to find refuge from light and how such behaviour might influence population persistence. Some research on nocturnal insects suggests that light-polluted areas may act as demographic traps, where immigration exceeds emigration, but courtship and reproductive success are diminished due to disrupted communication (Firebaugh and Haynes 2019). Larvae and female glow-worms, for example, appear especially maladapted to artificial light (Mbugua et al. 2019, Elgert et al. 2020), while adult fireflies may even be attracted to certain wavelengths like red light (Muharraran et al. 2026). A study suggested that increased canopy cover improves microclimatic moderation, especially in the leaf litter (Keppel et al. 2017), which could also result in localised dark patches for nocturnal insects such as fireflies.
To mitigate the effects of light pollution, conservation strategies often recommend planting trees as buffers against artificial light (Lewis et al. 2024). However, whether this recommendation is actually effective and to what extent, has received little empirical testing in the field. Previous work linking fireflies to vegetation structure has largely focused on mangrove habitats, where vegetation was associated with firefly presence, but canopy cover itself was not directly assessed (Jusoh et al. 2010). To our knowledge, no study has yet examined this relationship in an urban park setting, nor within Southeast Asia, where rapid urbanisation is placing increasing pressure on remaining firefly habitats (Lewis et al. 2024). In this study, we addressed this gap by investigating how firefly populations responded to artificial light and canopy cover in an urban park in Kuala Lumpur, Malaysia. By assessing firefly occurrence in relation to fine-scale spatial variation in both light exposure and canopy cover within the same park, we tested whether canopy cover provides functional dark refuges for fireflies under light pollution conditions. These findings may, in turn, provide insight into how urban green spaces in tropical cities can be managed to support pre-existing firefly populations, with implications that may extend to other urban parks facing similar light pollution pressures.
Materials and Methods
Study Site
The study was conducted in the urban park in Bukit Kiara, located in Kuala Lumpur, Malaysia [3.145748°N, 101.633559°E] as shown in Fig. 1. The park is situated between two major residential and commercial land-use areas (Taman Tun Dr Ismail to the west and Bukit Damansara to the east) and is bordered by primary roads (Penchala Link to the north and Jalan Sri Hartamas to the east) as shown in Fig. 2, making it an appropriate site to study the effects of artificial lighting and microhabitat features on firefly populations in an urban environment. Following the most recent re-discovery of Pteroptyx gombakia in Bukit Kiara (Tan and Jusoh 2025), a total of nine firefly species have been documented at this site. Following the classification of fireflies by signalling systems by Owens et al. (2022), these comprise two glow-worm fireflies (Lamprigera sp. and Pyrocoelia sp.), three dark fireflies (Drilaster sp., Stenocladius sp. and Lucidina sp.) and four roving flashing fireflies (Colophotia sp., Pygoluciola dunguna, Diaphanes sp. and Pteroptyx gombakia) (Friends of Bukit Kiara 2025). Unlike temperate regions, fireflies in Southeast Asia are not known to exhibit seasonal patterns, though this remains unstudied.
Bukit Kiara is a 466-hectare former rubber plantation situated near the Taman Tun Dr Ismail (TTDI) township, which has since been rehabilitated into a secondary forest (Jabatan Landskap Negara 2025). A paved tarmac road runs through the park, with streetlights positioned along its entire length, as shown in Fig. 3. Forest trails branch off the tarmac road and remain relatively undeveloped, consisting mainly of cleared pathways through the surrounding vegetation. Five forest trails and the tarmac road were surveyed for our study. The streetlights along the tarmac road were measured to emit an average intensity of 80.69 lux with a lux meter (Pros’Kit MT-4007 Light Meter), with an effective average range of 18.29 m. The spectral emission of the streetlights was measured between 410 – 750 nm using a portable spectrophotometer (Lightning Passport Spectrometer, Asensetek) as shown in Fig. 4.
Data Collection
Firefly surveys were conducted at Bukit Kiara between November 2023 and April 2024, using a modified Pollard walk method (Pollard 1977), whereby all fireflies observed within a 15 m x 15 m boundary were recorded and their GPS positions marked using a handheld GPS (Garmin GPSMAP 64s). This survey method was first adapted to evaluate firefly species diversity, distribution and abundance in Singapore (Chan et al. 2012) and has since been applied in recent urban firefly studies (Mbugua et al. 2019, Muharraran et al. 2026). To minimise double counting, only fireflies flying towards or across the observer within the 15 m boundary were recorded, while those observed outside this boundary or approaching from behind were excluded (Pollard 1977). Monitoring was paused when GPS coordinate recording was necessary and resumed upon resuming the walk. GPS coordinates were recorded at the point where each firefly was first observed upon entering the survey boundary. Survey started each night after sunset, which varies between 19:00 and 19:30 h. A steady walking pace was maintained throughout each survey. Where no fireflies were observed within a sampling unit during the allocated survey time, the count was recorded as zero.
To assess the impact of artificial lights, only the tarmac road, which will be referred to as a trail, within the park was sampled. The trail was divided into 100 m transects, totalling 21 transects along its length. The tarmac trail was surveyed over 31 nights, providing each transect with 31 replicates. The number of functional streetlights within each transect was recorded and the GPS position of each streetlight was noted using a hand-held GPS device. For each firefly observed, the distance to the nearest streetlight was calculated using the GPS coordinates of both the firefly sighting and the streetlight. The ambient light intensity at the location of the firefly was measured with a lux meter (Pros’Kit MT-4007 Light Meter).
To examine the influence of microhabitat characterisation, canopy cover was used as an indicator. All trails within the park were sampled. Canopy cover was measured at 25 m intervals along each trail, with each measurement location referred to as a canopy cover point. At each canopy cover point, an upward-facing photograph of the canopy was captured from the forest floor, using the same phone at a fixed focal length of 0.6x. Photographs were processed using ImageJ version 1.54 (Schneider et al. 2012) to calculate canopy cover as a percentage. Across six trails, a total of 154 canopy cover points were recorded. Measurements were conducted once in April 2024, with the assumption that canopy cover remained relatively stable throughout the study period. Surveys were conducted over 39 nights, with one trail surveyed per night and each trail surveyed between six and seven times. For each survey night, the number of fireflies observed within each 25 m segment was tallied. Fireflies observed outside the boundaries of a segment were excluded from the count.
Data Analysis
The relationship between environmental variables and firefly abundance was analysed using negative binomial generalised mixed models (GLMMs) fitted with the glmmTMB package (Brooks et al. 2017) in R version 4.5.2 (R Core Team 2025). A negative binomial distribution was selected over a standard Poisson distribution as the count data for both analyses exhibited variance exceeding the mean (Stoklosa et al. 2022b).
To examine the effect of artificial lighting on firefly abundance, total firefly count per transect per night was modelled as a function of the number of streetlights per transect (mean = 0.64; SD = 1.63; Suppl. material 1) as a fixed effect. To account for unmeasured environmental variation amongst transects and the repeated sampling of each transect across multiple survey nights, the transect was included as a random intercept. The survey date was also included as a random intercept to account for temporal variation in firefly activity across nights, driven by factors such as weather, temperature and lunar phase. Additionally, Spearman’s rank correlation through SPSS version 31 (IBM Corp 2024) was used to evaluate whether fireflies exhibited preferences for locations closer to or further from artificial light sources, based on the distance between each observed firefly and the nearest streetlight (Suppl. material 2).
To examine the effect of canopy cover on firefly abundance, total firefly count per canopy cover point per night was modelled as a function of canopy cover (%, z-standardised; mean = 80.4%, SD = 18.7%; Suppl. material 3) as a fixed effect. Three random intercepts were included to account for the nested and repeated structure of the data. Trail was included to account for shared environmental conditions amongst survey points within the same trail, a canopy cover point nested within the trail was included to account for unmeasured local site characteristics that may make repeated observations at the same point more similar than observations at different points and survey date was included to account for temporal variation in firefly activity across nights.
A model diagnostic for both GLMMs was evaluated using DHARMa v.0.4.7b (Hartig 2024) and simulated residuals in R version 4.5.2. It generates 1,000 simulated datasets for each fitted model and compares observed residuals with the simulated distribution to identify issues such as overdispersion and zero inflation.
Results
Overall, this study demonstrated that firefly sightings were more concentrated in areas with higher canopy cover, while areas with little to no canopy cover had fewer sightings. Within the tarmac trails, fireflies tended to concentrate in areas that did not overlap with streetlight coverage, although some overlap was observed (Fig. 5; Suppl. material 1).
Effect of Artificial Lighting on Firefly Abundance in Bukit Kiara
A total of 380 observations were recorded across 21 transects and 31 survey nights (Suppl. material 2). Firefly count per transect per night ranged from 0 to 13, with 72.4% of observations recording zero fireflies. The number of streetlights per transect ranged from 0 to 4. Ambient light levels recorded at firefly observed locations ranged from 0.01 to 73.40 lux (n = 191, mean = 0.58 lux, SD = 5.33), indicating that the majority of fireflies were observed in very low light conditions.
The negative binomial GLMM revealed a statistically significant negative association between streetlight density and firefly abundance (β = −0.46, SE = 0.18, z = −2.54, p = 0.011), as shown in Fig. 6. Incidence rate ratio indicated that each additional streetlight within a 100 m transect was associated with a 36.7% reduction in expected firefly count (IRR = 0.63, 95% CI: 0.44–0.90). Amongst the random effects, transect-level variation accounted for the larger source of variability (SD = 0.93) compared to survey date (SD = 0.34). Model diagnostic using DHARMa simulated 1000 residuals and indicated adequate model fit, with no evidence of overdispersion (dispersion = 0.75, p = 0.816) or zero-inflation (observed/simulated zero ratio = 1.03, p = 0.686).
Spearman’s rank correlation revealed a significant positive weak correlation between distance to the nearest streetlights and the number of fireflies observed (rs = 0.293, p = 0.004, n = 96, based on thresholds proposed by Cohen (2013), where |r| = 0.10 - 0.29 is considered weak), indicating that fireflies were observed more at locations further from artificial light sources, as shown in Fig. 7.
Effect of Canopy Cover on Firefly Abundance in Bukit Kiara
A total of 1,644 observations were recorded across 154 canopy cover points, six trails and 39 survey nights (Suppl. material 3). Firefly counts ranged from 0 to 37, with 69.8% of observations recording zero fireflies. Canopy cover across the sampled points ranged from 26.5% to 98.8% (mean = 80.4%, SD = 18.7%).
The negative binomial GLMM revealed a statistically significant positive association between canopy cover and firefly abundance (β = 0.56, SE = 0.16, z = 3.54, p < 0.001), as shown in Fig. 8. The incidence rate ratio indicated that, for each one standard deviation increase in canopy cover (18.7%), the expected firefly count increased by 75.1% (IRR = 1.75, 95% CI: 1.28 - 2.39). Amongst the random effects, trail-level variation accounted for the largest source of variability (SD = 1.34), followed by canopy cover point nested within trail (SD = 0.86) and survey date (SD = 0.54). Model diagnostics using DHARMa simulated 1,000 residuals indicated adequate model fit, with no evidence of overdispersion (dispersion = 0.03, p = 0.326) or zero-inflation (observed/simulated zero ratio = 1.31, p = 0.252).
Discussion
The results of this study indicate that artificial lighting is associated with reduced firefly distribution and abundance in urban parks, consistent with previous research (e.g. Owens and Lewis (2022), Owens et al. (2022)). Spatial mapping data of our surveys in Bukit Kiara revealed that fireflies were distributed in areas with lower artificial light presence and the negative binomial GLMM confirmed that each additional streetlight within a 100 m transect was associated with a 36.7% reduction in expected firefly count. The majority of fireflies were observed in very low light conditions, with a mean ambient light level of 0.58 lux at sighting locations. Furthermore, where fireflies were present, a significant positive correlation was found between the number of fireflies observed and their distance from the nearest light source. However, the strength of this correlation was weak (rs = 0.293). Notably, Muharraran et al. (2026) reported that adult males of Lamprigera sp., a genus present in Bukit Kiara, are attracted to red and warm-white light wavelengths, which would predict greater firefly activity closer to streetlights rather than further away. However, while the streetlights in Bukit Kiara do emit a peak at the red wavelength (641 nm), they also produce peaks at shorter wavelengths in the blue spectrum (424 and 472 nm), which Muharraran et al. (2026) speculated may have a deterrent effect on fireflies.
The moderate strength of the correlation between firefly observations and distance from streetlights may be attributed to several factors. Firstly, a greater distance from a streetlight does not necessarily equate to lower ambient light levels, as habitat features such as dense vegetation or built structure may shield light from reaching certain areas. This is supported by the finding that the majority of firefly observations were still recorded at very low ambient light levels, regardless of distance, suggesting that microhabitat characteristics mediate light exposure. Additionally, the moderate correlation may reflect a degree of tolerance to artificial light amongst fireflies. While fireflies generally thrive in low light environments, some individuals may tolerate moderate light levels, particularly at certain life stages or amongst certain species. A study on the behavioural responses of Lamprigera sp. to artificial light found that different life stages responded differently to the same light stimulus (Muharraran et al. 2026). The mixed spectral composition of the streetlights in Bukit Kiara, combining both attractive and potentially deterrent wavelengths, may further contribute to the moderate correlation by producing conflicting behavioural responses across species and life stages. Future studies should consider incorporating species identity and life stage as variables in the GLMM.
It remains unclear whether the reduced firefly abundance near artificial lights is driven by behavioural avoidance or physiological consequences of light exposure, such as reduced activity (Mbugua et al. 2019, Elgert et al. 2020, Owens and Lewis 2022, Owens et al. 2022, Muharraran et al. 2026), decreased mating (Elgert et al. 2020, Owens and Lewis 2022) and disrupted bioluminescent signalling (Owens et al. 2018, Owens and Lewis 2022). Previous studies have found that certain species of firefly larvae and female glow-worms exhibited spatial maladaptive responses, becoming immobilised when exposed to artificial light rather than moving away from the light source (Mbugua et al. (2019), Elgert et al. (2020), Muharraran et al. (2026)). However, Muharraran et al. (2026) observed soil-excavating behaviour under very intense white light and subtle signs of light-avoidable behaviour, with similar avoidance responses reported in female Photinus sp. (Owens and Lewis 2022). Furthermore, adult fireflies are capable of flashing more brightly in response to light (Owens et al. 2018) and appear to select for brighter signals (Elgert et al. 2021). These observations suggest that some degree of adaptive response persists despite the negative effects of artificial light exposure.
Regardless of the underlying mechanism, our study indicates that increased streetlight is associated with a reduced range of active habitat available to fireflies in urban parks. Thus, reducing the overall number of streetlights within urban parks or positioning lights away from areas where fireflies are likely to occur may help preserve a greater extent of usable habitats for firefly populations. Although firefly sightings within illuminated areas (up to 73.40 lux, equivalent to the average light intensity of the streetlights in Bukit Kiara measured at source) may suggest movement through lit areas is still possible, these observations are rare. This is further supported by the findings of Elgert et al. (2021), which found that, while mating selection still occurred under artificial light, mating likelihood was greatly reduced. However, we suspect that these instances may also be facilitated by habitat features that provide localised shading from artificial light (Straka et al. 2019) or may reflect species or life-stage-specific tolerance to higher light levels.
The role of habitat features in mediating the impact of artificial light on fireflies is further supported by the canopy-cover analysis. Using canopy cover as an indicator of microhabitat conditions, the negative binomial GLMM revealed that, for an 18.7% increase in canopy cover, the expected firefly count increased by 75.1%, indicating a strong positive association between overhead vegetation cover and firefly abundance. Canopy cover has been shown to reduce the brightening effects of skyglow and bright lunar phases (Straka et al. 2019, Sung 2022), both of which have been previously found to negatively affect firefly populations (Gunn and Gunn 2013, Jaikla et al. 2020). This pattern suggests that canopy cover and artificial light may not act as fully independent drivers of firefly abundance, although this was not tested directly as a statistical interaction in this study. Maintaining or restoring dense canopy cover, particularly in areas exposed to nearby lighting, may therefore serve as a practical means of protecting dark refuges within urban parks.
This study has several limitations that should be considered when interpreting its findings. First, firefly observations were not resolved to species level, despite nine species being documented at Bukit Kiara (Friends of Bukit Kiara 2025). Given that light tolerance and habitat use may vary considerably across species and life stages (Muharraran et al. 2026), pooling observations may obscure species-specific responses to artificial light and canopy cover. Second, while canopy cover was used as a broad indicator of microhabitat condition, other vegetation-related features, such as understorey density, leaf litter depth and ground-level vegetation structure, were not directly measured, despite their potential role in moderating microclimate and providing additional dark refuges (Keppel et al. 2017). Lastly, lunar phase, which has been shown to influence firefly abundance, was not accounted for in this study and may have contributed to unexplained temporal variation in firefly counts (Jaikla et al. 2020). Future studies incorporating species-level identification, finer-scale vegetation structure variables and lunar phase as model covariates would help clarify the relative contribution of these factors to firefly persistence in urban parks.
Our dataset had a high proportion of zero counts, which is expected given the urban park setting, where firefly presence might not be as common as in pristine habitats. Examination of the random effects revealed that the greatest variability in firefly counts in both models was attributed to differences between trails, suggesting that site-specific characteristics, such as localised habitat features, play an important role in determining firefly abundance. In contrast, survey date accounted for the least variability amongst the random effects. This is consistent with the tropical climate of Kuala Lumpur, where ambient temperatures and humidity within forested areas remain relatively high and stable year-round, reducing the influence of nightly weather variation on firefly activity. As this study was based in a single urban park, the transferability of these findings to other sites remains to be tested. However, the broader principle that canopy cover can help buffer the effects of artificial lights is likely to extend to other urban green spaces facing similar light pollution pressures.