The amount of artificial light at night is growing worldwide, impacting the behaviour of nocturnal organisms. Yet, we know little about the consequences of these behavioural responses for individual fitness and population viability. We investigated if females of the common glow-worm Lampyris noctiluca – which glow in the night to attract males – mitigate negative effects of artificial light on mate attraction by adjusting the timing and location of glowing to spatial variation in light conditions. We found females do not move away from light when exposed to a gradient of artificial light, but delay or even refrain from glowing. Further, we demonstrate that this response is maladaptive, as our field study showed that staying still when exposed to artificial light from a simulated streetlight decreases mate attraction success, while moving only a short distance from the light source can markedly improve mate attraction. These results indicate that glow-worms are unable to respond to spatial variation in artificial light, which may be a factor in their global decline. Consequently, our results support the hypothesis that animals often lack adaptive behavioural responses to anthropogenic environmental changes and underlines the importance of considering behavioural responses when investigating effects of human activities on wildlife.

Behavioural responses to light pollution

We collected glow-worm females in June 2017 from the surroundings of Tvärminne Zoological Station (N 59°51', E 23°14') in Southern Finland. We collected them by hand at night and transported them to the lab, where we placed them in individual vials (diameter: 8 cm) containing fresh moss and leaves. The vials were kept at room temperature, approx. 21°C, at a 20h: 4h light: dark cycle, as darkness is restricted to only 4 h during the height of the breeding season at the latitude.

We investigated the responses of females to artificial light on the night after capture. We used a 100 cm x 15 cm arena with a white light emitting diode, LED light (5 mm, cold white; peak intensity ~0.32 µW/nm (microwatts/nanometer)  at 660 nm (red) with a secondary peak intensity of ~0.26 µW/nm at 440 nm (blue), as measured with spectrophotometer and integrating sphere), at one of the short ends (Figure 1). Light intensity was 40 lx at the lit end, 1.5 lx in the middle, and 0.5 lx at the dark end of the arena, measured with an AIRAM UVM-B lx-meter. We covered the bottom of the arena with soil and placed a cardboard rectangle, 4 cm high, along the length of the arena (in the middle of it) for the female to perch on, as females usually climb up on suitable structures to enhance the visibility of the glow to flying males [28, 33]. A seashell was placed at each short end of the arena for the female to hide under. We alternated the position of the LED light between the two short ends among replicates.

We had two treatments: a light treatment and a control. We started each replicate at 11 PM by placing a female on the cardboard in the middle of the arena and turning on the LED light in the light treatment while leaving the LED light unlit in the control. We recorded the position of the female every 20 min for 2 h and noted whether she had settled down and initiated glowing. The borders of the arena had markings 10 cm apart, which we used to determine the position of the female and calculate the distance moved during each 20 min period. We defined females as settled when they had kept the same position over at least two consecutive observations and recorded the location and time of settling. If a female settled multiple times, we used the last settling location and time in the analysis. We estimated distance moved by summing distances moved during each 20 min period. This could underestimate the distance moved, as females may not move in a straight line, however, this estimation was consistent across trials and treatments. To investigate whether glow intensity influences responses to artificial light, we measured the maximum width of the pronotum (the structure that covers the dorsal surface of the thorax) of the female, as this correlates with lantern size, which roughly correlates with glow intensity [29].  We tested 26 females in the light treatment and 63 females in the control.

To analyse the impact of the presence of artificial light and the pronotum width (glow intensity) of the female on behaviours that were binary, we used logistic regression, with separate models for each behaviour. The behaviours were: (i) whether a female started to glow, (ii) the direction of movement in relation to the light source (towards or away), (iii) whether a female hid under the shell or within the soil, and (iv) whether a female settled down.

To analyse the impact of artificial light on continuous measures, we used ANOVA, with separate models for each measure. The measures were: (i) total distance moved during the 2 h of observation, and (ii) the settling location (for those that settled). Pronotum width (glow intensity) was included as a covariate in the initial models, but as the interaction terms and the main effects were non-significant, and the removal did not influence the significance of the other variables, it was removed in the final model. To analyse the influence of artificial light on the latency to glowing, we used a time-to-event analysis, Cox proportional hazards model [34]. The times were “right-censored” for females that failed to glow by the end of the experiment. We checked that the data fulfilled the requirements of the analyses. All analyses were performed using SPSS 25. Significance was designated as P < 0.05.

Impact on mate attraction

We conducted the experiment from June 6th to July 7th in 2017 in the surroundings of Tvärminne Zoological station. We selected six sites that lacked shadowing trees or bushes, and where we had detected glow-worms during earlier years. We manipulated the presence of artificial light by erecting a pole and attaching a white LED light (ANSI FL1 Standard: 35 lumen, beam distance 24 m) at the height of 1.7 m (Figure 2). The angle between the ground and the direction of the centre of the light was 55 degrees.

We placed two dummy females (LED lures, see construction in [29]) at two different distances, in a straight line from the pole (Figure 2). One dummy was placed in the inner position (within the cone of light from the pole when lit), at 1.1 m from the pole, at 15–20 lx (peak intensity ~0.08 µW/cm2/nm at 455 nm, measured with spectrophotometer and cosine corrector), which corresponds to light levels under common streetlights [2, 35]. The other dummy was placed in the outer position, at 2.3 m from the pole, at 1–1.5 lx (peak intensity ~0.004 µW/cm2/nm at 455 nm), which is slightly brighter than natural light levels at night in the area in the summer (0.1–0.6 lx, peak intensity 0.0003–0.0016 µW/cm2/nm at 460 nm, measured on May 31th and June 1st 2020 at 12.30 AM in an open area on both an overcast and a moonlit night). The dummies were designed to trap males within a plastic bottle [29]. The wavelength of the LED lures was 562 nm, similar to female glow-worms (550–570 nm) [33, 36]. To investigate the effect of glow intensity on mate attraction, we varied the glow intensity of the two dummies among replicates; peak glow intensity ~0.03, 0.06, and 0.13 µW/nm  (measured with spectrophotometer and integrating sphere). The paired dummies within a replicate had the same glow intensity. The differences in glow intensity reflected natural variation in the wild (AM Borshagovski, unpublished data on spectrophotometer measurements).

We started each trial when dusk began to fall (approximately 10 PM), by turning on the glow of the two dummy females, as well as the light from the pole in the artificial light treatment, while leaving poles unlit in the control. We checked the dummy females 3–4 h later (at 1–2 AM) for the presence of males and turned off all lights. Males are unlikely to escape from the traps (Elgert, personal observation).

We conducted 38 replicates of the artificial light treatment, with three dummy glow intensities (low: n = 13; medium: n = 14; high: n = 11, with two dummy females in each replicate) and 19 replicates of the control, with three dummy glow intensities (low: n = 6; medium: n = 5; high: n = 8, with two dummy females in each replicate). We distributed the treatments (presence of artificial light and glow intensities) equally among the six sites.

In the analyses, we used the presence or absence of males in each dummy female trap as the response variable, rather than the number of males caught, to rule out the possibility that the presence of one male had attracted additional males. We analysed the data using a GLMM with binomial error distribution and logit link function, with presence or absence of male(s) in each dummy trap as the binary response variable. We used light treatment (light on or light off), position with respect to the light fixture (inner position or outer position), and glow intensity as fixed factors, and date and site as random factors, with site nested within date. We started with a full model and deleted non-significant interaction terms and fixed terms when this was supported by the Akaike Information Criterion (AIC) and did not reduce the significance of other terms [37]. All analyses were performed using SPSS 25. Significance was designated as P < 0.05.

The data, legends and captions for both the laboratory and field experiment are provided in the same document, on separate tabs. Female F159 is missing size data.