can birds sleep with lights on

Artificial light at night (ALAN) is an increasing phenomenon associated with worldwide urbanization. In birds, broad-spectrum white ALAN can have disruptive effects on activity patterns, metabolism, stress response and immune function. There has been growing research on whether the use of alternative light spectra can reduce these negative effects, but surprisingly, there has been no study to determine which light spectrum birds prefer. To test such a preference, we gave urban and forest great tits (Parus major) the choice where to roost using pairwise combinations of darkness, white light or green dim light at night (1.5 lux). Birds preferred to sleep under artificial light instead of darkness, and green was preferred over white light. In a subsequent experiment, we investigated the consequence of sleeping under a particular light condition, and measured birds daily activity levels, daily energy expenditure (DEE), oxalic acid as a biomarker for sleep debt and cognitive abilities. White light affected activity patterns more than green light. Moreover, there was an origin-dependent response to spectral composition: in urban birds, the total daily activity and night activity did not differ between white and green light, while forest birds were more active under white than green light. We also found that individuals who slept under white and green light had higher DEE. However, there were no differences in oxalic acid levels or cognitive abilities between light treatments. Thus, we argue that in naive birds that had never encountered light at night, white light might disrupt circadian rhythms more than green light. However, it is possible that the negative effects of ALAN on sleep and cognition might be observed only under intensities higher than 1.5 lux. These results suggest that reducing the intensity of light pollution as well as tuning the spectrum towards long wavelengths may considerably reduce its impact.

Light pollution refers to the diminishing of darkness during night-time, caused by light from anthropogenic sources. Artificial light at night (ALAN) can threaten ecosystem dynamics through alterations in the biological timing of a wide range of species, with far-reaching consequences [1,2]. For instance, ALAN can lead to lethal consequences due to attraction to light sources, such as for hatching sea turtles [3] and migrating birds [4]. Night-time illumination can also have more subtle effects through changes in physiological processes and behaviour due to disruption of natural circadian rhythms and sleep, which in turn may affect the individuals health and ultimately fitness [1].

Birds use light cues for synchronizing their biological rhythms [5] and ALAN can alter their photoperiodic perception [6–9]. Consequently, ALAN can affect the timing of reproductive physiology and behaviour [8,10,11], timing of dawn singing [10,12,13] and sleep behaviour [14] of free-living passerine songbird species. Experimental studies on captive songbirds have confirmed work in the wild [7,15]. Blackbirds increase locomotor activity at night when roosting under light compared with darkness [15]. Similarly, great tits advance activity, delay activity offset, and move a higher proportion of their daily activity into the night when exposed to ALAN [7].

Although there is increasing evidence that ALAN alters biological rhythms, the consequences of such alteration are not always fully understood. ALAN can decrease melatonin production at night [16], increase blood inflammatory markers [17] and increase susceptibility to pathogens [18]. However, the increase in nocturnal activity due to ALAN could have a major impact also on energy consumption and sleep [19,20]. Energy is a crucial and limited resource for animals, and there is a trade-off between investment decisions on behavioural and/or physiological processes and these trade-offs are often associated with fitness [21]. A measurement of energy metabolism is daily energy expenditure (DEE). While DEE is mostly affected by body mass [22], it can also be influenced by environmental factors such as human disturbance, temperature and food availability [21,22]. In the context of light pollution, higher activity at night due to ALAN could potentially increase the energy expenditure of diurnal animals, with carry-over consequences on other physiological systems as well as fitness. However, in a recent field study on great tits (Parus major), we showed that a lower DEE was related to breeding in territories illuminated with white and green lights compared with dark areas [16]. This decrease in DEE could be explained by other ecological factors, such as the increase in food availability (insects) in artificially illuminated areas [16]. Furthermore, in forest areas, birds can avoid artificial lighting by choosing a distant nesting location [17], thereby possibly evading the negative effects of nocturnal light. As such, the direct effects of artificial light on DEE are yet unknown.

Other potential ecological costs of ALAN might arise through loss of sleep, as shown in humans [18]. Indeed, in birds previous studies suggested that ALAN is associated with nocturnal restlessness (i.e. activity bouts that are clearly distinguishable from sleep behaviour). For instance, female great tits exposed to ALAN for two nights in nest boxes slept less and had shorter sleep bouts compared with birds who roosted under darkness [14]. However, such short-term manipulation appeared to be transient as birds showed regular sleep behaviour when the exposure to light at night was stopped. Moreover, it is unclear whether such nocturnal restlessness really represents sleep disruption. Recently, reduced plasma levels of oxalic acid have been established as a biomarker of sleep disruption in humans and rodents [23]. This opens the possibility to measure sleep disruption in non-model organisms in the field. A recent study in great tits showed that a higher nocturnal activity due to ALAN was associated with a decrease in oxalic acid, thereby suggesting a negative effect of ALAN on sleep [24]. Sleep is a key state for the consolidation of memory, and thereby affects information use [25]. Information processing, retention and use is a part of cognition and important for behavioural decision-making processes, and cognitive abilities allow animals to detect danger, remember food resources and nesting sites based on environmental cues [25]. Studies with great tits show that they are able to memorize locations of cached food by observing other bird species and steal resources, indicating the importance of cognition on fitness [26]. In birds, cognition is affected by sleep, and thus cognitive abilities may be altered by sleep disruption due to ALAN [19]. In a recent study, birds kept under constant daylight showed a disruption in their activity patterns and a deterioration in their cognitive performance [27]. However, the effects of dim rather than constant bright ALAN on cognition remain unknown.

Although ALAN is increasingly associated with negative ecological effects, it is also necessary in human society for economic and safety reasons. Currently, there is an increase in the use of broad-spectrum light emitting diode (LED) lamps due to their cost-effectiveness [28]. As LED lights can easily be adjusted to different light spectra, this may offer the possibility of using a different light spectrum to decrease the ecological negative impact of light pollution. In birds, broad-spectrum white LED light seems to have major impacts, such as altering immune response [24], advancing reproductive activities [29] and increasing corticosterone levels [30] compared with control birds not exposed to ALAN. Experiments with blue tits (Cyanistes caeruleus) have shown that, at lower intensities, green light is less disruptive (compared with white and red light) on activity patterns [31].

While the effects of different light colours are yet to be fully appreciated, it is also unclear whether animals would prefer any type of light spectra when selecting for a roosting location, everything else being equal. Animals generally make behavioural decisions that maximizes their fitness, and therefore should choose for environments that satisfy their requirements the most [32]. On the one hand, animals might benefit from roosting in lit areas because they could forage at night, but on the other hand, they could suffer from increased predation risk and sleep disruption. These trade-offs may be modulated by light intensity and spectra. There has been some research in the poultry sector about the preference of chickens for artificial light of different colours, which showed that these birds seem to prefer light with high colour temperature (spectra) [33]. However, these studies were not carried out in the context of nocturnal lighting. Furthermore, even closely related species can show behavioural differences with regards to ALAN [34], and thus it is difficult to make generalizations. To date, there has not been any research into whether wild bird species prefer to roost in dark versus lit areas, and into whether a specific spectrum of ALAN would be preferred.

The aim of this study is to test the preference of birds for roosting in darkness versus different light spectra, and understand the physiological, behavioural and cognitive consequences of different spectra of ALAN. In a laboratory setting, we exposed male great tits to green light, white light (at similar intensities of 1.5 lux) or darkness. This light intensity is comparable to what wild birds are exposed to in light polluted areas [11,17] and in captive studies has been shown to have moderate effects on activity patterns of great tits [7]. We chose to use green light because there is considerable interest to find light colours that minimize the effects of light pollution on wildlife, and green has been suggested to be a potential option as it is also suitable for outdoor lighting [35]. We used birds both from urban and rural areas to assess whether urban birds respond differently to night light compared with forest birds, as previous research suggested that prolonged exposure to ALAN might alter sensitivity to light [11]. In a first experiment, we tested the preference of birds by giving them the choice of where to roost between pairwise combinations of darkness, green light or white light. In a second experiment, we forced birds to roost under a specific night light and measured daily activity patterns, DEE, plasma concentrations of oxalic acid, sleep behaviour and cognitive abilities. Our hypothesis is that birds prefer to roost under darkness than any light colour. White light, and to a lesser extent green light, will increase night time activity and cause sleep debt, thereby increasing the DEE and negatively affecting cognitive ability. We also expect that night light will have less disruptive effects on the physiology of urban birds as they might have developed tolerance to the presence of ALAN.

(b) General experimental set-up

We employed a within-individual design in both trials to ensure that every bird was exposed to every treatment. Every experiment had three treatment periods, with a randomized treatment order for both birds and rooms. Birds were exposed to full florescent spectrum light during the day at ±1000 lux (Activa 172, Philips), and LED lamps with either green light, white light, or darkness (no light) at night. Although full spectrum light is emitted by both green and white lights, green lamps emit more blue light and less red light (for spectra see supplement to [31]). At night, the intensity of the white and green lights was set to 1. 5 lux, measured at perch level. Before the trials began, we used a lux meter to test the lights at perch level to make sure that the levels of light in each cage were the same. The birds were allowed to recover for seven days in between the two trials. During this recovery period, birds were exposed to dark nights. In both trials, there was a 15-minute overlap between night and daytime lights in the morning and evening.

(d) Experiment 2: forced light exposure

In the second experiment, each bird was housed in a separate cage. The birds were subjected to three treatments at night: green light, white light, or darkness. Participants were assigned at random to one of six blocks spread over two rooms, as well as to a treatment group. Each block contained all three treatments and both origins. The birds underwent 11 nights of treatment and 3 nights of recovery for each treatment, which lasted a total of 14 days (electronic supplementary material, b). During the recovery week between the two experiments, two birds—one from a forest and the other from an urban area—died for unidentified reasons, making n = 33.

Every bird’s daily activity patterns were continuously measured using the same technique outlined in de Jong et al. [7]. We concentrated on the beginning and ending points of activity as well as overall and nocturnal activity. Refer to the electronic supplementary material for a comprehensive explanation of the measurement process.

We evaluated the minute movements of birds while they slept using camera traps. The cameras were programmed to take images upon motion detection and also at one-minute intervals. In order to study sleep behavior, we examined the intervals between the latest activity offset and the earliest activity onset. In every treatment, for each bird, one night was selected. For every two-week treatment period, our goal was to designate the same night—the seventh night following the start of the treatment—as an observatory unit. If that wasn’t feasible due to the bird’s poor visibility that evening, we chose the closest night that was available. The possibility of a bird being clearly visible in the camera frame was unaffected by the light treatment (p 1). Frame by frame, we analyzed the night recordings to evaluate the sleep behavior. “No movement”(0) was recorded if the bird remained in the same sleeping position, head tucked under the shoulder, for two consecutive frames. “Restlessness” (1) was recorded if the bird moved its head or changed its sleeping position during the time frame. This differentiation was predicated on earlier studies evaluating sleep behavior in large tits. We acknowledge that occasionally birds may also sleep with their heads outside of their feathers, particularly during REM-related sleep [36], but regrettably, we lacked the means to differentiate between these occurrences in the absence of matching EEG recordings.

A subsample of 11 birds had their DEE measured using the doubly labelled water (DLW) technique, which has been proven to be reliable in earlier research [16, 37]. Breath samples were collected for this purpose. We measured each of the 11 birds during each treatment period, yielding 33 DEE measurements in total. The order of treatments and origins were randomized. You can find a thorough description of this process in the electronic supplemental material.

We collected blood samples (a total of four per bird) prior to the start of experiment 2 and at the conclusion of each 12-day treatment in order to measure the plasma concentrations of oxalic acid. Sample procedures and laboratory test details are covered in [24] and the electronic supplemental material.

In experiment 2, cognition was assessed using a subsample of 22 birds (11 urban and 11 forest) using learning and memory tasks modified from Titulaer et al.’s dimensional shift paradigm. [38]. A dimensional shift paradigm looks at how people learn by looking at how they behave in response to cues from their surroundings. Six tasks in total were tested: two memory tasks and four learning tasks (see table S3 in the electronic supplementary material for a schematic). You can find a thorough description of these steps in the electronic supplemental material.

All data were analysed with SPSS statistics (v. 24, IBM SPSS), where ? = 0 is the significance level 05. For binary responses (preference for light, thought, and movement), we employed generalized linear mixed-effect models with logistic regression; for all other response variables, we employed linear mixed-effects models (LMMs). Assumptions for using linear models were met. To take into consideration the repeated measurements of the birds and the locations of the cages, individuals nested within blocks were added as random effects to all models. If an interaction term was significant, post hoc analyses using the Bonferroni correction were carried out. Step-by-step removal of non-significant terms—beginning with the interaction term—was used to select the model.

We divided and examined the data for each treatment (WG, DW, DG) in experiment 1. The models included fixed effects such as night lights in the sub-cage (green, white, and dark), origin, and cage position (left/right), with an interaction between the night light and origin. In the second experiment, we analyzed activity patterns using four different models. The four response variables that were employed in these models were total activity, nocturnal activity, activity offset, and activity onset. The three-way interaction of treatment, origin and treatment day (i. e. the treatment days (night lights) in every treatment period) were first added to every model. As fixed effects, origin, treatment, and their interaction were included in the analysis of DEE, change in oxalic acid, and sleep restlessness. We conducted a model for cognition that included the interaction of type × treatment × origin. The type of task (memory/learning) that birds had to perform in the cognitive test was referred to as type. If the interaction was significant we separated data by type.

(b) Experiment 2: forced light exposure

In contrast to darkness, ALAN disturbed the activity patterns of the birds in experiment 2, especially the forest birds that were roosting in white light (electronic supplementary material, figure S2; ). All variables were impacted by the interaction between origin and treatment, with the exception of activity offset. The biggest alterations were seen in the birds’ onset of activity. The most detrimental effect was caused by white light (urban: estimate = -148 min, s). e. = 8. 5; forest: estimate = ?158 min, s. e. = 8. 8), advancing onset almost by three hours. When compared to white light, birds that roost in green light also began their day earlier, but not as much (urban: estimate = ?123 min, s). e. = 8. 5; forest: estimate = ?117 min, s. e. = 8. 7). Additionally, a noteworthy treatment × origin effect was observed: urban birds exhibited a stronger response to green light compared to forest birds, whereas the opposite was observed for white light ( ; electronic supplementary material, figure S2). ALAN had a less pronounced and origin-independent impact on activity offset (electronic supplementary material, figure S2-B). For birds roosting in green light, it was highest (estimate: 31 min, s). e. = 3. 8), followed by white light (estimate = 20 min, s. e. = 3. 9) and then it got dark, with the offset being near to the lights going out (estimate = 4 min, s). e. = 3. 9). When compared to birds in the dark, nocturnal activity was greater in birds exposed to ALAN (DW, DG: p 001, electronic supplementary material, figure S2-C). There was a substantial difference in the nocturnal activity of forest birds between light spectra (p 001), as a result of the birds’ increased activity in white light (estimate = 118 min, s). e. = 8. 4) compared with green light (estimate = 86 min, s. e. = 8. 4). However, the difference was not significant (p = 0. 08) for urban birds. Likewise, overall activity was greater in the presence of ALAN than in the absence of it (DW, DG p 001; electronic supplementary material, figure S2-D). Forest birds exhibited a greater overall activity level (p 001) under white light (estimate = 481 min, s. e. = 18. 0) compared with green light (estimate = 442 min, s. e. = 18. 0), urban birds displayed identical levels of overall activity in relation to light spectra (p = 0). 07).

Additionally, treatment day had a significant impact on bird activity patterns, with the exception of nocturnal activity. While there were some similarities in the early treatment days, activity onset declined more under white light than under green light as time went on under both treatments (electronic supplementary material, figure S2-A). On the other hand, under ALAN, activity offset was delayed more at the beginning of treatment days and eventually drew closer to dark nights (electronic supplementary material, figure S2-B). Bird activity levels did not vary over the course of the relative night (electronic supplementary material, figure S2-C). Over the course of the first few treatment days, the total activity measured over a 24-hour period increased before plateauing (electronic supplementary material, figure S2-D).

ALAN had an impact on the percentage of nighttime bird movements. Specifically, white light (estimate = 0. 11, s. e. = 0. 03) induced more movement compared with darkness (estimate = 0. 04, s. e. = 0. 01, p = 0. 02), however there was no distinction between the green light (estimate = 0). 09, s. e. = 0. 04) and additional therapies (a; electronic appendix, table S2) ). But the change in blood oxalate levels was unaffected by ALAN (b; electronic supplementary material, table S2). Likewise, ALAN did not impair cognition (c; electronic supplementary material, table S2). The total number of trials needed to complete the task was only influenced by the type of task (p 001), as birds demonstrated greater speed (estimate = 3) in memory tasks. 3, s. e. = 0. 46), compared with learning tasks (estimate = 6. 6, s. e. = 0. 58), independent of treatment or origin (p > 0. 1 in both cases). DEE was significantly affected by treatment (p = 0. 002). Post-hoc analyses showed that birds exposed to ALAN had a higher DEE than the dark control group (DW, p = 0), independent of origin or light spectrum. 003; DG, p = 0. 011, WG, p = 0. 87) ( d; electronic supplementary material, table S2).

FAQ

Do birds need lights off to sleep?

Should I leave a light on for my birds?

Do birds need total darkness to sleep?

Can birds sleep with LED lights?