Artificial light at night causes an unexpected increase in oxalate in developing male songbirds

We quantified the effect of artificial light at night on oxalate, a potential marker of sleep debt, in free-living developing songbirds. Rather than a decrease, as was found in sleep-deprived mammals, oxalate increased in male nestlings. This research indicates a sex-specific physiological impact of ALAN that warrants further investigation.


Introduction
Artificial light at night (ALAN), also known as light pollution, has greatly altered the night-time environment. It is a widespread, ever increasing and important anthropogenic environmental pressure on wildlife (Falchi et al., 2016;Hölker et al., 2010). ALAN affects a wide variety of behavioural traits, such as reproduction, foraging and migration and has clear deteriorating physiological effects such as a supressed immune response to challenges and increased corticosterone levels (see e.g. Bedrosian et al., 2011;Russ et al., 2015Russ et al., , 2017 reviewed in Russart and Nelson, 2017). Disruption of the immature circadian system during development by ALAN may, due to its effects on the developing brain, influence adult behaviour, physiology, health, and disease and could have long-lasting and irreversible consequences (Fonken and Nelson, 2016). In wild developing great tits (Parus major) ALAN has been shown to affect physiology (Raap et al., 2016a(Raap et al., , 2016b and to cause begging at night (Raap et al., 2016c). Sleep disruption due to ALAN, similar as has been demonstrated in adult great tits (Raap et al., 2015(Raap et al., , 2016c, seems likely but is difficult to quantify in nestlings. 1 nestling period ALAN reduces female great tit sleep behaviour by ≈50% with some females not sleeping at all during the entire night. Furthermore, when nestlings were in natural darkness they never begged for food, while they showed substantial begging behaviour throughout the night when being exposed to ALAN (Raap et al., 2016c). This together with the severe sleep disruption in females gives a strong indication of sleep disruption and loss in great tit nestlings, but this is difficult to quantify as they do not exhibit sleep behaviour like adults. In adults, but not in nestlings, sleep behaviour can be defined as when an individual is in the classical sleep position, with the beak pointing backwards and tucked under the scapulars (Amlaner and Ball, 1983). Sleep is an important behaviour widespread across the animal kingdom (Cirelli and Tononi, 2008;Siegel, 2008) but difficult to study in the field (Rattenborg et al., 2017), especially in small animals.
Recently, a laboratory study showed that sleep loss reduced plasma levels of oxalic acid, also known as oxalate, in humans and in rats (Weljie et al., 2015). Ouyang et al. (2017) claimed that light pollution decreased oxalate levels in adult great tits. However, because of several severe methodological and statistical issues with this study, such as a high uncertainty that their light treatment was effective in terms of exposure and because they did not study sleep or sleep behaviour (Raap et al., 2017b), it is still unknown how ALAN affects oxalate in birds. If oxalate is a cross-species biomarker of sleep debt then this could be used to quantify whether ALAN causes nestlings to suffer from sleep disruption. Moreover, oxalate would be a particularly useful tool when working with free-living small animals as they are often too small to be fitted with modern data loggers and behaviour is necessarily used as a proxy for sleep and sleep debt (Aulsebrook et al., 2016). Data loggers are used to record brain activity which allows for a more precise and detailed measurement of sleep that cannot be obtained from behaviour. However, great tit nestlings are too small to be fitted with data loggers and sleep cannot be quantified behaviourally as they do not always exhibit the classical sleep position like adults do. Nonetheless, they are an important model system in evolutionary and environmental research, and are increasingly being used to study the effects of ALAN on behaviour and physiology (e.g. Da Silva et al., 2014;Kempenaers et al., 2010;Ouyang et al., 2017;Raap et al., 2017c;Sprau et al., 2017;Sun et al., 2017) and the impact of ALAN on nestling sleep would thus be of particular interest.
Oxalate is decreased in both sleep restricted rats and humans (Weljie et al., 2015) and as ALAN affects nestling behaviour and physiology it is important to test whether it also affects oxalate in developing animals. Sleep likely plays a crucial role in metabolic processes. Weljie et al. (2015) investigated cross-species consequences via comprehensive metabolite profiling and showed that oxalate may provide a potential link between sleep loss and metabolic dysfunction and serve as a biomarker for sleep deprivation. Blood oxalate can decrease through several mechanisms as outlined in Weljie et al. (2015): reduced synthesis, increased gut microbiota processing and/or increased urinary clearance. Increases could occur through degradation of ascorbate, production by the liver and red blood cells or by increased dietary oxalate (Marengo and Romani, 2008). To examine whether ALAN affects oxalate in developing birds, we experimentally exposed wild great tit nestlings to two nights of artificial light (3 lux) and compared these to nestlings which were not exposed to ALAN.

Methods
Plasma samples from a previous experiment were used in which free-living great tit nestlings were exposed to ALAN inside their nest box (see details in Raap et al., 2016aRaap et al., , 2016b. In brief, our experiment was performed during the 2015 breeding season (between 8 and 25 May) in a resident nest box population of great tits in the surroundings of Wilrijk (Antwerp), Belgium (51°9′44″N, 4°24′15″E). This population was established in 1997 and has been monitored ever since (see e.g. Eens et al., 1999;Van Duyse et al., 2000, 2005Janssens et al., 2001;Rivera-Gutierrez et al., 2010, 2012Vermeulen et al., 2016;Thys et al., 2017).
The previous experiment included two groups of each 16 nests; a control and a light treated group (see details in Raap et al., 2016aRaap et al., , 2016b. Nests with a similar hatching date and brood size (about seven) were used and otherwise randomly allocated to a treatment. In the light treated group, we obtained a blood sample (≤150 μl) from the brachial vein of all nestlings and subsequently weighed them (digital balance; Kern TCB 200-1) at baseline (Day 13 after hatching) and again after a two night exposure to ALAN (Fig. 1). Nestlings from the control 'dark' group were not exposed to ALAN, but were otherwise treated the same. Sampling was standardized for time and always occurred during the morning. The light treated group was exposed to 3 lux broad-spectrum white light as measured on the nest box bottom (ISO-Tech ILM-1335 light meter; Corby, UK). This intensity has been shown to disrupt sleep behaviour (Raap et al., 2017c) and affect nestling physiology (Raap et al., 2016a(Raap et al., , 2016b. We examined changes caused by ALAN on a within-individual level by taking two samples from each individual which we used to generate a single 'change in oxalate' measure (one metric per individual). This approach is especially important because there is high heterogeneity in physiological markers on the between-individual level within nests (see e.g. Vermeulen et al., 2015;Casasole et al., 2017). Moreover, effects of ALAN may differ greatly among individuals (see e.g. Raap et al., 2016c). Our field-based experimental approach with free-living animals (contrary to laboratory studies), may offer useful insights about possible physiological effects of ALAN on developing animals.
Oxalate was quantified in 4 μl of plasma using the manufacturer's instructions provided with the commercially available colorimetric assay (BioVision Inc., USA), similar to Ouyang et al. stored at −80°C but had also been defrosted twice for previous analyses (see details in Raap et al., 2016aRaap et al., , 2016b. In brief, we diluted samples with assay buffer to a total of 50 μl. In the same assay samples to generate a standard curve were included. Samples were subsequently incubated at 37°C for 1 h and absorbance was measured at 450 nm. We selected those samples of which sufficient plasma remained after previous analyses. Due to sample limitations we could not run duplicates although there was always some plasma left-over. All samples were within the assays limit. Samples of Days 13 and 15 from the same individual were kept on one plate, and two plates/assays were used in total. Overall we obtained repeated measurements (Days 13 and 15) of oxalate for 23 females and 21 males in the control group (14 nests) and 22 females and 22 males in the light group (13 nests). So, in total 88 great tit nestlings of known sex were sampled twice. Nestling sex was determined by molecular sexing (Griffiths et al., 1998).
All statistical analyses were conducted in R 3.3.2 (R Core Team, 2016). We compared changes in oxalate from Day 13 to Day 15, between the control and light group. Changes in oxalate (expressed as a percentage from baseline, similar to Weljie et al., 2015) were analysed by constructing a linear mixed model as data were normally distributed (lme4 package Bates et al., 2015). The change in oxalate is unrelated to the level of oxalate on Day 13. Nest identity (NestID) was included as random factor to avoid pseudoreplication. The full model contained weight on Day 13 (covariate), sex (factor), treatment (factor) and the interaction between sex and treatment as explanatory variables. We included the interaction with sex as there may be sex-specific differences in physiology (Giordano et al., 2015;Speakman et al., 2015) and environmental sensitivity (reviewed in Jones et al., 2009). P-values obtained by a stepwise backward model reduction (Zuur et al., 2009) are given in the results and Tukey HSD tests were used for post-hoc analyses (lmerTest package Kuznetsova et al., 2016).

Results
There was a sex-dependent effect of ALAN on oxalate (sex: treatment interaction: F = 4.994, P = 0.028; Tables 1 and 2). Light exposed males, but not females, showed increased levels of oxalate (≈15%; see Fig. 2 and Table 3). Male and female nestlings in a natural dark situation showed no change in oxalate from Day 13 to Day 15. Weight at Day 13 did not affect changes in oxalate (F = 0.050, P = 0.823).

Discussion
We found a sex-dependent effect of ALAN on oxalate, a potential biomarker for sleep debt (Weljie et al., 2015). The effect of ALAN on oxalate only manifested itself in male nestlings and in the opposite direction as we had expected, an increase instead of a decrease. In the following, we discuss these sex-dependent effects of ALAN on oxalate and the potential of oxalate as a proxy for sleep debt in developing birds.

Sex-dependent effect of ALAN on oxalate
Males and females differ in their behaviour, physiology and their response to the environment which may explain that   Bates et al., 2015). In bold the significant P-value is indicated (P < 0.05), N = 88 individuals. P-values obtained after a stepwise backward regression can be found in effects on oxalate were only found in males. Oxalate could increase through degradation of ascorbate, production by the liver and red blood cells or by increased dietary oxalate (Marengo and Romani, 2008) and there are several indications why oxalate could be affected in a sex-dependent manner. First, sleep behaviour (including its duration) is strongly sexdependent in great (Stuber et al., 2015a(Stuber et al., , 2017 and blue tits (Steinmeyer et al., 2010), and adult male and female great tits use different sleep strategies depending on their metabolic requirements (Stuber et al., 2015b). Contrary to females, males decrease their sleep duration with an increased basal metabolic rate. Second, male and female great tit nestlings may also differ in their physiology besides possible behavioural differences. For example, male nestlings have higher nitric oxide levels (Raap et al., 2017a), and nutritional conditions during development had a sex-dependent effect on the oxidative status of great tit nestlings, with females being more sensitive to nutritional stress indicating sex-specific allocation priorities (Giordano et al., 2015). Third, the response towards ALAN is potentially sex-dependent as adult females showed a slightly stronger reduction than males in sleep amount (% of sleep; amount divided by the total time spent inside the nest box) when subjected to ALAN in February (Raap et al., 2015). However, whether this is true during the breeding season and for developing nestlings remains to be tested. Generally speaking males are often more susceptible to environmental conditions than females although effects are small (reviewed by Jones et al., 2009). Overall, this implies that male and female nestlings may differ in their sleep behaviour, physiology and response to the environment, which could contribute to the observed differences in oxalate. Finally, we need to consider that as this is the first study in developing animals in the wild, our results may also be false positive. The effect is not particularly strong and in the direct opposite of predictions. However, we used a repeated measures design to look at changes within the same individuals, which increases statistical power (Seltman, 2013) and gives confidence in the obtained results. Nonetheless, further experimental and comparative work is needed to determine and validate our results.
Increased begging might have led to increased food provisioning by the parents and thus higher oxalate levels in males. When nestlings are exposed to ALAN they start begging for food which they never do during a dark night. Although this needs to be examined it could trigger females (and to an extent indirectly also males) to feed their nestlings more, especially during the morning. Our samples were obtained during the morning between 08:00 and 12:00. During this period nestlings could have received a substantial amount of food and because male nestlings are larger than females (average of Day 13-15, respectively, 15.9 ± 0.230 and 15.2 ± 0.231 g; Raap et al., 2016a) they might have received more food (e.g. Anderson et al., 1993) and thus more dietary oxalate. However, whether male nestlings did effectively obtain more food, that also contains oxalate, needs to be determined. Doing so will be challenging because, besides other practical issues, nestling sex can only be determined genetically. Another : ALAN increased oxalate in male nestlings. Raw data of the change in oxalate after two nights is shown for females (top panels) and males (lower panels), in a natural dark situation (left panels) and for light exposed animals (right panels). Lines indicate unique individuals; sample sizes: ♀ dark = 23; ♂ dark = 21; ♀ light = 22; ♂ light = 22 nestlings. In males there was a significant effect of artificial light at night on oxalate levels (t = 2.01, P = 0.048; Table 3

4
potentially contributing mechanism could be through the disruption of the gut microbiota by ALAN. This could have led to reduced processing and thus accumulation of oxalate in light exposed nestlings but this requires further investigation and does not explain the sex-dependent effect. To conclude, the increase of oxalate by ALAN may therefore be unrelated to any changes in sleep.

Oxalate as a proxy for sleep debt in developing birds
Because of the inability to directly measure sleep amount or disruption in nestlings we cannot be entirely certain about the relationship between sleep loss and oxalate levels in developing nestlings. While nestlings showed more begging at night when exposed to ALAN (Raap et al., 2016c), we cannot be sure that they spent more time awake. Awakenings at night are a normal part of the adult tits' sleep patterns, however, ALAN clearly disrupts adult sleep behaviour (Raap et al., 2015(Raap et al., , 2016c. It therefore seems very likely that great tit nestlings disturbed by ALAN (Raap et al., 2016c) also effectively slept less, but whether an increase in oxalate is related to sleep loss remains to be tested. Furthermore, whether oxalate can be used as a reliable biomarker of sleep loss in (developing) birds and whether it is direct or indirectly affected by ALAN will require further research.
Effects of sleep loss on oxalate in developing animals may differ from those in adults, where in rats and humans a decrease was found. Sleep differs between developing and adult animals in both mammals and birds (reviewed in Rattenborg and Martinez-Gonzalez, 2015), with for example the amount of rapid eye movement (REM) sleep being higher in developing animals than in adults. Following sleep deprivation a rebound mainly occurs in REM sleep. Sleep loss may thus be different for developing animals compared to adults. This difference may subsequently be reflected in differences in how oxalate is affected. Ouyang et al. (2017) examined effects of light pollution on oxalate levels in adult great tits. They claimed that great tits exposed to white light had higher nightly activity and linked this to a decrease in oxalate from March to May. However, there are several serious issues with these results (Raap et al., 2017b), warranting further study. For example, there is high uncertainty that their light treatment was effective in terms of exposure and the claims about a relationship between sleep loss (which was not measured) and physiological effects seems to be premature. Furthermore, Weljie et al. (2015) showed that in rats (but not in humans) the reduction of oxalate was related to increased oxidative stress due to sleep deprivation. We would thus expect that sleep deprivation in developing great tits might be associated with oxidative status as well as reduction in oxalate. However, we previously showed that our treatment of two nights ALAN did not affect the oxidative status of developing great tits (Raap et al., 2016a).
Our experimental treatment may not have restricted sleep sufficiently to be detected by a decrease in oxalate. The laboratory study by Weljie et al. (2015) revealed that oxalic acid is depleted after sleep restriction in male rats over a period of 5 days, while allowing them to sleep for only 4 h per day. In our study nestlings were exposed to ALAN continuously during two nights and the amount of sleep loss each night might be more severe than the sleep reduction in the study by Weljie et al. (2015) although their treatment was three days longer. Therefore, the reduction in sleep by ALAN over two nights could be insufficient to be detected using oxalate, although it affected body mass gain, haptoglobin and nitric oxide (Raap et al., 2016a(Raap et al., , 2016b. Future studies may, however, build upon our results and use longer periods of sleep restriction.

Conclusions
We have shown here for the first time that a brief exposure of only two nights to ALAN affects oxalate levels in developing wild animals. This provides further evidence of the physiological effects of ALAN during early-life, such as increased haptoglobin and decreased nitric oxide (Raap et al., 2016b), in addition to the behavioural effects, such as increased begging behaviour and likely disrupted sleep (Raap et al., 2016c). The sex-dependent effect on oxalate might indicate different physiological coping mechanisms by developing great tits, which requires further investigation. Experiments using humans and rats showed that oxalate is a potential biomarker of sleep loss (Weljie et al., 2015), and while its use in free-living birds and developing animals is promising it requires further investigation. The increasing illumination of the night is a serious threat as it disrupts circadian rhythms, physiology and behaviour. Early-life exposure could have long-lasting effects throughout adulthood, reducing survival and reproduction. With the progressive, Tukey HSD tests were used for post-hoc analyses (lmerTest package Kuznetsova et al., 2016). Estimates give the difference between oxalate from Day 13 to Day 15 as a % from baseline (Day 13) and P values indicate whether this differs from 0. Sample sizes: ♀ dark = 23; ♂ dark = 21; ♀ light = 22; ♂ light = 22 nestlings. In bold the significant P-value (P < 0.05) is indicated. Standard error (SE) and confidence interval (CI) of the estimates are given. See Fig. 2

5
worldwide increase of light pollution, the physiological effects of this stressor, especially during development, should receive greater attention. Urbanization is not only associated with light pollution but with many other anthropogenic stressors such as noise pollution, and new methods and approaches are necessary to understand the consequences of increasing human pressure on free-living animals.