• Keine Ergebnisse gefunden

implications of city life for daily and seasonal timing of songbirds

MATERIAL AND METHODS (a) Animals

Between May and July 2010 we caught adult male European blackbirds (urban = 20, rural = 20) with mist-nets set up at dawn in the city of Munich, in southeast Germany (48º 07´ N, 11º 34´ E; 518 m ASL) and in a rural forest near the village of Raisting (47º 53´ N, 11º 04´ E, 553 m ASL), 40 km southwest of Munich. Birds were placed in cloth cages and transported to our facilities in Radolfzell (47º 44´ N, 8º 58´ E, 404 m asl) where they were initially housed in outdoor aviaries. On November 26th, 2010, birds were

moved indoors into individual cages (width x height x length: 45x70x80 cm) in two separate rooms. Each room contained 10 city and 10 forest birds, all being initially exposed to light/dark (LD) cycles that simulated the natural variation of photoperiod in Radolfzell. One urban bird died on April 1st, 2011. Birds could hear but not see each other. Food (Granvit, Chemi-Vit, Italy) and drinking water were available ad libitum. All experimental procedures were carried out in accordance with the guidelines of the relevant German authorities. We used the same animals and the experimental set-up described below to test different hypotheses about the effects of light at night on daily and seasonal cycles of blackbirds. Some of the results have been already reported elsewhere (Dominoni et al. 2013).

(b) Light treatment

The experiment started on December 18th, 2010. Photoperiod followed the local natural variation of daylength in both treatment groups throughout the experiment. Control birds stayed under an LD cycle. Experimental birds were exposed to light/dim light cycles (LLdim). Daytime light intensity in both groups ranged from 250 to 1250 lux within each cage, and was provided by dimmable fluorescent white bulbs (Biolux 36 W, Osram, Germany) emitting light at wavelengths covering the human visible spectrum. Because lowest light intensity of dimmable fluorescent bulbs was still very high (~ 20 lux), we used a dimmable incandescent light bulb (SLV Elektronic, Germany, wavelength 450-950 nm) to simulate the low light intensities which free-roaming blackbirds experienced at night (Dominoni et al. 2013). We chose this type of light bulb because it is a common

representative of the spectrum that street lights deployed in the city of Munich emit (yellow-red lights). Light intensity at night in the experimental group was set at 0.3 lux, and represents the mean of measurements at all four perches. Control birds were exposed to a light intensity of ~ 0.0001 lux during the night, provided by the same light bulb type as that used for the experimental group. This light intensity of ~ 0.0001 lux was used to allow the birds to orientate in the cage at night. Each group was exposed to a twilight

To determine plasma melatonin concentrations, we took blood samples during the periods of July 25-29, 2011, and January 25-29, 2012. Each individual was sampled at four different times of the day during a period of six days. Samples were taken at least 24 h apart from each other to minimize the stress for the animals. The four samples were taken at the following times: at 12:00 (both in summer and winter), 30 minutes after the end of evening twilight (summer: 21:00, winter: 18:00), 24:00 (both summer and winter), and 45 minutes before the start of morning twilight (summer: 3:00, winter 6:00). All reported times are expressed as wintertime (GMT + 1 h). Each sampling session lasted for a maximum of 15 minutes, for all birds combined. We used headlamps with dim red light to catch and bleed the birds. Birds were bled outside the experimental rooms to minimize disturbance and stress to the other animals. We punctured the brachial vein with a 25-g

needle and collected 200 µ of blood in a hematocrit capillary. Blood was immediately stored on ice. Plasma was separated from red blood cells by centrifugation within 30 minutes after the end of the sampling and then stored at -80° C. The concentration of plasma melatonin was measured by Radioimmunoassay (RIA) (Goymann, Trappschuh,

& Fusani 2008). We ran four assays. All samples from one individual were included in one assay to reduce inter-assay variation. Inter-assay coefficient of variation was 6.9 %, while average intra-assay coefficient of variation was 4.7 % + 3.7 % (mean ± SD).

(d) Activity recordings

Locomotor activity was recorded continuously for the entire duration of the experiment through a passive infrared sensor mounted on each cage (Intellisense, CK Systems, Eindhoven, The Netherlands). Movements were counted and stored as two minute bins on a computer. For the purpose of this study we analysed activity data recorded for seven days prior to each blood sampling session (summer 2011: July 18-24; winter 2012:

January 18-24). For each individual we calculated the number of activity bouts for each hour of the day, averaged for the seven days of recording.

(e) Statistical analyses

Statistical analyses were conducted with the software R 2.15.0 (R Development Core Team. R: A language and environment for Statistical Computing 2011). All tests were two-tailed and we applied a significance level α = 0.05. All explanatory continuous variables were centred and standardized to facilitate interpretation of the estimates. In all

normality of residuals and homogeneity of variance. If non-significant interactions were present, they were sequentially removed. When linear mixed models (LMMs) were used, individuals were always included as random intercepts to account for non-independency of repeated measures. In these type of models we assessed the significance of model parameters using a Monte Carlo Markov Chain (MCMC) approach through the function pvals.fnc in the R package languageR (Baayen 2007). P-values (pMCMC) were calculated based on the posterior distribution of model parameters (50000 iterations).

When significant interactions were detected, we evaluated them by using 95 % confidence intervals (CI) for each estimated group mean. We considered two means to be significantly different from each other if the CI of one group mean did not overlap with another group mean (Cumming 2009).

We analysed the variation in absolute melatonin concentration with LMMs. Log-transformed melatonin concentration was the response variable. We modelled treatment (dark night/light at night), origin (urban/rural), time of day, season and their interaction as fixed effects.

Differences in diel melatonin amplitude were analysed with LMMs. We did not want to use the absolute maximum daily value of plasma concentration for each bird, because the results could have been biased by the fact that different birds may have different mean daily melatonin concentration. We rather calculated the difference in melatonin concentration between the maximum and minimum daily value for each individual. This parameter was then log-transformed and modelled as response variable in the LMM.

Season, origin, treatment, and their interaction were included as fixed effects.

We used LMMs to test for differences in activity levels across treatment groups and origin. Log-transformed mean activity was included as response variable. Time of day, treatment, origin, and their interactions were modelled as fixed effects. We first included season as fixed effect, too, but the model did not converge, so we ran two separated models for the winter and the summer. We included random slopes based on time of day for each individual to correct for different profile of daily activity between birds. Since the MCMC approach described above cannot be deployed for models with random slopes, we assessed the significance of each fixed effect with a Bayesian approach. We used the function sim in the R package arm (Gelman & Hill 2006). This function simulates values from the Bayesian posterior distribution of the model parameters and calculates confidence intervals (CI) for each of them. We considered one effect to be significant if the CI did not overlap with zero (Gelman & Hill 2006).

Finally, we used linear models to test for a relationship between plasma melatonin concentration and dawn activity. We used the change in melatonin concentration value between night and morning rather than the absolute morning value, to reduce bias due to different mean daily melatonin concentrations in different birds. Log-transformed mean activity level during the hour preceding morning twilight was modelled as response variable. Treatment, origin and change in melatonin levels from night (midnight) to morning (summer: 3 am; winter: 6 am) were included as fixed effects.

Fig. 1. Variation in diel melatonin concentration in winter and summer. Melatonin was measured at four different times of day for each bird, in both winter (a) and summer (b). X-axis represents the time of day at which melatonin samples were taken. The mid-night and mid-day sampling times were the same in both seasons, but we modified the evening and morning samplings in order to keep the time distance to the respective twilights equal. Y-axis represents log-transformed melatonin concentration. Black symbols depict birds under dark nights (circles, rural (N = 10); triangles, urban (N = 9). White symbols depict birds under light at night of 0.3 lux (circles, rural (N = 10); triangles, urban (N = 10). Error bars (solid, rural;

dashed, urban) represent SEM. Asterisks depict significant treatment effect (light at night vs. dark night).

RESULTS

In the analysis of the differences in plasma melatonin concentration, we detected a significant interaction between treatment and time of day (LMM, t = -2.98, pMCMC = 0.004, Table 1). In particular, melatonin concentration was significantly lower in the light

summer and winter (Fig. 1). In addition, in the summer birds exposed to light-at-night had lower plasma melatonin also in the mid-night sample (Fig. 1). There was no significant effect from origin (LMM, t = 0.62, pMCMC = 0.50, Table 1). Furthermore, there was a significant difference in melatonin concentration between seasons (LMM, t = -5.17, pMCMC < 0.001, Table 1). Indeed, melatonin levels were on average lower in winter than in summer (winter: mean ± SD = 1.65 ± 0.52; summer: mean ± SD = 1.99 ± 0.49).

Fig. 2. Variation in daily cycles of locomotor activity. We used one week of activity data before the start of the melatonin sampling, in both winter (a) and summer (b). X-axis represents time of day, Y-axis represents log-transformed mean activity level (counts per hour). For sample sizes and meaning of colours, symbols

Melatonin amplitude was 14.48 % significantly lower in winter than in summer (LMM, t

= -4.63, pMCMC < 0.001). Birds exposed to light at night showed a tendency to have a reduced melatonin amplitude compared to birds under dark nights, in both summer (-3.7

%) and winter (-13.13 %), but this effect was only marginally significant (LMM, t = 1.65, pMCMC = 0.056). There was no effect of rural or urban origin on melatonin amplitude (LMM, t = 0.04, pMCMC = 0.96).

The light treatment significantly affected the average locomotor activity in winter (mean

= 0.32, CI = 0.22, 0.44, Table S1). In particular, birds exposed to light at night showed higher levels of activity than individuals exposed to dark nights (Fig. 2a and Fig. S1a). In addition, if activity levels were higher in the urban birds under dark night compared to rural conspecifics in the same treatment group, the reverse was true for the birds exposed to light at night; here rural birds had overall higher activity levels than urban individuals, although this was only marginally significant (mean = 0.02, CI = -0.08, 0.13, Table S1, Fig. S1a). In summer the difference in locomotor activity levels between treatment and origin groups depended on the time of day (time of day*treatment: mean = 0.007, CI = -0.01, -0.0006; time of day*origin: mean = 0.009, CI = 0.002, 0.16, Table S1). Differences were particularly evident during the night and in the early morning (Table S4, Fig. 2b).

Overall, birds exposed to light at night showed significantly higher hourly activity levels than birds exposed to dark nights (mean = 0.36, CI = 0.17, 0.54, Table S1). While activity levels were higher in urban birds under dark night compared to rural conspecifics in the same treatment group, within the whole experimental cohort rural birds showed significantly higher activity levels than urban individuals (mean = 0.36, CI = 0.59, -0.12, Table S1, Fig. S1b).

The change in plasma melatonin concentration from night to morning was significantly related to the levels of locomotor activity in the morning, but only in the light at night group (LM, F7,70 = 5.48, t = 2.00, P = 0.049, Table S2, Fig. 3). There was no seasonal effect in the relationship between change in melatonin concentration and locomotor activity in the morning (LM, melatonin*season interaction, t = 0.61, P = 0.54). In addition, birds under light at night showed higher levels of morning activity than birds under dark nights, but only in winter (LM, treatment*season interaction, t = 2.48, P = 0.015, Table S2, Fig. 3). There was no significant effect of origin on activity levels in the hour before dawn (LM, t = - 0.65, P = 0.515, Table S2, Fig. 3). Some birds showed an increase in melatonin concentration between night and morning, both in winter (control:

N = 8, experimental: N = 0) and in summer (control: N = 3, experimental: N = 4).

Table 1. Variation in plasma melatonin concentration. Models are LMMs with subjects modelled as random factor. Reference for season is summer, for treatment is control group, for origin is rural birds. For each parameter we show the estimated mean, the lower and upper 95 % credible intervals and the p-value calculated after MCMC procedure.

Parameters estimate lower 95% upper 95% pMCMC

intercept 1.37 1.11 1.63 < 0.001

DISCUSSION

Our results show that exposure to light at night of very low intensity, simulating data obtained in the field using light loggers on free-ranging urban European blackbirds (Dominoni et al. 2013), leads to reduced melatonin secretion at night, during both winter and summer. Specifically, plasma melatonin concentrations were reduced in the beginning and at the end of the night, around the time of evening and morning twilight.

The effect was stronger in winter than in summer, which could possibly suggest a higher sensitivity to light at night in winter than in summer, but further experiments are needed to directly test this hypothesis. Furthermore, our results may suggest that the light at night treatment altered the perception of daylength in blackbirds. Indeed, in both seasons the suppression of melatonin in the late night/early morning correlated with the amount of activity shown at this time of day in the birds exposed to light at night: the higher the reduction in melatonin, the higher the amount of morning activity. We focused on morning activity because this is to be an important time of day for males for displaying and attracting a female (Poesel et al. 2006; Kempenaers et al. 2010). This result may suggest a physiological mechanism underlying the advanced onset of morning activity recorded in several urban bird species. Urban male songbirds, indeed, are known to sing earlier than rural conspecifics (Miller 2006; Fuller et al. 2007; Nemeth & Brumm 2009;

Kempenaers et al. 2010). If low light intensities at night were able to reduce melatonin concentrations in the late night, this could affect the perception of the light-on signal in the morning and therefore the time of activity onset (Dominoni et al. 2013).

Fig. 3. Relationship between melatonin and dawn activity. The change in melatonin concentration between the night and the morning sample was related to the average number of activity counts in the hour preceding morning twilight in both winter (a) and summer (b). X-axis represents change in melatonin between midnight and either 6 am (winter) or 3 am (summer), Y-axis represents log-transformed mean activity counts in the hour preceding the onset of morning civil twilight (winter: 6:45, summer: 3:45).

Dashed (light at night) and solid (dark night) black lines represent regression slopes from fitted linear models. Grey shaded areas represent 95 % CI. For sample sizes and meaning of colours and symbols see Fig. 1. Some birds showed an increase in melatonin concentration between night and morning, both in winter (control: N = 8, experimental: N = 0) and in summer (control: N = 3, experimental: N = 4).

The amplitude of the nocturnal melatonin secretion was found to be lower in winter than

(Fusani & Gwinner 2005) and in species living at arctic latitudes (Gwinner et al. 1997;

Hau et al. 2002). In migratory species, melatonin has been found to be functionally related to the amount of nocturnal restlessness, or Zugunruhe, such that during times when birds show Zugunruhe at night circulating melatonin concentrations decrease (Fusani & Gwinner 2004). Conversely, knowledge of seasonal change in diel melatonin in non-migratory songbirds is limited to the house sparrow (Passer domesticus) (Brandstätter et al. 2001). In this species, melatonin amplitude was found to be higher in spring and summer than in winter. Our results, therefore, confirm and expand previous urban population show earlier gonadal growth than rural counterparts (Partecke et al.

2005), and this has been recently linked to exposure to artificial light at night (Dominoni et al. 2013). Could the reduced release of nocturnal melatonin due to exposure to light at night, as observed in the present study, explain the advanced reproductive physiology of urban birds? Although the direct role of melatonin in photoperiodic time measurement in birds has been debated (Dawson et al. 2001; Kumar et al. 2007), there is evidence that it can affect some seasonal processes. For example, exogenous melatonin administration can down-regulate the volume of two brain areas involved in song production, HVc and area X (Bentley et al. 1999). In addition, melatonin up-regulates GnIH in both the brain

(Ubuka et al. 2005) and the gonads (McGuire, Kangas, & Bentley 2011) before breeding, thus keeping gonadal sizes and testosterone levels low. We suggest that the decrease in melatonin as a consequence of artificial light at night in winter may provide animals with the signal that the night is actually shorter, therefore speeding up reproduction (Dominoni et al. 2013). However the data we have right now do not allow testing this hypothesis.

Future research should aim at investigating the mechanisms responsible for the effects of light at night on the reproductive physiology of urban birds, focusing on the effects of the melatonin signal.

Light pollution, in the form of artificial outdoor and indoor lighting, has become a public health issue as it is now evident that exposure to light during the night can promote de-synchronization between internal timing and the natural light/dark cycle, with profound downstream negative effects (Navara & Nelson 2007). For example, exposure to light at night can alter metabolic processes in rodents, increasing body mass (Fonken et al. 2010) and promoting tumour progression (Dauchy et al. 1999), as well as reducing immuno-competence (Bedrosian et al. 2011). Furthermore, light at night has been linked to increased oxidative stress through the production of reactive oxygen species (ROS) (Tan et al. 2007). Melatonin is a well-known antioxidant, important for both the defence against ROS and the production of anti-oxidant enzymes (Rodriguez et al. 2004).

Constant light exposure can reduce melatonin levels and increase ROS in mammals (Baydaş et al. 2001), but not enough is known regarding the link between light at night, melatonin and oxidative stress in birds. Here we show that light at night can reduce melatonin release, and we believe that this finding offers new insights in the mechanisms

improved understanding of the potential physiological costs (seasonal processes, oxidative stress, metabolic disruption, immunological state) of the reduction in melatonin levels caused by artificial light at night.

Acknowledgements

J.P. was funded by Volkswagen Foundation (“Initiative Evolutionary Biology”).

Additional funding to D.D. was provided by the International Max Planck Research School for Organismal Biology. We thank K. Mortega, S. Kingma, C. Miranda, D.

Santos, T. Greives, A. Fudickar, E. Carmona-Wagner and E. Fricke for help during blood sampling, and Monika Trappschuh for hormone analysis.

SUPPLEMENTARY MATERIAL