Temporary caging results in reduced levels of circulating melatonin in migratory robins

Most migratory passerines are nocturnal migrants, in that they have usually diurnal patterns of activity but become active at night during migration. The reasons for this peculiar phenotype are still unknown, but several hypotheses have been proposed, ranging from improved thermoregulation to reduced predation risk (Berthold, 2001). The diurnal–nocturnal transition is a major behavioural and physiological switch, and at present, very little is known about its control. In the past two decades, we have conducted a series of studies illustrating that the hormone melatonin plays an important role in migratory physiology, especially in the expression of nocturnal locomotor activity, which in captive birds is called nocturnal restlessness or Zugunruhe. Melatonin has a well-known role in the physiology of day–night rhythms and has been called ‘the hormone of darkness’ because its levels are higher during the night than during the day, independently of the pattern of activity of the species (Reiter, 1991). Furthermore, this hormone plays an important role in controlling avian circadian activity. The main organ that releases melatonin, the pineal gland, possesses photoreceptors and is a main component of the circadian system (Gwinner et al., 1997).

Previous work showed that Zugunruhe and melatonin levels are inversely associated: higher melatonin levels are typically seen when Zugunruhe is low or absent (Fusani and Gwinner, 2001, 2004; Gwinner et al., 1993). When Zugunruhe in blackcaps (Sylvia atricapilla) is experimentally reduced through a food deprivation protocol, circulating melatonin increases (Fusani and Gwinner, 2004). The association seems to be specific, because blackcaps of resident populations (Cape Verde) may show some levels of nocturnal locomotor activity but these are not correlated with a reduction in melatonin levels (Fusani and Gwinner, 2001).

We hypothesized that melatonin controls the switch between non-migratory and migratory status, namely, from purely diurnal to diurnal/nocturnal. We tested this hypothesis in garden warblers (Sylvia borin) during spring migration by using transcutaneous application of exogenous melatonin (Goymann et al., 2008). However, we found no effects of the melatonin treatment on Zugunruhe (Fusani et al., 2011). This outcome might depend on the time of the year. In fact, previous work in our lab showed seasonal differences in the responsiveness of melatonin to experimental manipulation (Fusani and Gwinner, 2004), and a number of studies had shown that migratory behaviour is less sensitive to disruptive factors in spring than in autumn (Stanley et al., 2012; Wingfield et al., 1990). When we repeated our experiment during autumn migration, we obtained a strong reduction of Zugunruhe in blackcaps and garden warblers following the treatment (Fusani et al., 2013).

In sum, our previous studies demonstrated an involvement of melatonin in the physiological transitions from migratory to non-migratory status. A major methodological limitation of all previous studies is that melatonin levels were measured in birds that underwent temporary captivity. In fact, wild migratory birds are generally caught during the day, when melatonin levels are low or undetectable. Thus, to date we do not know how the levels measured in captive birds, even if only temporarily confined, compare with those found in birds that are actively migrating at night. Here, we addressed this question by taking advantage of a ringing station in the Italian Alps, Bocca di Caset, where migratory birds are trapped day and night. We aimed to determine how melatonin levels measured in birds temporarily housed in cages compared with those found in birds that are actively migrating. To achieve this objective, we compared birds that had been caught at the stopover site during the day and then housed in cages overnight with birds that had been caught at night during active migration and bled immediately after capture.

The study was conducted in September 2014 at the Bocca di Caset Ringing Station in the Italian Alps (45°51′ N, 10°41′ E; altitude: 1.618 m). During post-breeding migration, Bocca di Caset is an important pass for passerine birds crossing the Alps and flying southwards towards their wintering grounds (Pedrini et al., 2008). The ringing station has 7 m high mist-nets that allow the capture of nocturnally migrating birds that at Bocca di Caset fly close to the ground but higher than the reach of standard mist nets (3 m).

We trapped European robins (Erithacus rubecula Linnaeus 1758), the most common alpine nocturnal migrant at Caset, during the day and night. Upon capture, robins were ringed following standardized EURING methods, including scoring subcutaneous fat (0–8 scale) and recording body mass to the nearest 0.1 g. We used 113 robins to compare nocturnal levels of melatonin between migrating birds caught at night (hereafter ‘net’) and individuals caught during daytime and kept in captivity overnight (‘cage’). To measure melatonin levels, we blood sampled (150–200 µl) birds between 23:00 h and 03:00 h. In case exposure to white light at night affects melatonin production (de Jong et al., 2016), we used a red safelight during sampling. The net group consisted of 61 birds caught during the night between 23:00 h and 03:00 h and immediately bled. The cage group consisted of 52 birds caught during the daytime until 2 h before the civil sunset, and placed in individual custom-built fabric cages (50×30×30 cm) with water ad libitum under a natural photoperiod. Cage and net birds were bled at night in parallel. We assessed fat score and measured body mass at bleeding of all birds. We could house a maximum of six individuals per night and we centred blood sampling time at 00:00 h. Blood sampling procedures were completed by 01:00 h for CAGE birds and by 03:00 h for net birds. In each cage, an infrared sensor connected to a PC recorded locomotor activity as number of movements during 2-min periods. We calculated the intensity of diurnal and nocturnal activity as the average number of movements per 2-min interval from 12:00 h to sunset (diurnal activity) and from sunset to sunrise (nocturnal activity) based on civil twilight times (GMT +1). We used the extent of nocturnal migratory restlessness or Zugunruhe, as an indicator of migratory disposition (Fusani et al., 2009). Our cages are optimized for short-term housing of small migratory birds and have been used in a number of studies (Fusani et al., 2011, 2013). Permission to carry out the research was granted by the Provincia Autonoma di Trento (Deliberazione no. 635 Comitato Faunistico Provinciale, 27 June 2013).

We determined plasma concentration of melatonin using a commercial radioimmunoassay kit (Melatonin Serum Direct RIA, Labor Diagnostika Nord, Nordhorn, Germany, BA R-3300RIA). Beforehand, a liquid extraction was conducted overnight with 3 ml dichloromethane. The organic phase was freeze-decanted and then dried in a water bath at 43°C under a nitrogen stream. We resuspended the extracts in phosphate-buffered saline and allowed them to stabilize at 4°C overnight. Samples were then processed following the manufacturer's instructions. All samples were assayed in duplicate. The detection limit was 3 pg ml⁻¹; the intra-assay coefficient of variation (CV) was <12% and the inter-assay CV was <13%.

We performed all statistical analyses with R v.3.5.0 (https://www.r-project.org/) using a significance level of α<0.05. We calculated a scaled mass index at time of bleeding using equation 2 from Peig and Green (2009). For cage birds, we calculated an index of body mass loss between capture and blood collection as (body mass at bleeding–body mass at capture)/body mass at capture. We subdivided the variable time of bleeding into hourly intervals for all birds.

We used a general linear model (GLM) to compare melatonin levels between captive and actively migrating birds in the period 23:00–01:00 h controlling for the influence of scaled mass index at bleeding. We chose to include only samples collected during the period 23:00–01:00 h to have comparable times of bleeding between net and cage birds (time of bleeding: cage=23:00 to 01:00 h; net=23:00 to 03:00 h). The intensity of Zugunruhe (number of movements/2 min) in cage birds showed a large variability (mean=1.6, first quartile=0.478, third quartile=2.351; range 0 to 9.549). This variability reflects different dispositions to migrate among cage birds. To take this into account, we set a threshold of 1 for the intensity of Zugunruhe to subdivide the cage group in two groups: birds that would likely migrate in the wild (‘cage-active’, N=24) and birds that would likely prolong stopover (‘cage-inactive’, N=28). In the GLM, we compared melatonin levels between the three groups: net, cage-active and cage-inactive.

In addition, we studied the variation of melatonin in the cage and net datasets independently. We performed GLMs to relate the levels of melatonin to time of bleeding, body condition indices, and, for cage birds, locomotor activity and body mass loss. We excluded all cage birds that were housed after 12:00 h (N=12) as the measurement of diurnal locomotor activity in these birds was limited to a small number of hours. Furthermore, a more homogeneous time of caging across individuals allowed us to include body mass loss as an explanatory variable. To check for departures from regression model assumptions and fit, we used diagnostic plots (i.e. residuals versus fitted values, scale–location and normal QQ plots). We conducted model evaluation and selected the models that best fit our data following the Akaike information criterion (AIC) (Akaike, 1974) and conducting backwards selection of insignificant terms. For cage birds, we additionally studied the correlations between locomotor activity (diurnal and nocturnal=Zugunruhe) and nocturnal melatonin concentrations. Zugunruhe was not normally distributed and could not be normalized with data transformation. Diurnal activity was not normally distributed and was normalized by logarithmic transformation. Therefore, we calculated non-parametric (Spearman) correlation coefficients for the relationship between Zugunruhe and melatonin levels, and parametric (Pearson) correlation coefficients for the relationship between diurnal activity and melatonin levels.

The variation in melatonin was best explained by a model including only the group (net, cage-active and cage-inactive). Scaled mass index at bleeding was non-significant. An analysis of variance (ANOVA) on melatonin values yielded significant variation among groups (F=3.244, P=0.045, AIC=895.05, Fig. 1). Specifically, net birds showed higher melatonin compared with cage-active birds (Tukey test: P=0.030, Fig. 1). The difference between net and cage-inactive birds showed a clear tendency to significance (Tukey test: P=0.077, Fig. 1). The lack of a statistical significant difference is the result of one high melatonin value in the group cage-inactive. Melatonin levels did not differ between the two cage subgroups (Tukey test: P=0.894, Fig. 1).

In cage birds, the variation in melatonin levels was best explained by a model including only the variation in body mass at capture. Time of bleeding, Zugunruhe and diurnal activity were excluded from the best-fitting model. Although body mass at capture did not show a significant effect on melatonin levels (F=0.572, P=0.454, AIC=444.67), there was a clear tendency for melatonin levels to be lower when body mass at capture was larger (Fig. 2). In fact, the negative correlation becomes significant with the exclusion of one data point ([melatonin]: 436.2 pg ml⁻¹; body mass=18.8 g) (F=7.205, P=0.011, AIC=419.51). We did not find any significant correlations of Zugunruhe and diurnal activity with nocturnal melatonin concentrations (Zugunruhe: r_s=0.065, N=50, P=0.652; diurnal activity: r_p=0.077, N=50, P=0.597). In net birds, none of the predictor variables explained variation in melatonin concentrations.

In summary, we found higher nocturnal levels of melatonin in robins caught at the nets and sampled immediately compared with robins that were caught earlier during daytime and housed overnight in cages. All captive birds showed similarly low levels of melatonin regardless of the extent of nocturnal locomotor activity. Because nocturnal melatonin levels strongly depended on the captivity status (housed in cages versus free-living) but not on the extent of migratory disposition, we conclude that in robins, temporary captivity is associated with a decrease in the circulating levels of melatonin.

The results of the present study suggest that actual levels of melatonin in migrants could be generally higher than those reported by previous studies with captive birds. In most studies, the hormone was measured in samples taken from birds that were kept in cages. Thus, the inverse association between melatonin and migratory status found in several studies (Fusani and Gwinner, 2001, 2004; Gwinner et al., 1993) might be stronger in free-living birds because of the lack of dampening effects of captivity. The lack of a correlation between melatonin levels and nocturnal activity in our cage sample is not surprising, as previous studies showed that such correlation is not observed among animals in a similar migratory state despite variations in the extent of nocturnal restlessness (Fusani et al., 2011).

Effects of captivity have been described for other hormones such as testosterone (Dufty and Wingfield, 1986) and corticosterone (reviewed by Calisi and Bentley, 2009). The physiological changes consequent to transfer to captivity are quite complex, but in most cases they involve the activation of the hypothalamic–pituitary–adrenal (HPA) axis. Corticosteroids released by the adrenals are well known to affect the activity of other endocrine systems such as the hypothalamic–pituitary–gonadal (HPG) axis and thus are probably responsible for the reduced testosterone profile in captive male birds (Calisi and Bentley, 2009).

In a detailed study designed to address this question, Dickens et al. (2009) found that wild chukars (Alectoris chukar) entered a state of chronic stress during the first 5 days of captivity, after which they started recovering to reach ‘normal’ condition by the ninth day. Another study examined the effect of stress on corticosterone and melatonin in ring doves (Streptopelia risoria), and found that melatonin levels at night were reduced in birds subject to stress which had elevated circulating levels of corticosterone (Barriga et al., 2002). We did not measure corticosterone levels in our study; however, it appears likely that captivity affected melatonin levels in caged birds because of the activation of the HPA axis.

Nocturnal melatonin levels can be reduced by exposing birds to artificial light (Dominoni et al., 2013). However, the room in which our birds were housed had no artificial light and there was no significant source of light nearby. Nevertheless, the amount of light to which the birds kept in cages and those which were caught at night were exposed might have been different and an effect of illumination on melatonin levels cannot be excluded.

In conclusion, differences in the nocturnal circulating levels of melatonin between robins that had been caught during daytime and then kept in cages and robins that had been caught at night may be explained by captivity-induced stress. As birds appear to adapt to captivity after a few days, or at least to recover from the initial period of chronic stress, it is likely that such effects will be more evident in birds that have been just caged compared with those kept in cages for longer periods.

We thank Jacopo Bettin, Arianna Passarotto and Valentina Falchi, the ringers Alessandro Franzoi, Franco Rizzolli and Federica Bertola, Daniela Serafin and Elena Zamprogno of Caset Ringing Station for help during ringing and fieldwork.