Casimir V. Bolshakov & Victor N. Bulyuk

Transcription

1 Avian Ecol. Behav. 2, 1999: Time of nocturnal flight initiation (take-off activity) in the European Robin Erithacus rubecula during spring migration: direct observations between sunset and sunrise Casimir V. Bolshakov & Victor N. Bulyuk Abstract: Bolshakov, C.V. & V.N. Bulyuk (1999): Time of nocturnal flight initiation (take-off activity) in the European Robin Erithacus rubecula during spring migration: direct observations between sunset and sunrise. Avian Ecol. Behav. 2: Temporal distribution of migratory nocturnal departures of the European Robin Erithacus rubecula, from a stopover site on the Courish Spit of the Baltic Sea, was studied systematically by direct observations (visual and searchlight methods), from sunset to sunrise over 7 years across the whole period of spring migration. Time of first take-off in spring is not fixed, day-to-day variation reaching 5.5 hours. First birds do not depart until min after sunset. In 50% of cases spring nocturnal migration does not commence before the third hour after the sunset. The onset of nocturnal migration varies across the whole season, variation being 4.0 to 5.2 hours. The mean time of start of migration significantly shifts over the season by min towards sunset. During long nights (10-12 hours), in the first half of the migratory season, in 33% of cases the first birds do not depart until the fourth-sixth hour after sunset. During short nights (8-10 hours), in the second half of the season, in 95% of nights migration starts during first-third hour after sunset. Timing of nocturnal departures of Robins in spring is not synchronised and it is usually not possible to define the main departure period. Overall length of departure period reaches 9.5 hours, the latest take-off being recorded in the tenth hour after sunset. Only 26.1% of Robins initiate flight before the end of nautical twilight (ENT). Approx. 42.5% of birds depart during the deep night after the end of astronomical twilight (EAT). The mean departure time in spring is 191 (SD=121) min after sunset. Only in some nights at the end of the migratory period do nearly all Robins depart during a restricted period after sunset and during twilight. All temporal characteristics of take-off activity are subject to significant changes over the season when night duration falls from 12 to 8 hours. Overall departure period changes from 9.5 to 6 hours, the mean time of nocturnal departures shifts towards sunset by more than one hour, the proportion of birds starting flight after EAT decreases, and the proportion departing before ENT increases. Inter-annual variation of temporal distributions of nocturnal departures, are probably related to the variation in the timing of spring migration. Within the same part of season however, inter-annual variation of the mean departure time may be over 1 hour. Variation of the mean and median departure time between successive nights of one year may also reach hours. In spring Robins prefer to start migration under clear skies or slight cloud cover. As few as 9% of all birds departed without celestial cues visible under total overcast. Under cloudy skies, compared to clear weather, overall departure period is longer together with the variation of the timing of departures, no seasonal trend in the mean time of start of migration and in the mean departure time is recorded and nocturnal migration in the first half of the season starts significantly earlier. The mean departure time under clear skies and high cloud scores is however, not significantly different either over the whole spring, or over fractions of the migratory period. Large variations in the timing of nocturnal departures is supposed to be caused by birds taking off, not in response to any environmental stimulus, but in respect to their individual time programmes. These programmes are a complicated manner related to seasonal LD changes and possess their own seasonal variation. A crucial factor controlling the time of nocturnal departure in spring may be the distance to the goal of migration. Considerable variation in the departure time basically referring to individuals approaching the goal (less then one night of flight).

2 52 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin Avian Ecology and Behaviour Key words: European Robin, Erithacus rubecula, spring migration, searchlight and visual observations, time of nocturnal departure, temporal distribution of take-off activity, seasonal and inter-annual variation, impact of cloud cover. Address: C.V.B., V.N.B.: Biological Station Rybachy, Zoological Institute, Russ. Acad. Sci., Kaliningrad Reg , Russia Introduction In the majority of passerines, seasonal migrations over ecologically hospitable areas include a succession of migratory flights and stopovers. The flight is a succession of at least three forms of behaviour: pre-start activity on the ground and take-off, transit flight, reducing altitude and landing. For small nocturnal passerine migrants recent concepts suggest that the bulk of migrants take off soon after sunset, fly for several hours and land. Soon after midnight all passerine nocturnal migrants cease flight (Kerlinger & Moore 1989, Alerstam 1990, Berthold 1993). Evidence is available however, that the timing of departure in passerine nocturnal migrants may vary considerably in relation to species, latitude, and time of season (Åkesson et al. 1996). Furthermore, some data suggests a large intra-specific variation in this form of migratory behaviour (Cochran et al. 1967, Bolshakov 1992, Bolshakov & Rezvyi 1998, Åkesson et al. 1996, Moore & Aborn 1996). The timing of departure in small passerine nocturnal migrants may be determined not only by availability of celestial cues necessary for orientation during sunset and at twilight, but by variation in the atmospheric structure and energy condition of migrants (Moore 1987, Kerlinger & Moore 1989, Moore & Aborn 1996). It may reflect the interplay of endogenous time programmes with LD period changes over the season along migratory route, and also the position of birds in respect to the goal of their migration and migratory speed under different winds (Bolshakov & Rezvyi 1998). Unfortunately, the obvious lack of systematic evidence, especially on the specific level, does not allow an answer to the question of when small passerine nocturnal migrants commence their flight and which external and endogenous factors control take-off behaviour. As recently stated, the variation in take-off time of nocturnal migrants depending on species, latitude, and time of season, represents a fascinating problem, whose potential complexity we are just beginning to discern (Åkesson et al. 1996). In this paper we present the results of systematic observations of the time of take-off in the European Robin Erithacus rubecula during spring migration on the Courish Spit of the Baltic Sea. Robin populations from Fennoscandia, The Baltic States, and NW Russia are migratory. They spend their winter in SW Europe and the Mediterranean area (Rendahl 1966, Payevsky 1973, Roos 1984, Pettersson et al. 1986, Cramp 1988, Rezvyi et al. 1995). Nearly all authors believe that the Robin is a typical nocturnal migrant (Dorka 1966, Pettersson & Hasselquist 1985, Ehbom & Karlsson 1993, Martin 1990a, Åkesson et al. 1992, Szulc-Olech 1965, Bolshakov & Rezvyi 1998, etc.). For several decades the Robin has been a model species for field and experimental studies, mainly in the field of migratory orientation and stopover ecology. New information on migratory behaviour of this species in the field is therefore of considerable interest.

3 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin Study area, methods and material The departure time of Robins from a daytime stopover site on the Courish Spit of the Baltic Sea (see Fig. 1 in Payevsky, this volume) was studied during 7 spring migratory seasons , 1985, , using two methods. In a special visual method was applied that allows the detection of taking-off birds until darkness (Bolshakov & Rezvyi 1975, 1982, 1998). In other years, observations using the horizontal searchlight beam were conducted (Bolshakov 1992), or a combination of visual and searchlight methods was applied from sunset until sunrise, i.e. during the whole twilight and dark period. The method of recording take-off activity by the visual method was described in detail (Bolshakov & Rezvyi 1975, 1982, 1998). This method was less efficient at the Rossitten Cape on the Courish Spit, than in the eastern part of the Gulf of Finland in both seasons. There were three reasons. (1) Topographic features and the direction of the coastline, plus a lower concentration of birds on the shore at daytime and during pre-start period. (2) Dense vegetation, and the lack of an elevated place with good visibility after sunset. (3) A shorter twilight period due to a more southerly position (55 N), and an earlier spring migratory season. During the spring seasons of , this method was applied between -120 and +60 min relative to sunset (Tab. 1). For studying take-off activity in the darkness we applied a method similar to that suggested by Hebrard (1971). Taking-off birds were observed when they were crossing an area over the tops of trees illuminated by two searchlights (1kW each) (Bolshakov 1992). The searchlights were placed approx. 12 m high on the roof of the Biological Station on Rossitten Cape, Courish Spit (see Fig. 1 in Titov, this volume). The beams were oriented nearly horizontally ESE across the main spring migratory direction of the Robins. They formed an illuminated space over the tops of vegetation reaching the shore of the Courish Lagoon. Illumination of trees and scrub was avoided by screening the beams. Taking-off birds were observed by eye and through x8 binoculars. Orientation of beams and lamp capacity, were held constant throughout the study. To avoid disturbance by insects, that were numerous in some nights, the observer sat by the side of the searchlights ca. 3m away. Rossitten Cape, which juts out into the Courish Lagoon, is isolated from the rest of the Courish Spit by the village and a large meadow. It is an area where large concentrations of passerine nocturnal migrants, including Robins, occur at daytime stopovers especially in spring (Lyuleeva 1967, Titov 1998, 1999, this volume). The vegetation zone includes two main habitats: reedbed with willow (Salix spp.) and scrub admixed, with trees and bushes, incl. willow, bird cherry trees, black elder, hawthorn and mature trees (birch and ash). In 1985, and 1993 visual observations were conducted until sunset and during twilight, followed by searchlight observations during darkness. In 1994 only searchlight observations were conducted. The schedule of observations varied between years (Tab. 1).

4 54 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin Avian Ecology and Behaviour Table 1. Study season, methods and observation time of take-off activity. Year Period of spring Method Period of observations relative to SS and SR* start end /03 15/05 visual 120 min before SS 60 min after SS /03 20/04 visual 120 min before SS 60 min after SS /04 15/05 visual + searchlight SS 90 min after SS /03 25/05 visual + searchlight 30 min before SS SR /03 3/05 visual + searchlight 30 min before SS SR /03 1/06 visual + searchlight 30 min before SS 360 min after SS /04 30/05 searchlight 30 min after SS 240 min after SS Note: * SS sunset, SR sunrise. The first session of searchlight observations was started 30 min after sunset (in min after sunset). It lasted for 60 min (in 1985 for 70 min). Every following hour after sunset a 30-min session was conducted, the final session starting 60 min before sunrise. Before sunset, also during early dusk and late dawn (0-30 min after sunset and 30-0 min before sunrise), searchlights were switched off and visual observations conducted. The most intensive observations were done in the springs of 1990 and 1991 when every hour from sunset until sunrise was covered by 30-min sessions, together with the last hour before sunset. The season of observations also varied between years. The most systematic data are available for and 1993, when practically the whole spring migratory season of Robins in the study area was covered (Tab. 1, see also Tab. 2, 3). In a large illuminated space, take-off with subsequent climbing could easily be distinguished from other flight behaviour (horizontal flight, reducing altitude and landing). As during twilight, the mode of climbing was dependent on wind conditions (Bolshakov & Rezvyi 1982). Some taking-off birds, especially under strong winds, crossed the illuminated space nearly vertically. In the darkness passerines are much more prone to gain height immediately over the departure site than during twilight. Taking into consideration large illuminated volume and circular flight trajectories when climbing, birds may stay in sight for several dozens of seconds. On clear nights and under limited cloud cover when air humidity is normal, the illuminated space is difficult to detect by human eye even when taking a side view. When the moon is visible the beams are inconspicuous, even under near total overcast, and are practically invisible under clear skies. There is no evidence of any impact of the light on birds behaviour under such conditions. In autumn 1995, on 14 nights between October 1-17, we made a special study of the behaviour of nocturnal migrants at the same site, under clear skies and overcast, by night vision binoculars. We failed to detect any difference in birds behaviour in the absence of an illuminated area. We recorded the same forms of flight behaviour: departures from the ground with climbing, horizontal flights, reducing flight altitude and landing. During 54 observation sessions in the 2nd-5th

5 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin 55 hour after sunset, at least 52 take-off of passerines were recorded (Bolshakov, unpubl.). All this suggests that searchlight observations of take-off reflect genuine behaviour of at least some passerines (Turdus philomelos, T. iliacus, Erithacus rubecula & Regulus regulus) under natural nocturnal conditions. Table 2. Timing and temporal pattern of spring Robin migration on the Courish Spit estimated by different methods. Year Method Period of use First birds First wave Last wave Last bird Peak numbers 1977 Complex* 14/3 15/5 14/3 20/3 10/5 11/5 Rybachy-type 13/4 10/6 14/4 21/4 26/4 5/ / 4 traps** 1978 Complex 2/3 6/6 29/3 30/3 12/5 20/5 Rybachy-type traps 20/3 10/6 29/3 30/3 17/5 7/ / / Rybachy-type traps 4/4 10/6 5/4 5/4 10/5 30/ / / Complex 17/3 25/5 17/3 19/3 11/5 13/5 Rybachy-type 30/3 10/6 30/3 2/4 12/5 24/ / 5 traps 1991 Complex 19/3 3/5 19/3 3/4 30/4 3/5 Transect counts 19/3 3/5 19/3 3/4 30/4 3/5 Rybachy-type traps 9/4 10/6 10/4 12/4 11/5 7/ / Complex 21/3 29/5 21/3 5/4 27/4 27/4 Transect counts 19/3 27/5 19/3 5/4 27/4 8/ / 4 Rybachy-type traps 20/3 10/6 22/3 7/4 24/4 25/ / 4 Mist-nets*** 6/4 10/6 6/4 7/4 7/5 7/ / Rybachy-type traps 21/3 10/6 25/3 29/3 10/5 9/ / 4 Mist-nets 23/4 10/6 23/4 23/4 10/5 10/ / 4 Notes: * after Bolshakov 1977, 1981; ** field site Fringilla, 11 km SW from Rybachy. Data from all traps used from the onset of trapping until June 10; *** field site Rybachy, observation site. Most correct estimates printed bold.

6 56 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin Avian Ecology and Behaviour Table 3. Season of take-off activity of Robins on the Courish Spit during spring migration. Results of observations by different methods.* Year Period of spring Number of days with obs. Period of evening pre-start activity on the ground Period of take-off activity Number of obs. birds start end start end /3 15/ /3 11/5 26/4 11/ /3 20/ /3 20/4 31/3 16/ /4 5/5 35 NO NO 9/4 3/ /3 25/ /3 13/5 20/3 11/ /3 3/ /3 3/5 3/4 3/ /3 1/ /3 1/5 6/4 27/ /4 30/5 35 NO NO 28/4 1/5 6 Notes: * For observation methods see Tab. 1; ** after Bolshakov & Rezvyi 1975, 1982; *** only birds with certainty identified as Robins; NO no observation. When air humidity is high, during fog and rain the beams form a brightly-lit area, which is especially conspicuous under total overcast without moon. In such situations a number of passerines, (mainly thrushes T. philomelos, T. iliacus, T. merula, and Robins), produce more calls from the ground. Thrushes are also calling more actively under clear skies without fog, with a full moon and when the illumination level is high. Up to now however, no evidence is available that a brightly lit area which is formed at high humidity influences take-off activity of migrants (Bolshakov, unpubl.). During visual observations before darkness, taking-off Robins were identified by their characteristic shape, flight, and colour (Bolshakov & Rezvyi 1998). Not infrequently birds were identified still on the ground by their characteristic calls. During take-off Robins call extremely rarely (Bolshakov, unpubl.). In darkness, Robins crossing the illuminated area sometimes could not be identified by their usual field characters, although the contrast between light belly and dark (red) breast facilitates identification, especially from short distance. In this paper we discuss only birds positively identified as Robins. All dubious cases were excluded from the analysis or are discussed separately. The time of every departure was recorded in minutes or hours with respect to local sunset time. For each date, the time of sunset and sunrise, end of civil, nautical and astronomical twilight (Sun elevation -6, -12 and -18, respectively) and the period between sunset and sunrise were calculated using astronomic tables. When analysing take-off activity the following parameters were calculated: 1) Time of first departures. This parameter reflects the onset of nocturnal migration and corresponds to first echoes from small passerines on radar, or first silhouettes visible against the face of the moon and in the vertical searchlight beam (ceilometer).

7 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin 57 2) Temporal distribution of departures of all Robins in a particular night was computed on the basis of departures in each hour after sunset. 3) Temporal distribution of departures over segments of the migratory period and in a particular year. 4) Time of first departures and temporal distribution of departures under different cloud conditions. Cloud cover was scored from 0 (clear skies) to 10 (total overcast) during each observation session (hour after sunset). Table 4. Temporal characteristics of departure period (searchlight observations). Year Season of take-off activity Median date Number of birds First and last take-offs, min after SS Median time, min after SS Mean time ± SD, min after SS /3 11/5 29/ ± /4 3/5 7/ ± /3 27/4 15/ ± 72 Notes: 1) difference in distributions: : λ = 2.00, p < 0.001; : λ = 0.89, p > 0.05; : λ = 1.12, p > 0.05; 2) difference in mean time: : t = 4.1, p < 0.001; F = 13.2, p < 0.001; : t = 1.5, p > 0.05; F = 2.3, p > 0.05; : t = 3.0, p < 0.01; F = 7.2, p < Each distribution was characterised by the departure period (in hours), median and mean (with SD) departure time (min after sunset). Distributions were compared by Kolmogorov- Smirnov test, mean departure times by t-test and Fisher test and proportions were compared by t- test. To estimate the timing of spring migration of Robins over the Courish Spit in each year, we used: (1) Observations of nocturnal migration by a combined method (Bolshakov 1977, 1981, Bolshakov et al. 1981). (2) Daily trapping results in Rybachy-type traps in Fringilla, 11 km from the study site (all years), and in mist nets at the study site (1993). (3) Results of transect counts (4 km) near Rossitten Cape. Counts were made daily in the afternoon. In some days, thrushes excluded, Robins comprised over 80-90% of all small passerines recorded.

8 58 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin Avian Ecology and Behaviour 3. Results 3.1. Timing of spring migration Trapping in Rybachy-type traps, (for description see: Dolnik & Payevsky 1976), conducted on the Courish Spit annually since 1959, usually does not allow recording of the arrival of the first Robins in spring, as the trapping regimes start too late. Over the 7 years under consideration, it was possible only in 1978 and 1994 when the first Robins arrived on March 29th and 24th respectively. During the four years when direct observations of nocturnal migration and transect counts were started in mid March, Robins were already present on the Courish Spit (March 13, 1977; March 17, 1990; March 19, 1991; March 19; 1993). From , variation of first captures of Robins was 12 days, from March 30 to April 10, the mean date being April 2 (Sokolov et al. 1998). Seven-year data suggest considerable inter-annual variation between other temporal parameters of spring Robin migration in this region (Tab. 2). 1. Date of the first wave of migration varied by 20 days, from March 19 to April 7 (median March 30). 2. Date of the last wave of migration varied by 24 days, from April 24 to May 17 (median May 10). 3. Period of mass migration varies from 25 to 57 days with an average of 44 days (42 days between the mean dates of the first and the last wave). 4. Maximum numbers are recorded during approx. 30 days, in different years, from the second 5-day period of April until the second 5-day period of May. After estimates obtained by different methods, in over 90% cases mass passage occurs between April 16 and May 10, with maximum numbers recorded in the fifth 5-day period of April (April 21-25). After long-term trapping in Rybachy-type traps, the mean date of spring Robin passage is April 22, varying from April 12 until May 3 (Sokolov et al. 1998) Season of departures The section of spring when Robins take-off from the Courish Spit is at least 53 days, from March 20 until May 11, based on 6-year data (Tab. 3). The season of departures practically coincides with spring migratory period of Robins over the Courish Spit (Tab. 1, 2). First cases of takeoff during evenings and nights coincide with the first waves of migration, and last take-off with the end of mass passage Time of first departures (onset of nocturnal migration) Overall variation of the onset of nocturnal migration (first take-off) recorded by visual and searchlight methods is presented in Fig. 1 (n=55). Over 7 years, the earliest first take-off of Robins occurred 29 min after sunset. The latest first take-off was recorded in the sixth hour after sunset. These data show that the variation of the onset of spring nocturnal migration in Robins on the Courish Spit is ca. 5.2 hours. The most reliable data are available for and 1993 (Tab. 1, 3). At the end of the first hour after sunset Robin migration commenced in only 20% of cases (eight out of 40). Most frequently, first Robins departed in the second and third hour after sunset (30% of cases each). In

9 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin 59 20% of cases nocturnal migration started only in the fourth-sixth hour. Median departure time for the whole departure period (March 20 May 11) was 122 min after sunset (129 min for the most reliable data of and 1993) (Fig. 1). In more than 50% of cases spring nocturnal migration of Robins did not commence until third hour after sunset. Fig. 1. Time of first take-offs of Robins in spring on the Courish Spit. 1 visual observations ( ); 2 searchlight observations (1985, , ). Every dot represents the time of a single departure. Full lines time of sunset (SS) and sunrise (SR). Dotted lines end of civil twilight (ECT), end of nautical twilight (ENT), and end of astronomical twilight (EAT). MWT Moscow winter time. We calculated the data for three periods with sufficient data (April 1-10, n=12; April 11-20, n=12; April 25 May 11, n=11). 1. Earliest departures, (visual and searchlight observations over the whole study period), occur at a similar time for the whole spring, min after sunset. 2. For the whole spring no fixed period for the onset of nocturnal migration is recorded in Robins. Variation in the time of first departures within the three periods comprised 4.0 to 5.2 hours, with an average of 4.5 hours. 3. A trend towards earlier onset of nocturnal migration with respect to sunset is recorded from the start to the end of the season. After data from 1990, 1991 and 1993, median time of the onset of migration for the three periods was 138, 121 and 121 min after sunset, respectively.

10 60 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin Avian Ecology and Behaviour An even more explicit shift in the mean departure time towards sunset is recorded when the data for the first (March 20 April 13, n=21) and second (April 14 May 11, n=19) halves of spring migration period are compared. Median time for the onset of nocturnal migration (first takeoff) over 19 days from on average April 5 to April 24 shifted by nearly 45 min towards sunset (151 and 106 min after sunset, respectively). In the first half of the season, in 33% of cases, first Robins departed in the fourth-sixth hour. In the second half, in 95% of cases, migration started in the firstthird hour. Distributions of the timing of first departures in the first and second half of the season are however not significantly different (λ=1.02; p>0.05) Departure pattern during the night In and a total of 134 birds positively identified as Robins were observed during 45 nights with take-off activity. The latest departure was recorded in the tenth hour after sunset. Overall departure period in spring was 9.5 hours. According to the pooled data (March 20 May 11), 50% of Robins did not commence nocturnal flight until 2,5 hours after sunset (median of 152 min). The mean departure time was 180 ± 107 min after sunset. The highest number of birds departed in the second and third hour (26% and 29%, respectively), and by the seventh hour after sunset 90% of the birds had departed (Fig. 2). A total of 42.5% of Robins (51 out of 134) departed after the end of astronomical twilight (Sun lower than -18 ). From the moment of first take-off until the end of nautical twilight (Sun elevation more than -12 ) only 26.1% of all birds departed (34-35 out of 134) (Fig. 2). After the most reliable data obtained in , (where the whole period from sunset to sunrise was covered by observations), median take-off time was 154 min, mean time 191 (SD = 121) min. Mean departure time was not significantly different from the estimate based on pooled data (t=0.7; p>0.05; F=0.5; p>0.05). Temporal distribution over these two years (67.2% of all identified Robins) was also not significantly different from the complete data (Fig. 2) (λ=0.36; p>0.05). Calculations for the both halves of the season (March 20 April 13 and April 14 May 11) on the basis of 4-year data pooled showed the following: 1. Time of earliest departures is constant throughout the spring 29 min after sunset. 2. Overall duration of the departure period reduces from 9,5 to 6,5 hours. In the first half of the season Robins departed from the first to the tenth hour, in the second half from the first to the seventh hour. 3. Median departure time shifts by 34 min towards sunset in the second half of the season (176 and 142 min after sunset, respectively). Distributions are significantly different (λ=1.47; p<0.05). 4. Mean departure time shifts by 62 min towards sunset in the second half of the season (214 ± 132 min and 152 ± 70 min, respectively; t=3.3; p<0.01; F=11.6, p<0.001). 5. The proportion of Robins that take off after the end of astronomical twilight, significantly decreases from 65.6% to 23.3% (t=5.4, p<0.001). The proportion of birds that commence flight before the end of nautical twilight increases significantly from 18.0% to 35.6% (t=4.9, p<0.001).

11 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin 61 Fig. 2. Temporal distribution of take-offs of Robins over the night. Every small dot represents a single departure, large dot 5 departures. Other notations as in Fig Inter-annual variation in departure time As some Robins depart as late as the tenth hour after sunset, estimates of departure period are most reliable for 1990 and 1991 when observations covered the whole migratory period and the whole period between sunset and sunrise (Tab. 1, 3). Less reliable estimates are available for 1993 when take-off activity was recorded between the first and sixth hour after sunset. All distributions of take-off are insignificantly skewed to the left. The difference between mean and median values varies between 10 and 28 min, the median being close to sunset. The dates with peak numbers of departing Robins varied between years by 8, 14, and 22 days. Latest departures with respect to sunset were recorded in 1991 when peak numbers occurred on April 7. In 1990, when peak numbers were recorded on April 22, Robins departed on average 90 min earlier with respect to sunset. In 1993, with peak numbers recorded on April 15, both mean and median departure times were intermediate between 1990 and 1991 values. Shift of the median to sunset was 76 min in 22 days, a rate of min day -1. Shift in the mean departure time was 90 min, rate of min day -1. Departure period and variation in the take-off time of individual birds, was greatest in 1991 when peak numbers of migrating Robins were recorded earliest in spring.

12 62 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin Avian Ecology and Behaviour To compare temporal characteristics of departures in different years over the same part of season, we also computed the data for the period April 3-15 (middle part of the migratory season). Mean date of take-off activity differed by 8 days between 1991 and 1993, but mean and median departure time were both significantly different (a difference of 73 min and 55 min, respectively). Later in the season Robins started migration significantly earlier with respect to sunset. The rate of median and mean time change reached 7 min day -1 and 9 min day -1, respectively (Tab. 5). Table 5. Temporal characteristics of take-off activity over similar season (April 3-18) in 1991 and Year Median date Number of birds First and last takeoffs, min after SS Median, min after SS Mean time ± SD, min after SS Significance level, distributions Significance level, mean time / ± 141 λ=1.37 p < 0.05 t=3.0 p < / ± 74 F= 6.4 p < Day-to-day variation in departure time The number of birds positively identified as Robins on individual nights is usually small. Thus, it is difficult to analyse day-to-day variation in parameters of take-off activity. Data for individual nights with the largest number of Robins identified, and their proportion among other small passerine migrants that reached at least 82-97%, and are presented in Tab. 6. These data show that: 1) Take-off times of first and last birds, duration of departure period, and the pattern of departures during the night vary broadly even between adjacent dates of the same year. 2) Variation in median departure time between subsequent nights reaches 1 hour, in mean time 1,5 hour. 3) On some nights in the first half of the season, Robins commence flight on average considerably later with respect to sunset than in the second half Impact of cloud score on take-off activity In 29 cases out of 45 (data from 1985, and ) Robin migration started under a cloud score of 0 5.0, in 9 cases the score was , and in 7 cases it was (Fig. 3). A detailed weather analysis showed however, that in only three cases out of 7, total overcast was recorded during the whole night, and not just at the moment of first take-off. In four other cases cloud score was under 10.0 either before or after the first departure. This allows the assumption that in spring, Robins in 93.3% of cases initiated migration, having access to celestial orientation cues. When the most reliable data of and 1993 is included, this proportion is only slightly lower, 92.5% (n=40). A detailed analysis of take-off under total overcast revealed two features. Firstly, they were recorded only during migration waves, i.e. when motivation to fly was high

13 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin 63 (April 29/30, 1990, third night of mass migration; April 6/7, 1993, second night of mass migration; April 17/18, 1993, third night of mass migration). Secondly, in all cases cloud score during the previous night varied between 0 and 5, i.e. migration was happening under clear skies or limited cloud cover. Fig. 3. Time of first take-offs of Robins in spring under different cloud conditions. Data of searchlight observations in , Every dot represents the first take-off (n=45). 1 cloud score 0 5.0; 2 score ; 3 score Other notations as in Fig. 1. Obviously, tendency to initiate flight under clear skies impedes the analysis of cloud cover impact on the timing of first departures. Due to uneven distribution of the data, median time of first take-off were computed for two situations: 1. Clear skies or limited cloud cover (score 0-5.0, n=21). 2. Considerable cloud cover and overcast (score , n=16). Estimates over the whole spring suggest, that median times of start of migration were 143 and 109 min; in the first half of the season, (March 20 April 13) 181 min (n=12) and 111 min (n=9); and in the second half of the season (April 14 May 11) 106 min (n=9) and 106 min (n=7) after sunset. This suggests that under clear skies and limited cloud cover Robins start migration later with respect to sunset than under considerable cloud cover and overcast. During the first half of the migratory period the difference reaches 70 min, but it is completely lacking in the sec-

14 64 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin Avian Ecology and Behaviour ond half. In 11 cases out of 14 (ca. 79%), when migration started after the end of astronomical twilight (Sun lower than -18 ), the weather was clear or with limited cloud cover (Fig. 3). Before the end of nautical twilight (Sun above -12 ), migration started under a cloud score of in 58% of cases (18 out of 31) and under a cloud score of in 42% of cases (13 out of 31), the proportion being nearly equal. The available data do not allow other correct estimates. Table 6. Temporal characteristics of departure in individual nights. Date Cloud score Observed birds, N* Robins n** Percentage of Robins*** Fist and last take-offs, min Median min after SS Mean time ± SD, min after SS after SS 4/ ± 118 5/ ± 60 7/ ± / ± 74 18/ ± 37 1/ NO ± 19 Notes: NO no observations; in 1993 observations were made until 6th hour after SS. * number of small passerines detected at departure; ** number of identified Robins; *** percentage of Robins among passerine nocturnal migrants (thrushes excluding) recorded during transect count in the afternoon. The impact of cloud cover was estimated also for take-off activity of all Robins. Take-off under total overcast was recorded in 12 birds out of 134 (approx. 9.0%). All 12 birds departed in the above-mentioned nights during waves of migration, when cloud cover in the previous night was negligible, if any. The data concerning cloud impact on temporal parameters of take-off activity are presented in Tab. 7, 8. Temporal distribution of take-off activity and mean time of nocturnal departures under clear skies and cloud cover were significantly different neither over the whole spring, nor in the first or second half of the migratory period. Overall duration of departure period however, and variation in the timing of departure of individual birds were higher under cloudy skies (Tab. 7). Moreover, in the first half of the migratory period a larger proportion of Robins departed after the end of astronomical twilight under clear skies than under overcast. To the contrary, under a cloudy sky a larger proportion started flight before the end of nautical twilight than under clear skies. No significant difference was found in the second half of the migratory period, or over the whole spring (Tab. 8).

15 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin 65 Table 7. Temporal characteristics of take-off activity under different cloud cover (searchlight observations in and ). Season Cloud score First take-off, min after SS Period of take-off activity, hrs Median min after SS Mean time ± SD, min after SS Significance levels, distributions and mean time 20/3 11/ ± 84 t = 0.8; p>0.05 F = 0.7; p> /3 11/ ± 141 λ = 0.77; p> /3 13/ ± 94 t = 0.4; p>0.05 F = 0.1; p> /3 13/ ± 158 λ = 0.92; p> /4 11/ ± 66 t = 0.5; p>0.05 F = 0.4; p> /4 11/ ± 80 λ = 0.32; p>0.05 Table 8. Difference in proportions of birds departing under limited cloud cover (score 0-5.0) and significant cloud cover (score ). Season 20/3 11/5 20/3 13/4 14/4 11/5 Bird number after EAT, % t=0.63; p>0.05 t=1.90; p 0.05 t=0.94; p>0.05 Bird number before ENT, % t=0.81; p>0.05 t=1.9; p 0.05 t=0.54; p>0.05 The available data therefore, suggest that in the first half of migratory period nocturnal migration of Robins starts under a cloudy sky on average more than 1 hour earlier than under clear skies. Furthermore, under a cloudy sky more birds depart before the end of nautical twilight. Under clear skies, to the contrary, more birds depart during deep night after the end of astronomical twilight. In the second half of the season no similar difference between clear and cloudy nights was revealed. Over the whole spring the overall departure period and the variation in the timing of nocturnal departures are larger under a cloudy sky. We also analysed the seasonal change in temporal characteristics of take-off activity separately under clear and cloudy skies. Seasonal changes in the timing of first departures are obvious under clear skies and limited cloud cover: medians for the first and the second halves of the season are 181 and 106 min after sunset, respectively. Under considerable cloud cover no seasonal changes were recorded: (medians 111 and 106 min, respectively).

16 66 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin Avian Ecology and Behaviour Under clear skies temporal distributions of all take-off differed significantly between the first and the second halves of the season (λ=1.60; p<0.05). Median departure time shifted in the second half of the season towards sunset by 79 min, mean time shifted for 72 min (t=3.5; p<0.001; F=15.9; p<0.001). Under a cloudy sky seasonal changes were not significant (λ=0.59; t=1.4; F=1.5; p>0.05) (Tab. 7). The proportion of Robins that depart after the end of astronomical twilight decreased over the season under a clear sky (t=4.9; p<0,001), the difference being not significant under a cloudy sky (t=1.6; p>0.05). Similarly, the proportion of birds departing before the end of nautical twilight significantly increased only under a clear sky (t=2.0; p=0.05; and t=0.7; p>0.05, respectively). 4. Discussion 4.1. Variation in the time of first departures Until now most field studies have focused on the registration of the daily starting time of nocturnal passerine migration, especially on the starting time for the main exodus of migrants at dusk (Gauthreaux 1971, Casement 1966, Drury & Nisbet 1964, Parslow 1968, Alerstam 1976, Richardson 1978, Lindgren & Nilsson 1975, Bolshakov 1981). There are few studies, mainly radio-telemetry, that have investigated the temporal distribution of migratory departures throughout a major part of the night from sunset onwards (Cochran et al. 1967, Hebrard 1971, Åkesson et al. 1996, Moore & Aborn 1996, Bolshakov 1992). In this paper the temporal distribution of migratory departures of Robins was studied by direct visual observations in a systematic way throughout the whole night from sunset to sunrise, over several years and the whole spring migratory period. Furthermore, the two hours before sunset were covered by observations. This method allowed the following parameters to be estimated in at least one passerine species: 1. Variation of the onset of nocturnal migration in spring (first take-off). 2. Temporal distribution of migratory departures throughout the whole twilight and dark period. 3. Impact of cloud cover on take-off activity and temporal distribution of migratory departures. We found that the time of first departures in spring is not fixed with respect to sunset, daily variation reaching 5.2 hrs. On the Courish Spit first Robins do not depart until min after sunset. This is in agreement with data on other passerines in other locations in spring 22 to 40 min after sunset (Casement 1966, Parslow 1968, Gauthreaux 1971, Hebrard 1971, Cochran et al. 1967) or slightly earlier (Moore & Aborn 1996). The latest departure of first birds was recorded in the sixth hour after sunset. Considerable variation in the time of start of nocturnal migration is recorded in the Robin over the whole spring, reaching 4.0 to 5.2hrs. From the pooled data, in 50% of cases spring nocturnal migration did not start until third hour after sunset. Across the season the mean time of first departures was significantly shifting towards sunset, on average by min. During long nights (10-12 hrs) in the first half of the season, in 33% of cases first birds departed only in the fourth-sixth hour after sunset. During short nights (8-10hrs) in the second half of the season, in 95% of cases migration commenced in the first-third hour. Until now very limited evidence of variation in the time of start of nocturnal migration in different passerines has been available. Now, it is believed that only European thrushes (Turdus spp.) regularly start migration before sunset (Bolshakov 1992, Åkesson et al. 1996). In the Robin, such behaviour in spring is very rare, (two cases during 10 years of targeted observations), and oc-

17 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin 67 curs at the end of the migratory season (late May) when passage is arrested by headwinds (Bolshakov & Rezvyi 1998). Robins departed before sunset only during very short nights, ca. 7 hrs. In spring we recorded variation in the time of beginning migration of small passerines by the moon-watch method in western Central Asia. Although in 80-85% of cases migration commenced before the end of nautical twilight, time of first records of flying birds varied between the first and sixth hour after sunset (Dolnik & Bolshakov 1985). Start of migration in second-fourth hour is typical for the first half of the migratory season (Bulyuk & Shamuradov 1985, Shimov 1985, Popov & Bolshakov 1985), when birds migrate during longer nights. An obvious seasonal shift, towards earlier start of migration, was recorded in passerine nocturnal migrants using ceilometer observations in southern Louisiana in North America (Hebrard 1971, Fig. 3 in Åkesson et al. 1996). Available data suggest that considerable variation in the timing of onset of migration will also be found in other nocturnal passerine migrants that migrate in spring under varying lengths of night. Mass spring migration occurs during certain synoptic situations, limited both in space and time (Richardson 1978, 1990, Alerstam 1990). To react in time to a favourable change in meteorological conditions, passerines should not have a fixed period of starting flight, especially during the part of season when nights are long. Our data however, are not sufficient to analyse this problem in more detail Temporal distribution of take-off activity during twilight and darkness We found a very large variation in the timing of departure of individual birds. Overall departure period in spring on the Courish Spit reached 9.5hrs, with latest departures in the tenth hour after sunset. Our data suggest that in the Robin, timing of departures is not synchronised and it is usually not possible to define the main departure period. Mean and median take-off time are 191 ± 121 and 154 min after sunset respectively. In contrast with the widely distributed belief that the bulk of nocturnal passerine migrants depart within a short period after sunset (Kerlinger & Moore 1989, Alerstam 1990), only 26.1% of Robins departed before the end of nautical twilight. Moreover, 42.5% of birds departed during deep night after the end of astronomical twilight when only star information remained (Martin 1990b, Martin & Brooke 1991, Åkesson et al. 1996). Departure time of first and last birds, length of departure period and temporal distributions of nocturnal departures vary broadly even between adjacent dates of the same year. Our data show that the differences in median and mean departure time between subsequent nights reach 1.0 and 1.5 hrs, respectively. We found significant seasonal variation in all main temporal parameters of take-off activity when the dark period decreases from 12 to 8 hrs. Overall departure period decreases from 9.5 hrs during long hr nights to 6.5 hrs during short 10-8 hr nights. Mean and median departure times shift towards sunset by more than 1 hr during short nights. In the second part of the season, the proportion of Robins departing after the end of astronomical twilight decreases, and the proportion of birds departing before the end of nautical twilight increases. Only in the end of the migratory period on some nights, may one speak of a main departure period when nearly all birds take off during a short time period. Inter-annual variation in temporal distributions of nocturnal departures found in the Robin, is probably related primarily to considerable variation in the timing and pattern of spring migration in this species. Departure period and variation in the timing of departure were greatest, and birds

18 68 C.V. Bolshakov & V.N. Bulyuk. Time of nocturnal flight in the Robin Avian Ecology and Behaviour started migration most late, on average 3.5hrs after sunset, when peak numbers at stopover sites were recorded early in the season. When peak numbers occurred later the departure period was shorter by 1.5hrs, Robins departing on average 90 min earlier with respect to sunset. It is noteworthy however, that when the mean date of peak numbers differed by only 8 days (difference in night duration ca. 0.5hrs), mean and median departure time differed by more than 1 hour. Thus, not only the time of the season, but some migration characteristics in different years and nights govern variations in temporal characteristics of take-off activity Impact of cloud cover on take-off activity Our data unambiguously suggest, that in spring Robins prefer to depart under a clear sky or limited cloud cover when sun-related orientation cues (sunset position, horizon glow and skylight polarisation pattern) or stellar information are available. Nocturnal migration started under overcast in only 7% of cases. Overall not more than 9% of Robins departed for nocturnal flight under overcast without access to celestial cues. Data from the Courish Spit collected over the whole night, are in full accordance with the results of long-term visual observations between sunset and darkness in the eastern part of Gulf of Finland: in only 5% of cases did Robins depart under overcast (Bolshakov & Rezvyi 1998). In spite of little data being available, two features of take-off under overcast may by mentioned. Firstly, all cases occurred during waves of migration (mass passage and high numbers at stopovers), i.e. when motivation to fly was high. Secondly, in all cases on the previous night, birds were migrating during clear weather or under limited cloud cover. Thus they had access to various celestial information. The ability of birds to retain orientation notions (and possibly high motivation to fly), within first 24hrs after being denied access to celestial cues was shown both in the field and in experiment (Hebrard 1972, Helbig 1991, Moore 1987). Our results, and also our direct observations between sunset and darkness (Bolshakov & Rezvyi 1982, 1998), confirm the results of experimental studies that visual orientation cues, associated with sunset and star patterns, are important for nocturnal passerine migrants (Emlen 1975, 1980, Able 1980, Moore 1987, Wiltschko & Wiltschko 1991, Katz 1985, Helbig 1991). In our opinion, the prolonged stay (2.5-3hrs) of inactive birds on the ground during twilight, may allow them not only to make a more reliable decision whether to migrate or not, but also to specify migratory direction, possibly estimate the distance to the goal of migration, and to solve other orientation tasks. At the same time, the very wide scatter in departure time lends little support that departure time is adapted to the availability of a specific combination of celestial orientation cues at twilight (Åkesson et al. 1996). Variation in the timing of onset of migration and scatter in the time of nocturnal departures is probably controlled by other endogenous factors and environmental stimuli. Our data concerning the impact of cloud score on the temporal characteristics of departure are not unequivocal. The obvious trend to depart under clear skies results in unevenly distributed material and hampers analysis. At present, we can emphasise only three main facts: 1. Both overall departure period and temporal scatter of nocturnal departures in the Robin were larger under overcast than under clear skies. This was valid throughout the season. Cochran et al. (1967) also found greater variation and later departures under overcast skies compared to clear weather in North American Thrushes Hylocichla spp. Our data however, revealed no difference in the mean departure time in Robins neither over the whole spring nor in any half of the migratory period.