Female-biased obligate strategies in a partially migratory population

Authors

  • Adam M. Fudickar,

    Corresponding author
    1. Department of Biology, University of Konstanz, Konstanz, Germany
    Current affiliation:
    1. Department of Biology, Indiana University, Bloomington, IN, USA
    • Department of Migration and Immuno-ecology, Max Planck Institute for Ornithology, Radolfzell, Germany
    Search for more papers by this author
  • Andreas Schmidt,

    1. Department of Migration and Immuno-ecology, Max Planck Institute for Ornithology, Radolfzell, Germany
    Search for more papers by this author
  • Michaela Hau,

    1. Department of Migration and Immuno-ecology, Max Planck Institute for Ornithology, Radolfzell, Germany
    2. Department of Biology, University of Konstanz, Konstanz, Germany
    Search for more papers by this author
  • Michael Quetting,

    1. Department of Migration and Immuno-ecology, Max Planck Institute for Ornithology, Radolfzell, Germany
    Search for more papers by this author
  • Jesko Partecke

    1. Department of Migration and Immuno-ecology, Max Planck Institute for Ornithology, Radolfzell, Germany
    Search for more papers by this author

Corresponding author: E-mail: afudickar@gmail.com

Summary

  1. Partial migration occurs when a breeding population consists of seasonal migrants and year-round residents. Although it is common among birds, the basis of individual movement decisions within partially migratory populations is still unresolved.
  2. Over 4 years, we used state of the art tracking techniques, a combination of geolocators and radio transmitters, to follow individual European blackbirds Turdus merula year round from a partially migratory population to determine individual strategies and departure and arrival dates. The individual-based tracking combined with measures of energetic and hormonal (corticosterone) state enabled us to distinguish between obligate and facultative migration and to test several classical hypotheses of partial migration: the ‘Arrival Time’-, ‘Dominance’- and ‘Thermal Tolerance’-hypotheses.
  3. Two distinct periods of departures from the breeding grounds were observed during the study; one in early autumn, and another during the midst of winter. Although blackbirds that migrated in autumn were never observed overwintering within 300 km of the study site, four individuals that departed in the winter were observed within 40 km. Females were significantly more likely to migrate in autumn than males but there was no difference in the age or body size of migrants and non migrants in autumn. Just prior to autumn migration, migrants had higher fat scores than non migrants and tended to have higher concentrations of baseline corticosterone, but similar concentrations of triglycerides. Unlike autumn migrants, we found no difference between the tendencies of males versus females to depart in winter, nor did we find any difference in body size or age of individuals that departed in the winter.
  4. Autumn migration was sex biased and resembled obligate migration. Our results provide strong support for the ‘Arrival Time’ hypothesis for partial migration in the autumn. We found no clear support for the ‘Dominance’ or ‘Thermal Tolerance’ hypotheses. By tracking individuals year round, we were able to identify a second period of departures. Overall, these results suggest the co-occurrence of obligate autumn migrants, winter movements and sedentary individuals within a single population.

Introduction

Seasonal migration is a common trait in a diverse range of taxa that facilitates the exploitation of temporally pulsing resources in geographically distant habitats. Seasonal migration optimizes reproductive fitness during the critical phase of offspring rearing while increasing survival the remainder of the year. Even in the tropics, where seasonal environmental variability is less extreme compared to temperate latitudes, many species migrate annually (Greenberg 1980; Joseph 1997; Jahn et al. 2006). Migrations are often anticipatory: animals start migrating at times when environmental conditions are still benign where they are, escaping before conditions deteriorate. Despite the predictive nature of migration, many migratory species are flexible in their migratory behaviour, allowing for adjustments to unpredictable environmental conditions during migration (Jenni & Schaub 2003; Schmaljohann, Liechti & Bruderer 2009; Schmaljohann & Naef-Daenzer 2011).

Ongoing changes in habitat phenology and broadscale changes in weather patterns have resulted in variable responses by migratory populations. Some migratory populations have adapted to these changes and flourished while others are apparently limited in their ability to respond (Wilcove 2008). For migratory birds at North temperate latitudes, some populations have advanced the timing of their spring arrival in response to changing spring phenologies and in some cases the frequency, distances and even directions of migrations are changing (Hüppop & Hüppop 2003; Bearhop et al. 2005; Pulido & Berthold 2010). Given the rate at which the earth's biosphere is changing, it is clear that individual flexibility and population variability are critical for the survival of migratory populations globally.

Migratory patterns within populations can be variable. One common case of variable migration strategies within a population is partial migration. Partially migratory populations consist of both seasonal migrants and year-round residents. Therefore, partial migration is thought of as an intermediate stage between fixed migratory and fixed sedentary life histories (Berthold 2001). Increased knowledge on the interaction of migrants with the environment and the selective pressures that shape migratory life histories is important for our basic understanding of the evolution of migration but also for informing policy decisions directed at preserving migratory populations.

The mechanisms responsible for individual strategies within partially migratory bird populations are still mostly unresolved. Although the mechanisms most likely vary between species and populations, two general hypotheses have been advanced: genetically controlled and condition dependent partial migration. Genetic determination of migratory behaviour within partially migratory bird populations was proposed almost as early as the phenomenon was identified in the literature (Thomson 1921; Nice 1933; Lack 1944). Controlled lab studies over the past four decades have provided estimates for the heritability of nocturnal locomotor activity ‘Zugunruhe’ which coincide with the timing of autumn and spring migration for free living counterparts. Heritability estimates for onset, termination and intensity of Zugunruhe for partially migratory populations range from 0·16 to 0·67 (Newton 2008). Although these results are intriguing, it is difficult to interpret what Zugunruhe corresponds to in wild populations. Helm & Gwinner (2006) found that resident African stonechats Saxicola torquata express seasonal Zugunruhe in captivity. The authors' interpretation of their findings was that Zugunruhe could be a regular feature of the endogenous programme of birds that might act to facilitate periodic movements also in non-migrants.

If migratory behaviour is underlain by a genetic dimorphism, then for both migratory and sedentary morphs to be maintained within a population, differences in survival and reproduction should vary across years. If one strategy were to convey a fitness advantage, it should become fixed within the population rapidly (Pulido & Berthold 2010). Furthermore, if a genetic determination of the migratory strategy was based on a true dimorphism in a population, then there should be equal age and sex ratios of migrants unless the trait is either (i) also sexually dimorphic or (ii) ontogenetically plastic and changes throughout an individual's lifetime.

One concept that proposes that partial migration is condition dependent is the ‘Arrival Time’ hypothesis. This hypothesis posits that if the reproductive fitness of one sex is partly influenced by the acquisition of a territory in early spring, and high quality territories are limited, then it is advantageous for individuals of that sex to be at the breeding grounds as early as possible (Ketterson & Nolan 1976). If lifetime fitness is increased by remaining on the breeding grounds year round, then selection should lead to sedentariness in the sex that establishes a territory. Conversely, if over winter survival of the sex which chooses a mate based on territory quality is increased by leaving the breeding grounds during the non-breeding season and no net fitness advantages result from residency, then selection could promote migratory behaviour in that sex. Sex differences in the tendency to migrate are well documented in partially migratory populations (Schwabl 1983; Adriaensen & Dhondt 1990; Partecke & Gwinner 2007).

Conditional differences that result in differences in migration vs. residency within a population could also occur if the available breeding habitat can only support a fraction of the population during the non-breeding season. In this case, partial migration results from asymmetries in individual quality where migration is a conditional strategy and migrants are making the ‘best of a bad job’ (Lundberg 1987, 1988). The ‘Dominance’ hypothesis of partial migration, also referred to as the ‘Competitive Release’ hypothesis, proposes that competition for limited food sources forces subordinate individuals to leave the breeding grounds during the non-breeding season. Field-based studies on temperate breeding birds have provided support for the ‘Dominance’ hypothesis (Schwabl 1983; Lundberg 1985; Smith & Nilsson 1987; Nilsson, Alerstam & Nilsson 2008). Individual asymmetries that might lead to variation in migratory strategies are not limited to competitive ability; the ‘Thermal Tolerance’ hypothesis of partial migration, also referred to as the ‘Body Size’ hypothesis, predicts that individual differences in thermal efficiency result in differences in migratory tendency (Able & Belthoff 1998). Previous studies at North temperate latitudes have uncovered a relationship between body size and migratory tendency, with smaller individuals being most likely to migrate (Chapman et al. 2011). These results may indicate that larger individuals are more capable of surviving extreme temperatures and limited food supply during the winter due to a greater metabolic efficiency and are therefore less likely to migrate than smaller individuals.

In this study we tested predictions of the ‘Arrival Time’, ‘Thermal Tolerance’ and ‘Dominance’ hypotheses of partial migration by monitoring the movements of European blackbirds Turdus merula from a partially migratory population in southern Germany year round using automated radio telemetry and light-level loggers (geolocators; Table 1). By tracking individuals year round, we were able to accurately classify migrants and residents. Traditional methods, such as establishing transects to locate colour-banded individuals after the breeding season, risk overestimating the migratory fraction of a population due to local movements away from core breeding areas. Male blackbirds compete for high quality territories during the early spring (Lundberg 1985). Therefore, our prediction for the ‘Arrival Time’ hypothesis was that female blackbirds would migrate more frequently than males (Table 1). During the non-breeding season, a clear dominance hierarchy exists in European blackbird foraging flocks. Adults are dominant over juveniles and males are dominant over females during aggressive encounters over food (Lundberg & Schwabl 1983; Lundberg 1985). Thus, our prediction for the ‘Dominance’ hypothesis was that first year birds would migrate more frequently than adults, and females would migrate more frequently than males (Table 1). For the ‘Thermal Tolerance’ hypothesis we predicted that smaller individuals would migrate more frequently than larger individuals (Table 1). Using a combination of year-round radio telemetry and geolocators we were able to distinguish between true migration events and local dispersal. We also compared physiological correlates of the energetic state between autumn migrants and non migrants in the weeks preceding autumn departures from the breeding grounds to test whether or not blackbirds prepare for migration, a characteristic of obligate migration.

Table 1. Hypotheses and predictions
HypothesisDescriptionPrediction
Arrival time (Ketterson & Nolan 1976)Intra-sexual competition for breeding territory promotes residency among the territorial sex.Male blackbirds should migrate less frequently than females
Thermal tolerance (Ketterson & Nolan 1976)Individual differences in thermal efficiency leads to migration among the least efficient.The tendency to migrate among blackbirds should increase with decreasing body size
Dominance (Gauthreaux 1982)Competition for limited food leads to migration among subordinates.Resident ———————- migrant adult ♂ > adult ♀ > 1st year ♂ > 1st year ♀

Materials and methods

European blackbirds Turdus merula were captured over 4 years (2009–2012) in a mixed coniferous and/or deciduous forest in southern Germany (N 47°47′, E 9°2′). Birds were initially captured from March to September in the first 3 years of the study (2009–2011) using 5–12 mist nets (12 m wide × 3 m tall) opened between civil twilight and 12:00, when weather permitted. Nets were placed on the edge of breeding habitat next to known foraging areas. Nets were checked every 30 min. Age and sex of individuals were determined based on plumage differences (Svensson 1992). Prior to first pre basic moult, the sex of hatch year blackbirds cannot be determined based on plumage differences. A quantity of 50 μl of blood was collected from hatch year birds from the brachial vein by venipuncture for molecular sex determination. Tarsus length has been shown to be a reliable measure of body size in blackbirds as in many other species (Alatalo & Lundberg 1986; Richner 1989; Merilä 1997). We used tarsus length as a measure of structural body size and it was measured to the nearest 0·5 mm using dial callipers to compare the body size of migrants to non migrants. To reduce inconsistencies in measurements, all tarsus measures were taken by A.M. Fudickar.

The Mk 10S, Mk 12S, and 20s geolocators (≤1·2 g; British Antarctic Survey, Cambridge, UK) connected to radio transmitters (≤2·6 g; Sparrow Systems, Fisher, IL, USA) with heat shrink tubing (≤0·4 g), were attached to birds via leg-loop harnesses. A range of harness sizes were built from 1 mm elastic beading cord to fit the naturally occurring body sizes of blackbirds in the population (Naef-Daenzer 2007). Each backpack weighed < 5% of the mass of the individual that it was deployed on. Once a harness was fitted to a bird, it was inspected for appropriateness of fit. All birds were observed for as long as possible after release and throughout deployment to ensure normal behaviour. All transmitters and geolocators were manufactured to last at least 1 year. Beginning in March in the last 3 years of the study (2009–2012), birds tagged the previous year were recaptured for backpack removal and deployment of a new backpack consisting of a new transmitter and geolocator.

To identify presence of individuals at the breeding site and departure dates, all birds were tracked using radio telemetry whenever present. In the first year, all birds were located twice per week from the date of capture until 1 December 2009. Beginning 1 December, birds were tracked once per week by ground until recapture the following spring. In the second and third years birds were tracked twice per week by ground after capture until recapture the next spring except from 20 December to 10 January when birds were monitored from automated receivers exclusively. Ground tracking was done using the combination of either a handheld three element Yagi antenna (AF Antronics, Inc., Urbana, IL, USA) and AR 8200 MKIII handheld receiver (AOR USA, Inc., Torrance, CA, USA) or a handheld H antenna (Andreas Wagener Telemetry Systems, Köln, DEaaa) and a Yaesu VR 500 handheld receiver (Vertex Standard USA, Cypress, CA, USA). If an individual could not be located by ground tracking, aerial searches encompassing a 20 km radius (minimum) of the study site were performed using a Cessna airplane equipped with two H-antennas, one per wing, and two Biotrack receivers, one per antenna (Lotek, Newmarket, ON, Canada). Individuals were classified as migrants after at least two searches from the air without a signal.

Three to five stationary automated receivers (Sparrow Systems) were present at the study site throughout the study to monitor the presence of individuals, and departure and arrival dates (Kays et al. 2011; Mitchell et al. 2012). Each automated receiver searched for 16 frequencies every 60 s. Automated receivers were connected to H antennas (ATS, Isanti, MN, USA), mounted 3–6 m high. Not all birds were captured within range of an automated receiver; therefore, manual tracking was the only means for monitoring their presence. After a departure was identified, extensive ground and air tracking was done to confirm absences. After recapture, geolocators were used to confirm departure and arrival dates.

Before deployment, all geolocators were placed outdoors with a similar view of the horizon for a minimum of 7 days to affirm similar light sensitivity. After deployment, raw data were corrected for clock drift using Bastrak (British Antarctic Survey). Transitions for all geolocators were calculated using TransEdit2 and anomalous transitions were rejected from analysis. Anomalous transitions can be due to the light sensor being covered by a feather, heavy cloud cover at sunrise or sunset or any other unusual shading events. Sunsets were retarded by 10 min for Mk 10s and 12s geolocators and 2 min for 20s. Longitude was calculated using Locator (British Antarctic Survey). We did not include latitude estimates in our analysis because of the relatively high error for latitude estimates compared to the potential migration distance of birds from our population (Fudickar, Wikelski & Partecke 2012). A calibration angle is not needed to calculate longitude estimates using archived light values. To confirm departure dates, we identified the first date that longitude estimates diverged greater than 84 km from the breeding grounds and compared that date to the departure date identified by telemetry. We used 84 km because it is the known error in longitude estimates for geolocators on European blackbirds in the winter (Fudickar, Wikelski & Partecke 2012). If geolocator data were not available (in the case of failures, incomplete archives or when birds did not return the following spring), and if a departure was not recorded on an ARU, the departure date was identified as the average date of the last observed date and first date missing. To identify return dates, ARU data were scanned to identify first observation. If a bird was first observed using manual telemetry, the arrival date was identified as the average date of the last date not observed during ground tracking and first date observed. To confirm arrival dates with geolocators, we identified the first date that longitude estimates were within 84 km of the breeding grounds in the spring and compared that date to the arrival date identified by telemetry.

In each year, we observed a bimodal distribution in post breeding departures from our study site (one in early autumn and one in winter). We performed a k-means cluster analysis with an a priori criterion of two clusters to classify departures into one of the two categories (‘autumn migration’ or ‘winter movements’). We justified separating individuals into the two periods given that the periods were bimodal with autumn departures occurring immediately after post breeding moult and before the onset of winter conditions while winter movements occurred in the midst of winter. Birds that departed during the winter that were located away from the breeding grounds after departing were tracked throughout the winter, once per week, until their return to the breeding grounds the following spring.

To compare the energetic state of migrants and non migrants just prior to autumn departure, we recaptured tagged individuals (14 migrants and 22 non migrants) between 20 September and 20 October in 2010 and 2011 (referred to as the ‘pre-migratory period’). We chose this period because most departures in 2009 occurred in the first 3 weeks of October. To ensure that all individuals were in a similar resorptive state, only individuals captured between civil twilight and 09:30 were used for comparison.

Birds undergo a suite of endogenously controlled physiological changes to prepare for migration. Endogenously controlled fat accumulation, prior to migration, provides migrants with the fuel reserves required for their journey (Berthold 1996). All tagged individuals recaptured during the ‘pre-migratory period’ received a subcutaneous furculum fat score based on a scale of 0–5 (Helms & Drury 1960). To rule out the potential for observer differences in fat scores, all scores were assigned by AMF.

Immediately after capture, during the pre-migratory period, we collected 400–500 μl of blood from recaptured individuals from the brachial vein by venipuncture. The same catching procedure was used during the pre-migratory period as in the spring and summer, however, only one to six mist nets were opened and samples were collected ≤180 s of birds hitting nets. Blood samples were immediately stored on ice and transported to the Vogelwarte Radolfzell for centrifugation within 3 h of collection. Plasma was aspirated off and samples were stored at −70 °C until they were assayed. To compare the physiological state of migrants and non migrants during the pre-migratory period, we compared plasma levels of triglycerides and baseline corticosterone of migrants and non migrants during the ‘pre-migratory period’ (20 September–20 October).

Plasma levels of triglycerides (TRIG) have been used in avian studies to assess the physiological state and condition of free living individuals. Circulating plasma TRIG levels have been shown to increase during feeding and decrease during fasting (Jenni-Eiermann & Jenni 1998). Plasma TRIG levels were determined using a standard spectrophotometric assay (see Masello & Quillfeldt 2004). We added 6 μl of undiluted plasma to 600 μl Triglycerides reagent (n° 981786, Thermo Fisher Scientific, Waltham, MA, USA warmed to 37 °C). Standard curves were calculated for each run using the combination of 6 μl of sCal (n° 981831, Thermo Fisher Scientific) and 600 μl Triglycerides reagent. Plasma samples and standards were incubated for 5 min at 37 °C after addition to the Triglycerides reagent. Absorptions were then measured at 540 nm wavelength. All samples were run on the same day, but not in the same run. New standard curves were calculated for each run.

The glucocorticoid hormone corticosterone (CORT) is produced in the adrenal cortex of birds and at baseline concentrations it is involved in glucose metabolism while at elevated levels it serves to support an individual to cope with stressful stimuli (Nelson 2005). Baseline concentrations of CORT have been shown to be elevated in migrating birds (Piersma, Reneerkens & Ramenofsky 2000; Holberton, Boswell & Hunter 2008). Migration is thought to be an energetically demanding activity and elevated baseline CORT concentrations may act to mobilize energy reserves. We therefore compared baseline CORT concentrations of autumn migrants with those of residents during the pre-migratory period. We determined CORT concentrations from 7 μl plasma samples after a diethyl-ether extraction using enzyme immunoassays (Cat. no. 900-097; Enzo Life Sciences, Farmingdale, NY, USA, Ouyang, Hau & Bonier 2011). The intra-plate coefficient of variation of two replicate standards was 8·3%.

During our statistical analyses of the data we compared (i) autumn migrants with those that stayed behind through the autumn migration period and (ii) birds that moved in the winter with those that stayed through the winter. To determine if different age classes and sexes were more or less likely to migrate and/or move as predicted by the ‘Arrival Time’ and the ‘Dominance’ hypotheses we ran separate Fisher's exact tests for the autumn and winter periods. Separate multinomial logistic regressions were run to explore the correlates of departing in either autumn or winter including the variables year, age, sex, tarsus and the interactions between sex and tarsus and age and year using stepwise backward elimination. To test for differences in physiological measures of energetic state of autumn migrants and non migrants we ran three separate general linear models. The first model included migratory strategy (yes/no) as the binary fixed factor and baseline CORT concentrations (log transformed) as the dependent variable with time of day as a covariate. The second model included migratory strategy (yes/no) as the fixed factor and TRIG concentrations (log transformed) as the dependent variable with time of day as a covariate. The third model included migratory strategy (yes/no) as the fixed factor and fat score (log transformed) as the dependent variable with calendar date as the covariate. A priori, we excluded one TRIG and three CORT outliers from statistical analyses. All statistical analyses were performed in SPSS 15·0.

Results

Out of 153 blackbirds monitored, 66 (43%) departed after the breeding season. Fifteen birds which were included as non migrants in the autumn died between December and February and were therefore (conservatively) not included in the comparison between birds that moved in the winter and sedentary birds. The cluster analysis assigned 19 individuals, eight females and 11 males, to the winter movement group (mean: 30 December, range: 30 November–7 February) and 47 individuals, 28 females and 19 males, to the autumn migration group (mean: 15 October, range: 19 September–9 November; Fig. 1). Throughout the study, autumn migrants were never observed (even by aerial radio tracking) less than 300 km from the study site after their departure from and before they arrived on the breeding grounds. The results of the aerial tracking of autumn migrants were confirmed by geolocators that could be recovered from a total of 19 autumn migrants. Six autumn migrants that returned the following spring were not recaptured. Unlike autumn migrants, four birds that departed during the winter were located by airplane tracking (range: 6–38 km, heading: 173°–248° from the breeding grounds). Four birds that departed during the winter, which could not be located after departure, were recaptured in the following spring. Geolocator estimates indicate that two of the recaptured winter birds had moved > 400 km (±84 km) in one night and then remained sedentary at their new location for the remainder of the winter. Both these birds had moved west of the breeding grounds. Latitude could not be estimated for these individuals. The winter longitude estimates of the other two recaptured birds were statistically not distinguishable from that of the study site (< 84 km). However, tracking by air indicated that they had moved at least 20 km off the breeding site. In contrast, sedentary birds were typically observed within 500 m of the breeding grounds, but occasionally moved up to 1·5 km during the winter period.

Figure 1.

Departure, arrival, and return dates of European blackbirds from a partially migratory population in southwestern Germany over 4 years. Cluster analysis assigned individuals to one of two post breeding departure groups: autumn departures occurred from 19 September to 9 November (a; centre–15 October, n = 47) and winter departures occurred from 30 November to 7 February (b; centre–30 December, n = 19). Average return date for all migrants in the spring was 16 March (c; SD = 8.9 days, n = 24, range 23 February–29 March). Autumn migrant departure and return dates are indicated by white bars and winter departure and return dates are indicated by grey bars.

We found no difference in the tendency to migrate between hatch year and adult birds in the autumn (one-tailed Fisher's exact test, = 0·34; Fig. 2). Year, age, tarsus and all interactions were excluded from the model that best explained migratory tendency in the autumn. Sex was the only significant variable, with females being more likely to migrate than males (χ² = 6·969, d.f = 1, = 0·01). Autumn migrants had higher subcutaneous furculum fat scores in the pre-migratory period than non migrants (F1,36 = 16·050, = 0·00 adjusted R2 = 0·451; Fig. 3). Autumn migrants tended to have higher levels of baseline CORT than non migrants in the pre-migratory period (F1,14 = 4·330, = 0·06, adjusted R2 = 0·169). There was no difference in TRIG concentrations between autumn migrants and non migrants (F1,16 = 0·496, = 0·49, adjusted R2 = 0·179). In the winter, we found no difference between either age or sex in the tendency to depart (one-tailed Fisher's exact test, = 0·49, 0·62). The final model that best explained departure tendency in the winter excluded all of the predictor variables: age, sex, body size and interaction between sex and body size, indicating that none of the predictor variables contributed to the model.

Figure 2.

Per cent of European blackbirds from the study population that (a) migrated in the autumn (19 September–9 November) and (b) moved in the winter (30 November–7 February) separated by age and sex. Numbers in parentheses on the x-axis represent sample size. Females were significantly more likely to migrate in the autumn than males (one-tailed Fisher's exact test, = 0.01).

Figure 3.

Relationship between autumn migratory strategy and three physiological correlates of energetic state in European blackbirds sampled during the pre-migratory period (20 September–20 October): (a) Plasma concentrations of corticosterone (CORT), (b) fat score and (c) plasma concentrations of Triglycerides (TRIG). Circles indicate outliers. A priori, CORT and TRIG outliers were excluded from statistical analyses. Autumn migrant blackbirds had significantly higher fat scores than sedentary blackbirds (F1,36 = 16.050, = 0.00, adjusted R2 = 0.451). There was a tendency for autumn migrants to have higher CORT levels than non migrants (F1,14 = 4.330, = 0.06, adjusted R2 = 0.169). Plasma concentration of TRIG did not differ between migrants and non migrants (F1,16 = 0.496, = 0.49, adjusted R2 = 0.179).

Discussion

For a partially migratory population of European blackbirds in south-western Germany monitored over the course of 4 years we identified two distinct periods of post breeding departures (Fig. 1). In autumn, as predicted by the ‘Arrival Time’ hypothesis of partial migration, female blackbirds were significantly more likely to migrate than males, independent of age or body size (Table 1 and Fig. 2a). Like obligate migratory birds, autumn migrants prepared for the energetically taxing behaviour of migration. In the weeks preceding autumn departure, migrants had higher subcutaneous fat stores than non migrants and migrants tended to have higher levels of baseline corticosterone (Fig. 3a and b). In combination, these results suggest that autumn migration in the study population is a sex-biased anticipatory behaviour.

Sex differences in avian migration are commonplace and have been described both for differential migration, where females typically migrate farther than males and for long distance migration where the two sexes show phenological differences in their timing of migration (Newton 2008). Sex differences in partial migration, as we found in this study, could be linked to differences in reproductive strategies. Prezygotic contributions to reproduction differ between male and female blackbirds. Males incur costs associated with territorial establishment and maintenance while the greatest contributions by females are in egg production. Loss of eggs or offspring due to mistimed reproduction in early spring may come at a higher lifetime cost to females than males. Therefore, the average overall fitness benefit for male residents should be higher than for female residents.

We did not find strong support for either the ‘Dominance’ or ‘Thermal Tolerance’ hypotheses for autumn migration in European blackbirds (Table 1). Although female blackbirds migrated at a significantly higher rate than males as predicted by the ‘Dominance’ and ‘Thermal Tolerance’ hypotheses, we found no difference between hatch year and adult blackbirds in the tendency to migrate. Lundberg (1985) found that adult blackbirds are dominant over hatch year birds and males are dominant over females during the non-breeding season. The observed dominance hierarchy could explain the higher rates of females and juveniles previously observed to migrate from the area in the autumn (Schwabl 1983). However, previous methods for classifying free living songbirds as being migratory or non migratory, such as establishing transects to locate colour-banded individuals during the winter might overestimate the migratory fraction of the population. Local movements during the non-breeding season can result in individuals from core breeding habitat moving into areas not typically used during the breeding season (A.M. Fudickar, unpublished data). Hatch year individuals tracked in this study were commonly found during the non-breeding season 300–400 m from the study site in locations where they would likely not have been found without the use of radio telemetry. Adults were observed more frequently at core breeding areas.

Although female blackbirds are on average smaller than males, body size was not a significant factor in autumn migration as predicted by the ‘Thermal Tolerance’ hypothesis. Therefore, it is unlikely that differences in the tendency to migrate in the autumn are directly linked to differences in individual body size. Long term selection could lead to a greater tendency of females to migrate than males due to differences in thermal efficiency and over winter survival. However, there is considerable overlap in the body sizes of male and female blackbirds in central Europe (Svensson 1992). Increased over winter survival by migratory females, irrespective of body size, in addition to reproductive advantages for sedentary males, could result in sex differences in the tendency to migrate in the autumn. Similar conclusions have been reported for European blackbirds inhabiting urban areas (Partecke & Gwinner 2007). In particular, advanced timing of reproduction and increased survival probabilities due to milder microclimates in urban areas are thought to select for increased sedentariness in male blackbirds. Given the current results, future studies comparing sex differences in lifetime fitness for migrants and non migrants in partially migratory populations will be extremely valuable to further our understanding of the evolution of fixed migratory life histories. Furthermore, such studies would be informative for basic understanding of sex related differences in the scheduling of annual events.

In the midst of winter, we observed a second wave of departures from the breeding population (21% of the remaining population; Fig. 1b). Unlike for autumn migration, we found no support for the ‘Arrival Time’ hypothesis for the occurrence of winter movements nor did we find support for the ‘Thermal Tolerance’ or ‘Dominance’ hypotheses. Although we did not find support for any of these hypotheses for winter movements, our sample size might have been too small to detect differences. Given that winter movements typically occurred during periods of more extreme minimum temperatures and snow accumulation, it is likely that restricted access to food and differences in thermal efficiency could contribute to facultative movements away from the sedentary population by some individuals. Unlike for autumn migrants, aerial searches during the winter indicated that a portion of birds that departed during the winter moved less than 40 km from the breeding site before settling for the winter. Autumn migrants were never observed within 300 km of the breeding site between their date of departure in the autumn and their return in the spring. Four individuals that departed during the winter were located within 40 km west to southwest of the breeding site, a well-known general direction of migration used by blackbirds from the area (Schwabl 1983).

Given the differences in the timing and the distances between autumn migration and winter movements, we suggest that the two behaviours are driven by different mechanisms. Although autumn migration in this population resembles obligate migration in its anticipatory preparation before conditions decline such as pre-migratory fattening and a tendency for increased circulating CORT levels, winter movements could be a facultative response to harsh conditions. Escape behaviour by animals, termed ‘The Emergency Life History Stage’, during extreme conditions can act to increase survival (Wingfield et al. 1998). Interestingly, not all birds that departed in the winter moved only short distances. Estimates from geolocators indicate that two recaptured birds that departed during the winter moved >400 km (±84 km) overnight towards the west (latitude could not be estimated). These individuals did not return to the breeding grounds until the following spring. The ‘Environmental Threshold Model of Partial Migration’ posits that migration in partially migrant populations is a quantitative genetic trait with the behaviour of intermediate individuals (between sedentary and migratory) being partly influenced by the environment (Pulido 2011). Given the timing and considerable speed of the winter movements of the two individuals in our study, they might represent an intermediate phenotype as described by ‘The Environmental Threshold Model of Partial Migration’.

By testing hypotheses of partial migration using state of the art tracking techniques to follow individuals throughout their annual cycle and determining physiological measures of the pre-migratory energetic state, this study provides a detailed picture of multiple annual movement strategies within a population. We provide strong evidence for sex differences in obligate migration in the autumn. We also provide preliminary evidence for the co-occurrence of obligate autumn and facultative winter movements within the same population. Future studies which focus on the fitness consequences of the three observed strategies (autumn migration, winter movements, and residency) could provide insights into selective processes that lead to fixed migratory and sedentary life histories. In addition, much could be gained by comparing the energetic state of birds that move in the winter with those that stay through the winter. Finally, we feel that an increased focus on sex differences in migration could provide new insights into the evolution of migratory life histories.

Acknowledgements

We thank Dorris Matthes for help tracking birds. Evi Fricke and Nina Dehnhard provided assistance with lab work. Martin Wikelski provided inspiration and many hours throughout the years of tracking birds from the air. The Max-Planck-Society provided funding for the project. AMF was also supported by the International Max Planck Research School for Organismal Biology. The study was approved by the Baden-Württemberg, Regierungspräsidium Freiburg Abteilung Umwelt (Aktenzeichen 55-8853. 17/0).

Ancillary