Seasonal contrasts in individual consistency of oriental honey buzzards' migration

Abstract

Individual consistency in migration can shine light on the mechanisms of migration. Most studies have reported that birds are more consistent in the timing than in the routes or stopover sites during migration, but some specialist species showed the opposite patterns, being more consistent in spatial than temporal aspects of migration. One possible explanation for this contrast is that specialists rely on particular food or habitat resources, which restrict the migratory routes they can take, leading to high spatial consistency. If this is the case, the effect of specialist foraging should become apparent only when birds forage, instead of fasting and flying continuously. To test this effect, we analysed individual consistency in migration of the oriental honey buzzard (Pernis ptilorhynchus), a specialist raptor that feeds on honeybees and wasps, using a long-term tracking dataset. As honey buzzards make extended stopovers during which they forage in spring but not in autumn, the spatial consistency should be higher in spring than in autumn. Honey buzzards were highly consistent in both their migratory routes and stopover sites in Southeast Asia, but only during spring migration. Our results highlight an important link between species' migratory consistency and foraging ecology.

1. Introduction

Individual consistency, or how repeatable individual animals are, in various aspects of migration (e.g. [1–9]) helps to unveil the mechanisms of long-distance migration. As different mechanisms would result in varying patterns in migratory consistency, individual histories of migration would help in interpreting the roles of genetic variation and learning in shaping these patterns [2,10,11].

Most studies reporting the repeatability of migration found that birds are more repeatable in the timing of their migration than in their routes or stopovers (e.g. [3,4,6,7]). There were, however, a couple of exceptions including ospreys (Pandion haliaetus; [8]) and great reed warblers (Acrocephalus arundinaceus; [9]), both specialists relying on a particular prey or habitat (respectively, fish and reed beds). These two species were more repeatable in their routes and stopover sites than in the timing of migration. It has been suggested that one reason why these birds were more consistent spatially than temporally is that they have to travel through a specific type of habitat that may be scarce but widespread [8,9]. If this is the case, however, this pattern should only hold when birds forage, and not when they fast and travel non-stop to their destinations.

That specialist foraging constrains migration routes is consistent with the emerging consensus that foraging is intimately linked to various aspects of migration [12,13]. To address this link, we investigated the spatial and temporal consistency of migration in the oriental honey buzzard (Pernis ptilorhynchus), a specialist raptor foraging on honeybees and wasps, using a long-term tracking dataset. Oriental honey buzzards that breed in Japan migrate through East and Southeast Asia to spend winter in the Malay archipelago [14,15]. Honey buzzards rarely stopover during their autumn migration, but in spring, make stopovers that can last for weeks before reaching the Korean Peninsula [14,15]. This asymmetry in stopovers between the autumn and spring migrations provides an opportunity to test how foraging specialism affects the consistency of the migratory route. If specialist species, such as the honey buzzard, are constrained in their migratory routes and stopovers by their foraging requirements, then the spatial aspects of honey buzzard migration, such as the migratory route, should be more consistent in spring than in autumn. In addition, within the spring migrations, the degree of spatial consistency should be more pronounced around the stopover sites, at which birds are considered to forage for extended periods. To test these hypotheses, we examined whether the honey buzzard, a specialist feeder, is more consistent in spatial aspects in spring than in autumn, and if so, whether the high consistency corresponds to birds' stopover sites.

2. Material and methods

(a). Tracking dataset

To estimate individual consistency, we used long-term tracking data from 30 wild oriental honey buzzards (11 females; 19 males; all adults) with at least two journeys in the same season (autumn/spring). Each bird was tagged with a satellite transmitter (Solar Argos by North Star Science and Technology, LLC; Solar Argos or Solar Argos/GPS by Microwave Telemetry, Inc.) using harnesses with Teflon ribbons (for technical details, see [14]). Each tag plus harness weighed between 9.5 and 30 g (0.7−3.2% of the bird's body weight) and was tracked by the Argos system in Toulouse, France [16]. To exclude off-track data points, we processed the Argos data using the Douglas Argos filter [17]. The locations of 11 out of 30 tags were also calculated by the GPS (errors: ± 18 m [18]). For these, we used only the high-quality GPS-based location data for the analyses, as GPS and Argos data from the same birds showed similar results for crossing (see below), but GPS offered more frequent data points (mean sampling intervals for GPS and Argos: 5.17 and 14.55 h).

(b). Identifying longitudes and dates of latitude crossings

To calculate temporal and spatial consistency in migration, we identified the dates and longitudes on which the birds crossed the latitudes 0, 5, 10, 15, 20, 25 and 30° N [3]. This area encompasses three major regions through which the honey buzzards travel in both autumn and spring: the Malay Peninsula, Southeast Asia and Mainland China (figure 1). This area was chosen for data analyses to include data from as many individuals as possible, regardless of differences in their breeding/wintering sites and the seasonal difference in migratory routes. As for the crossings of 0, 5 and 10° N, we only included those that occurred on southern Thailand, the Malay Peninsula or Sumatra area, excluding crossings on Borneo and the Philippines on the way to different wintering grounds (figure 1). When birds crossed the focal latitude multiple times within the same season as they meandered in the area, we used the last crossing (i.e. when birds actually left the relevant area) for the analyses. To estimate the repeatability, we included birds that had two or more data points for crossing each latitude. As there were cases where the data points immediately before and after the latitude crossing were weeks apart, we excluded cases in which two data points before and after the crossing were more than 72 h apart, based on data distribution (for the sample size, see figure 2).

Figure 1.

Oriental honey buzzards' migratory routes in autumn and spring. Different colours indicate migratory routes by individuals. See electronic supplementary material, S2 for details of migration by ID115581 and 40729.

Figure 2.

Repeatability (r) estimates (electronic supplementary material, S3) of: (a) longitudes and (b) dates on which birds crossed 30, 25, 20, 15, 10, 5 and 0° N during autumn (aut) and spring (spr) migrations, and (c) longitudes and latitudes of centroid of stopover sites, and arrival and departure dates of stopovers. Bars show 95% confidence intervals. Asterisks indicate non-significant repeatabilities. (Online version in colour.)

(c). Identifying stopovers

To estimate consistency in honey buzzards' stopovers, we defined stopovers as days on which birds travelled less than 50 km between 0 and 30° N for longer than 7 days. To calculate daily travel distance, we chose a single data point per day with the best location accuracy and the earliest sampling time. We then calculated time difference and distance between two neighbouring points on the WGS84 ellipsoid, using the ‘difftime()' function and the ‘distGeo()' function in the ‘geosphere' package [19]. The daily travel speed was calculated by dividing the distance by the time difference. When there were days on which no location data were available during a stopover, the two points before and after those days were considered to belong to the same stopover if the distance between the two was less than 50 km. We included birds that had stopover data for two or more years into the subsequent analyses. Six out of 62 cases were additional stopovers by four birds: they had two stopovers, one in the area where they stayed in other years, and another in a distinct area, within the same year (electronic supplementary material, S1). For these, we used the stopovers in the area that they used in other years.

(d). Consistency in latitude crossings and stopovers

To quantify consistency in the latitude crossings and the stopovers, we first converted the dates of crossings and stopovers into Julian days (i.e. 1 January = 1). We then estimated and tested the significance of individual consistency (at α = 0.05), in either: the Julian day or the longitude of the latitude crossings, the arrival/departure Julian day of the stopovers, or the latitude/longitude of the centroids of the stopover sites, using the ‘rpt()' function in the ‘rptR’ package [20]. We used R [21] for all data analyses. Data for latitude crossings and stopovers are available via Dryad [22].

3. Results

(a). Consistency in longitudes and dates of latitude crossings

The most repeatable sections of the migration, in terms of the difference in longitude between years, differed between the spring and autumn migrations. The spring migration showed high repeatability at higher latitudes, peaking around 15–25° N, while the repeatability of the autumn migration became higher at lower latitudes close to their wintering grounds (figure 2a). The timing of the migration was significantly repeatable at all latitudes across years within each season (figure 2b). The timing of the spring migration appeared to be slightly more repeatable than that of the autumn migration, in particular at 20° N, but the differences were not as distinctive as in the repeatability estimates for longitudes.

(b). Consistency in stopover sites

The honey buzzards almost exclusively made substantial stopovers during the spring migration (61 out of 62 stopovers; electronic supplementary material, S1). The locations of these stopovers were exceptionally repeatable (figure 2c), showing that birds precisely returned to specific sites year after year. Individual stopover sites were spread between 8.59 and 23.69° N (mean ± s.d.: 17.13 ± 4.61° N), centring on the most repeatable sections of the spring migration (15−20° N; figure 2a). The timing of stopovers was also significantly repeatable (figure 2c), spanning between February and May (one stopover in October was excluded from the analyses; electronic supplementary material, S1).

4. Discussion

The most repeatable sections of oriental honey buzzards' migratory routes differed between seasons, with the spring migration being most consistent across years around 15–25° N in Southeast Asia, whereas the autumn migration was most consistent around 0–5° N in the Malay archipelago. Honey buzzards made long stopovers almost exclusively in spring. The locations of stopover sites spread across Southeast Asia, and were exceptionally consistent between years. Despite these differences in the repeatability of routes between seasons, the consistency in timing of migration did not show clear seasonal differences. The dates on which birds crossed different latitudes was repeatable across all regions within season. The arrival/departure dates of stopovers were also significantly repeatable.

Similar to other specialist species [8,9], oriental honey buzzards were more consistent in the routes and stopovers than in the timing of their migration, but only in spring (figure 2). This could be because of various factors. The lower repeatability in the early part of the autumn migration route may be owing to ocean crossing. In autumn, honey buzzards cross the 650 km-wide East China Sea from western Japan to reach eastern China (figure 1) in strong tail winds [23]. While crossing the ocean, birds may be susceptible to stochastically changing wind directions, and have no landscape features that they can use to follow specific routes. This combination of wind conditions and the lack of landmarks on the ocean may decrease spatial consistency in higher latitudes in autumn. At lower latitudes, in autumn as well as in spring, birds do not cross any large barrier (figure 1; [14]), and thus can land at any time and wait for better wind conditions. This may give birds better control over where they go, resulting in higher consistency.

The high spatial repeatability in Southeast Asia in spring may be owing to highly consistent spring stopovers in the area. Honey buzzards select stopover sites that contain suitable features for local honeybees, including the giant honeybee (Apis dorsata; [24,25]). While the high repeatability implies the general stability of local prey abundance in the stopover sites, interestingly, birds retained some degree of flexibility in stopovers, as birds made multiple stopovers in a handful of cases (electronic supplementary material, S1). This could be owing to birds responding to fluctuating food resources and competition with conspecifics. For instance, when birds made multiple stopovers within a migratory trip, each stop was shorter than the same bird's single stopover in other years (electronic supplementary material, S1), suggesting that birds made the second stopovers seeking for additional foraging opportunities. This flexibility, together with the high repeatability of individuals' stopover sites, suggests that honey buzzards might learn the precise location of the stopover sites, although further behavioural and environmental data are needed to understand the mechanisms involved in site selection.

Feeding specialists, including oriental honey buzzards, may rely on resources that might emerge in specific environments at specific times during migration [26], unlike generalists. Thus in prenuptial migration, specialists not only need to arrive at breeding sites on time but have to meet their requirement to forage on specific resources. These differences may lead to different developmental processes of migration between specialists and generalists. Specialists, for instance, may establish their highly consistent stopover sites early in life, by exploring and finding a productive site, and later refine the migratory route to the site through learning [2]. On the other hand, generalist migrants like black kites (Milvus migrans) might be less restricted to specific routes and stopover sites, but need to develop other abilities to time their departure appropriately, or to respond to weather conditions [7]. Large-scale datasets have started producing important insights into the effects of foraging ecology on migration, elucidating the mechanisms of migration [12] and demonstrating the resilience of migrant species [13]. To further our understanding of this link between foraging and movement ecology in migrants, it is crucial to gather long-term data about behaviour, and about spatial distribution of food across migratory routes.

Ethics

All fieldwork was licensed by Aomori and Yamagata Prefectural Governments in Japan (2003–2016).

Data accessibility

Data (Input data and R scripts) available from the Dryad Digital Repository at: https://doi.org/10.5061/dryad.1g1hh58 [22].

Authors' contributions

S.S. designed and carried out the analyses, and drafted the manuscript. H.H. provided the data, helped designing and interpreting the analyses, and edited the manuscript. Both authors agree to be held accountable for the content herein and approve the final version of the manuscript.

Competing interests

We declare we have no competing interests.

Funding

This project was supported by the Ministry of the Environment in Japan, and the Japan Society for the Promotion of Science Overseas Research Fellowship (to S.S.: H28/1018).