Very rapid long-distance sea crossing by a migratory bird

Landbirds undertaking within-continent migrations have the possibility to stop en route, but most long-distance migrants must also undertake large non-stop sea crossings, the length of which can vary greatly. For shorebirds migrating from Iceland to West Africa, the shortest route would involve one of the longest continuous sea crossings while alternative, mostly overland, routes are available. Using geolocators to track the migration of Icelandic whimbrels (Numenius phaeopus), we show that they can complete a round-trip of 11,000 km making two non-stop sea crossings and flying at speeds of up to 24 m s−1; the fastest recorded for shorebirds flying over the ocean. Although wind support could reduce flight energetic costs, whimbrels faced headwinds up to twice their ground speed, indicating that unfavourable and potentially fatal weather conditions are not uncommon. Such apparently high risk migrations might be more common than previously thought, with potential fitness gains outweighing the costs.

Scientific RepoRts | 6:38154 | DOI: 10.1038/srep38154 Iceland hosts important populations of several migratory wader (or shorebird) species 20 which winter in Europe and west Africa 21,22 . These species have to negotiate one of the longest continuous sea crossing of all Arctic and subarctic breeding landbirds 6 . Species migrating to west Africa (e.g. the Bijagós archipelago in Guinea-Bissau) could either undertake a non-stop oceanic flight (~5800 km) or, after an initial sea crossing (~800 km) to the UK, follow the continental land masses to the wintering grounds (~5200 km). Both alternatives result in similar distances and some Icelandic breeding waders are known to follow the coastline 21 , even those for which a single flight overwater is potentially feasible 23 . The extent to which either of these routes is used is not known and although the non-stop oceanic flight is potentially of higher risk, this will likely depend on wind conditions during migration. We deployed geolocator tags on Icelandic whimbrels (Numenius phaeopus islandicus Brehm, 1831), a species which is known to winter in West Africa 22 to investigate (1) if non-stop flights over oceanic waters between Iceland and West Africa are undertaken during autumn and spring migration; (2) the level of wind support encountered en route and how this affects flight speed.

Results and Discussion
Ten adult breeding Icelandic whimbrels were tagged with geolocators in June 2012 (Fig. 1), seven were recorded on their territories in June 2013, of which four tags with data were retrieved. During the post-nuptial migration in autumn 2012, all four whimbrels flew non-stop to their wintering areas in west Africa (Fig. 2), covering distances of ~3900 to 5500 km in 5 days (Table 1) and, on occasion, achieving the fastest recorded speeds for terrestrial birds on long-distance flight over oceanic waters (up to 18-24 m s −1 ). During the return migration, two of these birds stopped for 11 (Male 2) and 15 days (Female 2), covering a total distance of ~10,500 and 11,000 km, respectively. However, the remaining female and male completed the return migration in another continuous flight (Fig. 2)   Table 1). Maps created using R 3.1.2 using packages ggplot2, ggmap, raster and RgoogleMaps 33 (image data providers: US Dept. of State Geographer © 2016 Google) in WSG 48 coordinate reference system. for a total round trip of ~7800 (Female 1) and 11,000 (Male 1) km, respectively (Table 1). Completing such a long annual migration cycle in two long-distance flights is highly unusual for species where such studies have been undertaken. Only one other long-distance Arctic migrant wader, the Pacific Golden Plover (Pluvialis fulva), has been recorded undertaking such non-stop flights (ca. 9700 km total distance) between Alaska and Hawaii 9 . For this species however, no alternative route over coastal land masses is possible without considerably increasing flight distance.
In order to assess how wind conditions encountered on these journeys varied, particularly for flights with and without stopovers, we quantified wind support at each position just prior to and during the migratory flight. All four whimbrels departed Iceland in favourable wind conditions (i.e. tailwinds), but all four arrived in West Africa having faced headwinds, mostly in the later part of the journey (Fig. 3). Conversely, only one individual departed from the winter grounds in favourable wind conditions (Male 1, the earliest to depart), while all others departed in headwinds of 1.3-4.8 m s −1 , with one individual encountering headwinds for virtually the entire journey, including after stopping-over (Female 2, Fig. 3). A pre-nuptial migration strategy involving a stopover is likely safer regarding potentially unfavourable weather conditions encountered en route and upon arrival, but stopping to refuel will reduce overall migration speed. Indeed, and despite all tracked individuals departing the wintering grounds at similar times (22 nd -29 th of April), those flying non-stop arrived before those that undertook a stopover. This includes the last individual to depart which flew non-stop (Female 1) and arrived at the breeding grounds 6 to 10 days before the two individuals that made a stopover (of 11-15 days), thus overtaking its conspecifics 23 . The earlier arrival of the female and male that flew non-stop to Iceland could be advantageous if they capitalize by nesting early, as this is known to increase breeding success, particularly in Arctic and subarctic systems 24 . However, laying dates did not differ substantially between the individuals undertaking direct flights (20 th -31 st of May) or those that made a stopover (25 th May-7 th of June), suggesting that timing of breeding is constrained by other or additional factors, such as environmental conditions for nesting or timing of mate arrival ( Supplementary Fig. S1).
The very fast ground speeds achieved by migrating whimbrels were influenced by the wind speeds encountered en route, particularly at altitudes of 1500 m ( Table 2). Wind support at this altitude accounts for 4 to 36% of fastest speeds of each individual, with the highest wind assistance corresponding to the maximum recorded ground speed of 24.2 m s −1 (87 km h −1 ). Some individuals also reach very fasts speeds whilst facing headwinds which can be 2 to 40% of their ground speed, and in its most extreme case resulting in airspeed of 25.0 m s −1 (90 km h −1 ; Female 2). Average speeds for the entire continuous migratory flight are similar to those of other species crossing oceans (50-65 km h −1 5, 8,9 ) which are also strongly influenced by wind speed en route 5 . By flying at high speeds and non-stop over open ocean these species reduce the time on migration and might be using an "airspace corridor" to avoid predators, parasites and pathogens 5 . But if wind conditions at distant locations along the route are unfavourable, such migratory strategy can result in mortality by exhaustion, even after arriving at destination (TGG, pers. observation).
Our ability to track migration is providing new insights into the extraordinary capacity of birds to move extremely fast over very large distances, by continuously sustaining endurance exercise during several days. Wind support is crucial during such extreme journeys, but current predictions of changes in climatic patterns, specifically changes on regional scale wind patterns 25 , can potentially have a considerably disproportionate negative effect on those species that regularly undertake non-stop long distance flights over unsuitable habitats. Variation in migratory strategies within the same population will likely allow coping with potential changes, but predicting such responses requires an understanding on how these migration strategies can arise and are maintained. In addition, by linking different migratory strategies to associated fitness consequences will be key in our ability to anticipate demographic changes for migratory populations.     Technology, Ltd.) and "GeoLight" package in R 29 , assuming a sun elevation angle of − 6° based on in situ geolocator calibration prior to deployment. Total migration length, distance (great circle route) and speed were estimated between the last and first positions on land in the breeding areas (Iceland) or wintering areas (W Africa). As geolocator positions are only attained at a minimum of 12 hours intervals, flight speed (time taken to cover the distance between two sequential positions-in m s −1 ) was estimated for each 12 hour flight segment defined as two sequential positions between the first location outside the breeding, wintering or stop-over areas and the first location on land (at breeding, wintering or stop-over areas).
Wind support. Wind data at the location (± 2.5 degrees) and time of each geolocator recorded position was extracted from NOAA using the dataset Reanalysis by NCEP 30 . Headwind or tailwind vector between sequential positions was interpolated for three altitudes (1500, 3000 and 5500 m) using function "NCEP.interp" from package RNCEP 31 . To test at which altitude wind speed had an effect on ground speed, we built a GLMM with individual as random factor to control the non-independence and DFs were calculated using the Satterthwaite approximation in package lmerTest. All analysis and calculations were performed in R 2.15.0 32 .