What Are the True Costs of Sustained Energy Expenditure During Trans-Pacific Flights?

From a physiological perspective, the nonstop distance traveled by migrating birds is essentially irrelevant; what matters is the duration of continuous flight exercise. This can be explained with a simple thought experiment. A hypothetical migrant with a sufficient (imaginary) tailwind could travel thousands of kilometers in a matter of hours, or conversely with a strong headwind (equal to its maximum airspeed) could fly for many hours or days without traveling anywhere—just as birds do in wind tunnels (Guglielmo 2010). Trans-Pacific flights lasting over a week require having sufficient fuel, maintaining life support, and avoiding catastrophic internal damage. This means: (1) carrying enough energy (mostly fat; Jenni and Jenni-Eiermann 1998); (2) maintaining temperature, water, electrolyte, and nutrient (e.g., glucose) homeostasis (Jenni and Jenni-Eiermann 1998, Gerson and Guglielmo 2011); and (3) avoiding or repairing damage from physical and oxidative processes (Guglielmo et al. 2001, Costantini et al. 2007, Skrip et al. 2015, Dick and Guglielmo 2019). Here we focus on energetics to illustrate that, despite decades of research, we remain unable to satisfactorily explain these feats of endurance.

Satellite telemetry and geolocation devices have shown conclusively that medium to large shorebirds can sustain flights lasting at least 6–9 days (Conklin et al. 2017 and references therein). Based on flight durations and pre- and postflight body masses, Hedenström (2010) calculated that shorebirds like the Bar-tailed Godwit must have incredibly low, time-integrated rates of body mass loss and energy expenditure (chemical power input, P_chem) while in flight. These extremely low rates are about one-half or less of those predicted from allometric equations derived from over 50 years of studies on the energetics of bird flight; similar discrepancies are found for other trans-Pacific migrant shorebirds for which we now have empirical tracking data, with actual maximum flight durations about twice those predicted from the allometric equations (Box 1, Table 1). Thus, this simple physiological approach cannot explain trans-Pacific ultra-endurance flight. Flight energy costs during multi-day flights of shorebirds must be much lower than predicted, and since there is no evidence that shorebirds routinely soar or sail the winds during long flights (Senner et al. 2018), we must look for physiological explanations for their ability to maintain flapping flight for so long.

TABLE 1.

Actual and predicted maximum flight durations of trans-Pacific migrant shorebirds. Flight energy expenditure (P_chem, W) was predicted from the average of preflight and postflight body mass using y = 52.6M_b ^0.74 (M_b in kg, Bishop and Butler 2015). Maximum flight duration was predicted from the difference between preflight and postflight body mass assuming fat and wet protein, respectively, contribute 95% and 5% of energy for flight (0.12 g W^–1). Actual flight P_chem was estimated from the ratio of actual to predicted flight duration and the allometrically predicted P_chem. Ratio is the actual flight duration divided by the theoretically predicted flight duration. Data sources indicated below.

A more sophisticated approach to understanding the energetics of migratory flight is to use highly parameterized simulation models, such as program Flight (Pennycuick 2008). In this model, steady aerodynamic theory is used to predict mechanical power output (P_mech) from physical principles based on body mass, wing size and shape, characteristics of the flight environment, and various assumed parameters. Using default values for assumed parameters, simulations from the program for Great Knots (Calidris tenuirostris) and Bar-tailed Godwits predicted flight durations that were closer to but still a few days shorter than actual flights; however, slightly increasing the value for muscle efficiency or reducing the value for basal metabolic rate (BMR) would eliminate the discrepancies (Box 2). Thus, higher muscle conversion efficiency and lower BMR are two factors that could account for the longer than predicted flight durations in migratory shorebirds (Kvist et al. 2001, Hedh et al. 2020). In fact, we expect strong natural selection for energy efficiency in extreme nonstop migrant birds, which could lead to the evolution of greater muscle conversion efficiency, selective metabolic suppression of noncritical organ systems under extreme conditions, or other yet undescribed physiological strategies. The finding of extreme reductions in the metabolically active body parts of Bar-tailed Godwits starting trans-Pacific migratory flights compared with fueling birds (Piersma et al. 2021a), tallies with this expectation.

The physiological and biochemical adaptations that allow trans-Pacific migrants to successfully budget energy, water, and nutrients to maintain exercise for over a week remain to be discovered. The classic comparative approach using species and populations facing different flight challenges (Feder et al. 2000, Karasov and Martínez del Rio 2007), coupled with empirical physiological measurements of individual shorebirds during these incredible feats of endurance, will be required to solve the puzzle.

How Are Bodies Strategically Restructured Before and During Long-distance Flights?

Odum et al. (1964) argued that, in terms of fueling, migratory birds are like airplanes—structures with a constant shape but a load of fuel (fat) that is variable. However, subsequent analyses of the body composition of shorebirds revealed something quite different: not only the fat loads but also organ and muscle sizes vary greatly between and within seasons (Piersma 1990, Piersma and Jukema 1990, Battley and Piersma 1997, Piersma et al. 1999, Guglielmo and Williams 2003). The discovery of young Bar-tailed Godwits with tiny guts and livers upon southward departure from Alaska came as the clincher concerning restructuring of body components in long-distance migratory birds (Piersma and Gill 1998). The large amounts of subcutaneous and intraperitoneal fat, combined with the relatively small digestive organs (gizzards, livers, kidneys, and guts), equated to the highest relative fat mass (55% of body mass) recorded in birds. In this early study, Piersma and Gill (1998) concluded that “the small size of the nutritional organs of extremely fat Bar-tailed Godwits … is consistent with the suggestion that it is unprofitable and energetically too expensive to carry a digestive machinery over thousands of kilometers of open ocean (Piersma and Lindström 1997).” Sequential sampling has since confirmed that the small digestive machinery follows active atrophy of the organs (Piersma et al. 2021a for Alaska godwits; Landys-Ciannelli et al. 2003 for northward-migrating Bar-tailed Godwits in Europe).

Indeed, changes occur not only during the fueling (Piersma 1998, Battley and Piersma 2005) and flight phases (Battley et al. 2000) but also during the stationary and postflight phases (van Gils et al. 2007, Vézina et al. 2012). Such changes have invited seemingly straightforward interpretations about the importance of a large digestive apparatus to boost nutrient processing during fueling and then growing larger flight muscles, bigger hearts and thicker blood for sustained flight while simultaneously downsizing the digestive tract and other organs less required during the flight itself. Indeed, laboratory studies demonstrated some of the mechanistic processes in such rapid physiological changes, such as how pectoral muscles in Red Knots (Calidris canutus) flown in a wind tunnel change in size relative to body mass (Lindström et al. 2000) and how gizzards in captive Red Knots change in size relative to the quality of the diet (Bijleveld et al. 2014, Mathot et al. 2017).

Although research on temporal changes in organ sizes and fat stores in Pacific migrants has ebbed (but see Hua et al. 2013), there is great scope for further discovery, given the extreme nature of shorebird migration in this geographic arena. For example, in a test of the quantitative prediction of gizzard size of Red Knots from northwest Australia upon arrival in China (Battley et al. 2005), Yang et al. (2013) showed that the main bivalve food along the Chinese coast, Potamocorbula laevis (Zhang et al. 2019b, Piersma et al. 2021b), turned out to be heavy-shelled but easy to crush, unlike bivalves in other parts of the world. Illuminating the limits to adaptive organ size adjustments, Zhang et al. (2019a) demonstrated how severe food quality reductions at a main staging site in China, with a magnitude not reported elsewhere before, were associated with increased gizzard size in Great Knots. Similar comparisons of physiological plasticity on critical intertidal staging areas elsewhere in the Pacific and in other migratory systems may yield insights about limiting factors that shape migratory pathways.

Sensing of Time: Do Birds Have Seasonal Clocks?

Red Knots that were confined to aviaries, in which prolonged flight was not possible, still increased pectoral muscle mass at the appropriate seasons for migration, even without power training (Dietz et al. 1999). This hints that seasonal change in flight muscle in knots is regulated by an endogenous “calendar.” Experiments involving the withholding of timing cues from birds (especially daylight cycles) provide evidence for a capacity of annual time-keeping (see Gwinner 1986 and Helm et al. 2017 for early and recent reviews of this field, respectively). For shorebirds, this was established by the experimental work with Red Knots held under unvarying ambient temperature and photoperiodic conditions (Karagicheva et al. 2016, 2018).

No such experimental work has been carried out in populations crossing the Pacific, but, as a sense of time is a generic trait for all birds (see discussion in Piersma et al. 2005a), such endogenous calendars should surely exist. Indeed, seasonal clocks are likely to underpin the precise individual timing of migratory movements shown in Bar-tailed Godwits (Conklin et al. 2010, 2013), and provide the framework for building an understanding of the timing adjustments found among individual godwits during the nonbreeding season in New Zealand (Conklin et al. 2021). Different seasonal phenotypic traits (e.g., molt, mass gain) within an individual appear driven by clocks of different rigidity (Karagicheva et al. 2016). Evidence for a positive correlation between circannual-period lengths and species-specific adult annual survival rates suggests that long-lived species will benefit from being late chronotypes and initiating breeding at later, relatively safer times (Karagicheva et al. 2018). These findings highlight the considerable scope available for intraspecific and interspecific comparative analyses of calendar traits (both in captivity and in the wild), within and among individuals crossing the Pacific along various migratory pathways, originating at different latitudes with different photoperiodic conditions, and exhibiting different life-history strategies. In this search, the work by Parody-Merino et al. (2019) has suggested that annual-cycle timing is unlikely to be the result of any simple genetic pathway, but rather the sum of many traits plus ontogeny. In addition, seasonal timing may well be influenced by the ways that individual migrants assess the likely upcoming quality of staging habitats along the flyway, in interaction with assessments of the timing of key resources on the breeding grounds (Conklin et al. 2021).

Cognition of Space: How Do Migrants Orient and Navigate During Trans-Pacific Flights?

“Not all those who wander are lost …” (Tolkien 1954). Almost all vertebrate migrants require the cognitive abilities to know the direction in which they are headed (orientation) and how to reach a destination (navigation). Animal cognition includes all the ways in which “animals take in information about the world through the senses, process, retain, and decide to act on it” (Shettleworth 2001). The orientation and navigation mechanisms used by birds, particularly long-distance migratory shorebirds transiting the Pacific, are predicated on a navigational process that requires a map to define position and a compass to tell direction (Abel 1980, 2000; Åkesson et al. 2014, 2017; Mouritsen 2018, Kok et al. 2020).

Migrating for days on end across vast expanses of open ocean where geographic cues for guidance are limited, trans-Pacific migrant shorebirds clearly possess the tools needed for advanced forms of navigation. Many of the tracks of Bar-tailed Godwits (Battley et al. 2012) and Red Knots (Piersma et al. 2021b) followed over the past 15 years were not the shortest, nor the most direct, yet individuals could locate specific destinations (atolls, discrete estuaries, breeding sites) to which they are strongly site-faithful during stationary phases of their annual cycle (Y.-C. Chan et al. personal communication). Individual flight tracks reflect an array of strategies that vary from direct great circle routes (orthodromes) to what appear to be constant compass headings (loxodromes), while others exhibit marked arching tracks that deviate greatly from either orthodrome or loxodrome routes.

Perhaps the strongest evidence that trans-Pacific migrants know their location relative to their destination lies in numerous examples of adaptive wind drift (sensu Chapman et al. 2011), as demonstrated by tracks of Bristle-thighed Curlews, Pacific Golden-Plovers, and Whimbrels (Numenius phaeopus; Figure 3). The most pronounced example of this occurs among Bar-tailed Godwits during their migration from the Yellow Sea to Alaska (Figure 4). Here, godwits timed departures with favorable winds associated with synoptic (i.e. broad-scale weather) events embedded in the North Pacific storm track (Gill et al. 2014). Flying eastward along this storm track with strong tail winds, birds tended to track well south (1,000–2,600 km) of the direct 5,300-km-long route. After availing themselves of advantageous winds for some 4,500 km of their flights, godwits showed abrupt course corrections to reorient northeastward towards their goal.

Godwits also exhibited adaptive wind drift with corresponding course corrections when migrating north from New Zealand to the Yellow Sea and south from Alaska to New Zealand (Figure 4), but the deviations were less pronounced. During the southward flight, most birds left on south to southeast flows and drifted southeast for the initial 1,500–2,500 km of their flights (Gill et al. 2009, 2014). Such flows took godwits laterally up to 1,500 km from an orthodrome route to their goal and into or through the zone (approximately 50–30°N) of Westerlies, further promoting eastward drift. But once under the influence of the southwest-flowing trade winds, they reoriented accordingly and continued on a mostly orthodrome track to New Zealand and eastern Australia. This regular course deviation reflected a consistent behavioral choice in orientation to capitalize on wind drift through predictable alternating zones of winds.

Such pronounced course corrections suggest that birds are using orientation and navigation mechanisms other than the inclination compass, which would only allow them to distinguish between “poleward” and “equatorward” directions (Wiltschko and Wiltschko 1972, Åkesson et al. 2014, 2017). Birds in general also use celestial cues such as the stars, the sun, and patterns of skylight polarization as compass mechanisms (Åkesson et al. 2014, 2017), and many species appear to be able to transfer information from one type of compass to another (e.g., Cochran et al. 2004). Navigational maps are less well understood but may be based on physical features or cues from the Earth's magnetic field, wave and wind patterns, and olfactory and infrasonic stimuli (Alerstam 1976, Åkesson and Hedenström 2007, Hagstrum 2013, Thiebot et al. 2020). Maps can be considered mosaic in nature, such as a series of landscape features or other cues that have become recognized through past experience, or gradient in nature, such as isolines of magnetic field intensity and inclination where they form a bi-coordinate grid (Wallraff 1974, Åkesson et al. 2014).

Hedenström (2010) used the movements of Bar-tailed Godwits (Gill et al. 2009, Battley et al. 2012) for a case study of the likely cues for orientation and navigation used by transoceanic migrants. Given the largely southward route that godwits take from Alaska to nonbreeding areas in New Zealand and eastern Australia, he postulated that godwits could simply be using an inclination compass for orientation except for near the equator, where they would need to use other, possibly celestial, cues. However, an inclination compass would not work for the godwits' other two migratory legs (i.e. New Zealand and Australia nonbreeding to Yellow Sea staging, Yellow Sea staging to Alaska breeding; Battley et al. 2012), during which they would need some form of a bi-coordinate map to guide direction (Åkesson et al. 2014). A geomagnetic map would work, especially if the angular difference between isolines for field intensity and inclination is large, which is the case for the last half of the migration of godwits from New Zealand to the Yellow Sea and the initial migration from the Yellow Sea to Alaska (Boström et al. 2012). However, for the initial southward migration from Alaska to Hawaii (∼3,500 km) the angular differences along the migration corridor are so small that bi-coordinate navigation would be impossible or extremely difficult (Boström et al. 2012:1,044) and birds would likely have to rely on a different navigational tool.

Given that most trans-Pacific migrations involve multi-day flights that span large blocks of latitude and longitude, it is likely that godwits and other trans-Pacific migrants use time-compensated celestial cues (stars and sun compass) for their primary map, as found in numerous other species of migratory birds (Åkesson et al. 2014). Additional information could be gained from wave patterns, which in northern and southern parts of the Pacific are driven by seasonally structured macro-scale wind fields (Crowe 2018). Especially during approaching storms or when birds are engaged in adaptive wind drift (Green et al. 2004, Chapman et al. 2011), the wave patterns could help visual orientation (Alerstam 1976) or yield infrasound cues, likely from both prevailing winds (Figure 2) and those associated with synoptic systems (Hagstrum 2013, Weimerskirch and Prudor 2019, Thiebot et al. 2020, Zeyl et al. 2020). Skylight polarization could also be used (Åkesson et al. 2014), but it remains unclear how such information could be used with a time-compensated sun compass (Muheim et al. 2014).

Trans-Pacific migrants are true navigators and, in fact, almost always behave as if they have a Geographical Positioning System (GPS) on board (cf. Thorup and Holland 2009), the nature of which represents one of the great mysteries in biology (Hedenström 2010, Hedenström and Lindström 2014, Muheim et al. 2014, Mouritsen 2018, Barrie 2019, Kok et al. 2020). Advances in remote-tracking technologies and improved interfaces for integrating migration tracks with relevant environmental factors (e.g., Kemp et al. 2012, Fernandez-Lopez and Schliep 2019) will surely contribute to the unraveling of this mystery. This provides a rich scope for comparison between deviations from expectations during cross-ocean flights and those along geographic leading lines such as coasts and over land, both within and among populations and across species. Such observational studies would provide fertile ground for further physiological and experimental work (i.e. blocking specific sensors, etc.).

How Do Marathon Migrants Cope with (a Lack of) Sleep?

In a review of sleep in flying animals, Rattenborg (2017) concluded, “[I]f for some reason sleep is not possible during flapping flight, then species that primarily rely on this flight mode (e.g., brant, shorebirds, and songbirds) might not sleep on the wing.” A lack of proper sleep is indeed suggested by the observations made on shorebirds after their arrival in New Zealand. Citing Beth Brown (in OSNZ News 37, December 1985), Schwilch et al. (2002) recounted, “[O]n 21 September newly arrived godwits and knots kept separate from those which had overwintered. While the latter fed on the falling tide, two flocks of recent arrivals slept on. Next day, many more had arrived and the pattern was repeated.” Such observations suggest that, after long flights, migrants need not only to refuel but also to catch up on sleep (see also Fuchs et al. 2009, Nemeth 2009). The Red Knots and Bar-tailed Godwits that “slept on” as the intertidal feeding areas became exposed seemed to have prioritized sleep over feeding and vigilance. In this way, they “paid” for catching up on something of which they had apparently incurred a shortage. This would represent an important consequence of the extremely long migrations by trans-Pacific migrants. However, it is also clear that we have overestimated minimum short-term sleep requirements of animals, as demonstrated by weeks-long periods of little or no sleep during the intensely demanding reproductive season of Arctic-nesting shorebirds (Lesku et al. 2012).

It has been commonly proposed that multi-day flights might involve uni-hemispheric sleep, which has been shown to occur in soaring flight (Rattenborg et al. 2016) and is associated in other taxa with some continuous activities, such as swimming (Rattenborg et al. 2017). We expect, however, that the more constant demands of transoceanic powered flight, which include not only near-constant flapping but engagement with critical systems for orientation and weather monitoring, might exceed the capabilities afforded by uni-hemispheric sleep. Thus, transoceanic flights might be another instance of “adaptive sleep loss,” involving no sleep at all for the duration. The recent miniaturization of sleep-recording devices for use on free-living birds is poised to help answer fundamental questions about the needs and processes of sleep during migration. The Pacific theater, given the potential extreme length of sleepless periods even relative to other transoceanic flights, presents us with great research opportunities (Rattenborg et al. 2017, van Hasselt et al. 2020).

How Do Individual Migration Routines Develop Across Time and Space?

As Geertz (1973) profoundly observed, “[W]e all begin with the natural equipment to live a thousand kinds of life but end in the end having lived only one.” What makes one life different from another, even in identical twins? How and why can individuals born from the same parents, or in the same area, develop such a great variety of seasonal routines (Verhoeven et al. 2019, 2021a)? To what extent does ecological context cause this type of variation (Badyaev 2009, Duckworth 2010)?

FIGURE 4.

Wind rose plots for selected points along migration corridors of 24 satellite-tracked (2006–2010) Bar-tailed Godwits (L. lapponica baueri) that show daily mean wind conditions that occurred during each migration period: (A) New Zealand to the Yellow Sea, (B) Yellow Sea to Alaska, and (C) Alaska to New Zealand and eastern Australia (Gill et al. 2014). Plots depict near-surface (10 m above sea level) wind data (Risien and Chelton 2006, 2008; from Climatology of Global Ocean Winds at  http://numbat.coas.oregonstate.edu/cogow/) that have been vector-averaged and spatially smoothed over 0.5° latitude and 0.5° longitude grid cells for all dates between first godwit departures and last arrival at the destination between 1999 and 2009. Colors indicate wind speeds, and the length of the sectors extending radially from the center of each rose reflects the relative frequency of winds from that direction. For example, a sector converging to the center of the rose from the northwest indicates the wind is blowing from the northwest. The scale of wind roses varies for legibility, but open circles around each wind rose are scaled to depict 10% frequency of occurrence. The black arrow emanating from the center of each rose represents the geographic bearing from that location to the destination; the white arrow is the average track bearing of all birds nearest that location. Sample points were selected 300 km past the departure site, 300 km before the destination, and at intervening locations where migratory corridors bisected major wind fields and convergence zones (see Figure 2). Red lines denote great circle routes. Lightly stippled areas define the migration corridors of the tracked godwits (Gill et al. 2014). Map is a Mercator projection. Original illustration of Bar-tailed Godwit by Keith Woodley.

The development of birds from hatchling to fledgling to mature adult is probably best seen as a strongly intertwined web of reciprocal interactions between a developing organism and its environment: “incremental development” (Bateson 2005, Duckworth 2009, 2010, Frankenhuis and Panchanathan 2011, Panchanathan and Frankenhuis 2015). So far, empirical investigations of the mechanisms to help steer young birds (i.e. those not flying in family groups) along a migration track from the area of birth to a nonbreeding area have focused on genetic aspects. These studies provided evidence for the inheritance, or genomic correlates, of a clock for the timing of fueling and flight (e.g., Pulido 2007, Maggini and Bairlein 2010, Pulido and Berthold 2010, Mueller et al. 2011, Bazzi et al. 2015, Saino et al. 2015) and a compass for the direction of migration (Helbig 1991, 1996)—the two mechanisms assembled in an inherited clock-and-compass system (Berthold 1996, Mouritsen 1998). To explain seasonal migration in birds that supposedly receive no parental guidance during their initial migration away from the area of birth, such as in the Common Cuckoo (Cuculus canorus; Willemoes et al. 2014), it was proposed that specific geographic goal areas are genetically inherited (Rabøl 1978, Thorup and Rabøl 2001, Thorup et al. 2003). Yet Weimerskirch et al. (2015), grappling with the interindividual variation of annual routines in Wandering Albatrosses (Diomedea exulans), suggested that, rather than being a heritable character, “the individual strategy probably develops from experience during the long immature stage when birds have to learn foraging skills and locate productive waters to which they return subsequently.” That the well-explored genetic inheritance pathway is insufficient to explain individual patterns of variation in seasonal migration is especially clear for birds migrating in family groups. Among them, geese and cranes may fully rely on social learning of the relevant ecological conditions during the development of their migration routines (Harrison et al. 2010, Mueller et al. 2013).

The multiple migratory flyways that occur within the Pacific Basin provide some unique opportunities to investigate where shorebirds fall along this gradient between inheritance and learning. Precise descriptions of the individual development of a seasonal migration routine by birds of particular populations, including the variation among individuals (Verhoeven et al. 2019) and how juveniles may differ from older birds (Meyburg et al. 2017, Verhoeven et al. 2018), are mostly lacking for shorebirds migrating across the Pacific, with one notable exception. Sharp-tailed Sandpipers (Calidris acuminata) breed on the east Siberian tundra, but only juveniles appear to move east (Handel and Gill 2010) for rapid fueling in southwestern Alaska (Lindström et al. 2011). Most adults appear to leave the Siberian tundra to reach the southern nonbreeding grounds (ephemeral freshwater wetlands in Australia) more directly along a coastal flyway across East Asia (Handel and Gill 2010). The juveniles, which accrue huge fat stores in Alaska (Lindström et al. 2011), reach Australia after a trans-Pacific migration. That juveniles fly separately from adults hints at a “cuckoo-like” story of inherited factors being more important than information learned from experienced conspecifics. How can we find out and would it be different in other species? Battley et al. (2020) showed that the individually consistent timing of northbound migration by Bar-tailed Godwits depends on wintering location and latitude, with juveniles “finding” this location by exploring during early life. Are the first southward migration and subsequent migrations also shaped by social learning and other experiences? For want of in-depth studies, at this point, we turn to some insightful anecdotes.

Consider Alaska Peninsula estuaries, where each autumn three godwit species stage for migration following nesting in Alaska, often in mixed flocks, and where juveniles usually depart after adults (Gill and McCaffery 1999, Ruthrauff et al. 2020, D. R. Ruthrauff and R. E. Gill, Jr., personal observation). The three species have very disparate migration routes, which lead them to nonbreeding areas in New Zealand/Australia (Bar-tailed Godwit; Battley et al. 2012), South America (Hudsonian Godwit Limosa haemastica; Senner et al. 2014), and the Pacific coast of the United States (Marbled Godwit Limosa fedoa; Ruthrauff et al. 2019). Surprisingly, Hudsonian Godwits are vagrants to the Hawaiian Archipelago (Pyle and Pyle 2017) and irregular visitors to eastern Australia and New Zealand, well west of their normal migration corridor. Because their flights to and from South America are similar in distance to those of Bar-tailed Godwits, Hudsonian Godwits staging with flocks of Bar-tailed Godwits should have lipid stores sufficient to accompany their congeners successfully on 10,000-km-long flights across the Pacific. Joining southward flights of Bar-tailed Godwits likely also explains the 4 records of Marbled Godwits in the Hawaiian Archipelago (Pyle and Pyle 2017). Marbled Godwits are medium-distance migrants and reaching Hawaii is probably at the limit of their flight range (∼4,000 km); this would account for the absence of records farther south in Oceania. Scattered reports of both Bar-tailed and Hudsonian godwits within the wintering range of Marbled Godwits (Hamilton et al. 2007; californiabirds.org; oregonbirding.org/oregon-bird-records-committee/) likely reflect similar congeneric associations. Juveniles predominate among autumn records of vagrant godwits for all three species. These occurrences suggest that social learning, especially among young birds, can override an inherited clock- and-compass system. It is easy to imagine the occasional godwit of any of these species becoming integrated into late-staging flocks of congeners. Despite the three godwit species having different migration destinations and routes, the social nature of flocking and the abrupt, en masse departures may be stronger motivators for the onset and orientation of migration than any endogenously programmed stimulus (Helm et al. 2006).

Even if most adult Bar-tailed Godwits and other shorebirds are gone before the juveniles set off, it is possible that a few late adults may provide guidance for some or all of the juvenile flocks. If a successful first migration across the Pacific would indeed require guidance by experienced flock mates, this would put a premium on juveniles being prepared physiologically to migrate with adults. Early-departing young would have much better chances to survive long-distance flights than those departing later after most adults had migrated. Further studies of tagged juveniles in various social groups throughout the season will help elucidate the mechanisms through which individual migratory routines become established.

How Do Individuals Assess and Adjust to the Pacific's Complex Aerosphere?

The importance of winds as a selection factor in bird migration cannot be overstated (Alerstam 1979, 2011, Åkesson and Hedenström 2000, Shamoun-Baranes et al. 2017), and probably nowhere on Earth is the aerosphere as dynamic and yet structured as over the Pacific Basin (Ahrens 2016). At the macro-scale this structure begins with 3 well-defined and essentially closed circulation cells symmetric about the equator, each spanning about 30°of latitude and reaching heights of 10–15 km at the equator but becoming progressively shallower towards the poles. The circulation within and between these cells produces several discrete zonal belts of wind and pressure (e.g., northern and southern hemisphere Westerlies and Trade Winds; Figure 2, after Gill et al. 2014) throughout the lower (<10 km) troposphere, where migration occurs. Within the Pacific Basin, these belts of wind and pressure have greater linearity compared to other regions of the Earth in terms of fluid dynamics, resulting in flows that are considered more predictable by atmospheric scientists (Lee and Kim 2003). But does this predictability hold true for birds as well, and, if so, at what temporal and spatial scales?

Extreme shorebird migrations across the Pacific Basin traverse between 3 and 6 of these zonal wind belts, and success of a nonstop flight is dependent upon an individual's decision as to when and where to fly. Within each zonal belt winds can change (1) daily through convection, especially near the tropics, (2) weekly to monthly through recurring and shifting patterns of pressure and circulation (e.g., Pacific–North American, Western Pacific, Eastern Pacific–North Pacific patterns; subtropical jet stream; and Madden–Julian and Arctic Oscillations), and (3) interannually or decadally in association with phases of large-scale modes of oceanic–atmospheric variability (e.g., Western Pacific Monsoon, El Niño–Southern Oscillation [ENSO], and Pacific Decadal Oscillation [PDO]; Mantua and Hare 2002, Di Lorenzo et al. 2010, Chand and Walsh 2011, Smith et al. 2012, Wang et al. 2012, Feldstein and Franzke 2017).

Atmospheric patterns that affect changes in weather at weekly to monthly time intervals are particularly important for migrating birds, especially the Madden–Julian Oscillation (tropical Pacific; Zhang 2013) and the subtropical jet stream (North Pacific; Lee and Kim 2003). Both of these features affect the location, strength, and structure of broad-scale storms (disturbances), especially in the North Pacific between May and October. The strength of the jet determines the predictability of where and when disturbances will occur (Lee and Kim 2003), but cyclogenesis generally originates in the warm Japanese current and storms gain in intensity as they move eastward across the Pacific into the Gulf of Alaska (Mesquita et al. 2010). Such movement creates patterns of strong general airflow accompanied by troughs and/or ridges and, if birds time their departure from staging areas appropriately, they can capitalize on favorable wind assistance during both eastward (spring) and southward (autumn) migrations (Gill et al. 2014).

Indeed, use of winds by birds has been shown to be highly adaptive in terms of timing between and within seasons (Gill et al. 2009, 2014), routes (Newton 2008, Åkesson and Bianco 2015, Kranstauber et al. 2015), and altitude (Gill et al. 2014, Senner et al. 2019, Lindström et al. 2021), at least to the extent to which birds can predict and discern conditions favorable for long-distance migration. Properly timed decisions will maximize wind profitability and thus reduce energetic costs of migration; conversely, poor decisions or unpredictable weather events will threaten survival (e.g., Watts et al. 2021). Thus, regular long-distance, transoceanic migrants must have the cognitive abilities necessary to assess the variability in atmospheric patterns, evaluate the spatial and temporal extent of these patterns and their associated winds, and incorporate this information into decisions that minimize the risk and optimize the benefits of their annual flights. Studies have shown that Bar-tailed Godwits and other shorebirds do in fact integrate such cues when transiting the Pacific. Not only did satellite-tracked godwits select favorable winds at all their primary departure sites during their annual cycle (Gill et al. 2009, 2014, Conklin and Battley 2011), but the days they selected for departure also conferred the greatest possible overall wind assistance throughout the remainder of their flights (Gill et al. 2014). Such behavior suggests that there exists an unexplained cognitive mechanism that allows godwits and other transPacific migrants to assess changes in weather conditions that are linked (teleconnected) across widely separated atmospheric regions (Gill et al. 2014, Feldstein and Franzke 2017). Landbird and shorebird migrations across large expanses of the Atlantic Basin are likely similarly linked to birds' cognition of atmospherically teleconnected winds (Richardson 1978, Nigam and Baxter 2015, Carneiro et al. 2020, Watts et al. 2021).

To minimize the cost of transport, trans-Pacific migrant shorebirds should also exhibit adaptive flexibility in their response to en route conditions vertically in the air column (Alerstam and Lindström 1990, Hedenström 2010, Hedenström and Åkesson 2017, Senner et al. 2019, Lindström et al. 2021). For Bar-tailed Godwits previously tracked across the Pacific, technology was not yet sufficient to determine altitude of the birds during flight. Models suggested, however, that optimal flight efficiency would have depended significantly upon active shifts in altitude to maximize wind assistance, and that overall net wind assistance would have been obtained at flight altitudes at or above 3,000 m (Gill et al. 2014). Recently, studies by Senner et al. (2018) of Black-tailed Godwits (Limosa limosa) and Lindström et al. (2021) of Great Snipe (Gallinago media) migrating between Europe and Africa confirmed such shifts, with diel changes of over 5,000 m altitude.

The behavioral responsiveness of trans-Pacific migrants to spatial shifting of storm tracks, temporal periods of cyclogenesis, and cues associated with stochastic events at departure sites does, however, appear to leave room for improvement. This was evident from case studies of 2 Bar-tailed Godwits that departed with favorable winds but were overtaken by rapidly developing cyclones in which they flew through prolonged periods of strong headwinds (Gill et al. 2009). In both instances, the birds eventually continued on their migrations but with apparent carryover effects (see also Watts et al. 2021 for Whimbrel in the Atlantic). One northward-migrating godwit did not nest upon reaching the breeding grounds but did successfully migrate back south, while the other southward migrant stopped its migration on a small atoll in the tropical Pacific, 1,600 km short of New Zealand, where its transmitter went silent and the bird presumably died.

Many intriguing questions remain to be answered about how extreme migrant shorebirds choose when to depart and which route to take in navigating vast expanses of ocean. Some birds departed from Alaska when local winds were in opposition over the first few hundred kilometers of their flight, but beyond that they entered long stretches of favorable winds over the North Pacific, resulting in net wind assistance for the entire flight (Figure 4). This implies cognition of not only large-scale seasonal atmospheric patterns but also the variation in synoptic features (e.g., shifting storm tracks) associated with these teleconnected patterns. What are the cues and their limits that govern departure decisions and how is learning involved to fine-tune this complex decision-space? To what extent have the godwits and other extreme endurance migrants buffered the range and duration of their flights (cf. Hedenström 2010) through adjustments of fuel deposition rates, catabolism of protein, water balance, altitudinal wind selection, metabolic costs, and so on? How will projected future changes in atmospheric patterns and their teleconnections affect the frequency and strength of broad-scale weather systems at departure and throughout the flight corridors across the Pacific and over other expansive ecological barriers? The Pacific Basin, by virtue of its complex aerosphere, provides a unique laboratory in which to address such questions for multiple species.

Do Long-distance Migrants Show Reversible State and Other Carry-over Effects?

In a photographic study of uniquely marked Bar-tailed Godwits, Conklin and Battley (2012) showed that individuals arriving late in New Zealand after crossing the Pacific delayed the initiation of wing molt but partially compensated for the delay by increasing the growth rate of the feathers. Delays in arrival (and wing molt) of more than a month did not affect the extent of breeding plumage at northward departure half a year later, nor the timing of this departure. In this example, the timing of southward migration, therefore, carried over to affect the wing molt but not preparations for the subsequent migration and breeding. Carry-over effects denote evidence for circumstances encountered sometime in life affecting subsequent performance (Harrison et al. 2011, O'Connor et al. 2014). These include effects at the scale of the life cycle, such as the circumstances experienced at hatch and differences in the ecological and social contexts during development (Senner et al. 2015), as well as effects occurring at the annual scale, subject to year-to-year variation in environmental conditions (i.e. seasonal interactions; Marra et al. 1998), to which individuals can respond by repeatedly altering phenotypes (Senner et al. 2015). Problems incurred sometime in life may be dissipated so that later performance is not affected (Inger et al. 2010, Legagneux et al. 2012), but earlier life conditions may also lead to a more permanent stratification in higher- and lower-quality individuals (Studds and Marra 2005, Sedinger et al. 2011). To understand this process, the precise developmental origins of individual differences need scrutiny (Senner et al. 2015).

Conklin and Battley (2012) tried to assess the individual adjustments to environmental conditions that would, alone or cumulatively, affect trade-offs and fitness (Senner et al. 2015). Despite these adjustments apparently being common and widespread (Harrison et al. 2011), such “reversible state effects” (defined as the “reversible changes in functional traits resulting from life-history trade-offs during adulthood that affect fitness”; Senner et al. 2015) are surprisingly rare in the extreme migrants of concern here (Conklin and Battley 2012, Senner et al. 2014). To explain this puzzling fact, Conklin et al. (2017) proposed that “severe natural selection, particularly during early life, … quickly removes low-performing individuals and results in populations with a narrow and elevated spectrum of individual quality.” Swift et al. (2020) showed that individual Hudsonian Godwits using intertidal nonbreeding habitats in Chile with high godwit densities had comparably better predeparture body condition than individuals foraging in bays with lower densities, a difference that was associated with higher return rates to the Alaska breeding grounds and higher breeding success. This work, however, did not unambiguously show reversible state effects: such individual variation could still have had developmental origins. The exciting field of organismal cross-seasonal interactions thus calls for more such detailed longitudinal studies of individual performance and development, as well as for comparative studies using the worldwide variation in migration strategies found in species such as the Red Knot and Bar-tailed Godwit (Piersma 2007). Studies of the Pacific migrants, which offer the most extreme cases, will be particularly enlightening.

Can Phylogeography Reveal the Evolution of Trans-Pacific Migrations?

The range of avian migratory behavior forms an unbroken continuum, from sedentary through short-distance migration, to extreme long-jump strategies (Piersma 1987). Migration is thought to evolve through gradual dispersal into regions with different resources and conditions (Salewski and Bruderer 2007, Winger et al. 2019), and may be subject to extension and contraction amid long-term climate oscillations (Zink and Gardner 2017). For this reason, we generally assume that both the need and capacity for extended nonstop flights arose incrementally. Two unique aspects of the Pacific make this an intriguing region to explore the origin of such flights. First, unlike most Arctic regions, Beringia (which includes breeding ranges of all trans-Pacific migratory shorebirds) was largely ice-free during the LGM (Kaufman and Manley 2004), and therefore could have been continuously occupied throughout recent glaciations. Second, southward migration routes from these breeding areas soon encounter at least 3,000 km of land-free ocean expanse. These two aspects mean that, unlike continental migration routes, recent northward expansion of breeding areas was unlikely to contribute to the origin or upscaling of nonstop trans-Pacific flights.

For Bar-tailed Godwits, several conceivable historical scenarios could have led to a direct Alaska–New Zealand flight. Hedenström (2010) envisioned two such scenarios: (1) a continental Palearctic population expanded its breeding and nonbreeding ranges eastward into the Pacific, or (2) an Alaska-breeding population initially migrated to east Asia but gradually shifted its nonbreeding grounds southeastward into Australasia. Both scenarios involve incremental colonization, with each step requiring ever-longer overwater flights and imposing increasing selection for endurance-flight capabilities.

Mayr (1953) invoked a different type of incremental change, arguing that “it seems inconceivable that a migration into or through the central Pacific should have evolved through natural selection if the configuration of land and water had always been exactly the same as today.” He envisioned a Pacific Ocean with many archipelagos, now smaller or totally submerged, that previously provided “stepping stones” across the Pacific, and the disappearance of these fostered the evolution of long nonstop flights along historical migration routes. Thus, endurance-flight capacity arose in situ, rather than through colonization. Mayr (1953) specifically mentioned periods of the Tertiary (>2.6 million years ago) when such stepping stones were probable, predicting an ancient origin for present-day trans-Pacific migrations. However, periods of dramatically lower sea levels, and greatly enlarged land masses in both Beringia and the South Pacific, also occurred during more recent periods of global cooling, such as the LGM, implying that some upscaling of flights could have occurred in just the last 12,000–18,000 years of rising sea levels. This could have additionally involved the gradual southward drift of nonbreeding areas, if land masses of the mid-Pacific were previously more suitable for intertidal specialists such as Bar-tailed Godwits. Alternatively, Newton (2008) proposed an in situ hypothesis that would require no change in the landforms of the Pacific. For example, a population originally migrating from Alaska to Australasia on an indirect, short-hopping route along coastal Asia could have undergone progressive corner-cutting, involving increasingly longer oceanic flights and a gradual abandonment of previous stepping stones, eventually resulting in a single, direct nonstop flight.

These various scenarios all assume that the essential capacity for trans-Pacific flights was a direct product of the geography and environmental conditions of the Pacific theater. However, we see that in some shorebirds, extended nonstop flights occur without any discernible barrier such as oceans or other inhospitable terrain (e.g., Klaassen et al. 2011, Lislevand et al. 2017). In fact, at least nine species make nonstop flights >5,000 km along the largely land-based flyways of the Americas, compared with seven species crossing from Beringia to islands in the Pacific (Conklin et al. 2017). This implies that nonstop flights can be an equivalent or preferable solution even if intermediate options are available. Therefore, it is possible that the extreme nonstop flights required to cross the Pacific were simply “imported” from another region. For example, Hudsonian Godwits make multiple nonstop flights of 6,000–10,000 km on a journey that follows landmasses of the Americas, even well into the continental interior (Senner et al. 2014), overflying wetland habitats used for stopover and wintering by congeneric Marbled Godwits (Olson et al. 2014). One can imagine these birds expanding their breeding or nonbreeding ranges westward and applying their existing long-jump migration strategy to a new transoceanic route. Such an innovation might occur suddenly, by one or a few migratory flocks being blown off course to discover a new and sustainable migration. Here, we require no incremental scenario specific to the Pacific theater because the essential upscaling occurred elsewhere.

Trapped in the present as we are, how can we discern among possible evolutionary hypotheses? For this sort of inquiry, we require tools that reach into the past, such as combining molecular methods with reconstructions of historical climate and habitat conditions (i.e. phylogeography; Avise et al. 1987, Knowles and Maddison 2002). Ongoing developments in genomic methods provide ever-increasing resolution for exploring recent historical processes (Brito and Edwards 2009). Hedenström (2010) mentions that a phylogeographic study would shed light on our question; indeed, if Alaska-breeding baueri Bar-tailed Godwits were a young offshoot of the eastern Palearctic subspecies menzbieri, it would argue for his eastward-expansion scenario, and calculation of a population divergence time might then represent a ceiling on how long godwits have been crossing the Pacific.

However, there are several reasons to expect the question is not so easily answered. As stated earlier, Beringia was unglaciated at the LGM, and perhaps continuously occupied by migratory species for millions of years (McLaughlin et al. 2020). Conversely, this does not necessarily imply that trans-Pacific migration itself is very old or was slow to evolve. Evidence is growing that birds can pioneer new migration routes quickly and have done so quite recently. For example, the Red Knot subspecies rufa and islandica breed in areas of Arctic Canada that were glaciated and unavailable at the LGM; these populations are virtually indistinguishable in neutral genetic variation (Buehler and Baker 2005, Conklin et al. 2022), yet they use different flyways and have a nearly three-fold difference in migration distance (Piersma 2011a). This implies colonization of current breeding areas followed by discovery of a new trans-Atlantic migration route to Europe during just the last few thousand years (Buehler et al. 2006, Conklin et al. 2022).

Phylogeography would be best applied to the evolution of transoceanic flights in a comparative framework, including both widely distributed species that have at least one trans-Pacific migratory population (e.g., Bar-tailed Godwits and Ruddy Turnstones), and those for which the migration might have imposed isolation associated with speciation (e.g., Bristle-thighed Curlew vs. continental Numenius species). Such inquiry should involve genomic approaches coupled with reconstructions of historical climate and habitat conditions (ecological niche modeling; e.g., Wauchope et al. 2017). Assuming that a species' ecological requirements are relatively conserved over time (Peterson et al. 1999), we may then discover how long circumstances associated with contemporary migrations have existed, and when and where similar conditions might have occurred in the past. This may also reveal whether trans-Pacific migration relies on rare, potentially temporary, circumstances for its appearance and persistence, or if it is a routine development, robust to environmental changes and open to discovery by multiple evolutionary paths.

How Have Trans-Pacific Migrations Been Shaped by Cross-hemispheric Food and Predator Landscapes?

Although we may have given a different impression in this review so far, in ecological terms long-distance migration is more about fueling than about flight (Lindström et al. 2019). Migratory birds fuel their migrations by consuming excess food (Lindström 2003, Piersma et al. 2005b, van Gils et al. 2005, Rakhimberdiev et al. 2018) and storing the energy from the food as fat (Jenni and Jenni-Eiermann 1998). Access to predictably abundant, high-quality food resources around the Pacific Rim undoubtedly facilitated the evolution of long-distance migration in the region. For many shorebirds and waterbirds, such resources take the form of benthic invertebrates that are exploited along the region's extensive intertidal margins (MacKinnon et al. 2012). Given the impressive scale of the nonstop avian migrations in the Pacific region, one might assume correspondingly impressive attributes related to the benthic food resources that fuel these migrations. Surprisingly, beyond a comparative study on food availability for Western Sandpipers (Calidris mauri) along the Pacific shores of the Americas (Mathot et al. 2007) and the important discovery that in the Pacific Basin biofilm may be grazed by sandpipers (Kuwae et al. 2008, 2012), there is scant empirical evidence on this topic. Site-specific assessments notwithstanding (e.g., Powers et al. 2002, Battley et al. 2011, Hickey et al. 2015, Mathot et al. 2018), there is no synoptic perspective on the abundance and distribution of benthic invertebrate resources in the Pacific Basin, and a notable paucity of information from certain key sites (e.g., Alaska, eastern Australia, New Zealand) hampers such efforts. Numerous site-specific studies elsewhere have clearly established the connection between the quality and abundance of benthic resources and shorebird distributions in general (e.g., Quaintenne et al. 2010, Piersma 2012, Bijleveld et al. 2016), but such studies are rare in the Pacific Basin (Piersma et al. 1993, but see Mathot et al. 2007). Typically (but not always, see Bom et al. 2018), research focuses sampling efforts only where birds occur (e.g., Ruthrauff et al. 2013, Choi et al. 2017), and often the presence of birds alone is taken as evidence of suitable food resources (e.g., Gill and Handel 1990).

Complicating the assumption that a presence of shorebirds reflects suitable benthic resources, is the role that predators play in shaping the distribution of long-distance migrants throughout their annual cycle (Lank et al. 2003, Ydenberg et al. 2007). In fact, it was studies in the Pacific region, notably on Western Sandpipers, that first articulated how falcon (Falconidae) predators may shape the timing of migration and the use of sites (Ydenberg et al 2004, Pomeroy et al. 2006). Thus, does an absence of foraging shorebirds at seemingly suitable migratory staging sites mean that food resources are not present to support these migrants, or instead that the risk from predation outweighs the benefits accrued while foraging (see discussion in van Gils et al. 2004)? Rich intertidal areas in western Alaska have relatively low predation pressure during autumn staging (Handel and Gill 1992, Lindström et al. 2011), and a notable characteristic of intertidal sites in New Zealand is a lack of avian predators, such as eagles and hawks (Accipitridae) and falcons, that might hunt and disturb shorebirds. An absence of predation pressure would ensure that exhausted Bar-tailed Godwits arriving from Alaska could recuperate and regain body condition in a safe, benign environment, and would also facilitate foraging for fueling up for nonstop flights from New Zealand to the Yellow Sea. Interestingly, individuals of the menzbieri subspecies of Bar-tailed Godwit embark on nonstop flights of 4,000–5,000 km (Battley et al. 2012) despite an abundance of avian predators on the staging area in northwestern Australia. Such contrasts between foraging opportunities, predation danger, and time constraints would provide rich grounds for insightful ecological comparisons.

What Is the Future of Trans-Pacific Migrations Given Rapid Environmental Change?

We now raise our final question: What might the future hold for trans-Pacific shorebird migrations? Conklin et al. (2017) stated that “the most extreme shorebird migrations rely on abundant but difficult-to-access resources; the high minimum individual performance required for survival predicts that degradation of these resource hot-spots will propel rapid population collapse, rather than incremental declines in condition or performance.” Recent large-scale intertidal land claim schemes along the coast of East Asia (Ma et al. 2014, Murray et al. 2014, Melville et al. 2016) are unintentionally providing sobering insights into the theory of “high minimum performance” requirements in Pacific migrants. The reclamation and degradation of sites in the Yellow Sea region correspond with declining vital rates (Conklin et al. 2016, Piersma et al. 2016) and declining numbers (Studds et al. 2017, Wang et al. 2018) of the migratory birds in the region.

The monitoring of long-distance migratory shorebirds and their benthic food resources in the Yellow Sea at Yalu Jiang, China, over a 6-year period, demonstrated a sudden near-complete disappearance of the shorebirds' primary prey resources (Zhang et al. 2018). The surprising finding that the use of the site remained relatively unchanged was interpreted to reflect a lack of alternative sites available to the birds (Zhang et al. 2018). Zhang et al. (2019a) further demonstrated that Great Knots increased the size of their gizzards over this same period to facilitate the consumption of the low-quality, thick-shelled prey that now dominate the site. Thus, long-distance migratory shorebirds to some extent manage to accommodate drastic changes in the abundance or quality of food resources, but such changes may nevertheless come at the expense of decreased survival (Rakhimberdiev et al. 2018).

How can the migrants of the Pacific, and indeed the migratory birds of the world, cope with rapid anthropogenically driven loss and change of staging areas throughout their annual cycles (Iwamura et al. 2013)? Similarly, how will global sea level rise affect the integrity of these critical areas (Murray et al. 2019)? How do birds deal with changes in the temporal and spatial characteristics of their Arctic breeding habitats (Wauchope et al. 2017), in conjunction with simultaneous changes in staging and nonbreeding areas (van Gils et al. 2016, Rakhimberdiev et al. 2018)? Such questions become ever more pressing in the face of unprecedented changes in especially the East Asian part of the Pacific Basin (e.g., Barter 2002, Ma et al. 2014, MacKinnon et al 2012, Murray et al. 2014, Melville et al. 2016, Zhang et al. 2018).