Migratory network reveals unique spatial-temporal migration dynamics of Dunlin subspecies along the East Asian-Australasian Flyway.

Determining the dynamics of where and when individuals occur is necessary to understand population declines and identify critical areas for populations of conservation concern. However, there are few examples where a spatially and temporally explicit model has been used to evaluate the migratory dynamics of a bird population across its entire annual cycle. We used geolocator-derived migration tracks of 84 Dunlin (Calidris alpina) on the East Asian-Australasian Flyway (EAAF) to construct a migratory network describing annual subspecies-specific migration patterns in space and time. We found that Dunlin subspecies exhibited unique patterns of spatial and temporal flyway use. Spatially, C. a. arcticola predominated in regions along the eastern edge of the flyway (e.g., western Alaska and central Japan), whereas C. a. sakhalina predominated in regions along the western edge of the flyway (e.g., N China and inland China). No individual Dunlin that wintered in Japan also wintered in the Yellow Sea, China seas, or inland China, and vice-versa. However, similar proportions of the 4 subspecies used many of the same regions at the center of the flyway (e.g., N Sakhalin Island and the Yellow Sea). Temporally, Dunlin subspecies staggered their south migrations and exhibited little temporal overlap among subspecies within shared migration regions. In contrast, Dunlin subspecies migrated simultaneously during north migration. South migration was also characterized by individuals stopping more often and for more days than during north migration. Taken together, these spatial-temporal migration dynamics indicate Dunlin subspecies may be differentially affected by regional habitat change and population declines according to where and when they occur. We suggest that the migration dynamics presented here are useful for guiding on-the-ground survey efforts to quantify subspecies' use of specific sites, and to estimate subspecies' population sizes and long-term trends. Such studies would significantly advance our understanding of Dunlin space-time dynamics and the coordination of Dunlin conservation actions across the EAAF.

Introduction

For millions of Arctic-breeding shorebirds, seasonal migrations span thousands of kilometers and present survival risks that can in turn affect future productivity, and thus the growth trajectory of a population [1-3]. Determining the extent to which individuals are spatially and temporally connected to particular flyway areas is, therefore, an important component of coordinated conservation strategies designed to halt or reverse population declines [4-7]. For small Arctic-breeding shorebirds (< 100 g), archival tracking devices facilitate the estimation of population-level migratory connectivity [5, 8] between breeding, migration, and wintering areas, and are a useful tool for studying drivers of population declines [9, 10]. Studies employing archival tracking devices often focus specifically on understanding spatial and temporal connectivity between breeding and wintering areas [9-12]. However, understanding the spatial and temporal dynamics of flyway areas that connect breeding and wintering areas is also important for determining drivers and consequences of population declines [13-15].

Constructing a migratory network from animal tracking data is one approach to assess spatial and temporal connections among flyway areas and the migrants that use them [16-19]. Adapted from network theory, a migratory network combines movement data from multiple individuals to graphically summarize how breeding and nonbreeding areas (i.e., network nodes) are interconnected via immigration and emigration (i.e., network edges; [20]. Network nodes and edges may also be characterized (i.e., weighted) by the quantity of individuals that use them. Once constructed, a migratory network is a powerful framework for evaluating how populations use a flyway in space and time [13, 21], estimating the tradeoffs associated with various conservation actions [6, 7], and helping to understand drivers of population trends [16, 22]. However, there are comparatively few examples where spatially and temporally explicit migratory networks have been constructed to evaluate the migration dynamics of a bird population across its entire annual cycle [e.g., 18, 19, 23].

Study system

The East Asian-Australasian Flyway (EAAF) has the greatest proportion of threatened and near-threatened migratory bird species of any global flyway [24]. At least 60% of Arctic-breeding shorebird populations have declined by up to 8% annually [2, 25], and the available evidence suggests declines are primarily linked to anthropogenic habitat degradation at key stopping and wintering sites in the Yellow Sea [2, 26–28]. However, for many species on the EAAF, prioritizing flyway conservation actions has been difficult due to limited information regarding population migration patterns [29-31].

The Dunlin (Calidris alpina) is a species of sandpiper that spends the winter in coastal and interior wetlands of East Asia and migrates to Arctic and sub-Arctic breeding areas in eastern Russia and northern Alaska. Within the Beringia breeding region, there are 5 recognized subspecies. Each subspecies has their own distinct breeding range (Fig 1), and population sizes vary from ~2,000–500,000 [32, 33]. Dunlin that breed in northern Alaska (C. a. arcticola) appear to have concerningly low adult survival rates (S = 0.54; [34]), and surveys indicate Dunlin populations have likely declined in the Republic of Korea (H.-J. Kim, pers. comm.), the People’s Republic of China (28% from 1996–2014; [35; but see 36]), and Japan (up to 80% from 1975–2008; [37]). However, prioritizing flyway conservation actions for specific populations has been difficult because Dunlin subspecies are visually indistinguishable in the field and, therefore, we have an incomplete understanding of where and when on the EAAF each subspecies occurs [33, 38, 39].

Fig 1

Breeding ranges of the 5 Dunlin subspecies that occur in the North Pacific, and the location of each field site (a–h) where light-level geolocators were deployed. See Table 1 for site info. The pacifica subspecies does not migrate and winter along the East Asian-Australasian Flyway and is not discussed in this paper.

Table 1

Location, subspecies, and number of Dunlin equipped and later recaptured with light-level geolocators at 8 field sites along the East Asian-Australasian Flyway.

Site locations are in Fig 1.

Site id	Field site	Latitude, longitude	Subspecies	Deployment–recapture year	Total deployed	Geolocator model	Total recoveredaC | P
a	Utqiaġvik, Alaska	71.2652, -156.6359	arcticola	2010–11	51	Mk12	15 | 3
				2016–17	46	Intigeo-W65	16 | 0
				2017–18	8	Intigeo-W65	1 | 0
				2018–19	40	Intigeo-W65	10 | 1
b	Ikpikpuk River, Alaska	70.5525, -154.7309	arcticola	2010–11	35	Mk12	4 | 1
c	Canning River, Alaska	70.1180, -145.8506	arcticola	2010–11	22	Mk12	3 | 2
				2016–17	13	Intigeo-W65	3 | 0
d	Chaun Delta, Russia	68.7750, 170.5495	sakhalina	2013–14	35	Intigeo-W65	4 | 6
e	Belyaka Spit, Russia	67.0647, -174.5000	sakhalina	2011–12	10	Mk12	4 | 1
				2013–14	15	Intigeo-W65	5 | 1
				2016–17	14	Intigeo-W65	6 | 1
f	Meinypilgyno, Russia	62.5833, 177.0300	sakhalina	2014–15	5	Intigeo-W65	3 | 0
				2016–17	7	Intigeo-W65	4 | 0
G	Cape Pogodny, Russia	56.2645, 162.5815	kistchinski	2017–18	20	Intigeo-W65	5 | 0
H	Chaivo Bay, Russia	52.5000, 143.2833	actites	2016–17	18	Intigeo-W65	1 | 0

aC = number of tags with complete migration tracks, P = number with partial migration tracks.

By compiling band recoveries, Lagassé et al. [40] provided the first detailed information on the migration patterns of the 4 Dunlin subspecies that migrate and winter along the EAAF (C. a. actites, arcticola, kistchinski, and sakhalina). In their analysis, the authors found that Dunlin in Japan are predominantly C. a. arcticola, while Dunlin migrating and wintering in other areas of the EAAF may comprise all 4 subspecies. The authors also found that continued habitat degradation at key sites in the Yellow Sea would likely have a strong negative effect on all 4 Dunlin subspecies, because 21–50% of subspecific migration recoveries were connected to the region. Although these findings are a useful step toward understanding Dunlin migration dynamics on the EAAF, Lagassé et al. [40] failed to locate any recoveries of the kistchinski subspecies and warned of the likely effects of regionally biased observer effort on regional recovery patterns. The authors were also unable to determine how birds moved between initial capture sites and recovery sites, and consequently, lacked information on the spatial and temporal migration dynamics of the 4 subspecies. These knowledge gaps continue to impede identification of the network of sites visited by Dunlin subspecies, their relative use, and therefore, the prioritization of flyway conservation actions for declining Dunlin populations on the EAAF [7, 38, 41].

Here, we aggregate geolocator-derived migration tracks of 84 Dunlin and construct a migratory network to evaluate the spatial and temporal migration dynamics of the 4 Dunlin subspecies that migrate and winter along the EAAF. Consistent with earlier work on Dunlin migration patterns in the western Palearctic [42, 43], we predicted the distribution of Dunlin subspecies along the EAAF would reflect the geographic distribution of the subspecies on their breeding grounds (i.e., exhibit a parallel migration pattern; Fig 1). For example, we expected C. a. arcticola, which breeds farthest north and east on the flyway (i.e., Alaska), to winter farthest north and east (i.e., Japan). Similarly, we expected C. a. actites, which breeds farthest south and west on the flyway (i.e., Sakhalin Island), to winter farthest south and west (i.e., the South China Sea). Following this pattern, we expected C. a. sakhalina and kistchinski, which have breeding ranges between the 2 other subspecies, to winter in intermediate regions, such as the Yellow Sea and East China Sea, respectively. Finally, following Holmes [44] and Tomkovich [45], we predicted Dunlin subspecies would migrate asynchronously, with subspecies breeding farther north departing/arriving on breeding, migration, and wintering grounds later, and subspecies that breed farther south departing/arriving earlier. We expected to find this pattern because spring phenology is later at northern breeding sites used by the 4 subspecies, and because Dunlin generally breed once favorable conditions become available in the spring and migrate to intertidal staging areas soon after breeding is complete [44, 45].

Methods

Geolocator deployment and recapture

Nests were located at 8 Arctic and sub-Arctic breeding sites using systematic area searches or by rope dragging suitable habitats (Table 1 and Fig 1; [46]). Adults were captured at nests during incubation or while attending broods using a bow net or mist net, respectively [47]. Once captured, we attached a unique metal band, a Darvic leg flag with an affixed light-level geolocator, and a unique combination of color bands to the tibiotarsus and tarsometatarsus. The total weight of the leg flag with affixed geolocator was < 3% of the mean body mass of the smallest subspecies [48]. Although our tags were within suggested weight limits [49], leg-mounted geolocators were found to decrease individual return rates by 43% at 2 of 3 sites where C. a. arcticola were tagged in 2010 (Table 1; [50]). However, the same analysis found return rates were unchanged in 3 other Dunlin subspecies (C. a. hudsonia, pacifica, and schinzii), and the authors suggested that future studies could mitigate impacts of tags by minimizing the use of additional markers [50]. We minimized the use of additional markers after 2010, and tags were recovered from individuals that successfully migrated and exhibited typical breeding behavior the following year (i.e., as they incubated eggs or attended broods); suggesting that carrying a tag likely did not alter the behavior of the birds presented here. Geolocators were placed on both adults of a breeding pair, when possible, to maintain an even ratio of males and females within our tagged populations. Adults were also measured (wing, tarsometatarsus, total head, culmen) and had feathers and/or blood collected for archival purposes [51].

Ethics statement

Permits to capture and tag C. a. arcticola were approved by the U.S. Geological Survey Bird Banding Laboratory (permit 23269, 23566), the U.S. Fish and Wildlife Service (permit MB-085371), the Alaska Department of Fish and Game (permit 10–044, 11–018, 16–111, 17–102, 18–160, 19–154), the North Slope Borough, and the Ukpeaġvik Iñupiat Corporation. Trapping and handling procedures were carried out in accordance with Institutional Animal Care and Use Committee protocols (Bishop’s University BUACC 2009–07; U.S. Fish and Wildlife Service Alaska Region IACUC 2016–005 and 2019–008). Within Russia, permits were not required to capture and tag C. a. kistchinski or C. a. sakhalina. Trapping and handling of C. a. actites was approved by the Russian Federal Service for Supervision of Natural Resources (permit 2016–62, 2017–42).

Geolocator analysis

We used 2 models of light-level geolocators (Mk12, British Antarctic Survey, and Intigeo-W65, Migrate Technology Inc.) across the 9-year study period (2010–2019; Table 1). Geolocators measured ambient light levels every minute and recorded the maximum value across either a 2- (Mk12) or 5-minute (W65) sampling period. Mk12 geolocators truncated the maximum light value on an arbitrary scale from 0–64, and therefore, only detected coarse-scale changes in ambient light levels (e.g., dusk and dawn). In contrast, W65 geolocators did not truncate light values and detected fine-scale changes in ambient light levels throughout each day. Ambient light levels were downloaded from recovered tags and any offset between the geolocator’s internal clock and Greenwich Mean Time (i.e., clock drift) was linearly corrected [52]. We were unable to address potential clock drift for tags that were non-functional upon recovery (16 of 100 tags; Table 1).

Migration tracks were generated from ambient light level readings using the TwGeos (v0.0–1; [53]) and FLightR (v0.4.9; [54]) packages in program R (v3.5.2; [55]). First, we used the findTwilights function to identify the time of each sunrise and sunset (i.e., twilights; [56]). These twilights were defined as the Greenwich Mean Time when light levels crossed an arbitrary threshold of 5 (Mk12) or 12.5 lux (W65). We then used the twilightEdit function to identify and discard incorrect twilight assignments due to periodic shading of the light sensor (e.g., if an adult roosted with the geolocator tucked among its body feathers). A twilight was considered incorrect and discarded if it was > 45 minutes different than the corresponding twilights that occurred in the surrounding 4 days (2 days before and 2 days after; [56]). Second, we calibrated individual geolocators using twilight data collected at a known location before geolocators were deployed or after they were recovered (i.e., “rooftop” calibration), or while geolocators were attached to birds on their breeding grounds (i.e., “in-habitat” calibration; [57]). We preferentially used an in-habitat calibration except at breeding sites north of 66.7˚N where 24-hour sunlight precluded the identification of twilights in the light intensity data. For geolocators that were not calibrated using either method, we used the calibration parameters from a geolocator that was of the same model and manufactured in the same year (e.g., [58]). This calibration step was necessary to calculate tag-specific parameters that described the difference between the theoretical and observed light-levels recorded by each geolocator [59, 60].

Twilight periods and calibration parameters were then used in the state-space hidden Markov model in FlightR to generate twice-daily location estimates. This approach used observed light-levels, an uncorrelated random walk movement model with migratory and sedentary behavioral states, and a spatially explicit behavioral mask to generate the most probable location estimates according to available information on Dunlin migration ecology [59, 60]. Because Dunlin cannot rest or forage in deep water, we parameterized the movement model so that location estimates over land or over ocean were equally likely if ambient light levels indicated an individual was in a migratory state, but were weighted toward land if light levels indicated an individual was in a sedentary state. Parameter values for the movement model and the spatial behavioral mask were the same for all individuals. We used the “on-the-fly” outlier detection algorithm to discard unrealistic location estimates [59]. In total, 10% of possible location estimates were typically discarded per complete migration track (median: 10%, IQR: 8–11%, n = 84). Finally, we used the stationary.migration.summary function to aggregate daily location estimates and estimate where and when a bird was stationary for 2 days or longer (hereafter, stationary estimates; [60]). These stationary estimates include stopover and staging sites [61], and were used in all further analyses. We chose to use FlightR because daily location estimates, and derived stationary estimates, are less affected by tag shading than a traditional threshold analysis [59], and it provides more reliable estimates in the weeks surrounding the vernal and autumnal equinox [62].

After generating an initial migration track, we re-calibrated each geolocator using the median latitude, longitude, and light level data from the bird’s longest estimated stationary site (typically during the boreal winter). We chose to re-calibrate because calibration parameters from a nonbreeding site are better at accounting for environmentally and behaviorally induced noise in light-level recordings at nonbreeding sites than calibration parameters from a rooftop calibration, or while a bird was on its breeding grounds [56, 57]. This re-calibration approach was necessary because the truncated light-level recordings collected by Mk12 tags precluded our ability to use other unknown-location-based calibration methods [e.g., 60, 63]. After re-calibrating, we used the re-calibration parameters to re-estimate the migration track and stationary estimates. Re-estimated migration tracks were typically more precise, having less variability between consecutive location estimates and typically half as many daily location estimates discarded by the outlier detection algorithm (median: 5%, IQR: 4–6%, n = 84). Other than increased track precision, re-estimated migration tracks exhibited no major changes in spatial or temporal itineraries of individual birds (S1 Appendix).

Refining migration tracks

We further refined stationary estimates and subsequent migration tracks using a 3-step process. First, because of the general inaccuracy of solar geolocation [62, 64, 65], we merged all sequential stationary estimates that were < 250 km apart by averaging geographic coordinates and combining arrival/departure dates. We also discarded stationary estimates prior to a bird travelling > 250 km from their known breeding site (i.e., capture/re-capture site). This distance is a conservative buffer for the geographic resolution of stationary estimates returned by FlightR [62] and functions to aggregate routine movements that may occur during a stationary period (e.g., daily movements between roosting and foraging sites; [66]. Second, because solar geolocation performs poorly at high latitudes [67], we discarded stationary estimates prior to a bird travelling south of 66.7˚N. Finally, because FlightR can generate erratic stationary estimates due to noisy light-level data [59], we discarded stationary estimates that had a turning angle of < 60˚ (i.e., locations comprising an angle of < 60˚ between their prior and subsequent stationary estimate; [68] but combined their arrival/departure date with their nearest neighbor. We did not follow this procedure for the stationary estimate that was farthest from the breeding site and had a stationary period ≥ 42 days (S1 Appendix). This approach assumes that an individual migrated without reversing direction until they departed their most distant winter site to migrate north to breed [e.g., 69]. It also assumes that an individual would stop at its farthest winter site for ≥ 42 days, a minimum winter duration supported by prior Dunlin tracking studies [58, 70] and repeat band resightings of Dunlin on the EAAF [71]. Although these assumptions might not be fully met (e.g., a bird could exhibit north-south movements during migration or winter), spatial inaccuracies in estimating latitude preclude finer resolving of the tracks.

Defining migration parameters

To describe the migration ecology of individual Dunlin, we divided each migration track into 4 periods: breeding, south migration, winter, and north migration. We then estimated the following parameters: migration initiation date and arrival date on breeding and wintering grounds, migration duration, minimum migration distance, migration speed (km/day), total number of stationary estimates, and days spent at each stationary estimate. If a bird’s first or last stationary estimate was > 250 km from their breeding site (e.g., due to breeding north of 66.7˚N), their south migration initiation/north migration arrival date was back/forward calculated by dividing the minimum geographic distance between their first/last stationary estimate and their breeding site by an assumed travel rate of 58 kilometers per hour [72]. This analysis was unable to consider post- and pre-breeding stationary periods that occurred north of 66.7˚N or were within 250 km of a bird’s breeding site and, therefore, southward initiation and northward arrival dates are the latest and earliest dates possible, respectively.

South migration began when an individual departed for a stationary estimate that was south of 66.7˚N and > 250 km from its breeding site. South migration ended, and the winter period began, when an individual arrived at a stationary estimate that was south of 45.5˚N and that lasted for ≥ 42 days. We chose 45.5˚N and ≥ 42 days as the migration-winter threshold because Dunlin on the EAAF overwinter at latitudes this far north [40], are generally stationary through the winter [58, 70, 71], and typically stop at migration sites for < 42 days [58, 70, 73]. The winter period ended, and north migration began, when an individual departed for a stationary estimate north of 45.5˚N (i.e., outside of the typical wintering range), or to a stationary estimate(s) south of 45.5˚N and for < 42 days before migrating north of 45.5˚N. North migration ended when an individual arrived at a stationary estimate that was < 250 km from its breeding site or, if a bird’s last stationary estimate was > 250 km from their breeding site (e.g., due to breeding north of 66.7˚N), was forward calculated by estimating the time it would take to travel between its final stationary estimate and its Arctic breeding site (see above). Finally, we defined minimum migration distance as the sum of geographic distances between sequential stationary estimates within south migration, winter, and north migration periods. We did not estimate the distance of the actual route that the bird flew. We estimated migration speed by dividing the minimum migration distance by the total days spent in each period (i.e., migration duration). Our approach to estimating migration speed included stationary fueling periods, and therefore, slower migration speeds by individuals that migrated farther, and/or crossed major ecological barriers (e.g., the Bering Sea), may reflect a non-linear increase in fueling demands [74] and not necessarily individual differences in migration strategy [75]. To avoid potentially misinterpreting differences in migration speed between individuals, we only assessed within-individual changes in migration speed between south and north migration.

Evaluating subspecific migration dynamics

Annual migration tracks were combined across individuals to construct a migratory network map that was used to evaluate subspecies’ spatial and temporal migration dynamics. Partial migration tracks, resulting from geolocators malfunctioning, were included in the characterization of subspecies’ migration ecology (e.g., migration initiation and arrival dates, etc.), but were not included in the migratory network. If a bird was tracked over multiple years, we only used the first year of data.

To construct the migratory network, we first clustered stationary estimates into flyway regions using the partitioning around medoids (pam) function in the raster package (v2.5–8; [76]) in program R. This function required that the number of clusters (i.e., flyway regions) be decided a priori. We optimized the number of clusters by iteratively clustering stationary estimates until the median cluster diameter was < 700 km. We selected a 700 km threshold because resightings of Dunlins carrying geolocators indicated that stationary estimates were typically accurate to within 700 km of an individual’s actual location (n = 5; location error = 45 km, 260 km, 349 km, 1067 km, 1246 km). This approach allowed us to objectively define flyway regions at a fine enough spatial resolution to be biologically relevant but coarse enough to capture much of the geographic uncertainty in stationary estimates [62, 64, 65]. We then determined the proportion of each subspecies that occurred in each flyway region on each ordinal day (i.e., day 1–365, independent of year) during south migration, north migration, and winter periods.

Finally, we evaluated the spatial and temporal migration dynamics of Dunlin subspecies using multiple approaches. First, we used pairwise two-tailed Fisher’s exact tests to statistically compare the proportion of each subspecies that occurred in each flyway region during south migration, north migration, and winter. We limited these analyses to subspecies with data from ≥ 25 individuals (i.e., C. a. arcticola and sakhalina). Second, we used pairwise Wilcoxon rank sum tests to evaluate seasonal differences between subspecies in migration initiation date, arrival date, migration duration, minimum migration distance, migration speed (km/day), total number of stationary estimates, and days spent at each stationary estimate. Here, we included all subspecies with ≥ 5 individuals (i.e., C. a. arcticola, sakhalina, and kistchinski).

Results

In total, 339 geolocators were deployed at 8 sites from 2010 to 2018 (Table 1 and Fig 1). The number of tags deployed ranged from 12 to 145 per site, and 18 to 215 per Dunlin subspecies. One-hundred and seven (32%) tags were recovered, including 84 that recorded complete migration tracks and were used to construct a migratory network, 16 that recorded partial migration tracks and were combined with the above 84 to characterize subspecies’ migration parameters (Table 1), 4 that were excluded because they were from a previously tracked bird, and 3 that failed to collect useable data.

Subspecific migration dynamics

We found that the timing of south migration differed across the subspecies, with C. a. kistchinski (n = 5) initiating south migration 35–51 days earlier than C. a. sakhalina (n = 35) and 48–68 days earlier than C. a. arcticola (n = 59; Wilcoxon rank-sum tests: 95% CI reported above; Table 2). Similarly, C. a. sakhalina initiated south migration 10–20 days earlier than C. a. arcticola (Table 2). This staggered migration pattern is consistent with our prediction that Dunlin breeding at lower latitudes would initiate south migration earlier than those breeding at higher latitudes (Table 2 and Fig 1). This pattern was violated, however, by the 1 tracked C. a. actites, which bred farthest south but migrated later than the other subspecies (Table 2). Subspecies’ south migration distances and durations were also similarly staggered (Table 2). The 1 tracked C. a. actites had the shortest migration distance and duration (Table 2).

Table 2

South migration characteristics for each subspecies of Dunlin on the East Asian-Australasian Flyway.

Reported is the median value and interquartile range.

Subspecies	arcticola	sakhalina	kistchinski	actites	Wilcoxon rank-sum pairwise comparisonsa
South migration	n = 59	n = 35	n = 5	n = 1	a	b	c
Initiation	31 Aug (22 Aug–9 Sep)	15 Aug (12 Aug–23 Aug)	5 Jul (29 Jun–9 Jul)	4 Sep	< 0.01	< 0.01	< 0.01
Winter arrival	2 Nov (28 Oct–13 Nov)	18 Sep (10 Sep–2 Oct)	19 Jul (16 Jul–3 Aug)	8 Sep	< 0.01	< 0.01	< 0.01
Total duration (days)	65 (51–76)	32 (27–47)	20 (15–21)	4	< 0.01	< 0.01	0.01
Distance (km)	7,067 (6,433–7,777)	5,038 (4,786–5,684)	4,164 (4,110–4,420)	2,359	< 0.01	< 0.01	0.01
Number of stationary estimates	4 (3–5)	3 (3–5)	1 (1–2)	0	0.14	< 0.01	< 0.01
Stationary estimate duration (days)	8 (4–17)	7 (5–11)	9 (6–13)	–	0.16	0.98	0.60
Speed (km/day)	109 (97–137)	155 (115–185)	250 (196–278)	590	–	–	–

a“a” indicates a pair-wise comparison between arcticola and sakhalina; “b”: arcticola and kistchinski; “c”: sakhalina and kistchinski. No statistical comparisons were made with the actites subspecies due to a low sample size. P-values are reported.

We identified 12 flyway regions that were used only during migration (i.e., migration regions); 2 of the 12 (regions 24 and 25 in W Alaska; Fig 2) were used only during south migration. Regions in W Alaska (regions 24 and 25) and the NW Sea of Okhotsk (region 18) supported a greater number of C. a. arcticola (27–56%; n = 52) during south migration than C. a. sakhalina (0–12%; n = 26; pairwise Fisher’s exact tests: p = < 0.01–0.01), whereas the NE Sea of Okhotsk (regions 19 and 20) supported a greater number of C. a. sakhalina (42% & 69%) than C. a. arcticola (10% & 10%, respectively; p = < 0.01; Fig 2). In contrast, the majority of tagged C. a. arcticola (56%), sakhalina (77%), and kistchinski (100%) occurred in N Sakhalin Island (region 15; Fig 2). This is also the only region where C. a. actites are known to breed (Fig 1). Although, during south migration C. a. arcticola, kistchinski, and sakhalina occurred in many of the same regions (Fig 2), they generally did not occur in the same region at the same time (Fig 3). Differences in subspecies’ south migration initiation dates and durations (see above) were reflected in the subspecies having staggered peak passage dates across migration regions (e.g., region 15, Fig 3), and were consistent with staggered arrival dates on the wintering grounds (Table 2).

Fig 2

Migratory network depicting south migration movements made by Dunlin subspecies along the East Asian–Australasian Flyway, and (D) south migration stationary estimates color-coded by flyway region.

Network nodes and edges are weighted by the proportion of individuals that stopped in each flyway region, and the proportion that migrated between flyway regions, respectively.

Fig 3

The proportion of each Dunlin subspecies that occurred in each flyway region along the East Asian-Australasian Flyway by day of year.

See Fig 2 for the location of each region.

During the winter period, Dunlin were generally stationary (Table 3), but C. a. kistchinski (n = 5) spent 41–88 more days on its wintering grounds than C. a. sakhalina (n = 26), and 82–112 more days than C. a. arcticola (n = 52; Wilcoxon rank-sum tests: 95% CI reported above; Table 3). Similarly, C. a. sakhalina spent 26–46 more days on its wintering grounds than C. a. arcticola (Table 3). However, we found that C. a. arcticola were typically more stationary than C. a. sakhalina; flying 528–1,455 fewer km, having 1–2 fewer stationary estimates, and spending 14–67 more days at each stationary estimate (Table 3).

Table 3

Winter characteristics for each subspecies of Dunlin on the East Asian-Australasian Flyway.

Reported is the median value and interquartile range.

Subspecies	arcticola	sakhalina	kistchinski	actites	Wilcoxon rank-sum pairwise comparisonsa
Winter	n = 52	n = 26	n = 5	n = 1	a	b	c
Total duration (days)	192 (179–200)	230 (202–242)	289 (277–292)	206	< 0.01	< 0.01	< 0.01
Distance (km)	0 (0–721)	1,403 (536–1,953)	349 (0–1,321)	1,744	< 0.01	0.29	0.29
Number of stationary estimates	1 (1–2)	3 (2–3)	2 (1–2)	2	< 0.01	0.36	0.28
Stationary estimate duration (days)	128 (56–182)	66 (49–103)	102 (81–202)	47 & 158	< 0.01	0.57	0.02
Speed (km/day)	0 (0–4)	7 (2–9)	1 (0–5)	8	–	–	–

a“a” indicates a pair-wise comparison between arcticola and sakhalina; “b”: arcticola and kistchinski; “c”: sakhalina and kistchinski. No statistical comparisons were made with the actites subspecies due to a low sample size. P-values are reported.

We identified 13 flyway regions that were used during the winter (i.e., winter regions). We also found that the S Korean Peninsula and central Japan (region 11 and 12, respectively) supported more C. a. arcticola (15% & 14%) than C. a. sakhalina (0% & 0%; pairwise Fisher’s exact tests: p = 0.05 & 0.09, respectively), whereas inland China, the NW Yellow Sea, and N China (regions 5–7, respectively) supported more C. a. sakhalina (19–50%) than C. a. arcticola (0–19%; p = < 0.01–0.01; Fig 4). However, many winter regions supported similar proportions of the subspecies (Fig 4). For example, when assessing major geographic regions, we found that the Yellow Sea (regions 4, 6, 9–11) supported the majority of tagged C. a. kistchinski (60%), arcticola (63%), and sakhalina (92%; including the 1 C. a. actites). The East and South China seas (regions 1–3) also supported many C. a. sakhalina (31%), arcticola (48%), and kistchinski (100%; including the 1 C. a. actites; Fig 4 and S1 Dataset). No individual Dunlin that wintered in Japan also wintered in the Yellow Sea, China seas, or inland China, and vice-versa (Fig 4). Where subspecies used the same winter region, they generally occurred in the region at the same time (Fig 3).

Fig 4

Migratory network depicting winter movements made by Dunlin subspecies along the East Asian–Australasian Flyway, and (D) winter stationary estimates color-coded by flyway region.

Network nodes and edges are weighted by the proportion of individuals that stopped in each flyway region, and the proportion that migrated between flyway regions, respectively.

We found that north migration parameters were similar across the Dunlin subspecies. For example, C. a. kistchinski, sakhalina, and arcticola had similar north migration initiation dates and durations (Table 4). However, like south migration, subspecies migrated different distances, consistent with the minimum distances required to travel between winter regions and subspecific breeding sites (Tables 2 and 4). We also found that individuals differed in how they migrated during north versus south migration. Individuals migrating north typically traveled 1.7–5.4x faster, had 0.2–0.7x as many stationary estimates, and spent 0.3–1.1x the number of days at each stationary estimate than when they migrated south (IQR reported above, n = 83; Tables 2 and 4). However, the 1 tracked C. a. actites showed the opposite pattern during north migration, migrating slower and stopping more often and for longer durations than during south migration (Tables 2 and 4).

Table 4

North migration characteristics for each subspecies of Dunlin on the East Asian-Australasian Flyway.

Reported is the median value and interquartile range.

Subspecies	arcticola	sakhalina	kistchinski	actites	Wilcoxon rank-sum pairwise comparisonsa
North migration	n = 52	n = 26	n = 5	n = 1	A	b	c
Initiation	17 May (7 May–19 May)	17 May (29 Apr–21 May)	16 May(19 Apr–22 May)	2 Apr	0.84	1.00	0.87
Breeding arrival	30 May (27 May–3 Jun)	30 May (27 May–2 Jun)	27 May (25 May–29 May)	18 May	0.53	0.05	0.10
Total duration (days)	13 (10–25)	13(10–31)	9 (5–40)	46	0.83	0.58	0.61
Distance (km)	6,308 (5,806–6,897)	5,152 (4,944–5,616)	4,612 (4,562–4,705)	4,254	< 0.01	< 0.01	< 0.01
Number of stationary estimates	2 (1–2)	2 (1–3)	1 (1–2)	2	0.70	0.39	0.34
Stationary estimate duration (days)	5 (3–9)	5 (3–10)	3 (3–25)	11 & 33	0.96	0.53	0.47
Speed (km/day)	432 (270–615)	400 (172–494)	542 (112–922)	92	–	–	–

a“a” indicates a pair-wise comparison between arcticola and sakhalina; “b”: arcticola and kistchinski; “c”: sakhalina and kistchinski. No statistical comparisons were made with the actites subspecies due to a low sample size. P-values are reported.

We identified 10 migration regions that were used during north migration. Across the 10 regions, C. a. arcticola and sakhalina occurred in similar proportions (Fig 5), had similar peak passage dates (e.g., region 15; Fig 3), and had similar arrival dates on their breeding grounds (Table 4). Over the entire nonbreeding period, Dunlin subspecies exhibited the highest degree of spatial and temporal overlap during north migration, a moderate degree during the winter, and the lowest degree during south migration (Tables 2–4 and Figs 2–5).

Fig 5

Migratory network depicting north migration movements made by Dunlin subspecies along the East Asian–Australasian Flyway, and (D) north migration stationary estimates color-coded by flyway region.

Network nodes and edges are weighted by the proportion of individuals that stopped in each flyway region, and the proportion that migrated between flyway regions, respectively.

Discussion

Patterns of spatial and temporal flyway use can have a profound effect on individual fitness, and thereby, where and when a population experiences decline [13-15]. For example, the timing and degree to which Dunlin subspecies use an area may affect individuals’ access to optimal foraging conditions [77], exposure to predation pressure [78, 79], subsequent reproductive success [1, 77], and survival rates [1]. Our migratory network provided an informative framework for objectively delineating flyway regions and describing population-specific migration patterns in space and time. We found that Dunlin subspecies used many of the same core flyway regions (e.g., the East China Sea, Yellow Sea, and N Sakhalin Island; Figs 2–5), but that C. a. arcticola and sakhalina segregated along edge flyway regions; with C. a. arcticola occurring more along the eastern edge of the flyway (e.g., migrating and wintering in western Alaska and central Japan) and C. a. sakhalina occurring more along the western edge of the flyway (e.g., wintering in N China and inland China; Figs 2 and 4). No individual Dunlin that wintered in Japan also wintered in the Yellow Sea, China seas, or inland China, and vice-versa (Fig 4). This apparent east-west divide, combined with evidence that C. a. arcticola exhibit strong interannual site fidelity to specific wintering sites in Japan and elsewhere [40], suggests that wintering-ground effects on population survival rates likely operate independently among C. a. arcticola that winter in Japan, and among Dunlin that winter elsewhere. Future efforts to compare seasonal survival rates of C. a. arcticola that winter in Japan to those of Dunlin that winter outside of Japan could enable researchers to better parse the many factors potentially driving C. a. arcticola population declines [26, 27, 80, 81].

During south migration, we found that C. a. arcticola, sakhalina, and kistchinski staggered their migration initiation and winter arrival dates; the southernmost (kistchinski) and northernmost (arcticola) breeding subspecies migrating first and last, respectively (Table 2 and Fig 1). Although we expected this pattern to result from earlier breeding phenology at lower latitudes [44, 45], it may also reflect differing strategies for how subspecies minimize their exposure to predation by Peregrine Falcons (Falco peregrinus); C. a. kistchinski timed their south migration before Peregrine Falcons typically migrate south [82], and C. a. arcticola migrated after (e.g., that seen in Western Sandpiper [Calidris mauri] and Dunlin on the East Pacific Flyway, [78]; Table 2). Alternatively, subspecific differences in south migration timing may reflect differing strategies for where and when subspecies undergo flight feather molt; C. a. arcticola and sakhalina undergo flight feather molt on their breeding grounds (mid-June to late August; [45, 83], whereas, C. a. kistchinski initiate flight feather molt in mid-July (on their breeding [45] or nonbreeding grounds [84] and complete it on China Sea and Yellow Sea nonbreeding grounds (e.g., that seen in Dunlin in the western Palearctic, [85]; Table 2, [84]. If the south migration initiation dates we observed are widespread among C. a. kistchinski, efforts to identify and conserve important China Sea and Yellow Sea molting sites may be more critical to the persistence of this subspecies than previously recognized [33, 86].

During the winter, we found that C. a. sakhalina flew farther, stopped more, and spent fewer days at each stop than C. a. arcticola (Table 3). The more mobile wintering behavior exhibited by C. a. sakhalina likely reflects the subspecies unique use of freshwater wetlands in inland China (region 5; Fig 4) and annual hydrological dynamics in the Yangtze River Floodplain (YRF), where most Dunlin in inland China occur [87]. Indeed, we found that 46% of tagged C. a. sakhalina wintered in inland China. All arrived after monsoonal flood waters typically recede and reveal abundant shorebird habitats (i.e., October), and all generally departed before the spring rains (i.e., April; Fig 3, [88, 89]); requiring individuals make additional migrations between inland China and alternate wintering regions in the fall and spring (e.g., that seen in C. a. pacifica on the East Pacific Flyway, [90]; S1 Dataset). In addition to flying farther and stopping more, C. a. sakhalina likely comprises much of the > 45,000 Dunlin that occur in inland China [87], and is likely the subspecies most threatened by habitat degradation/loss from human modifications to the natural hydrological regime of the YRF (e.g., the Three Gorges Dam; [88, 89, 91]. However, many waterbodies in the YRF are connected through a series of sluices, and therefore, coordination of hydrological management actions that optimize wetland habitat quality and seasonal availability could significantly support C. a. sakhalina populations in the region [92].

During north migration, we found that C. a. arcticola, sakhalina, and kistchinski departed their wintering sites and arrived at their breeding sites on similar dates (Table 4), despite differences in their migration distances (Table 4) and breeding phenologies [44, 45]. However, arrival dates for C. a. arcticola and sakhalina that bred north of 66.7˚N may have been 3–10 days later than estimated [93] because we were unable to identify pre-breeding stationary periods in areas with 24-hour sunlight (see above). Nonetheless, the similar departure dates suggest that Dunlin subspecies use similar social and/or environmental cues, such as annual changes in day length [94], to time their north migrations.

Limitations and future directions

The migration patterns presented here are for adult Dunlin that returned to the same breeding site and were recaptured in a following year. Therefore, our results do not include the migration patterns of birds that died, juveniles, or adults that emigrated from their original capture site. Uneven sampling effort across subspecies and breeding sites is another limitation of our findings. For example, the migration patterns we found for C. a. actites and kistchinski are from 1 and 5 individuals, respectively, and each subspecies was only studied at a single breeding site (Table 1). Lastly, the geographic uncertainty associated with geolocator-derived location estimates [62, 64, 65] required coarse spatial and temporal interpretation [56].

Combining migration tracking data with on-the-ground survey techniques may be an effective approach to refine our understanding of subspecies’ migration patterns and to overcome the limitations of our findings. For example, C. a. actites is classified as vulnerable under the International Union for Conservation of Nature’s regional Red List criteria, due to its small population size [39]. Despite conservation concerns, targeted conservation efforts have not been possible because, until recently, little was known of the subspecies’ migration dynamics [39, 40]. Our tracking data and previously published band recoveries [40, 95] indicate the subspecies migrates through the Yellow and East China seas and primarily winters in the South China Sea (Fig 3). Due to C. a. actites’ genetic and morphological distinctness among Dunlin subspecies [96, 97], it is possible to combine capture and sampling methodologies with flock counts [e.g., 43, 98] to estimate how many C. a. actites likely occur at particular South China Sea sites, and at sites in other regions where C. a. actites occur [40, 95]. Collectively, such efforts may significantly advance our understanding of C. a. actites space-time dynamics and our ability to implement conservation actions for this vulnerable subspecies.

Combining migration tracking data with on-the-ground survey techniques may also be an effective approach to estimate subspecies’ population sizes and long-term trends. For example, we found that subspecific migration phenologies were generally asynchronous during south migration, and that the pattern of asynchrony was consistent across migration regions (Fig 3 and S1 Table). By understanding how subspecies migrate in temporally distinct waves, on-the-ground survey efforts (e.g., daily flock counts) may be combined with migration tracking data (e.g., individual turnover rates, population peak passage dates) to estimate the number of individuals of each subspecies’ that use a particular site [99, 100]. Such survey efforts may also be applied across years and across key regions to estimate each subspecies’ population size [87, 101] and population trends [2, 102, 103]. Such a monitoring design could provide specific information necessary to inform more comprehensive conservation plans for Dunlin on the EAAF [33, 38, 39].

Conclusion

Our migratory network, constructed using geolocator-derived migration tracks of individual Dunlin, provided an informative framework for objectively delineating flyway regions and describing population-specific migration patterns in space and time. We found Dunlin subspecies exhibited unique patterns of spatial and temporal flyway use on the EAAF. Spatially, C. a. arcticola predominated in regions along the eastern edge of the flyway (e.g., western Alaska and central Japan), whereas C. a. sakhalina predominated in regions along the western edge of the flyway (e.g., N China and inland China; Figs 2 and 4). No individual Dunlin that wintered in Japan also wintered in the Yellow Sea, China seas, or inland China, and vice-versa (Fig 4). However, similar proportions of the 4 subspecies used many of the same regions at the center of the flyway (e.g., N Sakhalin Island and the Yellow Sea; Figs 2–5). Temporally, Dunlin subspecies staggered their south migrations and exhibited little temporal overlap among subspecies within shared migration regions (Table 2 and Fig 3). In contrast, Dunlin subspecies migrated simultaneously during north migration (Table 4 and Fig 3). South migration was also characterized by individuals stopping more often and taking more days to complete their migration (Table 2) than during north migration (Table 4). Taken together, these spatial-temporal migration dynamics indicate that Dunlin subspecies may be differentially affected by regional habitat change and population declines according to where and when they occur. By understanding how subspecies migrate south in temporally distinct waves (S1 Table), we suggest on-the-ground survey efforts (e.g., daily flock counts) may be combined with migration tracking data (e.g., individual turnover rates, population peak passage dates) to estimate the number of individuals of each subspecies’ that use a particular site [99, 100]. Such survey efforts may also be applied across years and across key regions to estimate subspecies’ population sizes [87, 101] and long-term trends [2, 102, 103]. Such studies would significantly advance our understanding of Dunlin space-time dynamics and the coordination of Dunlin conservation actions across the EAAF.

Light-level geolocator analyses.

Annotated R code of steps taken to generate geolocator-derived stationary estimates and refine Dunlin migration tracks along the East Asian-Australasian Flyway.

(HTML)

Click here for additional data file.

Light-level geolocator location data.

Geolocator-derived stationary estimates comprising 100 Dunlin migration tracks along the East Asian-Australasian Flyway. A ReadMe tab is included to help guide the user.

(XLS)

Click here for additional data file.

South migration timing of Dunlin subspecies by migration region.

(PDF)

Click here for additional data file.

(TIF)

Click here for additional data file.

(TIF)

Click here for additional data file.

(TIF)

Click here for additional data file.

2 May 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Peng Chen, Ph.D.

Academic Editor

PLOS ONE

Journal Requirements:

1. When submitting your revision, we need you to address these additional requirements.

Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. Thank you for stating in your Funding Statement:

(This work received financial and/or logistical support from the American Ornithological Society–Alexander Wetmore Research Award, NGO Amur–Ussuri Center for Avian Biodiversity (Vladivostok, Russia), Arctic Landscape Conservation Cooperative, Arctic Shorebird Demographics Network, BirdsRussia, British Petroleum Exploration (Alaska) Inc., Bureau of Land Management, Calvin J. Lensink Fund, Kinross Gold Corporation–Kupol Mine, The MacArthur Foundation, Manomet Inc., National Fish and Wildlife Foundation, Neotropical Migratory Bird Conservation Act Grant Program, The Nuttall Ornithological Club–Blake-Nuttall Fund Grant, The Trust for Mutual Understanding, University of Alaska Fairbanks, University of Colorado Denver, University of Missouri Colombia, U.S. Fish and Wildlife Service (Avian Influenza Program, Migratory Bird Management Division, National Wildlife Refuge Challenge Cost Share Program, National Wildlife Refuge Division), Wildlife Conservation Society–Arctic Beringia Regional Program, and the Wilson Ornithological Society–Paul A. Stewart Grant. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.)

Please provide an amended statement that declares *all* the funding or sources of support (whether external or internal to your organization) received during this study, as detailed online in our guide for authors at http://journals.plos.org/plosone/s/submit-now.  Please also include the statement “There was no additional external funding received for this study.” in your updated Funding Statement.

Please include your amended Funding Statement within your cover letter. We will change the online submission form on your behalf.

3. Thank you for stating the following in the Acknowledgments Section of your manuscript:

(Please include your amended statements within your cover letter; we will change the online submission form on your behalf.

This work received financial and/or logistical support from the American Ornithological Society–Alexander Wetmore Research Award, NGO Amur–Ussuri Center for Avian Biodiversity (Vladivostok, Russia), Arctic Landscape Conservation Cooperative, Arctic Shorebird Demographics Network, BirdsRussia, British Petroleum Exploration (Alaska) Inc., Bureau of Land Management, Calvin J. Lensink Fund, Kinross Gold Corporation–Kupol Mine, The MacArthur Foundation, Manomet Inc., National Fish and Wildlife Foundation, Neotropical Migratory Bird Conservation Act Grant Program, The Nuttall Ornithological Club–Blake-Nuttall Fund Grant, The Trust for Mutual Understanding, University of Alaska Fairbanks, University of Colorado Denver, University of Missouri Colombia, U.S. Fish and Wildlife Service (Avian Influenza Program, Migratory Bird Management Division, National Wildlife Refuge Challenge Cost Share Program, National Wildlife Refuge Division), Wildlife Conservation Society–Arctic Beringia Program, and the Wilson Ornithological Society–Paul A. Stewart Grant. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.)

We note that you have provided funding information that is not currently declared in your Funding Statement. However, funding information should not appear in the Acknowledgments section or other areas of your manuscript. We will only publish funding information present in the Funding Statement section of the online submission form.

Please remove any funding-related text from the manuscript and let us know how you would like to update your Funding Statement. Currently, your Funding Statement reads as follows:

(This work received financial and/or logistical support from the American Ornithological Society–Alexander Wetmore Research Award, NGO Amur–Ussuri Center for Avian Biodiversity (Vladivostok, Russia), Arctic Landscape Conservation Cooperative, Arctic Shorebird Demographics Network, BirdsRussia, British Petroleum Exploration (Alaska) Inc., Bureau of Land Management, Calvin J. Lensink Fund, Kinross Gold Corporation–Kupol Mine, The MacArthur Foundation, Manomet Inc., National Fish and Wildlife Foundation, Neotropical Migratory Bird Conservation Act Grant Program, The Nuttall Ornithological Club–Blake-Nuttall Fund Grant, The Trust for Mutual Understanding, University of Alaska Fairbanks, University of Colorado Denver, University of Missouri Colombia, U.S. Fish and Wildlife Service (Avian Influenza Program, Migratory Bird Management Division, National Wildlife Refuge Challenge Cost Share Program, National Wildlife Refuge Division), Wildlife Conservation Society–Arctic Beringia Regional Program, and the Wilson Ornithological Society–Paul A. Stewart Grant. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.)

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Partly

Reviewer #3: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

Reviewer #3: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Thanks for sending me this interesting manuscript PONE-D-22-05779 by Benjamin Lagasse et al. titled “Flyway network model reveals unique spatial-temporal migration dynamics of Dunlin

subspecies along the East Asian-Australasian Flyway”. I am glad to see the data from this amazing long-term collaborative tracking effort being analyzed and written up. The authors summarized the migration characteristics of four Dunlin subspecies along the flyway. Despite being a common species along the flyway, we know very little about the migration pattern among different subspecies, especially when observing them in the non-breeding grounds. This study also provides important insights in monitoring and conservation effort and I added a few comments below which could be used to improve the manuscript.

General comments:

1. I agree with the authors that this work is useful for guiding on-the-ground survey efforts. It would be even more useful if the authors can be more specific here, either in the main text or as supplementary materials, to state when is the good time for a particular region to conduct its Dunlin survey (probably the time when only one subspecies occur?) to achieve long-term monitoring goals.

2. Line 468 – does this mean no sakhalina occured in the two regions? It's unclear if this percentage represent the subspecies composition within one region, or regional composition of one subspecies. I believe it's the latter, and if that's the case, then the presentation using region as the subject in a sentence to report the percentage is confusing (Sorry if I get this wrong). The same applies to the rest of this paragraph, it would help to clarify this at the beginning of this paragraph.

3. Line 565 – The departure date seems to be different to some published records (https://cms.hkbws.org.hk/cms/component/phocadownload/category/25-reports-of-shorebird-monitoring) and my personal observation in southern China such as Leizhou Peninsula. Dunlin in southern China 'disappear' in March or even earlier. It would be good to mention this phenomenon and the potential reasons behind the difference in those records with this geolocator study. Base on the definitions used in this study and the resolution of geolocator data, could a short move from southern China to the Yellow Sea be captured in the analysis?

4. Lines 633-635 – note that this seems to contradict to what was said Lines 484-487.

Sorry if I misunderstood something here.

Specific comments:

Lines 77: This sentence appears in the manuscript several times but please consider reword it a bit to explain clearly what was meant to say. An individual can 'winter' only in one place isnt'? Perhaps 'No individual Dunlin demonstrated movement between ... within the same winter'?

Lines 104, 109: Consider ‘population decline’

Line 131: instead of 'migration sites', consider 'stopping and wintering sites'

Line 132: Consider citing Zhang et al's work, which showed a different way of degradation than the three references cited (doi:10.1017/S0959270917000430)

Line 140: please also report the rates

Line 142: instead of M. Tian pers. comm, now there is published record for this “a 28.4% decline in the Chinese Yellow Sea within about a decade” (doi.org/10.1016/j.biocon.2022.109547)

Line 171: is there a name for this pattern? Parallel migration?

Line 178: Should reword "which breed between the 2 other subspecies" as this sounds a bit strange.

Lines 533, 534: Perhaps ‘times’ instead of ‘timing’?

Reviewer #2: First, I would like to congratulate the authors presenting a comprehensive analysis of the migration routes and timing and wintering movements of several subspecies of Dunlins in the East Asian-Australasian Flyway (EAAF). This is a key contribution towards conservation of this species in the EAAF.

One thing that really puzzles me is the title and framing of the manuscript as presenting a ‘flyway network model’, since I could not find any network analyses being presented in the manuscript. For a manuscript about a network model, I would expect the calculation of network metrices such as relative node strength and betweenness centrality (see Jacoby and Freeman 2016 for a good overview of methods applicable in movement ecology). In the second paragraph of the Introduction, several papers that applied network theory to evaluate patterns of migratory connectivity, identifying important sites, etc. were cited. Therefore, I am surprised not finding the analyses.

Unless the authors would like to conduct additional analyses based on network theory, I would suggest framing the manuscript differently. The results presented are valuable to understand the migration ecology of this species and highly informative in its conservation. I suggest deleting the ‘flyway network model’ in the title. For example, ‘Migration patterns of four Dunlin subspecies along the East Asian-Australasian Flyway’ would be a more appropriate title. Accordingly, the Introduction would need to be heavily revised, as well as other parts of the manuscript.

Further comments:

Line 72: delete ‘to construct a flyway network model’ and rewrite

Line 104: I don’t get why ‘however’ is used. There are some studies that focuses on spatial connectivity, but there are also a lot of studies about both spatial and temporal aspects of migration.

Line 126: The subtitle ‘study system’ in the Introduction is not necessary.

Line 154: It is good to mention results of Lagasse et al. (2020a), however too many details are included. This paragraph can be shortened substantially and combined with the previous one. For example, you can try to summarise it in one sentence starting with, ‘Recent analysis of band recoveries reveals…’

Line 166: I think there needs to be more elaboration on why knowledge on migration patterns is informative in developing flyway conservation actions. You can introduce the importance of identifying stopover sites here, as this is a main result of the manuscript but not much is said about it in the Introduction.

Line 169: The authors should rethink what is the main aim and objectives of this manuscript and revise this paragraph accordingly.

Line 263: Instead of stating that Dunlin is a ‘terrestrial species’, it is better to specify that Dunlins cannot rest or forage in deep water.

Line 267: ‘same for all Dunlin’: say ‘same for all individuals’ instead.

Line 271: which R package is this ‘stationary.migration.summary’ function from?

Line 273: Please elaborate the biological meaning of ‘stationary estimates’, either here or at another part of the manuscript. This is used a lot in the Results and it will help interpretation of your findings to link it to more common terms, such as a stopover site or staging site.

Line 298-300: I don’t understand this sentence, what exactly are you trading-off?

Line 314: I do not understand this procedure. Why do you discard stationary estimates that had a turning angle of <60 degrees? If you assume an individual migrated without reversing direction (line 319), then you should discard points that have a large turning angle instead? I could not find this procedure described in the reference given - Edelhoff et al. 2016 (line 316). I wonder how discarding these stationary estimates would affect the resulting flyway regions.

Line 363: It is hard to understand the phrase ‘to avoid potentially misinterpreting differences in migration speed between individuals’ – what types of misinterpretation are you avoiding? I think it made sense to look at within-individual changes in speed between north and south migration, maybe just rewrite the reasoning.

Line 368: I do not think the analyses described here is about constructing a flyway network model. It is just a method to cluster the stationary estimates.

Line 375: In this paragraph, you described how ‘flyway region’ is extracted, but ‘flyway region’ is an uncommon term and you should explain what it means in terms of stopover ecology. I imagine it could be a cluster of stopover sites since a ‘normal’ stopover site would have a much smaller diameter.

Line 413: do you know the sex of this C. a. actites individual? Perhaps it is the sex that takes care of the young. The timing differences between the other three populations that you found could also be due to differences in breeding success and the sexes that were tracked. Do you know the sex of the birds and whether some tagged birds had nest failures? Failed breeders might leave for southward migration earlier. I think an analysis that takes these two factors into account is necessary to really say something about differences between the subspecies. Hatching success should be able to be determined from the geolocator data by extracting the incubation period (e.g. Verhoeven et al. 2020).

Fig 2: This is not a flyway network model. It does not describe the relationships between the flyway regions statistically or mathematically (e.g. based on graph theory) other than plotting a line between them. The caption would be fine if you start with ‘General (A) south migration…’.

Also, the part ‘migrating and wintering birds’ in the sentence ‘see methods for determining migration and wintering regions…’ seems redundant.

Among figure 2A to D, I find only figure 2D is informative. The circles/triangles/squares in Fig A to C are quite misleading as the actual stationary estimates that each shape represents covers a much larger area, according to Fig 2D. Instead of these diagrams, I would really like to see the migration tracks used by different subspecies of Dunlin plotted with a different colour to show the differences and overlaps between subspecies’ migration routes (the colours can be slightly transparent to show overlaps). Such a map will illustrate what you described in line 512-519.

Line 497: To be consistent with the wording, should it be ‘flyway regions’ instead of ‘migration regions’?

Line 532: do you have any information on migration phenology of Peregrine Falcons in the EAAF? If not, there is really no support for this guess. Even Lank et al. (2003) have presented alternative hypotheses that are not danger-based. I suggest changing this sentence to reflect that there are several hypotheses (not mutually exclusive) that can explain these patterns.

Line 574-578: The logic of this sentence does not make sense. Your results agree with the general predictions that migrants used a ‘time minimization strategy’ during northward migration and ‘energy minimization strategy’ during southward migration (Alerstam and Lindström 1990). But this strategy itself does not lead to optimal timing of arrival, which is achieved by migrating at the right time, e.g. by using environmental cues that predict breeding ground food availabilities. Also, the strategy does not lead to higher stress levels. The papers you cited at line 577-578 are about species that are facing habitat degradation during northward migration. These negative impacts are not expected in situations with good stopover habitat quality.

Line 581: In this paragraph you should also discuss that the southward migration timing could be a result of breeding success and the sex of the individuals tracked.

Line 612: if the subspecies are not distinguishable from each other during field observations, it seems impossible to estimate, from flock counts and tracking data, the number of individuals of each subsepecies that use a particular site. To achieve that would also require knowing the population size of each subspecies and the proportion of the population that uses a particular stopover site. The latter might be able to be derive from tracking data, but that would require a very large sample size.

Line 614, 639: To estimate population size and trends, the obvious method would be counting at wintering regions. I wonder how that could be done at stopover sites (as suggested in these lines) where several subspecies co-occur and you cannot visually distinguish them from each other. If you have a valid method in mind, please elaborate more.

Line 638: It would be better to specify which part of the results is useful in guiding on-the-ground survey efforts. I think spatially it is not really useful as the area indicated (including the uncertainty of 200 km) is way too large. Only the timing of when Dunlins occur at a specific area would be useful. From the point of view of somebody doing ground surveys, it is quite difficult to derive the timing from Fig. 4. It will be much easier if the range of timing is provided in a supplementary table.

References:

Alerstam, T., and Å. Lindström. 1990. Optimal Bird Migration: The Relative Importance of Time, Energy, and Safety. In Bird Migration, 331–51. Springer Berlin Heidelberg.

Jacoby, D. M., & Freeman, R. 2016. Emerging network-based tools in movement ecology. Trends in Ecology & Evolution, 31(4), 301-314.

Lank, David B., Robert W. Butler, John Ireland, and Ronald C. Ydenberg. 2003. Effects of Predation Danger on Migration Strategies of Sandpipers. Oikos 103 (2): 303–19.

Verhoeven, M.A., Loonstra, A.H.J., McBride, A.D., Macias, P., Kaspersma, W., Hooijmeijer, J.C.E.W., van der Velde, E., Both, C., Senner, N.R. and Piersma, T. (2020), Geolocators lead to better measures of timing and renesting in black-tailed godwits and reveal the bias of traditional observational methods. J Avian Biol, 51

Reviewer #3: In their study, Lagasse and colleagues provide a flyway network model based on tracking data from different breeding populations and subspecies of Dunlins migration through the East Asian-Australasian Flyway. Using the network approach, the authors characterized migration strategies (routes, phenology) and compared the different breeding populations and subspecies. First, I would like to congratulate all authors on collecting such an impressive collection of tracks from this species and for putting them together allowing to compare the subspecies and reveal some of the larger scale patterns of migration in the EAAF. The paper is very well written and easy to understand. The methods include one of the most elaborate and transparent description of light level geolocation I have seen and the work the authors put into the track estimates allowing robust conclusions. While reading the paper, I of course though that information on population dynamics would make the study so much more valuable. I am aware that that information is not yet available (and that there are plans already installed in will likely be installed in the future to get these urgently needed information). The lack of these data is discussed in the “Limitations and future directions” section, and I am sure that the network developed in this study can be used on the future and may even can inform effective monitoring plans.

In my opinion, the paper is ready to be published. If I can suggest a slight improvement, it would be to also put some of the values presented in the tables into figures (e.g. departure/arrival/wintering duration etc. for the different populations). I think that this would make it easier to see the differences/communalities. I got a (tiny) but frustrated while trying to extract and understand the dates/periods and picture them myself in the head. I ended up visualizing them for myself to get a better impression.

Two additional very minor comments:

L95: Sounds like migration only presents risks. Maybe add that migrations evolved due to the opportunity to exploit optimal conditions across regions.

L124: What about Morrick et al. 2022 Conservation Science and Practice

Again, congratulations on the data collection and the synthesizing analysis across populations. I am sure that this study is of interest to the migration/shorebird community and that the methods described will be appreciated and repeated in future studies.

Simeon Lisovski,

Alfred Wegener Institute, Potsdam

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: Yes: Chi-Yeung Choi

Reviewer #2: No

Reviewer #3: Yes: Simeon Lisovski

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

17 Jun 2022

See attached cover letter and response to reviewers

Submitted filename: Response to Reviewers.pdf

Click here for additional data file.

22 Jun 2022

Migratory network reveals unique spatial-temporal migration dynamics of Dunlin subspecies along the East Asian-Australasian Flyway

PONE-D-22-05779R1

Dear Dr. Lagasse,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Peng Chen, Ph.D.

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

27 Jul 2022

PONE-D-22-05779R1

Migratory network reveals unique spatial-temporal migration dynamics of Dunlin subspecies along the East Asian-Australasian Flyway

Dear Dr. Lagassé:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Peng Chen

Academic Editor

PLOS ONE