Shorebirds have experienced a precipitous reduction in abundance over the past four decades. While some threats to shorebirds are widespread (e.g., habitat alteration), others are regional and may affect specific populations. Lesser Yellowlegs (Tringa flavipes) are long-distance migrants that breed across the North American boreal biome and have declined in abundance by 60–80% since the 1970s. The documented harvest of Lesser Yellowlegs in the Caribbean and northeastern South America during southward migration is a possible limiting factor for the species, but it is unknown to what extent birds from different breeding origins may be affected. To address the question of differential occurrence in harvest zones during southward migration, we used PinPoint GPS Argos transmitters to track the southward migrations of 85 adult Lesser Yellowlegs from across the species' breeding range and 80° of longitude from Anchorage, Alaska, USA, to the Mingan Archipelago, Quebec, Canada. We classified migratory locations as inside or outside three zones with high levels of harvest (Caribbean, coastal Guianas, and coastal Brazil) and then fit generalized additive mixed models to estimate the probability of occurrence of Lesser Yellowlegs in harvest zones according to their breeding origin. Individuals from the Eastern Canada population had a higher probability of occurrence within one or more harvest zones and remained in those zones longer than individuals breeding in Alaska and western Canada. Linear regressions also suggested that longitude of the breeding origin is an important predictor of occurrence in harvest zones during southward migration. Lastly, our findings, combined with other sources of evidence, suggest that current estimated harvest rates may exceed sustainable limits for Lesser Yellowlegs, which warrants further investigation.
LAY SUMMARY
The Lesser Yellowlegs is experiencing a steep population decline. Addressing longstanding knowledge gaps, such as the potential impact that unregulated harvest has on specific breeding populations, helps scientists and managers develop and implement effective conservation actions for this vulnerable species.
Using GPS telemetry, we tracked the southward migration of 85 adult Lesser Yellowlegs across the Western Hemisphere to establish the specific populations that migrate through areas with high harvest.
Lesser Yellowlegs originating from the eastern part of their migratory range were more likely than western-breeding birds to occur within jurisdictions in the Caribbean and northeastern South America where shorebird harvest occurs.
Without considering differential occurrence of Lesser Yellowlegs within harvest regions, their decline will likely continue, resulting in the loss of biodiversity and an important cultural resource.
INTRODUCTION
Shorebirds are biological indicators of ecosystem health (Piersma and Lindström 2004, Ogden et al. 2014) and are of high conservation concern due to their extensive annual migrations (Myers et al. 1987), dependence on vulnerable wetland, grassland and coastal habitats (Bildstein et al. 1991, Donaldson et al. 2000), and low reproductive potential (Brown et al. 2001, Watts et al. 2015). Over the past 40 years, North America's breeding shorebirds have experienced a 37% reduction in abundance (Rosenberg et al. 2019). The potential limiting factors leading to shorebird declines are complex and overlapping and include wetland conversion for urban development (Studds et al. 2017), habitat alteration through agricultural intensification (Kleijn et al. 2010), agrochemical contamination (Strum et al. 2010), and climate change (Galbraith et al. 2014). Additionally, shorebird harvest pressures within parts of the Atlantic Americas Flyway are a risk to the sustainability of some North American breeding populations (AFSI 2016).
Shorebird harvest occurs at varying levels across a large swath of the Americas from eastern Canada to northeastern South America (AFSI 2020). In many jurisdictions and territories, shorebird harvest is a cultural practice for subsistence and recreation; shorebirds are also sold in markets as a means of supplemental income (AFSI 2016). Furthermore, illegal harvest and sub-lethal hazing occasionally occur as a preventive measure to reduce crop and aquaculture damage by shorebirds (AFSI 2017). Although hunting is widespread and there is uncertainty in the number of birds taken annually, data indicate that annual harvest rates are highest in Barbados, St. Martin/Guadeloupe, Martinique, Guyana, Suriname, French Guiana, and northern Brazil (AFSI 2020). Harvest in other South American countries has not been identified as a significant threat, or as only a small local threat, in the development of shorebird conservation plans for Colombia (Johnston-González et al. 2010), Ecuador (Ágreda 2017), Argentina (Ministerio de Ambiente y Desarrollo Sostenible et al. 2020), and southern Chile (Delgado et al. 2010). The timing of harvest is variable across political boundaries and a subset of countries and departments (e.g., Barbados, Guadeloupe, and Martinique) have implemented seasonal protections (Watts and Turrin 2016). An estimate of 111,000–243,000 shorebirds are harvested annually from the Caribbean and northeastern South America, based on hunter surveys and logbooks (AFSI 2020). In Barbados, 9,000–10,000 shorebirds are currently harvested annually (B. Andres personal communication) and about 11,000–18,000 shorebirds are harvested in Guadeloupe and Martinique combined (AFSI 2020). About 60% of the harvest on Barbados are Lesser Yellowlegs (Tringa flavipes) (Reed 2012), which is likely a similar proportion on Guadeloupe and Martinique. Harvest appears to be much higher in Suriname and Guyana, where several tens of thousands of shorebirds are harvested each year (Ottema and Ramcharan 2009, AFSI 2020), although the proportions of Lesser Yellowlegs is uncertain.
The Lesser Yellowlegs is a monotypic, Neotropical migrant that breeds in the boreal forest of Alaska and Canada and spends the nonbreeding period in Central and South America and the Caribbean (Clay et al. 2012). The species has experienced a substantial population decline since the early 1970s. Data from North American migration surveys suggest a statistically significant population decline of 60–80% (Bart et al. 2007, Andres et al. 2012, Rosenberg et al. 2019); most migration data come from the eastern USA and Canada (see https://www.manomet.org/iss-map/). The North American Breeding Bird Survey (BBS) suggests a long-term, but variable, decline of 1.87% yr–1 (95% confidence interval [CI]: –5.96, 1.27; Sauer et al. 2017). Estimates from the roadside BBS and off-road Alaska Landbird Monitoring Surveys (ALMS) suggest 5.3% (95% CI: 8.5, –2.2) and 9.2% (95% CI: –15.0, –0.6) annual declines within the Northwestern Interior Forest Bird Conservation Region (BCR; Handel and Sauer 2017). Abundance surveys at nonbreeding sites in the Caribbean and South America also indicate a decline in Lesser Yellowlegs. In Suriname, a repeat aerial survey indicated an 80% population decline at a single coastal site between the 1970s and early 2000s (Ottema and Ramcharan 2009). Additionally, declines in Lesser Yellowlegs numbers between 1973 and the late-2000s have been reported at Mar Chiquita in Córdoba and Santiago del Estero Provinces, Argentina (Nores 2011). While many factors could be contributing to the species long-term decline, harvest on the nonbreeding grounds may be one of the most significant threats.
The most recent Lesser Yellowlegs population estimate is 660,000 individuals (Andres et al. 2012). This population estimate is based on a historical BBS analysis (Morrison et al. 2001); however, BBS coverage of the boreal forest, where Lesser Yellowlegs breed, is scant (Van Wilgenburg et al. 2015), and the population size was notably estimated with high uncertainty. As a means to help regulate the annual mortality of Lesser Yellowlegs, a Potential Biological Removal (PBR) model has been used to estimate the maximum mortality for a population without leading to long-term declines (Wade 1998). The PBR value incorporates all sources of mortality a population experiences. Watts et al. (2015) estimated a PBR for Lesser Yellowlegs of 79,000, based on the continental population size.
Despite the prevailing threat of unsustainable harvest, little research has been conducted on the relative occurrence of different Lesser Yellowlegs populations within jurisdictions practicing shorebird harvest. Reed et al. (2018) inferred the natal origins of juvenile Lesser Yellowlegs harvested in Barbados using stable hydrogen isotopes derived from their feathers. Results suggested a highly probable natal origin in the James Bay region of eastern Canada. A range-wide assessment of Lesser Yellowlegs populations' occurrence within shorebird harvest locations will help managers in the Caribbean and northeastern South America determine if harvest could be a driver of the overall population decline or is instead specific to some segments of the breeding populations.
Inferring the connectivity between an individual's origin and migratory movements, including stopover and staging locations, can help provide a better understanding of the species' susceptibility to harvest. The main objectives of our study were to identify (1) the probabilities of occurrence within shorebird harvest zones in the Caribbean and northeastern South America of individuals from geographically disparate breeding populations during southward migration, (2) the period(s) when birds from separate populations experience the highest probabilities of occurrence in these zones, and (3) factors associated with occurrence and the duration of stay within harvest zones. We hypothesize that adult Lesser Yellowlegs from geographically disparate breeding origins differentially occur within jurisdictions that practice shorebird harvest. Based on isotopic evidence from Reed et al. (2018), we anticipate that birds originating from the Eastern Canada population will have a higher risk of harvest than birds originating in Alaska, Yellowknife, and Churchill. Further, Lesser Yellowlegs migrating from eastern Canada should spend the nonbreeding period farther east as a result of the theory of longitudinal parallel migration (Newton 2008). Therefore, we suspect that Eastern Canada populations most likely occur within harvest zones during migration because harvest zones are located within the eastern portion of the species nonbreeding range ( Supplementary Material Figure S1 (duab061_suppl_supplemental_material.pdf)). If specific breeding populations are likely to occur within jurisdictions practicing shorebird harvest at a greater frequency, in terms of spatial and temporal overlap within known hunting areas, this occurrence can have important implications for refining sustainable mortality estimates and implementing targeted conservation actions.
METHODS
Study Area
We tracked adult Lesser Yellowlegs from four breeding sites: Anchorage, Alaska, USA (61.2181°N, 149.9003°W; n = 48); Kanuti National Wildlife Refuge, Alaska, USA (66.4167°N, 151.8333°W; n = 10; the Anchorage and Kanuti sites are combined and known hereafter as the “Alaska population”); Yellowknife, Northwest Territories, Canada (62.4540°N, 114.3718°W; n = 11); Churchill, Manitoba, Canada (58.7684°N, 94.1650°W; n = 20); and two stopover and staging sites during the early North American period of the birds' southward migration (hereafter, “Eastern Canada population” because known breeding origin is within the eastern Canadian provinces of Ontario, Quebec, and Labrador): James Bay, Ontario, Canada (53.5369°N, 80.5457°W; n = 7); and Mingan Archipelago, Quebec, Canada (50.2205°N, 63.6242°W; n = 14; Figure 1).
Data Collection
We deployed PinPoint global positioning system (GPS) Argos-75 satellite tags (hereafter, GPS tag; Lotek Wireless, Newmarket, Ontario, Canada) on adult Lesser Yellowlegs. We captured Lesser Yellowlegs during brood-rearing periods in May to August, using mist nets and chick call playback methods described in Johnson et al. (2020). Additionally, we installed mist nets within foraging habitats and used foraging call playbacks and shorebird decoys to capture birds. We fitted each bird with an alphanumeric leg flag (dark green flag with white lettering for USA and white flags with black lettering for Canada) above the joint on the left leg, a plastic color band corresponding to the study site (Anchorage = dark green, Kanuti = dark blue, Yellowknife = light blue, Churchill = black, James Bay = dark green, and Mingan = white) above the joint on the right leg, and a Bird Banding Office metal band below the joint on the right leg. We recorded standard morphometrics and determined sex using the CHD gene (dos Remedios et al. 2010). Birds were fitted with 4.0 g GPS tags using a modified leg-loop harness (Rappole and Tipton 1991, Sanzenbacher et al. 2000) made of 1.0-mm diameter elastic cord (Stretch Magic, Pepperell Braiding Company, Pepperell, Massachusetts, USA). Leg-loops were secured in position using 0.5-mm diameter brass jewelry crimps and instant adhesive gel (Loctite 454, Henkel Corporation, Düsseldorf, Germany). Total weight of the auxiliary leg bands and GPS tag including the harness material was between 3.9% and 5.5% of the bird's body mass (87.7 ± 6.8 g). We used GPS tags because they are light-weight, collect accurate locations (∼10 m), and transmit data remotely through the ARGOS system, eliminating the need to recapture the bird (Scarpignato et al. 2016).
FIGURE 1.
Deployment locations of PinPoint GPS Argos. Numbers in parentheses indicate the total GPS tags deployed in 2018, 2019, and 2020. (A) Anchorage, Alaska, USA; (B) Kanuti National Wildlife Refuge, Alaska, USA; (C) Yellowknife, Northwest Territories, Canada; (D) Churchill, Manitoba, Canada; (E) James Bay, Ontario, Canada; and (F) Mingan Archipelago, Quebec, Canada.
We programed GPS tags to transmit location data from mid-May through late October. In Alaska, transmissions began the day after tag deployment in May and June, whereas in Yellowknife and Churchill transmissions began July 1 because the breeding season in western and central Canada starts and concludes later than Alaska. Similarly, transmissions began in mid-July for the Eastern Canada population, because the breeding season for shorebird species extends into late summer in eastern Canada (Meltofte et al. 2007). Location collection schedules varied across years; we collected data more frequently during migration in 2019 than in 2018 and 2020. During southward migration (defined here as the departure date of birds from breeding population location through October 21) GPS tag transmissions were spaced every 2 days in 2019 and every 4 days in 2018 and 2020.
Data Preparation
The data collected by GPS tags were filtered through the Lotek Argos GPS Data Processor (Lotek Wireless, v4.2). We only included location data with 2 or 3-dimensional fixes that passed cyclic redundancy checks, which detect errors in raw data (Sobolewski 2003). On rare occurrences, the data processor passed locations with incorrect dates or longitudes, which we removed from our dataset. For all individuals tracked to 23°N latitude in any one year of the study, we estimated the bird's occurrence during migration through the Caribbean and northeastern South America in zones of high harvest (hereafter “harvest zones”). This latitude represented the furthest north any tracked bird spent the nonbreeding period (i.e. Salado, Nayarit, Mexico) across all 3 years of this study. Tracked birds either remained at 23°N for the nonbreeding period or migrated south to nonbreeding locations in the Caribbean and Central and South America. Additionally, we selected 23°N as a threshold to eliminate any biases imposed by tag failures during southward migration. For example, a bird whose tag stopped transmitting in the northeastern United States did not have the chance of occurring within a harvest zone because the bird could no longer be tracked by satellite. A bird may have occurred within a harvest zone, but this cannot be determined without GPS data. Therefore, we removed birds from our analyses that died or had potential GPS tag transmission failures during southward migration and prior to reaching 23°N latitude (n = 25). The northern boundary of the Caribbean and South American countries and jurisdictions that have reported shorebird harvest is 18°N; however, we included birds that spent the nonbreeding period north of this latitude because they represented the variability in the geographic distribution of Lesser Yellowlegs during southward migration. For example, all tracked birds in this study were capable of flying to a harvest zone; however, some birds spent the nonbreeding period in Mexico, Central America, and western portions of South America, and therefore, did not occur within a harvest zone. The data collected for these individuals were critical for calculating the probability of occurrence in harvest zones for each tracked population.
We restricted our analysis to southward migration, ending on October 21 for 3 reasons. First, harvest policies in the jurisdictions of the harvest zones we examined (see Defining Harvest Zones) indicate that most Lesser Yellowlegs harvest occurs during southward migration when flocks of shorebirds stop in the Lesser Antilles and northeastern South America due to seasonal weather events such as hurricanes and tropical depressions (Watts and Turrin 2016, Andres 2017). Second, the predefined GPS tag schedules limited the number of transmissions received after October 21 in 2018, 2019, and 2020. GPS tags were scheduled to optimize battery performance and longevity; therefore, transmission frequency decreased after October 21. Of the 69 (82%) individuals tracked throughout the nonbreeding period, none of the individuals occurred in harvest zones after October 21 that had not already been previously observed there. This indicates that differences in the southward migration initiation dates among tracked populations were not a significant source of bias. For our modeling analyses, we pooled tracking data from all 3 years.
Defining Harvest Zones
Watts and Turrin (2016) completed a review of harvest policies in jurisdictions (35 independent nations and 22 dependent territories or other entities) within the Western Hemisphere. Of the 57 jurisdictions evaluated, 96.5% follow an international treaty to protect migratory bird species; however, treaties differ by type of protections (i.e. species, seasons, and bag limits) and geographic extent. Of the 11 jurisdictions that allow hunting of shorebirds in formal policy, 10 are located within the Atlantic Flyway (excluding Nicaragua). To investigate spatial and temporal use patterns and to predict the probability of occurrence within jurisdictions practicing shorebird harvest, we defined 3 harvest zones: the Caribbean, coastal Guianas, and coastal Brazil. We divided the geographic regions into 3 zones based on policy, policy compliance, and degree of knowledge. Hunters on Caribbean islands are organized and hunting policies, either voluntary or government-dictated, are in place. The Caribbean zone comprises the Lesser Antilles, which includes St. Martin/Guadeloupe, Martinique, Barbados, and additional islands within the archipelago (Figure 2). Although harvest does not occur regularly in all countries within the Caribbean Islands, the entire archipelago was considered because islands are near one another. Lesser Yellowlegs may move among islands within the same day and could potentially occur in harvest zones without our detection. The Guianas zone comprises Guyana, Suriname, and French Guiana. Harvest in the Guianas is either unregulated (Guyana), regulations are not enforced (Suriname), or shorebird hunting is not perceived as being widespread (French Guiana). The coastal Brazil zone includes the coastal and estuarine habitats of Amapá, Pará, and Maranhão states of Brazil (Figure 2). Access and the extensive coastline in Brazil limit the ability to generate a regional perspective, although some information on harvest is becoming available and shows that hunting of Lesser Yellowlegs does occur (Bosi de Almeida et al. 2018). We defined the coastal Guianas and Brazil zones as the area between the coast and the rainforest boundary using 2020 aerial satellite imagery and ArcGIS software (ArcMap 10.6; ESRI 2018), and only considered the coastal and estuarine habitats as part of the harvest zone because the majority of harvest occurs along the Atlantic coastline (AFSI 2016). These 3 harvest zones are located within the northeastern portion of the Lesser Yellowlegs nonbreeding range (Fink et al. 2020; Supplementary Material Figure S1 (duab061_suppl_supplemental_material.pdf)).
Track Processing
Although we programed tags to collect location data with a predetermined schedule, GPS tags sometimes failed to receive or transmit GPS positions as scheduled. These failures resulted in data gaps for some individuals ( Supplementary Material Figure S2 (duab061_suppl_supplemental_material.pdf)). Given these data gaps and variation in sampling frequency across years, we used a movement model to estimate locations during data gaps and to equalize sampling through time across individuals. Regularizing locations across time ensured that each individual contributed the same number of positions to subsequent analyses. We thus used foieGras 0.4.0 in program R to fit a continuous-time state-space random walk model (Codling et al. 2008) to the GPS data to estimate most probable locations for each individual (Jonsen et al. 2020, R Core Team 2021). We predicted locations every 24 hr (1 location per day) from July 1 to October 21 for each bird.
Statistical Analysis
Probability of occurrence in harvest zones. To determine if adults from different breeding locations were more or less likely to occur in harvest zones during southward migration, we first classified each individual's model-estimated position as within or outside each of the 3 harvest zones using the over function in R package sp 1.4-4 (Pebesma et al. 2020). Next, we used generalized additive models (Wood 2006) to estimate how probability of occurrence in harvest zones varied across time (July 1 through October 21). With each population, we addressed 2 questions: (1) what was the probability of occurrence in any (combined) harvest zone and (2) what was the probability of occurrence in each harvest zone?
Probability of occurrence in any harvest zone. First, we fit separate generalized additive models (binomial error, logit link; Wood 2019; R package mgcv 1.8-31) to estimate the daily probability that a tracked bird from each source location occurred in any of the harvest zones between July 1 and October 21. The response variable was whether a bird was in a harvest zone with the predictor variable population (Alaska population, Yellowknife, Churchill, and Eastern Canada population); individual bird was included as a random intercept effect.
Probability of occurrence in each harvest zone. Second, we estimated the daily probability that a tracked bird from each population would occur in each of the three harvest zones between July 1 and October 21. To estimate the probability of occurrence in each zone we attempted to use a multinomial logistic regression—an extension of a generalized additive model (Wood 2006, Wood et al. 2016). The multinomial model predictions can be interpreted as the probability that a randomly selected individual from each tracked population was located in or outside each harvest zone on a given day (see Harrison et al. 2018 for a similar application of this method).
To estimate the probability of occurrence within harvest zones, the multinomial modeling approach was most appropriate and forces the predictions of occurrence within all harvest zones and outside harvest zones to sum to 1 on any given day. However, when random effects were included, models failed to converge, which did not allow us to represent the hierarchical structure of our data (i.e. individuals nested within populations). Multinomial models are complex, and failure of convergence was likely due to small sample sizes in some zones. Therefore, to examine whether the predicted occurrence patterns would remain when our models accounted for the structure of the data, we fitted separate binomial generalized additive mixed models for each population and harvest zone with a random intercept of individual. Binomial models were fit using R package mgcv 1.8-31 (Wood 2019). Fitting a sequence of independent binomial models is less appropriate for our question (i.e. total probabilities of occurrence on any given day are not constrained to a sum of 1), but the independent binomial approach allowed us to examine the influence of random effects on predicted occurrence patterns and corresponding confidence intervals.
Predictors of occurrence within harvest zones. We explored the influence of population origin (i.e. capture location), body mass, and year on the occurrence of Lesser Yellowlegs in harvest zones and the total number of days a bird was present inside a zone. To model occurrence in harvest zones, we used a binomial generalized linear model using the R package MASS 7.3-53; Venables and Ripley 2002, R Core Team 2021). Each model included all individuals tracked south of 23°N latitude. Birds present inside a harvest zone for one or more days were categorized as 1 and birds absent from harvest zones were defined as 0. Covariates included year, sex, capture mass, longitude of capture location, and departure date from each capture location. We used Pearson's Product-Moment Correlation to evaluate correlations among predictor variables; when high correlations (r > 0.70) were detected, we used the small-sample bias-corrected version of Akaike's Information Criteria (AICc; Burnham and Anderson 2002) to select which of the correlated predictors was more strongly associated with duration of stay. The covariate with the smallest AICc value was then used in subsequent models. To assess model fit, we visually inspected residual plots and standard errors from the model output. After eliminating correlated predictors, models with different predictors were compared using AICc. Finally, we defined duration of stay as the total number of days spent by each bird in all harvest zones, using the predicted daily locations from the random-walk movement models (see Track Processing). For modeling occurrence duration, we followed a modeling process similar to the methods described for occurrence in harvest zones, except we used a generalized linear model assuming a negative binomial error distribution and log link function. All values presented in the results are means ± one standard error (SE).
RESULTS
Migratory Flyway Patterns
We successfully tracked 85 birds south of 23°N latitude prior to October 21 (Table 1, Figure 3). Individuals from the Eastern Canada population, Yellowknife, Churchill, and the Alaska population followed different migratory flyways within the Western Hemisphere, suggesting differential occurrence in harvest zones during migration (i.e. July to October; Figure 3) while passing through the Caribbean and northeastern South America. Birds from Alaska (n = 44) followed portions of the Pacific Americas and Midcontinent Americas Flyways, except for 2 individuals that followed the Atlantic Americas Flyway. Individuals in the Yellowknife and Churchill (n = 30) populations followed the Midcontinent Americas and Atlantic Americas Flyways, whereas individuals from the Eastern Canada population (n = 11) followed the Atlantic Americas Flyway exclusively (Figure 3).
Probability of Occurrence in Harvest Zones
Of the 85 individuals tracked south of 23°N latitude, 26 entered at least 1 harvest zone prior to October 21, and 9 entered 2 or more harvest zones. The probability of occurrence within harvest zones was lowest for birds of the Alaska population compared to birds of the Eastern Canada population (Table 1). For the Alaska population, 5% of birds that migrated south of 23°N latitude entered 1 or more of the 3 harvest zones; both had been captured in Anchorage (Figure 3). Of the birds originating in Yellowknife and Churchill, 45% and 53% of birds entered one or more harvest zones. For birds of the Eastern Canada population, 82% of birds entered a harvest zone. The Caribbean and coastal Guianas zones had 10 and 18 individuals enter the harvest zones, respectively, whereas only 7 individuals entered the coastal Brazil zone. Across all populations, 54% of birds detected within a harvest zone were male, whereas 46% were female. The male to female sex ratio of birds originating from each capture location was 0.5:1 for Anchorage, 0.7:1 for Kanuti, 1.8:1 for Yellowknife, 1.2:1 for Churchill, 1.3:1 for James Bay, and 0.8:1 for Mingan Archipelago. The overall sex ratio for the 85 birds tracked south of 23°N latitude was 44% male and 56% female.
TABLE 1.
Lesser Yellowlegs proportion of occurrence within each harvest zone. Number of transmitting birds corresponds to the number of individuals that migrated south of 23°N prior to October 21. The (n) indicates the total number of birds that occurred within the specified harvest zone. Number of individuals exposed indicates the total number of individuals occurring within one or more harvest zones. Proportions are not necessarily additive because a bird may occur in more than one harvest zone. Average duration indicates the average total number of days birds from each population were present within the specified harvest zone
FIGURE 3.
Occurrence of Lesser Yellowlegs within harvest zones. Populations include Alaska (Anchorage and Kanuti National Wildlife, Alaska, USA); Yellowknife, Northwest Territories, Canada; Churchill, Manitoba, Canada; and Eastern Canada (James Bay, Ontario, Canada; and Mingan Archipelago, Quebec, Canada). The colored dots represent GPS fix locations within harvest zones. All birds that migrated south of 23°N latitude were included. The n value for green (2018), blue (2019), and orange (2020) dots indicates the number of birds present in the harvest zones. The n value for gray dots indicated how many birds were detected south of 23°N latitude but were not present within a harvest zone.
Probability of occurrence in any harvest zone. Results from the generalized additive models (binomial) suggest that Lesser Yellowlegs populations did not have an equal probability of occurrence within harvest zones (Figure 4). Alaska birds had the lowest probability of occurrence in any harvest zone throughout migration, whereas the Eastern Canada population experienced the greatest probability of occurrence from mid-August through October. Birds from the Yellowknife population experienced the second-lowest probability of occurrence, which peaked at 0.08 on August 26. The probability of occurrence for birds originating from Churchill peaked at 0.20 on August 18. Finally, the birds from the Eastern Canada population had the highest probability of occurrence with two distinct peaks: 0.90 in the first week of September and 0.65 in the last week of September (Figure 4).
Probability of occurrence in each harvest zone. The probability of occurrence of individuals from the Alaska population in the Caribbean or Guianas was <10% for any given day during migration (i.e. July to October; Figure 5). No individuals from Alaska or Yellowknife were detected in Brazil during the period of interest. For the Yellowknife
FIGURE 4.
Generalized additive model (binomial) with random effect of individual. Probability of a single adult Lesser Yellowlegs occurring in any harvest zones (n = 85). Lines represent the estimated effect of day of year on the probability of a randomly selected individual from each tracked population occurring in a harvest zone. Shading around each line represents the simulated interquartile range of estimates from the posterior distribution of the model parameters. The probability of occurence for the Alaska population was <2% on any given day.
