Evaluating environmental drivers and synchrony of Arctic shorebird demographic rates to inform conservation management

Abstract

Many Arctic-breeding shorebirds are assumed to be declining, yet obtaining reliable estimates of species' demographic rates and trends is difficult because of challenges collecting data in remote breeding regions and throughout the annual cycle. For many vulnerable species, data limitations impede efforts to determine appropriate management actions in the face of ongoing environmental change. Integrated population models (IPMs) offer an approach to maximize the utility of available data by providing a framework for estimating demographic rates and environmental drivers of population change, while also accounting for sources of uncertainty. Here, we used an IPM to estimate demographic rates, synchrony, and population trends of Arctic-breeding shorebirds within the context of climatic and management-related changes. We estimated species-specific breeding population sizes, adult survival rates, number of adults gained into the breeding population through recruitment (i.e., the sum of immigration and reproduction), as well as the effects of environmental drivers on demographic traits for three shorebird species nesting near Utqiaġvik, Alaska, over an 18-year study period (2005-2022). We found that the annual number of adults recruiting into the breeding population was important for maintaining local populations, and that local environmental factors and management regimes had strong effects on demographic rates. The timing of snowmelt had a notable effect on (1) fecundity, (2) the number of adults recruited for two of the three species, and (3) adult survival during the following year for one species. Predator removal increased fecundity of all three species but had limited effects on subsequent local population sizes. The Pacific Decadal Oscillation, a broad-scale climate metric, affected adult survival differently across species, with a positive and negative effect for one species each, and a negligible effect for the other. Unlike adult recruitment and fecundity that varied synchronously among species, annual adult survival varied asynchronously. Our results suggest that differences in survival were likely related to conditions experienced during nonbreeding periods arising from dissimilar migratory routes, stopover sites, and nonbreeding season ranges. Future work should focus on incorporating additional environmental factors on the nonbreeding grounds to determine when and where these species could benefit most from management interventions.

Keywords: Arctic‐breeding shorebirds; demographic rate synchrony; integrated population model; nonbreeding season.

FIGURE 1

(a) Annual range of American golden‐plovers (burgundy) and the arcticola subspecies of dunlin (brown). (b) Annual range of semipalmated sandpipers (navy). In panels (a) and (b) the black dot indicates the location of Utqiaġvik, Alaska, where the data were collected during 2005–2022. (c) Directed acyclic diagram of the Arctic shorebirds life cycle and the integrated population model (IPM) framework. There are two categories of adults that make up the breeding population in a given year (N _ad): (1) adults recruited through reproduction and immigration (R) and (2) adults who were previous breeders at the study site and returned (S). The total breeding population is estimated with a state‐space model using the annual count of breeding pairs (y), determined by the number of nests. The true number of breeding pairs in the population is linked to the observed survey data through an observation process that incorporates imperfect detection during surveys (${\mathrm{\tau }}_{\mathrm{obs}}$). Fecundity is estimated outside of the joint likelihood of the IPM with covariates including predator removal in each year (Fox) and annual snowmelt date (Snow). Annual estimates of demographic parameters in the IPM include the expected number of new recruits to the population ($\mathrm{\gamma }$; yellow), which is estimated from nest and chick data with covariates including snowmelt date (Snow) and fecundity from the previous season t − 1 (f); and annual recapture/resighting probability (p) and adult survival (${\mathrm{\phi }}_{\mathrm{ad}}$; blue), which is estimated with a Cormack‐Jolly‐Seber model using the mark‐recapture data (capture histories: Ch) with snowmelt date (Snow) and the Pacific Decadal Oscillation (PDO) covariates. Data are shown in light gray. Solid black arrows depict how the demographic data inform model parameters. Dashed arrows depict how the covariates are incorporated in the model, and colored arrows (green, yellow, and blue) show how the model parameters are connected. American golden‐plover image (a) by Peter Wilton, used under a Creative Commons Attribution 2.0 license. Dunlin image (a) by Jevgenis Slihto, used under a Creative Commons Attribution 2.0 license. Semipalmated sandpiper image (b) by Mdf/Wikimedia Commons used under a Creative Commons Attribution‐ShareAlike 3.0 license.

FIGURE 2

Demographic parameters for American golden‐plovers (burgundy column), dunlin (brown column), and semipalmated sandpipers (navy column) estimated from a long‐term study at Utqiaġvik, Alaska, 2005–2022. (a–c) Estimates of shorebird population size (breeding pair abundance estimates, gray line) and 95% credible intervals (light gray shading) with observed pair counts (nest counts excluding known renests, black line), (d–f) annual number of adults (pairs) recruited to the population, (g–i) annual apparent adult survival probabilities, and (j–l) fecundity (chicks per breeding pair) estimated from chick counts. In (j–l), dark green indicates fecundity estimates in years with high fox removal, and light green shows fecundity estimates in years with low or no fox removal. Solid points with vertical lines indicate mean annual estimates with 95% credible intervals. American golden‐plover image (a) by Peter Wilton, used under a Creative Commons Attribution 2.0 license. Dunlin image (b) by Jevgenis Slihto, used under a Creative Commons Attribution 2.0 license. Semipalmated sandpiper image (c) by Mdf/Wikimedia Commons used under a Creative Commons Attribution‐ShareAlike 3.0 license.

FIGURE 3

Demographic parameters for American golden‐plovers (AMGP), dunlin (DUNL), and semipalmated sandpipers (SESA) in relation to snow melt, Pacific Decadal Oscillation (PDO), and fox removal efforts estimated from a long‐term study at Utqiaġvik, Alaska, 2005–2022. (a) Expected number of adults recruited (new breeding pairs) into the population as a function of the Julian date of 20% snowmelt on the breeding grounds for American golden‐plover (burgundy) and dunlin (brown). Earlier snowmelt increased recruitment until ~5 June (Julian date = 156) for American golden‐plover and ~7 June (158) for dunlin. Shading shows 95% credible intervals. (b) Estimated number of adults recruited as a function of fecundity in year t − 2 for semipalmated sandpiper (navy); shading shows 95% credible intervals. Points show the estimates of annual number of adults recruited with 95% credible intervals. Semipalmated sandpiper recruitment was positively correlated with fecundity in the breeding season two years earlier. (c) Estimated apparent adult survival as a function of the PDO for American golden‐plover (burgundy) and semipalmated sandpiper (navy). Apparent adult survival was positively correlated with PDO in American golden‐plovers and negatively correlated with PDO in semipalmated sandpipers. Shading shows 95% credible intervals. (d) Estimated apparent adult survival as a function of the Julian date of 20% snowmelt at the breeding site for semipalmated sandpiper (navy); shading shows 95% credible intervals. Points show the annual apparent adult survival estimates with 95% credible intervals. Semipalmated sandpiper apparent adult survival was negatively correlated with snowmelt date. (e) Boxplots showing posterior medians, interquartile range and 95% credible intervals of fecundity estimates (chicks per breeding pair) under high (gray) versus low or no (white) fox removal efforts for American golden‐plover, dunlin, and semipalmated sandpipers. High predator removal effort resulted in substantially higher fecundity compared to years with low or no fox removal. (f) Estimated fecundity as a function of the Julian date of 20% snowmelt at the breeding site for dunlin (brown), and semipalmated sandpiper (navy), shown in years with low/no fox removal. Earlier snowmelt resulted in increased fecundity until ~5 June (Julian date = 156) for semipalmated sandpiper, and dunlin fecundity decreased linearly with later snowmelt. Shading shows 95% credible intervals.

FIGURE 4

Posterior means of the annual population growth rate (λ _t ) at Utqiaġvik, Alaska, (2005–2022) plotted against (1) the posterior means of annual adult survival for each species (left column); (2) number of adults recruited for each species (middle column); and (3) fecundity (right column). Posterior means of the correlation coefficients (r) with associated 95% credible intervals and the probabilities that estimates are positive p(r > 0) are given for each parameter correlation. Horizontal and vertical lines show the extent of 95% credible intervals for each annual estimate. American golden‐plover image (top row) by Peter Wilton, used under a Creative Commons Attribution 2.0 license. Dunlin image (middle row) by Jevgenis Slihto, used under a Creative Commons Attribution 2.0 license. Semipalmated sandpiper image (bottom row) by Mdf/Wikimedia Commons used under a Creative Commons Attribution‐ShareAlike 3.0 license.

FIGURE 5

Pairwise species comparisons of American golden‐plover, dunlin, and semipalmated sandpiper demographic rates at Utqiaġvik, Alaska, 2005–2022, including means and 95% credible intervals for adult recruitment (yellow; a–c), apparent adult survival (blue; d–f), and fecundity (green; g–i). The pairwise correlation coefficients (r; means and 95% credible intervals) as well as the probabilities that estimates are positive p(r > 0) are given for each species pair.