Introduction
Arctic ecosystems are at high risk of environmental change, from both direct and indirect consequences of changing climate (ACIA 2004; Christensen et al. Reference Christensen, Barry, Taylor, Doyle, Aronsson and Braa2020; IPCC Reference Shukla, Skea, Calvo Buendia, Masson-Delmotte, Pörtner and Roberts2019). Arctic areas have experienced the greatest climate warming worldwide, and consequently significant ecological changes have already been observed (Gao et al. Reference Gao, Ma, Li, Sun, Liu and Xing2023; IPCC Reference Shukla, Skea, Calvo Buendia, Masson-Delmotte, Pörtner and Roberts2019; Nolet et al. Reference Nolet, Bauer, Feige, Kokorev, Popov and Ebbinge2013; Schmidt et al. Reference Schmidt, Kankaanpää, Tiusanen, Reneerkens, Versluijs and Hansen2023). Governments of Arctic countries have therefore called for better information on Arctic biodiversity and its change to be able to respond through appropriate conservation and management actions (CAFF 2013; Convention on Biological Diversity 2012). Data relating to groups of species or environmental variables collected from the Arctic have been widely used to evaluate the state of Arctic biodiversity (e.g. MacRae et al. Reference McRae, Zöckler, Gill, Loh, Latham and Harrison2010). However, while the need for effective long-term monitoring information on Arctic ecosystems and biodiversity is clear, there are major logistic challenges in surveying this area (Christensen et al. Reference Christensen, Barry, Taylor, Doyle, Aronsson and Braa2020).
A considerable number of migrant waterbird species breed within the Arctic but spend the non-breeding season in other areas (CAFF 2010, 2013; Deinet et al. Reference Deinet, Zöckler, Jacoby, Tresize, Marconi and McRae2015). These include geese, ducks, and waders that are considered highly vulnerable to climate change impacts, as well as facing other threats such as habitat loss and localised anthropogenic disturbance (Nagy et al. Reference Nagy, Breiner, Anand, Butchart, Flörke and Fluet-Chouinard2021; Wauchope et al. Reference Wauchope, Shaw, Varpe, Lappo, Boertmann and Lanctot2017). This makes it important to understand their population dynamics and ecological drivers of change in the context of developing conservation policies (Pearce-Higgins et al. Reference Pearce-Higgins, Brown, Douglas, Alves, Bellio and Bocher2017). There are also long-standing, recognised conservation needs to improve the monitoring of migrant bird populations throughout their annual cycle (Burton et al. Reference Burton, Daunt, Kober, Humphreys and Frost2023; CMS 2017; Culp et al. Reference Culp, Cohen, Scarpignato, Thogmartin and Marra2017; Piersma and Lindström Reference Piersma and Lindström2004; Rakhimberdiev et al. Reference Rakhimberdiev, Duijns, Karagicheva, Camphusen, Castricum and Dekinga2018; Reneerkens et al. Reference Reneerkens, Versluijs, Piersma, Alves, Boorman and Corse2020). Yet the vast extent and remoteness of breeding areas used by Arctic birds presents a substantial economic cost to monitoring, and further practical issues such as local capacity, feasibility, and accessibility can limit the number of sites and species which can be covered (López-Blanco et al. Reference López-Blanco, Topp-Jørgensen, Christensen, Rasch, Skov and Arndal2024; Mallory et al. Reference Mallory, Gilchrist, Janssen, Major, Merkel and Provencher2018; Soloviev and Tomkovich Reference Soloviev and Tomkovich2024). Furthermore, with more northerly latitudes, the breeding season is more compressed than more southerly latitudes, affording shortened and earlier survey windows prior to breeding to assess numbers (Jarrett et al. Reference Jarrett, Lehikoinen and Willis2024; Meltofte Reference Meltofte2001).
A solution may therefore be to monitor the status of these species at more accessible non-Arctic sites in the expectation that changes in their populations provide signals of environmental changes on breeding areas (Stroud Reference Stroud, Zubrow, Meidinger and Connolly2019). Grouping monitoring information from species using the same northern areas as “multi-species” (composite trend) indicators can help communicate desired information about changes to ecosystems, providing conservation decision-makers and stakeholders a cost-effective way to visualise aspects of environmental change and track progress towards goals (Fraixedas et al. Reference Fraixedas, Lindén, Piha, Cabeza, Gregory and Lehikoinen2020; Pereira and Cooper Reference Pereira and Cooper2006; Reynolds et al. Reference Reynolds, Thompson and Russell2011).
Multi-species indicators aim to present combined population trends across several species and provide broad measures of biodiversity change (Butchart et al. Reference Butchart, Walpole, Collen, van Strien, Scharlemann and Almond2010) – for example, see the Living Planet Index (Ledger at al. Reference Ledger, Loh, Almond, Böhm, Clements and Currie2023; WWF 2022). To be successful, however, they ideally should integrate information across species in a manner that is easy to interpret (such as relative population change over time), easy to measure, and have a known and low variability in response to additional environmental changes and anthropogenic stresses (Dale and Beyeler Reference Dale and Beyeler2001; Gregory et al. Reference Gregory, Noble, Field, Marchant, Raven and Gibbons2003). Although indicators can mask differing trends in individual species, and therefore potentially hide species-specific conservation issues, there are many examples of the successful implementation and application of composite trend indicators that track population sizes of species at national (Gregory et al. Reference Gregory, Noble, Field, Marchant, Raven and Gibbons2003), continental (Gregory et al. Reference Gregory, van Strien, Vorisek, Gmelig Meyling, Noble and Foppen2005; de Heer et al. Reference de Heer, Kapos and Ten Brink2005), and global scales (Ledger et al. Reference Ledger, Loh, Almond, Böhm, Clements and Currie2023). Birds are often seen as useful measures of ecosystem function, biodiversity, and habitat productivity through being at a high trophic level, widespread, diverse, and sensitive to environmental change (Durant et al. Reference Durant, Hjerman, Frederiksen, Charrassin, Le Maho and Sabarros2009; Gregory et al. Reference Gregory, Noble, Field, Marchant, Raven and Gibbons2003). Consequently, data from national monitoring schemes are routinely used within indicators such as the European Wild Bird Indicators (Gregory and van Strien Reference Gregory and van Strien2010). Within the Arctic, a programme to develop a more comprehensive set of biodiversity indicators covering species, habitats, and ecosystem processes was initiated by the Circumpolar Biodiversity Monitoring Program (CBMP) of the Arctic Council’s Conservation of Arctic Flora and Fauna (CAFF) (Barry and Helgason Reference Barry and Helgason2019; CAFF 2013; see also Christensen et al. Reference Christensen, Barry, Taylor, Doyle, Aronsson and Braa2020 and Taylor et al. Reference Taylor, Lawler, Aronsson, Barry, Bjorkman and Christensen2020 for summaries). This work has led to the production of an Arctic Migratory Birds Index covering four main global areas (Deinet et al. Reference Deinet, Zöckler, Jacoby, Tresize, Marconi and McRae2015).
Migratory waders that occur on the eastern seaboard of the Atlantic Ocean (in western Europe, northern and western Africa) use the “East Atlantic Flyway” (Birdlife 2015; Smit and Piersma Reference Smit, Piersma, Boyd and Pirot1989). In this paper we refer to the East Atlantic Flyway in terms of its use by all migratory waterbirds (i.e. wildfowl and waders) that were monitored. Waterbirds breed throughout the Arctic terminus of this flyway over an extensive area from Canada (to 122oW) to mid-northern Siberia (to 122oE). Many biogeographical populations of species that use the East Atlantic Flyway overwinter in western Europe, while others pass through in spring and autumn to spend the non-breeding season in northern and western Africa (Delany et al. Reference Delany, Scott, Dodman and Stroud2009; Scott and Rose Reference Scott and Rose1996; van Roomen et al. Reference van Roomen, Citegetse, Crowe, Dodman, Hagemeijer and Meise2022). Most countries along the East Atlantic Flyway (and indeed globally) contribute count data on non-breeding waterbirds to the International Waterbird Census (IWC) (Nagy and Langendoen Reference Nagy and Langendoen2020; Pavón-Jordán et al. Reference Pavón-Jordán, Abdou, Azafzaf, Balaž, Bino and Borg2020). There is also extensive monitoring of annual breeding success (i.e. productivity) of goose and swan populations in core wintering areas in north-west Europe (Beekman et al. Reference Beekman, Koffijberg, Wahl, Kowallik, Hall and Devos2019; Fox et al. Reference Fox, Ebbinge, Mitchell, Heinicke, Aarvak and Colhoun2010; Laubek et al. Reference Laubek, Clausen, Nilsson, Wahl, Wieloch and Meissner2019; Madsen et al. Reference Madsen, Cracknell and Fox1999), and age ratios are determined from hunting bag data of ducks in some countries as a proxy for annual reproductive success (Christensen and Fox Reference Christensen and Fox2014). Productivity measures can indicate likely changes in abundance in long-lived species before they are discernible from count data (Cook et al. Reference Cook, Dadam, Mitchell, Ross-Smith and Robinson2014; Minton et al. Reference Minton, Jessop and Hassell2012; Robinson et al. Reference Robinson, Clark, Lanctot, Nebel, Harrington and Clark2005). In some waterbird populations, productivity is correlated with overall population size, for example positively in the Dark-bellied Brent Goose Branta bernicla bernicla despite density dependence operating (Ebbinge et al. Reference Ebbinge, Blew, Clausen, Günther, Hall and Holt2013; Nolet et al. Reference Nolet, Bauer, Feige, Kokorev, Popov and Ebbinge2013), linked in some cases to warmer spring and summer weather in Arctic-breeding areas and a longer growing season (Descamps et al. Reference Descamps, Aars, Fuglei, Kovacs, Lydersen and Pavlova2017), impacting prey abundance during breeding (Nolet et al. Reference Nolet, Bauer, Feige, Kokorev, Popov and Ebbinge2013). Hence there are considerable opportunities for the existing monitoring efforts away from the Arctic “ex situ” to contribute to status assessments of Arctic migratory waterbirds (Christensen et al. Reference Christensen, Barry, Taylor, Doyle, Aronsson and Braa2020; Stroud Reference Stroud, Zubrow, Meidinger and Connolly2019; Zöckler Reference Zöckler2005), and in doing so develop indicators within the East Atlantic Flyway at finer geographical scales. Indeed, much focus has been directed at understanding drivers of wintering abundance and distribution on estuaries, including climate change (Anderson et al. Reference Anderson, Fahrig, Rausch, Martin, Daufresne and Smith2023; Burton et al. Reference Burton, Daunt, Kober, Humphreys and Frost2023; Maclean et al. Reference Maclean, Austin, Rehfisch, Blew, Crowe and Delany2008). Among taxonomic groups, separate consideration is also given to wildfowl and wader species within existing wild bird indicators (Gregory and van Strien Reference Gregory and van Strien2010). However, pressures may be acting differently within global flyways. For example, within the East Atlantic Flyway, birds may use migratory pathways stretching from Greenland and eastern Canada in the west to Russia in the east, and due to the different pressures operating in these areas may show contrasting non-breeding population trends (Harvey and Heubeck Reference Harvey and Heubeck2012). The non-breeding abundance of species breeding in more northerly Arctic or southerly sub-Arctic areas may reflect different latitudinal responses of ecosystems to climate (IPCC Reference Shukla, Skea, Calvo Buendia, Masson-Delmotte, Pörtner and Roberts2019; McRae et al. Reference McRae, Zöckler, Gill, Loh, Latham and Harrison2010). Given the pace of Arctic warming (Post et al. Reference Post, Alley, Christensen, Macias-Fauria, Forbes and Gooseff2019), there is a further need to evaluate these indicator trends over different time periods.
To demonstrate how existing monitoring activities in non-breeding areas can contribute to the further development of Arctic biodiversity indicators, we used data from the UK and the Netherlands to present a series of composite trends in abundance and “productivity” (i.e. annual juvenile recruitment of individuals into the non-breeding population). These indicators summarise the status of a selection of Arctic-breeding waterbirds using different parts of the East Atlantic Flyway. To differentiate composite indicators for different Arctic areas, we then generated finer-scale composite trends based on abundance by grouping species by migratory pathway, based on the best available knowledge of migratory movements of species between breeding and wintering grounds (Spina et al. Reference Spina, Baillie, Bairlein, Fiedler and Thorup2022). Specifically, we examined the variability in trends between: (1) birds using Eastern and Western migratory pathways and from Arctic and sub-Arctic breeding areas; (2) birds using UK and the Netherlands non-breeding locations; (3) wader and wildfowl taxonomic groupings, all over distinct temporal phases. We also aimed to stimulate progress in the wider development of indicators for migratory species.
Methods
Selection of species/populations for incorporation into abundance indicators
Several steps were taken to ensure that the indicators were representative of Arctic-breeding waterbird populations. The general “Arctic” area of the East Atlantic Flyway (encompassing high-Arctic and sub-Arctic areas) was defined as the northerly geographical regions of Svalbard, arctic Fennoscandia, Greenland, Canada, and Russia, as well as areas that are biologically classified as sub-Arctic, such as the Faroe Islands and Iceland (see CAFF 2013; Christensen et al. Reference Christensen, Barry, Taylor, Doyle, Aronsson and Braa2020; Hohn and Jaakkola Reference Hohn and Jaakkola2010). To account more appropriately for species of high-Arctic and sub-Arctic breeding origin based on latitude, we therefore used modified vegetation community-based geographical boundaries (Christensen et al. Reference Christensen, Barry, Taylor, Doyle, Aronsson and Braa2020) to include these refined high-Arctic and sub-Arctic definitions (see Figure 1). We further restricted our analyses to species in the East Atlantic Flyway (Delaney et al. Reference Delany, Scott, Dodman and Stroud2009) that are predominantly (>50% of population) confined to the Arctic in the breeding season (Christensen et al. Reference Christensen, Barry, Taylor, Doyle, Aronsson and Braa2020). Summaries of waterbird movements, e.g. in Cramp and Simmons (Reference Cramp and Simmons1977, Reference Cramp and Simmons1983), Delaney et al. (Reference Delany, Scott, Dodman and Stroud2009), Scott and Rose (Reference Scott and Rose1996), and Stroud et al. (Reference Stroud, Davidson, West, Scott, Hanstra and Thorup2004), together with mapped ringing data within the Eurasian African Bird Migration Atlas (Spina et al. Reference Spina, Baillie, Bairlein, Fiedler and Thorup2022), were used to determine whether species used Eastern and/or Western migratory pathways and whether the majority of their breeding range was in the vegetation community-based high-Arctic or sub-Arctic. For species with only part of their breeding distribution in the Arctic, it is difficult to separate Arctic breeders from boreal or temperate birds during counts in the non-breeding season. Consequently, species with a large latitudinal range in their breeding distribution were excluded, such as Eurasian Teal Anas crecca, Tufted Duck Aythya fuligula, and Eurasian Curlew Numenius arquata (Cramp and Simmons Reference Cramp and Simmons1977, Reference Cramp and Simmons1983). To reduce analytical problems associated with scarce species, we included species with winter populations that surpass the international 1% threshold (Wetlands International 2023) at one or more UK or Dutch sites. Species that had greater uncertainty in wintering locations and/or likely latitudinal breeding origins were included in more than one indicator to represent this potential variation; this was the case for Greater Scaup Aytha marila (Western and Eastern, high-Arctic, and sub-Arctic), Redshank Tringa totanus (Netherlands data, Western and Eastern), Grey Plover Pluvialis squatarola (high-Arctic and sub-Arctic), and Purple Sandpiper Calidris maritima (Western and Eastern, high-Arctic, and sub-Arctic). We also excluded data collected for the annual census of Icelandic Greylag Goose Anser anser in the UK due to increased overlap with the British Greylag Goose population at wintering sites currently precluding confident allocation of birds to the Iceland population (Hearn and Frederiksen Reference Hearn, Frederiksen, Boere, Galbraith and Stroud2006). Following these selections, our paper covers seven biogeographical populations of seven migratory wildfowl species, and 12 biogeographical populations of 12 migratory wader species (Table 1).
Map depicting the high-Arctic and sub-Arctic regions (adapted from CAFF 2013; Christensen et al. Reference Christensen, Barry, Taylor, Doyle, Aronsson and Braa2020; Hohn and Jaakkola Reference Hohn and Jaakkola2010). The area to the north of the approximated black line is defined as “high-Arctic”, and between the approximated black and grey lines is “sub-Arctic”; our definition of the sub-Arctic was also “modified” (grey dashed line) to also include the Faroe Islands (see main text). The East Atlantic Flyway is also depicted redrawn from Delany et al. (Reference Delany, Scott, Dodman and Stroud2009), and the UK and Netherlands where wintering waterbird data on abundance and productivity were collected. Several key monitoring sites are also depicted; three important sites (Waddenzee, the Netherlands, The Wash, UK, and Severn Estuary, UK) are directly labelled, and others are depicted as points for the Netherlands, i.e. Oosterschelde, Westerschelde, and Noordzeekustzone, and the UK, i.e. Thames Estuary, Humber Estuary, Dee Estuary, Mersey Estuary, Solent and Southampton Water, Islay, Caerlaverock, Firth of Tay, Inner Moray Firth, Montrose Basin, Loch of Strathbeg, Loch Leven, Strangford Lough, Lough Neagh, and Lough Beg.
Allocation of Arctic-breeding waterbirds to abundance indicators: East Atlantic Flyway (1), wildfowl (2), wader (3), Western (4), Eastern (5), high-Arctic (6), sub-Arctic (7), Western wildfowl (8), Western wader (9), Eastern wildfowl (10), Eastern wader (11), Western high-Arctic (12), Western sub-Arctic (13), Eastern high-Arctic (14), Eastern sub-Arctic (15), wildfowl high-Arctic (16), wader high-Arctic (17), wildfowl sub-Arctic (18), and wader sub-Arctic (19).
UK = United Kingdom; NL = The Netherlands; EAF = East Atlantic Flyway.
European Red List status included as superscript after species name: EN = Endangered; VU = Vulnerable; NT = Near Threatened; LC = Least Concern (BirdLife International 2021).
For indicator derivation, UK data from WeBS except for †where Goose and Swan Monitoring Programme (GSMP). All WeBS-based data relate to UK, except for *Northern Ireland only, $Scotland only, £East England only.
Allocation of species/biogeographical populations to flyways, migratory pathways, and seasons (data periods) is based on Cramp and Simmons (Reference Cramp and Simmons1983), Spina et al. (Reference Spina, Baillie, Bairlein, Fiedler and Thorup2022), and Robinson and Clark (Reference Robinson and Clark2014), and other species-specific studies including: 1Scott and Rose (Reference Scott and Rose1996), 2Branson and Minton (Reference Branson and Minton1976), 3Davidson and Wilson (Reference Davidson and Wilson1992), 4Reneerkens et al. (Reference Reneerkens, Versluijs, Piersma, Alves, Boorman and Corse2020), 5Nicoll et al. (Reference Nicoll, Summers, Underhill, Brockie and Rae1988), 6Hallgrimsson et al. (Reference Hallgrimsson, Summers, Etheridge and Swann2012), 7Lopes et al. (Reference Lopes, Hortas and Wennerberg2008), 8Summers et al. (Reference Summers, Nicoll, Underhill and Petersen1988), 9Branson et al. (Reference Branson, Ponting and Minton1978), 10Beale et al. (Reference Beale, Dodd and Pearce-Higgins2006), 11Dick et al. (Reference Dick, Pienkowski, Waltner and Minton1976), 12Atkinson (Reference Atkinson1996), 13Engelmoer (Reference Engelmoer2008), 14Piersma et al. (Reference Piersma, Rogers, Boyd, Bunskoeke and Jukema2005), 15Boyd and Piersma (Reference Boyd and Piersma2001), 16Burton et al. (Reference Burton, Dodd, Clark and Ferns2002).
Abundance indicators of Arctic-breeding waterbirds
Data sourced from UK and the Netherlands
We used abundance data collected since 1975, from (1) the UK’s British Trust for Ornithology (BTO), Royal Society for Protection of Birds (RSPB), and Joint Nature Conservation Committee (JNCC) Wetland Bird Survey (WeBS), and (2) the Dutch waterbird monitoring scheme (coordinated by Sovon) that forms a part of the Network Ecological Monitoring programme. Both the UK (Frost et al. Reference Frost, Calbrade, Birtles, Hall, Robinson and Wotton2021) and the Netherlands (Hornman et al. Reference Hornman, Koffijberg, van Oostveen, van Winden, Louwe Kooijmans and Kleefstra2024) support a large proportion of the non-breeding populations of many waterbird species/populations in north-west Europe (Wetlands International 2023). Data for all species were available from both countries for the period 1975 to 2017. WeBS “core counts” are undertaken monthly according to standardised protocols, on a predetermined synchronised date, for up to 3,000 coastal and inland wetlands, with the number of sites surveyed increasing over time (Frost et al. Reference Frost, Calbrade, Birtles, Hall, Robinson and Wotton2021). The BTO/JNCC/NatureScot Goose and Swan Monitoring Programme monitors winter abundance and juvenile recruitment of migratory goose and swan populations (except for Dark-bellied Brent Goose, Svalbard Light-bellied Brent Goose Branta bernicla hrota, and European White-fronted Goose Anser albifrons albifrons, whose abundances are monitored by WeBS). In the Netherlands, standardised monthly counts on predetermined synchronised dates of all non-breeding waterbirds are undertaken at key wetlands including all EU Special Protection Areas (SPAs), important freshwater bodies, and additional agricultural areas that support relevant migratory goose and swan numbers (Hornman et al. Reference Hornman, Koffijberg, van Oostveen, van Winden, Louwe Kooijmans and Kleefstra2024). At most sites, this is undertaken from September to May, at a selection of sites over the entire year. For migratory geese and swans, productivity data are collected annually during autumn and early winter.
As part of the survey schemes, the data are used to calculate modelled annual “indices” in abundance for over 70 species/populations of waterbird in each country (as per indexing methods described in Frost et al. Reference Frost, Calbrade, Birtles, Hall, Robinson and Wotton2021; Hornman et al. Reference Hornman, Koffijberg, van Oostveen, van Winden, Louwe Kooijmans and Kleefstra2024; Woodward et al. Reference Woodward, Calbrade, Birtles, Feather, Peck and Wotton2024).To minimise the potential limitations that exist within the data collected from survey schemes, such as number of sites monitored, survey habitat coverage, and how well scarce species are monitored (Austin et al. Reference Austin, Banks and Rehfisch2007; BTO 2017), we only included species that are well-monitored and representative of national patterns, such as those principally estuarine species, and geese species (Table 1).
Calculation of annual indices in abundance
To ensure a consistent approach across data sets, we standardised the annual index value for a given species as the number of birds recorded in a specified period of the year expressed relative to the number present in the same period of a reference year. We used different monthly calendar periods relating to main non-breeding periods for individual species, accounting for species-specific life histories (Table 1). Monitoring within UK and Dutch schemes dates back to 1960, however monitoring across a wider species suite started from the mid-1970s. We therefore used 1975 as the reference year within indicators to mitigate against potential sample size biases in trends that could arise from considering too few species at too few sites early in the time series. We used generalised additive models (GAMs) with Poisson error terms to fit a smoothed trend curve to the annual counts, modelling count as dependent on year and adopted a pragmatic n/3 degrees of freedom, which produces a level of smoothing that, while removing temporary fluctuations not likely to be representative of long-term trends, captures aspects of trends that may be considered important (Atkinson et al. Reference Atkinson, Austin, Rehfisch, Baker, Cranswick and Kershaw2006; Fewster et al. Reference Fewster, Buckland, Siriwardena, Baillie and Wilson2000). Owing to the large variation in the size and number of wetlands where waterbirds are counted by WeBS in the UK, and thus their representativeness as sample sites, it is not possible to use a boot-strapping technique with the GAM to derive meaningful confidence limits.
Presentation of abundance indicators
We derived each multi-species abundance indicator as the geometric mean of the unsmoothed annual index values for study species or biogeographical population with the value in the base year, the first in the time series, set at one. The smoothed trend was then fitted through each resulting multi-species index. Between-year variation is highlighted by the unsmoothed index values, whereas smoothed trends better indicate the trajectory of changing status through time (e.g. Gregory et al. Reference Gregory, van Strien, Vorisek, Gmelig Meyling, Noble and Foppen2005). Geometric means rather than arithmetic means were used because an index change from 100 to 200 is equivalent (but opposite) to a decrease from 100 to 50. All species were weighted equally in each indicator. When positive and negative changes of indices are in balance, we expect their mean to remain stable. If more species decline than increase then the mean should go down, and vice versa. Thus, the index mean is considered a measure of change in migratory waterbird abundance.
Refinement of abundance indicators according to migratory pathway
A major challenge in deriving robust area-based indicators that use “non-Arctic” data is being able to confidently attribute breeding origin to waterbird species. We adopted a relatively conservative approach and allocated species and biogeographical populations to pathways based on an extensive body of evidence including published mark–recovery data, such as Cramp and Simmons (Reference Cramp and Simmons1977, Reference Cramp and Simmons1983), Delany et al. (Reference Delany, Scott, Dodman and Stroud2009), Scott and Rose (Reference Scott and Rose1996), Stroud et al. (Reference Stroud, Davidson, West, Scott, Hanstra and Thorup2004), and Spina et al. (Reference Spina, Baillie, Bairlein, Fiedler and Thorup2022) (see Table 1 and footnote; see also Supplementary material Table S1.1).
Having produced an indicator using all species/populations, separate trends were derived for wildfowl and waders, for distinct Western and Eastern migratory pathways, reflecting species’ different breeding origins: Canada/Greenland/Iceland/Faroes (Western), and Russia/Svalbard/Fennoscandia (Eastern), respectively, and high-Arctic and sub-Arctic, representing overall latitudinal breeding origin (Table 1). We further separated Western and Eastern indicators into “Wildfowl” and “Wader” given groupings of species as shown in Table 1. Second, we split Western and Eastern indicators by latitudinal breeding origin; for the Western pathway this equated to high-Arctic Canada/Greenland and sub-Arctic Iceland/Faroes, and for the Eastern pathway this equated to high-Arctic Russia/Svalbard and sub-Arctic tundra of Russia/Fennoscandia. Last, we separated “Wildfowl” and “Wader” indicators by latitudinal breeding origin, i.e. high-Arctic or sub-Arctic (Table1).
Analysis of abundance indicator values
For all abundance indicators, we calculated the natural log-ratio of change from smoothed trend values during four periods: five years (2011 to 2016), 10 years (2006 to 2016), 25 years (1991 to 2016), and 41 years (1975 to 2016), excluding the final year of the smoothed trend production that can be more sensitive to the GAM. Generalised linear mixed models (GLMMs) with a Gaussian distribution were used to test whether changes in smoothed indicator trend values varied consistently with: (1) taxonomy (wildfowl vs wader), (2) longitude of breeding origin (Western vs Eastern pathway), (3) latitude of breeding origin (high-Arctic vs sub-Arctic area), and (4) breeding location (UK vs the Netherlands). Models were fitted using R package glmmTMB (Brooks et al. Reference Brooks, Kristensen, van Benthem, Magnusson, Berg and Nielsen2017). In separate models for each period, we treated each species/biogeographical population as a random intercept and examined fixed effects of taxonomy, longitude, latitude, and non-breeding location as two-level factors in GLMMs. We also tested for interactions between longitude and latitude, group and latitude, and group*direction, finding these to be non-significant in each case (P >0.05), and are therefore not presented to ensure main effects are interpretable. To assess the contribution of variance partition for random effects and fixed effects within each period, we calculated the Intraclass Correlation Coefficient (ICC) using the performance package (Lüdecke et al. Reference Lüdecke, Ben-Shachar, Patil, Waggoner and Makowski2021), carried out for main effects models and models fitted with a random-effect only (ICC_re). There were no violations of model assumptions (heteroscedasticity and dispersion) using simulated residuals the R package DHARMa (Hartig Reference Hartig2022). Post-hoc model tests were used to calculate estimated marginal means for main effects, carried out using R package emmeans (Lenth Reference Lenth2024). All GLMM analyses were conducted using R 4.3.2 (R Core Team 2023).
Wildfowl productivity indicators
In contrast to most waders and other wildfowl, families of geese and swans associate together in winter and, because they are large-bodied birds and moult towards adult plumage proceeds progressively during the first winter, it is relatively easy to determine juvenile recruitment by counting the proportion of juveniles within autumn flocks (e.g. BTO 2023; Fox et al. Reference Fox, Ebbinge, Mitchell, Heinicke, Aarvak and Colhoun2010; Wood et al. Reference Wood, Newth, Hilton, Nolet and Rees2016), albeit that this does not account for unknown mortality during the southwards post-breeding migration, which can be large (Gupte et al. Reference Gupte, Koffijberg, Müskens, Wikelski and Kölzsch2019). We developed indicators of juvenile recruitment in species of Arctic-breeding geese and swans by calculating the geometric mean of the proportion of juveniles in populations in the Western and Eastern pathways (Table 2). Zero values of proportion are possible for species productivity indices but are not mathematically valid within the calculation of the geomean. Therefore, where zeros occurred, these were set as an arbitrary 0.01 value. As with the abundance indicators, we excluded data for Greylag Goose due to current confusion between Icelandic-breeding and British-breeding populations (Hearn and Frederiksen Reference Hearn, Frederiksen, Boere, Galbraith and Stroud2006). Although some information on juvenile proportions was available as early as 1958 (Table 2; Madsen et al. Reference Madsen, Cracknell and Fox1999), we started this indicator at 1985/86 to: (1) allow sufficient time for species/site coverage to establish within monitoring schemes; (2) balance species availability in both east and west to avoid undue bias in either indicator; (3) allow sufficient, and as much as possible unbiased, introduction of new species in both Eastern and Western indicator groups. For two species, European White-fronted Goose and Dark-bellied Brent Goose, two data sources from UK and the Netherlands were available for the same Eastern indicator (Table 2); these species were first averaged (geometrical mean) before taking the indicator value across all species. Productivity indicators were not split further for high-Arctic and sub-Arctic as only three species (Whooper Swan Cygnus cygnus, Pink-footed Goose (Iceland) Anser brachyrhynchus, and Tundra Bean Goose Anser serrirostris/A. fabalis rossicus) were allocated to the sub-Arctic category, thus the sample size was too low for meaningful indices. The productivity indicators for juvenile recruitment were graphically represented similar to the abundance indicators. Given that there were only two productivity indicators, trends were assessed for significance using a simple linear model (t-test) for each indicator over time using R 4.3.2 (R Core Team 2023).
Allocation of Arctic-breeding swans and geese to productivity indicators Western and Eastern, split further by data source from the UK or the Netherlands (NL); also shown are the start years for the trends available.
Results
Abundance indicators
Trends in abundance of species over the longer 25-year and 41-year time periods were more positive than more recent time periods as indicated by positive coefficients from GLMMs (Table 3), whereas trends were mostly negative within the 10-year period and mixed for the most recent 5-year period (Table 3, see also Figures 2 and 3). Over the 41-year and 25-year periods, trends were significantly more negative for wader than wildfowl species (Table 3 and Figures 2 and 4; see also Figure S1.2). However, this pattern was reversed in more recent time periods, with trends over five years being significantly more negative for wildfowl (and also more variable, Figure S1.1) than waders (Figures 2 and 4). Trends from the Western migratory pathway were consistently more positive than from the east over all time periods (Table 3 and Figures 2 and 5), significantly so across 41 years, with differences being more indicative (P <0.1) across 5 and 10 years. There was also an indication that variation in species’ trends over 25 and 41 years differed more with latitude (Figure 6) than the more recent period, with trends being less positive from the high-Arctic as opposed to sub-Arctic areas, significantly so for the 25-year period and a non-significant weak tendency (P = 0.1) for 41 years (Table 3 and Figure 6). Although not reaching significance in interaction (taxonomy*latitude, P >0.05), there was evidence for steeper declines over the more recent 5-year period for wildfowl using the Eastern pathway than the Western pathway (“West” ß-coefficient: 0.23 ± 0.07, P = 0.002; Table 3 and Figure 7). There was no significance shown for the interaction between taxonomy and latitude (P >0.05). However, there was some indication that trends were more negative for Eastern high-Arctic than Western high-Arctic species, with the sub-Arctic Eastern and Western indicators showing more similar trajectories (Figure 8), and in turn also reflected in an apparent greater decline for the high-Arctic wildfowl indicator (Figure 9). Trends over the longer two time periods were also biased lower for UK trends compared with the Netherlands, but this pattern was not significant (Table 3). The species’ population random effect term failed to account for any variance in the 5-year or 10-year trend model, but did in the 25-year (ICC = 0.431) and 41-year (ICC = 0.548) models, with a slightly greater variance accounted for in the 25-year model through species’ population random effects alone (ICC_re = 0.403) than in the 41-year model (ICC_re = 0.364). Consequently, in the latter two models, a greater partitioning of variance to the random effects reduced the independence of data points from the same species’ population, contributing to wider error margins for fixed effects for these two models (Table 3 and Figure 2).
Generalised linear mixed models (GLMMs) of Arctic-breeding waterbirds investigating the effects of taxonomy (wildfowl, wader), longitude (west, east), latitude (Arctic, sub-Arctic), and non-breeding location (UK, the Netherlands) over four different time periods. Reference categories are given alongside each effect, for example, for taxonomy being wildfowl, where a positive population coefficient indicates a more positive trend than wader; statistic designators are z-tests, and coefficients are given as beta (β) values ± 1 standard error; significance of terms (P) are given as: . P < 0.1; * P < 0.05, ** P < 0.01 and *** P < 0.001
Graphical representation of the main effects from the analysis of proportional change from GLMMs; each panel represents a modelled period, plotted as post-hoc estimated marginal means for the main effects of longitude, taxonomy, and latitude, with significance shown by asterisk notation (P <0.05 = *); black arrows represent 95% confidence intervals and red arrows indicate significance if not-overlapping.
East Atlantic Flyway Arctic-breeding waterbirds indicators, based on species annual indices, with smoothed trend; vertical lines represent GLMM analytical time periods for smoothed trends, as illustrated by horizontal arrows – see Table 2 for model output.
Indicators for Arctic-breeding wildfowl and waders in the East Atlantic Flyway, based on species annual indices, with smoothed trends; the East Atlantic Flyway Arctic-breeding waterbirds is shown for comparison; vertical lines indicate GLMM analytical time periods of smoothed trends (see also Figure 1).
Indicators for Western and Eastern pathways in the East Atlantic Flyway, based on species annual indices, with smoothed trends; the East Atlantic Flyway Arctic-breeding waterbirds is shown for comparison; vertical lines indicate GLMM analytical time periods of smoothed trends (see also Figure 1).
Indicators for Arctic and sub-Arctic regions within the East Atlantic Flyway, based on species annual indices, with smoothed trends; the East Atlantic Flyway Arctic-breeding waterbirds is shown for comparison; vertical lines indicate GLMM analytical time periods of smoothed trends (see also Figure 1).
Indicators for the Arctic region in the East Atlantic Flyway for wildfowl and waders, and the Eastern pathway for wildfowl and waders, based on species annual indices, with smoothed trend; vertical lines indicate GLMM analytical time periods of smoothed trends (see also Figure 1).
Indicators based on species indices (solid lines) and smoothed trend values (dashed lines for waterbirds breeding in Western Arctic and sub-Arctic and Eastern Arctic and sub-Arctic; vertical lines indicate GLMM analytical time periods of smoothed trends (see also Figure 1).
Indicators for the Arctic region within the East Atlantic Flyway for wildfowl and waders, and the sub-Arctic region for wildfowl and waders, based on species annual indices, with smoothed trend; vertical lines indicate GLMM analytical time periods of smoothed trends (see also Figure 1).
Productivity indicators
Composite productivity trends were highly fluctuating. Following placement of species in Eastern and Western migratory pathways (Table 3), the trends since 1985/86 suggested no clear long-term trend in the indicator for the Eastern pathway (East: t = -0.50, df = 31, P = 0.620, R 2 = 0.008), i.e. those from Russia, Svalbard or Fennoscandia (Figure 10), but showed a significant decline in the indicator for the Western pathway (West: t = -2.14, df = 31, P = 0.041, R 2 = 0.128).
Indices of productivity (juvenile recruitment into non-breeding adult populations) in Arctic-breeding geese and swans in Western and Eastern pathways of the East Atlantic Flyway.
Discussion
Abundance indicators
While difficult to generalise across all species, our abundance indicators suggest a recent change in the long-term trend for some Arctic-breeding waterbirds using the East Atlantic Flyway. This was perhaps most pertinent for some populations originating in Arctic Russia (Figure 6) using the Eastern migratory pathway, where there has been a downward trajectory since the mid-2000s (Figure 8). This pattern may be partly driven by wildfowl species within the Eastern migratory pathway (Figure 7), particularly those from the Eastern high-Arctic (Figures 8 and 9), that showed significantly more negative trends over the most recent five-year (2011/12 to 2016/17) period. Within the Western migratory pathway, declines were also apparent for a short period in recent years for some species from sub-Arctic areas (Figure 8); this mostly reflected concurrent declines (or temporary cessation of increases) between 2006/07 and 2011/12 for Icelandic-breeding populations of Pink-footed Goose, Black-tailed Godwit Limosa limosa and Golden Plover Pluvialis apricaria in the UK. However, as shown in Figure 8, the sub-Arctic Western indicator reverted to a positive trend in 2012, related to improved annual indices for those populations (Frost et al. Reference Frost, Calbrade, Birtles, Hall, Robinson and Wotton2021). The changes in the composite population trends represented in the indicators reflect demographic responses to pressures operating throughout species annual cycles, not only on their breeding grounds, but at stop-over sites and in non-breeding areas (Pearce-Higgins et al. Reference Pearce-Higgins, Brown, Douglas, Alves, Bellio and Bocher2017; Vickery et al. Reference Vickery, Ewing, Smith, Pain, Bairlein and Škorpilová2014). Indeed, for some species/populations included in our study (e.g. Iceland/Greenland Pink-footed Goose) there is evidence that changes in populations have been primarily driven by factors operating during the winter rather than the breeding season (Fox et al. Reference Fox, Madsen, Boyd, Kuijken, Norriss and Tombre2005). It is thus important to recognise that these indicators reflect factors operating throughout species’ annual cycles (Culp et al. Reference Culp, Cohen, Scarpignato, Thogmartin and Marra2017; Rakhimberdiev et al. Reference Rakhimberdiev, Duijns, Karagicheva, Camphusen, Castricum and Dekinga2018; Reneerkens et al. Reference Reneerkens, Versluijs, Piersma, Alves, Boorman and Corse2020; Zurell et al. Reference Zurell, Graham, Gallien, Thuiller and Zimmerman2018) and thus may not be a direct proxy for environmental conditions in the Arctic.
Wildfowl productivity indicators
For productivity, our indicators showed a decline for the Western pathway yet indicated relative stability for the Eastern pathway. Contrary to our results, for Arctic-breeding geese, Fox et al. (Reference Fox, Ebbinge, Mitchell, Heinicke, Aarvak and Colhoun2010) identified declines in juvenile recruitment for the Eastern pathway, whereas declines were significant in this paper only for the Western productivity indicator; however, Fox et al. (Reference Fox, Ebbinge, Mitchell, Heinicke, Aarvak and Colhoun2010) used a longer period than the start year of 1985 in this paper and, importantly, a more extensive set of species. Further scrutiny of these trends and differences within the flyway is therefore needed. The productivity indicators also revealed strong cyclical patterns; cyclical changes have also been recorded in reproduction and population density reflecting predator–prey population dynamics (Hanski et al. Reference Hanski, Hansson and Henttonen1991; Summers et al. Reference Summers, Underhill and Syroechkovski1998). This pattern can indirectly determine the size of waterbird populations via mutual predator abundance (Bêty et al. Reference Bêty, Gauthier, Giroux and Korpimakki2001; Blomqvist et al. Reference Blomqvist, Holmgren, Akesson, Hedenstrom and Pettersson2002). However, unlike other research (Aharon-Rotman et al. Reference Aharon-Rotman, Soloviev, Minton, Tomkovich, Hassell and Klaassen2014; Nolet et al. Reference Nolet, Bauer, Feige, Kokorev, Popov and Ebbinge2013), our productivity recruitment indicators for geese and swans (Figure 10) suggest that cyclical patterns of Arctic productivity may be on-going. This is in keeping with other large-scale assessments that suggest lemming cycles (Lemmus spp) are still taking place despite changing Arctic conditions (Gauthier et al. Reference Gauthier, Ehrich, Belke-Brea, Domine, Alisauskas and Clark2024); this would in turn explain continued cyclical changes in waterfowl productivity, although further formal analysis is required to confirm this effect.
Understanding drivers of change in the context of indicators
It was beyond the scope of this paper to analyse the specific factors driving changes in populations of Arctic-breeding waterbirds for individual species within the indicators. However, as noted, the patterns further underlying the indicator trends may be driven by changing ecological processes occurring throughout the species’ annual cycle (Piersma and Lindström Reference Piersma and Lindström2004).
The documented long-term increases in many waterbird populations in the UK (Woodward et al. Reference Woodward, Calbrade, Birtles, Feather, Peck and Wotton2024) and the Netherlands (Hornman et al. Reference Hornman, Koffijberg, van Oostveen, van Winden, Louwe Kooijmans and Kleefstra2024) up to the mid-late 1990s (see Figures 3 and 4) are thought to reflect the effects of multiple conservation policies to improve the quality of non-breeding habitats as well as land-use changes and positive response to agricultural change for some species (Fox and Abraham Reference Fox and Abraham2017). Such policies potentially include the establishment and management of protected areas (Gaget et al. Reference Gaget, Pavón‐Jordán, Johnston, Lehikoinen, Hochachka and Sandercock2021; Ostermann Reference Ostermann1998; Wauchope et al. Reference Wauchope, Jones, Geldmann, Simmons, Amano and Blanco2022), establishment of disturbance-free refuges within such sites (Fox and Madsen Reference Fox and Madsen1997), managed re-alignment (Mander et al. Reference Mander, Scapin, Thaxter, Forster and Burton2021), introduction of legislation to reduce former chronic and acute pollution (Osborn and Bull Reference Osborn and Bull1982), and the recovery of populations following protection of many species from hunting, especially following the national implementation of the 1979 EU Birds Directive (Tubbs Reference Tubbs1991, Reference Tubbs1996; Tubbs et al. Reference Tubbs, Tubbs and Kirby1992).The changes seen in waterbird populations and distributions also reflect demographic responses to changing climatic conditions both on the breeding grounds and in non-breeding areas. There is increasing evidence of climate change-driven changes in the abundance (Pearce-Higgins et al. Reference Pearce-Higgins, Dennis, Whittingham and Yalden2010) and phenology (Tulp and Schekkerman Reference Tulp and Schekkerman2008) of the invertebrate food supplies of high-Arctic and sub-Arctic breeding waterbirds that have the potential to impact their breeding success. However, there is uncertainty as to whether potential temporal mismatches between prey abundance and the hatch dates of waterbirds (Versluijs et al. Reference Versluijs, Zhemchuzhnikov, Kutcherov, Roslin, Schmidt and van Gils2024) may be offset by overall increases in the availability of prey (Chagnon-Lafortune et al. Reference Chagnon-Lafortune, Duchesne, Legagneux, McKinnon, Reneerkens and Casajus2024). Timing of breeding and subsequent productivity can also be directly affected by poor weather as well as the date of thaw of snow cover, particularly for waders (Boyd Reference Boyd1966; Green et al. Reference Green, Greenwood and Lloyd1977; Rakhimberdiev et al. Reference Rakhimberdiev, Duijns, Karagicheva, Camphusen, Castricum and Dekinga2018; Schekkerman et al. Reference Schekkerman, van Roomen and Underhill1998; Schmidt et al. Reference Schmidt, Reneerkens, Christensen, Olesen and Roslin2019). The links between timing of food abundance/phenology and fitness may be complex (Durant et al. Reference Durant, Hjermann, Anker-Nilssen, Beaugrand, Mysterud and Pettorelli2005; Zhemchuzhnikov et al. Reference Zhemchuzhnikov, Versluijs, Lameris, Reneerkens, Both and van Gils2021), for example, related further to duration and level of food abundance surrounding peaks and bird density (see Reneerkens et al. Reference Reneerkens, Schmidt, Gilg, Hansen, Hanson and Moreau2016).
There is also considerable evidence of changes in the non-breeding distributions of waterbirds in response to milder winters, reflecting increased survival and/or juvenile recruitment. For waterbirds from the Eastern high-Arctic, it is probable that indicators will reflect both the combined negative impacts of a changing Arctic climate and local declines in the UK and the Netherlands brought about by the changing distributions of wintering waterbirds in north-west Europe towards those breeding areas (Lehikoinen et al. Reference Lehikoinen, Jaatinen, Vähätalo, Clausen, Crowe and Deceuninck2013; Maclean et al. Reference Maclean, Austin, Rehfisch, Blew, Crowe and Delany2008; Nagy et al. Reference Nagy, Breiner, Anand, Butchart, Flörke and Fluet-Chouinard2021; Pavón-Jordán et al. Reference Pavón-Jordán, Clausen, Crowe, Dagys, Deceuninck and Devos2018, Reference Pavón-Jordán, Abdou, Azafzaf, Balaž, Bino and Borg2020). Improved monitoring along the flyway (Fox et al. Reference Fox, Nielsen and Petersen2018) and further collaborative flyway-scale research is needed to assess the impacts of global climate change and any possible implications for waterbird recruitment in order to influence relevant conservation policies.
Recommendations
We have demonstrated a novel way of generating indicators of the status of Arctic-breeding waterbirds in the East Atlantic Flyway by using non-breeding season monitoring data from the UK and the Netherlands for species using different migratory routes. In doing so, the indicators provide an indicative summary of the status of waterbirds from different parts of the Arctic. The International Breeding Conditions Survey on Arctic Birds (ABBCS) (http://www.arcticbirds.net/; Soloviev and Tomkovich Reference Soloviev and Tomkovich2024) has provided a standardised monitoring protocol that has enabled the collation of information on the environmental conditions at the breeding grounds of Arctic-nesting birds. However, it is not possible to independently verify trends through direct monitoring of these species’ breeding populations, owing to the logistical infeasibility of surveying huge areas of the high-Arctic and sub-Arctic where these species breed, typically at very low densities. This, indeed, is the rationale for the current approach, using data from the non-breeding season when the species are aggregated at high densities at relatively few, well-monitored sites. For many waterbird species that have been studied in detail, the derived trends are in accord with current understanding of past and current demographic changes.
As noted, the indicators reflect demographic responses to pressures operating throughout species annual cycles, and it is valuable to understand the relative importance of the drivers of population change at different times of the year, whether in the Arctic or in non-breeding areas. For example, this could be attempted through modelling drivers of populations at different times of the year to help reveal residual Arctic climatic patterns at the species and indicator level. However, we also acknowledge that there are further ways the composite trend indicators shown in this paper could be improved; some possible approaches are given below.
Broaden geographical scope
All countries along the East Atlantic Flyway are internationally important for waterbirds, either during migration and/or the non-breeding season. Many of these countries operate long-term waterbird monitoring schemes and contribute data to the IWC. Hence, additional waterbird data from outside the Arctic could be readily integrated into assessments of Arctic biodiversity. Given that Arctic-breeding waterbirds can shift their breeding (Rakhimberdiev et al. Reference Rakhimberdiev, Verkuil, Saveliev, Vaisanen, Karagicheva and Soloviev2011) and non-breeding (Lehikoinen et al. Reference Lehikoinen, Jaatinen, Vähätalo, Clausen, Crowe and Deceuninck2013; Maclean et al. Reference Maclean, Austin, Rehfisch, Blew, Crowe and Delany2008) distributions, composite trend indicators should use data from sufficiently large geographical areas to ensure that conclusions based on changes in trends are robust and representative. Here, we have initiated development of flyway-wide collaboration by combining data from the UK and the Netherlands, where concentrations of waterbirds are among the largest in Europe. Therefore, expansion of this collaborative effort incorporating information from other countries could be valuable to derive more representative indicators and increase the chance of detecting changes in abundance due to shifts in winter range.
For instance, this approach would better account for annual variation in the distribution of the total populations of some geese. This could be achieved by using data from Ireland for Greenland White-fronted Goose Anser albifrons flavirostris, Greenland Barnacle Goose Branta leucopsis, and Canadian Light-bellied Brent Goose, and data from Denmark and Belgium for Svalbard Pink-footed Goose. The accuracy/sensitivity of the indicators may be improved by expanding the spatial range of information included. A too narrow scope could compound issues of turnover of birds within and outside the area as currently presented, as some birds may winter further south (e.g. France, Iberian Peninsula) or birds may be shot and replaced by others further north. Further, major shifts in the distribution of waterbird species can occur within Europe during periods of severe winter cold (Ridgill and Fox Reference Ridgill and Fox1990), which could result in a misinterpretation of species trends if only considering counts from the UK and the Netherlands. However, pertinently, the IWC is undertaken in January only, thus being sensitive to extreme weather events and focuses on occurrences in wetland areas only. These aspects may limit the further potential sensitivity of indicators shown in this paper. Further evaluation would be needed to understand whether consistent results would be produced if the indicators produced here were based on January counts alone. If suitably sensitive data for wildfowl and waders existed for other countries (i.e. collected throughout the calendar year, or at least during multiple months), the indicators could be expanded to the whole flyway, or regional levels such as those in Deinet et al. (Reference Deinet, Zöckler, Jacoby, Tresize, Marconi and McRae2015), and would in turn help address any potentially confounding effects of range shift through shortening of the migratory corridor (“short-stopping”; Elmberg et al. Reference Elmberg, Hessel, Fox and Dalby2014).
Broadening the species pool and refinement of species information
Broadening geographical scope could lead to the inclusion of additional species such as seaducks (Grebmeier et al. Reference Grebmeier, Overland, Moore, Farley, Carmack and Cooper2006), thereby resulting in a larger group of species contributing to each indicator. However, consideration may be needed as to whether existing monitoring schemes provide adequate data for these species or whether additional scheme development would be needed (Cook et al. Reference Cook, Thaxter, Wright, Moran, Burton and Andrews2012; Hornman et al. Reference Hornman, Koffijberg, van Oostveen, van Winden, Louwe Kooijmans and Kleefstra2024; Meltofte et al. Reference Meltofte, Durinck, Jakobsen, Nordstrøm and Rigèt2006). While the breeding areas of waterbird species were sufficiently well described for the purposes of the classifications used in this study, there is also potential to improve knowledge of their migratory pathways through satellite telemetry and light-level geolocation (Minton et al. Reference Minton, Gosbell, Johns, Christie, Fox and Afanasyev2010; Catry et al. Reference Catry, Correia, Gutiérrez, Bocher, Robin and Rousseau2024; McDuffie et al. Reference McDuffie, Christie, Taylor, Nol, Friis and Harwood2022). Consequently, it may be possible to link non-breeding populations and sub-populations to breeding areas with increased confidence, thereby enhancing the sensitivity of these indicators. Furthermore, being able to increase the accuracy of links between breeding and non-breeding locations will help to better inform evaluation of drivers of change across the year.
Incorporation and improvement of productivity measures
The recruitment indicators in this paper are composite trends based on established goose and swan monitoring in the UK and the Netherlands, and there is potential to develop comparable programmes for other species groups. This highlights a long-recognised need (and opportunity) for coordinated and collaborative efforts across flyways to attempt to measure productivity in other Arctic-breeding wildfowl and waders (AEWA 2002; Davidson Reference Davidson1995; Davidson et al. Reference Davidson, Stroud, Rothwell and Pienkowski1998). Many waterbirds can be aged in the field and all species can be aged in the hand. For example, there have been several studies where juvenile recruitment data on waders collected during bird-ringing operations (Clark et al. Reference Clark, Robinson, Clark and Atkinson2004) and by field observations (Lemke et al. Reference Lemke, Bowler and Reneerkens2012) have been used to understand population change (Atkinson et al. Reference Atkinson, Clark, Clark, Bell, Dare and Ireland2003; Boyd and Piersma Reference Boyd and Piersma2001). A primary objective should be to establish a coordinated collection of juvenile recruitment data on waders and other waterbirds across a range of sites (Robinson et al. Reference Robinson, Clark, Lanctot, Nebel, Harrington and Clark2005; van Roomen et al. Reference van Roomen, Delany and Schekkerman2013), as recognised by the African-Eurasian Waterbird Agreement (AEWA) as a priority. The development of flyway-scale indicators summarising juvenile recruitment across a broad suite of species could also incorporate predictions of the future trajectory of population trends (especially when combined with estimates for annual survival in integrated population models (e.g. Johnson et al. Reference Johnson, Zimmerman, Jensen, Clausen, Frederiksen and Madsen2020). This would be helped by development of a shared online data repository for collating information from different sources, such as ringing information from participating ringing stations and observations of juvenile proportions by birdwatchers. Although such a data portal would be a considerable undertaking needing resourcing, perhaps under the auspices of IWC, it would represent important progress and complement the existing ABBCS (Soloviev and Tomkovich Reference Soloviev and Tomkovich2024). Similar endeavours that collate data, such as the US MAPS project, highlight the value of wider-scale collaborative efforts (Institute for Bird Populations 2023) that are invaluable for conservation. Pooling such information could also help in understanding biases associated with measuring waterbird productivity (Clark et al. Reference Clark, Robinson, Clark and Atkinson2004).
It should also be noted that the ratio of juvenile to adults, indicative of annual juvenile recruitment into the non-breeding population, depends on the number of adults in the population, i.e. if a population grows but has the same annual reproduction, the adult/juvenile proportion would decrease, but annual absolute reproduction would be the same. Further indicator developments could therefore also consider annual juvenile counts, but would need careful consideration of temporal and spatial heterogeneity in such data due to migration and further biases in site use by age groups, mindful also of potential identification issues of juveniles after their first moult when monitoring at distance in the field. Biases in crude age ratio assessments from the field could also be dealt with in more comprehensive analyses to avoid biased field observations (Jensen et al. Reference Jensen, Johnson and Madsen2023). Further, although waterbird productivity measures can indicate abundance (e.g. Minton et al. Reference Minton, Jessop and Hassell2012), large inter-annual population-level variation in reproductive success may nonetheless occur that could mask such patterns. For long-lived species such as waterbirds, annual adult survival could yield a more sensitive metric than productivity to indicate future population change (Robinson et al. Reference Robinson, Burton, Clark and Rehfisch2007).
Developing techniques to present indicators and assess representativeness
The lack of ability to generate confidence limits for this data set using conventional statistical bootstrapping approaches limits the certainty for increases or decreases in our indicator values. If methods could be developed to overcome this challenge in the future, then that would enable a more robust statistical analysis of changes in the indicator through time. Secondly, to assess representativeness of the species indicators, we propose that variation in the annual index values of individual species could be compared to those of the derived indicator. Anomalous species would thereby show up as being unrepresentative of the general situation.
Conservation importance and policy relevance
The development of composite trend indicators for Arctic waterbirds can be increasingly relevant to inform conservation policies, by identifying recent changes and potential responses to policy interventions such as effectiveness of climate change mitigation. They further meet the need for non-Arctic countries that host migratory species to share relevant data and information to inform Arctic assessment processes as explicitly requested by the Convention on Biological Diversity (2012) and as a fundamental obligation of Contracting Parties to AEWA. Indicators for migratory Arctic waterbirds fill a gap among existing European programmes and indicators, such as those describing the status of birds in farmland (Gregory et al. Reference Gregory, van Strien, Vorisek, Gmelig Meyling, Noble and Foppen2005), forest (Gregory et al. Reference Gregory, Vorisek, van Strien, Gmelig Meyling, Jiguet and Fornasari2007), and montane habitats (Lehikoinen et al. Reference Lehikoinen, Green, Husby, Kålås and Lindström2014), and those evaluating aspects of climate change (Newson et al. Reference Newson, Mendes, Crick, Dulvy, Houghton and Hays2009). Our indicators can therefore help support Arctic conservation strategies in two ways: (1) providing initial insight signalling potential changes in Arctic ecosystems and their ecological conditions among species biogeographical populations that warrant further investigation, complementing existing Arctic conservation initiatives and indicator development, and (2) initiating further discussion for indicator improvement methodologically, statistically, and ecologically. Multi-species indicators offer practical “ex situ” solutions to monitoring changing Arctic conditions.
Acknowledgements
Above all, we thank the thousands of enthusiastic volunteers who have contributed data to the count schemes and acknowledge the vital information on population structures generated by volunteer bird ringers over the years, with special thanks to Larry Griffin and Malcolm Ogilvie. Thanks also to Colette Hall of the Wildfowl & Wetlands Trust (WWT) for assistance with data requests. This manuscript has also benefited from comments provided by Rob Robinson, Jacquie Clark, Diana de Palacio, and Andy Musgrove. This work was funded by the UK’s WeBS; a partnership between BTO, RSPB, and JNCC. The UK’s GSMP was funded by the WWT, JNCC, and Scottish Natural Heritage (SNH). Waterbird and goose and swan monitoring in the Netherlands is coordinated by Sovon in association with Statistics Netherlands and funded by the Ministries of Agriculture, Nature and Food Safety and Infrastructure and Water Management and the 12 provinces. Additional funding support was provided by BTO.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0959270925100063.
Open access
