Introduction

‘Stopover sites’ are all sites where migrating birds are resting or feeding during migration (Warnock 2010). Introduced for Arctic-breeding waders, Piersma (1987) defined three different migration strategies: ‘hopping’, ‘skipping’, and ‘jumping’. Hopping involves short-distance hops between successive stopover sites, skipping refers to longer, medium-distance skips, and jumping involves non-stop jumps covering thousands of kilometres. Hopping is energetically less demanding and allows for greater flexibility in responding to, e.g. adverse (weather) conditions, by spontaneously extending stopovers or interrupting migratory flight (Grönroos et al. 2012). However, the less flexible jumping strategy is predominant among long-distance migrants breeding in the Arctic along the East Atlantic Flyway (EAF) (Piersma 1987; Zwarts et al. 1990; Catry et al. 2024); but there are also hoppers along this flyway, who minimise costs rather than migration duration through numerous and prolonged stopovers, resulting in short daily travel distances (Klaassen et al. 2012). Depending on the migration strategy, stopovers will probably serve different functions (Linscott and Senner 2021) and, thus, stopover patterns may vary. Hoppers and skippers are expected to exhibit short stopovers with low refuelling rates, whereas jumpers tend to display longer stopovers with higher refuelling rates (Warnock 2010).

Sex can also be a factor causing variation in stopover patterns. If males and females follow different migration strategies, it may result in, e.g. deviating stopover durations (Baert et al. 2018). In Pied Avocets (Recurvirostra avosetta; hereafter referred to as Avocet), older chicks are only cared for by one parent (Glutz von Blotzheim et al. 1986), which may have an influence on the departure from the breeding colony, with the non-caring parent leaving earlier.

The Avocet is a common breeding bird in the Wadden Sea area (Hagemeijer and Blair 1997; Koffijberg et al. 2022). A few Avocets also overwinter there, but most overwinter along the Atlantic east coast from the United Kingdom in the north to Guinea in the south (Glutz von Blotzheim et al. 1986; Blomert et al. 1990; Hötker 1998; Dietrich 1999). The wintering distribution has shifted in recent decades, with increasing numbers of Avocets wintering in France, Great Britain, and the Netherlands, and decreasing numbers in Portugal and Senegal (Blomert et al. 1990; van Roomen et al. 2020); a trend interestingly shared with other species, e.g. Eurasian Spoonbill (Platalea leucorodia; Lok et al. 2013). This trend towards more northerly wintering areas results in shorter migration distances, thereby increasing the time budget available for energetically less demanding hops and potentially facilitating a hopping migration strategy.

The hopping strategy requires the availability of high-quality stopover sites that can be used by the birds along their migratory routes (Piersma 1987; Rakhimberdiev et al. 2018). The Wadden Sea, a central hub for migratory birds along the EAF, is such a core site. Its high productivity, fuelled by the high nutrient influx from the North Sea and adjacent river systems, offers a rich perennial food supply, making it a critical roosting, moulting, and wintering area for seabirds and waders (Laursen et al. 2010). During autumn, 35%–53% of the entire EAF Avocet population uses the Wadden Sea as a stopover site (Laursen et al. 2010; Kleefstra et al. 2022).

The Wadden Sea is designated as UNESCO World Heritage Site (World Heritage Committee 2009) and thus is protected under the highest conservation category. However, it is not the only important site used by Avocets as a migrating bird species during its annual cycle. Therefore, we investigated the conservation status, as well as the habitat type, of Avocet stopover sites to determine their habitat preferences and to evaluate the connectivity of important sites against the background of international nature conservation measures.

In this study, we analysed the temporal and spatial stopover patterns of Avocets breeding in the German Wadden Sea during autumn migration along the EAF. We examined differences in departure dates related to sex and breeding site. We investigated the duration, usage, and characteristics of various stopover sites to identify critical sites for Avocets along their migration routes and assessed potential sex-related differences in migration patterns. We explored differences among stopover sites regarding the number of individuals using each site and the duration of stay. Our analysis covered both site-level and individual-level stopover patterns to detect inter-individual variability in migration strategies. Finally, we aimed to determine the predominant migration strategy of Avocets, including the number of stopovers and the duration of each stopover.

Materials and methods

Study area

The tracking data originated from two Avocet breeding areas in the Schleswig-Holstein Wadden Sea (Fig. 1). Neufelderkoog (8°59′14″E, 53°53′39″N) is a salt marsh area characterised by sheep grazing, where Avocets breed in vegetation structures of the upper salt marsh (Eskildsen 2022). The Sönke-Nissen-Koog (8°52′21″E, 54°36′46″N) is a conservation polder, connected to Hamburger Hallig, with a hands-off management strategy, where Avocets breed on the shore of a small pond between the dyke and the Hallig, and on the most-dykewards part where vegetation density is low due to sheep grazing.

Fig. 1
figure 1

Use of stopover sites in the Wadden Sea. Size of dot indicates mean site stopover duration; colour gradient indicates stopover ratio (n = 19); stars indicate Avocet breeding areas where tagging took place: Neufelderkoog and Sönke-Nissen-Koog (very close to HH)

Full size image

Global positioning system (GPS) tracking

A total of 23 adult Avocets were caught on their nests during incubation using walk-in-traps in 2020–2022. Each individual was weighed to the nearest gram, measured (tarsus, wing, and head-bill), ringed (colour code, metal ring), and equipped with a GPS-Global system for mobile communications (GSM) datalogger with a solar panel for battery charging, to enable long-time deployment (OrniTrack-9, Ornithology and Telemetry Applications (Ornitela), UAB, Vilnius, Lithuania). Loggers were fixed on the bird’s back using a body harness (Thaxter et al. 2014; see Borrmann et al. 2019). The GPS device (excluding harness weight) weighed 9 g, equivalent to 2.7% of the mean body mass of the caught Avocets (mean body mass ± standard deviation (SD), 331 g ± 31.7 g), which was less than the commonly accepted maximum of 3% additional weight to avoid behavioural effects of the device (Barron et al. 2010; Geen et al. 2019). In addition, feather samples were taken for molecular sexing, conducted by Tauros Diagnostik GbR, Bielefeld, Germany.

The data loggers recorded date, time, geographical position, and velocity at a changeable time interval. Communication with the devices via the GSM network allowed remote-controlled adjustment of time intervals and other device settings according to battery status, depending on light availability (time interval: 1 fix per minute − 1 per day). The data were sent directly via the GSM network to a database for storage (www.movebank.org, Max Planck Institute of Animal Behaviour, North Carolina Museum of Natural Sciences, University of Konstanz, Kays et al. 2022).

We lost connections with four of these individuals shortly after tagging, and data for 19 Avocets (11 males and eight females) were therefore used for the analysis, including 15 individuals at Neufelderkoog and four at Sönke-Nissen-Koog. Signals for three individuals were lost before leaving the Wadden Sea area and these were excluded from most analyses. All 19 individuals were included in the analysis of departure dates and the detailed map of the Wadden Sea stopover sites (Fig. 1), because adequate data were available for these analyses.

Definition of migratory variables

To analyse the stopover patterns of autumn migration, only GPS fixes assigned to autumn migration (max. range: 19 June–27 December, earliest departure from breeding site—latest arrival at wintering site) for the years 2020–2022 were considered. Migration tracks for 2 consecutive years were recorded for two of the 19 birds, but only the track of the first autumn migration was considered, to avoid bias due to multiple use of data of these two individuals. All other data points for all individuals, i.e. those assigned to breeding, wintering site, or spring migration, were removed from further analyses. Given that we aimed to analyse Avocet distribution between leaving the breeding and reaching the wintering site, we did not differentiate between moulting and staging when defining stopovers. Since birds with active wing moult usually do not migrate (de Boer et al. 2024), both behaviours might be considered in the analysis of stopover patterns.

Departure date was the last data point at the breeding area before the start of migratory flight. A movement pattern was defined as a stopover if the individual interrupted migratory flight for more than 1 h at a single site (Pederson et al. 2022). All stopovers between departure from the breeding area and arrival at the wintering site were considered. Stopover duration was calculated by defining arrival as the first data point at the stopover site after migratory flight ended, and departure from each stopover as the last data point at the stopover before the migratory flight started again. Stopover duration was the time between arrival at and departure from the respective stopover. Cumulated stopover duration was defined as the total time an individual stayed at all stopover sites used during its autumn migration. The mean individual stopover duration was calculated by dividing the cumulated stopover duration by the number of stopovers per individual, i.e. the average time an individual stayed at one stopover. The mean site stopover duration was defined by calculating the mean of all stopover events per stopover site. The stopover ratio represents the proportion of individuals that used a specific stopover site in relation to the total number of tagged individuals.

Use of habitat types and conservation status of stopover sites

The locations and borders of nature conservation sites along the EAF were identified using web databases: European Environment Agency (2022) for Natura 2000 sites in the European Union (EU), The Ramsar Sites Information Service (2024) for Ramsar sites, and BirdLife International (2024) for Important Bird and Biodiversity Areas (IBAs) worldwide. The conservation status of each stopover site was defined by intersecting the locations of the stopover sites with the borders of the conservation sites. We quantified the conservation statuses of Avocet stopover sites along the EAF by calculating the proportions of stopover sites assigned to the respective conservation category relative to the total number of stopover sites. We also calculated the proportion of stopover events occurring within each conservation category, as well as the proportion of total stopover duration spent within each category. Furthermore, stopover sites were categorised in 15 different classes of habitat types, such as estuary, river, or polder, by visual remote sensing (see Table 1 for a detailed list of habitat types).

Table 1 Use of different habitat types as stopover sites (n = 16)
Full size table

Statistical analysis

We examined sex-specific differences in stopover patterns in relation to the above parameters: number of stopovers, cumulated stopover duration, and duration per stopover. The influence of sex on stopover number and cumulated stopover duration was analysed by Mann–Whitney–Wilcoxon test.

We also generated a generalised linear mixed model (GLMM) using the R packages lme4 (Bates et al. 2015) and glmmTMB (Brooks et al. 2017), to test the influence of the predictor variable sex on the duration per stopover as response variable, with bird ID as a random factor. We examined differences in stopover duration between inland and coastal wetlands using a GLMM with wetland type as predictor variable. We ran a generalised linear model (GLM) to test the impact of sex and breeding site and its interaction as predictor variables on the response variable departure date. The most appropriate probability distribution was selected, based on the Akaike information criterion (Akaike 1998). Gaussian distribution was used for the variable departure date and Gamma distribution was used for the variable stopover duration.

We defined migration distance as the distance between the point of departure from the breeding site and point of arrival at the wintering site and calculated the great circle distance using Vincenty’s formulae in the R package geosphere (Vincenty 1975; Hijmans 2022). We tested if migration distance or stopover duration was correlated with number of stopovers using Spearman’s rank correlation.

All statistical analyses were conducted in R (R Core Team 2023). Data are presented as mean ± SD. Graphics were created using the R package ggplot2 (Wickham 2016) or Excel (Microsoft Corporation 2019) and maps were created in ArcGIS Pro 3.0.0 (ESRI 2022).

Results

Individual departure dates and stopover patterns

Autumn migration started between 18 June and 3 August (mean: 12 July ± 12 days). There was no significant difference in the start of migration between the two observation sites (p = 0.123; GLM; Fig. 2), but on average, individuals from Neufelderkoog (13 July ± 12 days) departed a few days earlier than individuals from Sönke-Nissen-Koog (8 July ± 7 days) (Fig. 2).

Fig. 2
figure 2

Differences in departure dates with respect to sex (A) and breeding site (B). Solid lines: medians; boxes: quartiles; whiskers extending up to 1.5 times the interquartile range from the median; crosses: means. n.s.: not significant, *p < 0.05, **p < 0.01, ***p < 0.001

Full size image

A total of 147 stopovers were counted for the 16 individuals (9.1 ± 5.1 stopovers per individual). On average, individuals spent 123.0 ± 45.2 days at stopovers during autumn migration (cumulated stopover duration). The number of stopovers varied from two to 23 stopovers per individual, and the cumulated stopover duration accordingly showed a wide range (25.5–164.7 days). Two different strategies were noted: some individuals made a few long stopovers, whilst others made more but shorter ones. One female with only two stopovers showed the highest mean individual stopover duration (41.8 ± 15.6 days). However, there was no significant correlation between cumulated stopover duration and number of stopovers (ρ = 0.083; p = 0.380; Spearman’s rank correlation), or between migration distance (mean: 1987.0 ± 1273.0 km) and number of stopovers (ρ = −0.125; p = 0.677; Spearman’s rank correlation).

Focusing on the most important stopover sites along the EAF, i.e. sites where more than two individuals stopped, showed that the choice of stopover sites was not consistent among individuals (see Fig. S5 in Electronic Supplementary Material). The migration routes, i.e., the successive use of stopover sites, as well as stopover duration at single sites differed markedly. No clear pattern was detected, except that most individuals spent most of their cumulated stopover duration in the Wadden Sea area (mean: 82% ± 19%) using different stopover sites on their westward migration at the early stages of autumn migration.

Females left breeding areas significantly later than males (females: 17 July ± 10 days; males: 9 July ± 11 days; p = 0.041; GLM). There was no difference between males and females regarding the number of stopovers (males: 7.0 ± 2.1 stopovers; females: 11.3 ± 6.4 stopovers; W = 46; p = 0.138; Mann–Whitney–Wilcoxon test) or cumulated stopover duration (males: 143.5 ± 22.5 days; females: 102.6 ± 53.9 days; W = 18; p = 0.161; Mann–Whitney–Wilcoxon test), but males stayed significantly longer at each specific stopover site (males: 20.5 ± 32.4 days; females: 9.0 ± 15.2 days; p = 0.001; GLMM) (Fig. 3).

Fig. 3
figure 3

Stopover patterns in relation to sex: A Number of stopovers per individual (n = 16). B Cumulated stopover duration per individual (n = 16). C Mean individual stopover duration (n = 19). The following outliers were removed for reasons of clarity: female: 14 outliers (range: 25.7–59.9 days); male: 12 outliers (44.9–139.2 days). For boxplots and significance level, see Fig. 2. For definitions of migratory variables, see Methods

Full size image

Characteristics of stopover sites

Overall, 63 different stopover sites were approached by the Avocets, all located along the EAF from the Ho Bugt (DK; 55°35′23″N, 8°16′45″E) in the north to the Roneraie de Dionguer (SN-S; 12°55′38″N, 15°55′07″W) as the southernmost site (Fig. 4). On average, stopover duration was 13.4 ± 23.8 days, with wide variations from 1 h to 140 days (Jade Bay, DE-LS; 53°27′57″N, 8°06′15″E). The shortest mean site stopover duration was observed at Valdeduey O Navajos River (ES-NW; 0.04 days; 42°08′56″N, 5°03′38″W) and the longest was observed at Senegal River (SN-N; 59.9 days; 16°27′49″N, 15°57′38″W).

Fig. 4
figure 4

Use of stopover sites along the EAF. Size of dot indicates mean site stopover duration; colour gradient depicts stopover ratio (n = 16). Stopover sites pooled in regions. SH Schleswig–Holstein, LS Lower Saxony, N North, NW North–West, C Central, S South. Abbreviations of the countries refer to ISO 3166–1

Full size image

The Wadden Sea areas of Schleswig-Holstein, Lower Saxony, and the Netherlands were frequently used as stopover sites, mainly as the first stopovers after leaving the breeding area (Fig. 1). Two Avocets (stopover ratio: 12.5%), however, started their migration northwards and stayed in the Danish Wadden Sea for 23.3 ± 19.2 days before migrating southwards following the EAF (Fig. 5). The most frequently used stopover sites in the southern Wadden Sea were the Weser Estuary (81.3%; mean site stopover duration: 24.8 ± 34.2 days; 53°32′31″N, 8°31′19″E), Jade Bay (68.8%; 35.6 ± 36.0 days), Ems-Dollart (56.3%; 8.5 ± 15.9 days; 53°15′51″N, 7°10′18″E), and Westhoek/Zwarte Haan (Dutch Wadden Sea) (56.3%; 37.9 ± 37.4 days; 53°19′01″N, 5°36′46″E) (see Table S1 Electronic Supplemntary Material).

Fig. 5
figure 5

Absolute stopover durations of tagged individuals (n = 16) at different stopover regions, sorted from north to south. SH Schleswig–Holstein, LS Lower Saxony, N North, NW North–West, C Central, S South. Abbreviations of the countries refer to ISO 3166-1

Full size image

Further south, the French Atlantic coast was an important stopover region, with seven of 16 individuals (43.8%) that started migration at breeding areas stopping here (4.6 ± 7.8 days), of whom five did not proceed and overwintered in France. Baie de l’Aiguillon (46°16′03″N, 1°09′01″W) was the most frequently used stopover site at the French Atlantic coast (31.3%; 6.2 ± 7.8 days). Stopover patterns then diversified crossing the Iberian Peninsula, where six individuals overwintered, including four in Portugal and two in Spain. Only three Avocets followed the EAF further south to overwinter in Western Africa (see Fig. S1 in ESM). Maps of the detailed stopover sites at France, Iberian Peninsula, and West Africa are shown in Online Resource 1 (Figs. S2–S4)

Estuaries played an important role as stopover sites for Avocets during autumn migration. Twelve different estuaries along the EAF were used 43 times, accounting for nearly a third of all recorded stopovers, with a stopover duration of 12.9 ± 24.2 days (Table 1). Other important habitat types used as stopover sites were bays (33 stopovers; 22.2 ± 28.9 days) and seacoasts (29 stopovers; 14.7 ± 27.4 days). The most frequently used inland habitat type was river (10 stopovers; 9.7 ± 18.5 days). Classifying all stopover sites as inland wetland (n=16) or coastal wetland (n = 47) based on their geographical location revealed a preference for coastal compared with inland wetlands as stopover sites (see Table S1 in ESM). Of the overall 147 recorded stopovers, 130 were made at coastal wetlands. Duration per stopover, however, was similar for the two wetland types (coastal: 14.4 ± 24.7 days; inland: 6.1 ± 14.4 days; p = 0.054; GLMM; Fig. 6).

Fig. 6
figure 6

Duration per stopover according to wetland type. Number of stopovers in brackets. The following outliers were removed for reasons of clarity: coastal: 20 outliers (range: 38.9–139.2 days); inland: 4 outliers (range: 10.2–59.9 days). For boxplots and significance level, see Fig. 2

Full size image

Fifty-eight of the 63 stopover sites were declared as IBAs (92.1%). Avocet is defined as an important species for 68.3% of these areas, and 63.5% of the stopover sites are Ramsar sites. Considering only stopover sites within the EU, 41 of 49 were Natura 2000 sites (83.7%). All the sites were within the Birds Directive, which aims to preserve, restore, and recreate habitats of rare and endangered bird species (European Parliament 2019), and included Avocet as an important species. These proportions were even higher for the other stopover variables: 96.6% of all stopovers were taken at IBAs and 99.7% of the cumulated stopover duration was spent at IBAs. For the two categories IBAs with Avocet focus (83.7%, 98.0%, respectively) and Ramsar sites (82.3%, 94.6%, respectively) the difference between these two proportions was even more distinct (Table 2).

Table 2 Use of conservation sites as stopover sites (n = 16 Avocets)
Full size table

Discussion

Migratory strategies

Departure timing of Avocets in the present study is in accord with the study by Glutz von Blotzheim et al. (1986), who reported autumn migration departure dates beginning in mid-June and peaking in mid-July. In contrast, a population along the East Asian-Australasian Flyway (EAAF) had a much later start of autumn migration, with a mean of 23 October (Wu et al. 2022). This difference is likely due to the more southerly location of the breeding area at 39°N compared with 54°N for the Wadden Sea population and consequently shorter mean migration distance to suitable wintering areas (1124 km EAAF vs 1987 km EAF), allowing for a much later start of migration.

Avocets from the EAAF population spent 5–43 days at stopover sites (Li et al. 2024), staying an average of 36 days at their main stopover site, Lianyungang (Wu et al. 2022). The stopover duration of 13.4 days per stopover for the EAF population observed in the present study falls within this range, as does the stopover duration at the main stopover sites in the Wadden Sea (e.g. Westhoek/Zwarte Haan: 37.9 days). The individual that made the longest stopover at Senegal River (59.9 days) stayed there from 5 October to 4 December and then flew westward from this more-inland site to the Atlantic coast, and its signal was lost 4 days later. It therefore cannot be certain that this was a stopover; it might already have been in its wintering area and returned there after a short westward trip. The second longest stopover was at Westhoek/Zwarte Haan, as mentioned above.

The GPS fixes per day decreased for all individuals to the end of the year due to a lower charging rate. In the Wadden Sea area a median of 220.2 GPS fixes per day per individual was recorded, after leaving the Wadden Sea the median decreased to 66.9 GPS fixes per day. Since some devices were not able to maintain a sub-hourly sampling interval, we may not have observed every single stopover event of each individual, and therefore slightly underestimated temporal stopover parameters. Nevertheless, our findings reveal more stopovers (9.1 stopovers) and longer cumulated stopover durations for Avocets compared with other species. For instance, Lesser Black-backed Gulls (Larus fuscus) made 3.4 stopovers during autumn migration, staying 22.2 days per stopover (Klaassen et al. 2012), and Grey Plovers (Pluvialis squatarola) spent only 8–15.8 days in total at 3.3–4.8 stopovers during autumn migration (Catry et al. 2024). Eurasian Curlews (Numenius arquata) made more but very brief stopovers (5.4 stopovers; 31.8 h per stopover), probably due to their short migration period of 10.9 days (Pederson et al. 2022). This contrasts with findings by Wijethunge et al. (2024) for the EAAF population of Avocets, which just occasionally make 1–2 stopovers during autumn migration. The comparisons between our results and the referred studies have to be interpreted cautiously, due to deviating definitions of stopovers (depending on tagging intervals).

Moreover, we observed some pre-migratory movements possibly related to moulting, e.g. to Rømø dam, a well-known moulting area for Avocets with peak abundance in early August (Hötker and Frederiksen 2001). However, because many waders migrate to appropriate stopover sites for moulting and moulting is therefore an essential part of migration (Kjellén 1994; Hötker and Frederiksen 2001; Cimiotti 2023), we explicitly included moulting migration in the analyses. Our findings strongly suggest that Avocets make more stopovers and spend more time at stopover sites in total than some other (wader) species, probably related to their migration strategy.

Avocets displayed two different types of migration: some tended to make a few long stopovers and others made more but shorter ones. Both strategies, however, have in common that all tagged Avocets spent a high proportion of their cumulated stopover duration (82%) in the Wadden Sea region. Piersma (1987) described these different migration strategies as jumping and hopping. Hopping is energetically cheaper because it does not require birds to carry all the fat reserves needed for a long non-stop flight; however, jumping and skipping (mid-distance flights with short stopover periods) are the predominant strategies among long-distance migrants breeding in the Arctic, due to a lack of high-quality stopover habitats along their migration routes (Piersma 1987; Zwarts et al. 1990; Catry et al. 2024). Notably, this did not seem to hold true for Avocets: the higher number and longer duration of stopovers compared with other waders suggests good availability of high-quality stopover habitats for Avocets in Europe, allowing them to perform a hopping migration.

Stopovers serve various purposes, including energy accumulation (Lindström 1991), physiological recovery (Skrip et al. 2015), avoiding adverse environmental conditions for migration (e.g. headwinds; Grönross et al. 2012; Carneiro et al. 2020), and social interactions or information gathering (Bauer et al. 2020; Linscott and Senner 2021). The relative importance of these purposes may differ between hoppers and jumpers. Jumpers exhibit high refuelling rates during their comparatively long stopovers (Warnock 2010). In contrast, hoppers, who exhibit low refuelling rates, may rely more on other drivers like favourable weather conditions. Consequently, the necessity for multiple high-quality stopover habitats along the migration routes, offering rich food resources, appears to be less critical for hoppers. Lisovski et al. (2024) predicted that hoppers could buffer habitat loss by redistributing stopover sites to novel or less-used sites. The high cumulated stopover duration spent in the Wadden Sea suggests that other stopover sites along the EAF are less critical and therefore more interchangeable in the event of habitat loss. Avocets are known to display certain flexibility in choosing their breeding sites in response to habitat transitions (Glutz von Blotzheim et al. 1986), and a hopping migration strategy may enable them to extend this flexibility to their migration behaviour.

Stopover patterns, however, are influenced by factors other than migration strategy, such as the prevailing wind regime or other weather conditions. For example, departure timings from stopovers and thus stopover durations for Bar-tailed Godwits (Limosa lapponica) and Dunlins (Calidris alpina) depended on the prevailing wind (Grönroos et al. 2012). Grey Plovers wintering in Portugal or France took no additional stopover between the Wadden Sea and their wintering sites, whilst individuals wintering in Guinea-Bissau made intermediate stopovers in Iberia or West Africa (Catry et al. 2024). This suggests that migration distance also influences migration strategy. In contrast, this was not the case for Avocets, which showed no correlation between cumulated stopover duration or migration distance and the number of stopovers.

In general, we detected wide variations in all the examined stopover patterns in Avocets, indicating high inter-individual variability. Inter-individual variation is common in bird ethology, and birds are often considered to have individual personalities (Nicolaus et al. 2012; Cockrem 2013), which might influence migration timing. For instance, slower-exploring Red Knots (Calidris canutus) arrived approximately 10 days earlier at marine non-breeding areas compared to faster-exploring individuals, and non-stop flying individuals had lower exploration scores than those using other migration strategies (Ersoy et al. 2024). The still comparatively short duration of the current dataset means that we were unable to determine intra-individual repeatability in Avocet stopover patterns; however, Avocets have shown high breeding-site fidelity (Dietrich 1999). If this fidelity is also applicable to their stopover sites, it supports the assumption that high inter-individual variability might explain the wide range of stopover patterns.

Within the Wadden Sea, Avocets clearly adopted a hopping migration strategy, moving between nearby sites. After leaving this region, no consistent pattern could be discerned. We thus propose that Avocets may not strictly conform to one of the migration strategies defined for long-distance migrants breeding in Arctic regions (Piersma 1987), and are strongly influenced by high inter-individual variation in stopover patterns.

Sex-related differences of stopover patterns

In general, biparental care or male care is a common practice in shorebirds (Szekely and Reynolds 1995), leading to the assumption of no sex-related differences in departure dates or an earlier departure of females from breeding sites. In contrast, we observed a significantly later departure for females compared with males. In Avocets, older chicks are only cared for by one parent (Glutz von Blotzheim et al. 1986). Consequently, our results suggest that females primarily undertake this responsibility. Wu et al. (2022), however, found no sex-related difference in migration timing or distance for Avocets migrating along the EAAF, whilst other studies have shown varying patterns of sex-related differences in migration timing. For example, female Red Knots arrive around 8 days earlier at their most important stopover in the Wadden Sea area (Ersoy et al. 2024). Conversely, male Lesser Black-backed Gulls arrive at their wintering sites 15 days earlier and depart earlier (though not significantly) from the breeding grounds than females (Baert et al. 2018). Sex-related variation in departure dates may thus depend on the species’ chick-rearing system, and may not be generalisable across species.

In addition, we found that males spent longer per stopover on average than females, which might be explained by the later departure of females. A later departure shortens the time available to reach their wintering sites, potentially reducing stopover duration. Female Avocets tended to make more stopovers but have slightly shorter cumulated stopover durations than males, although the differences were not significant. Wright et al. (2022) observed a similar sex-related migration pattern in Dunlins (C. a. hudsonia), attributing it to males adopting a more relaxed autumn migration compared to the rushing migration strategy used to reach breeding grounds quickly in spring. However, since Hötker and Frederiksen (2001) found no sex-related difference in the time Avocets spent at Rømø for moulting, and no sex-related differences in stopover number and duration were observed for Eurasian Curlews (Pederson et al. 2022), these patterns could also reflect high individual variability, as Muraoka et al. (2009) assumed for Wood Sandpipers (Tringa glareola). Interestingly, Lesser Black-backed Gulls exhibited an inverted sex-related stopover pattern, with females having a higher mean individual stopover duration during autumn migration (Baert et al. 2018); however, females also showed a later departure, similar to our observations for Avocets, contradicting the explanation of a shortened time contingent.

The current results thus found no clear sex-related pattern, suggesting that other, as yet unknown factors, besides departure date, must influence sex-related stopover behaviours. A larger sample size of GPS-tagged individuals would be necessary to evaluate possible sex-related differences in Avocet’s stopover patterns in greater detail.

Connectivity of wetland areas along the EAF

Our results highlighted the critical importance of the entire Wadden Sea for the Avocet. The four most frequently used stopover sites were located within this area, with each individual making at least one stopover here. The current telemetry data thus confirm what counting data and ring recoveries already indicated: 35%–40% of the whole EAF Avocet population utilises the Wadden Sea for breeding, moulting, or staging (Kleefstra et al. 2022). This intensive use by Avocets aligns with the behaviour of many other migrating shorebird species, e.g. Grey Plovers had their highest cumulated stopover duration at Wadden Sea sites (Catry et al. 2024), and all important stopover sites for Lesser Black-backed Gulls were also in Northern Europe, but not primarily in the Wadden Sea (Baert et al. 2018).

The most important stopover sites for Avocets were the Weser Estuary, Jade Bay, Ems-Dollart, Dutch Wadden Sea, and Baie de l’Aiguillon, which were utilised by at least 30% and up to 80% of the tagged individuals. Grey Plovers showed a similar pattern, using the Schleswig-Holstein, Lower Saxony, and Dutch Wadden Sea areas, as well as the Tejo Estuary (Portugal) and Diawling National Park (Mauritania) (Catry et al. 2024). Although the Tejo Estuary also served as stopover and wintering site for some Avocets, they did not use Diawling National Park, but did use three nearby sites. Despite slightly different habitat demands (Glutz von Blotzheim et al. 1984, 1986; Bauer et al. 2005), Grey Plovers and Avocets show a distinct overlap of stopover sites, which might suggest a scarcity of high-quality stopover habitats.

The diversification of stopover patterns from the Iberian Peninsula onwards might be partially explained by potential data distortion caused by six individuals whose signals were lost between 25 September and 16 January at the end of autumn migration. Given that they had already reached an area known to host wintering Avocets and stayed there for a prolonged period before their signals were lost, we included these individuals in our analyses; however, additional data might have shown that some of these individuals continued their migration to West Africa, using the same stopover sites as the other two individuals that overwintered there.

Estuaries emerged as the most critical habitat type for Avocets during autumn migration, accounting for one third of all recorded stopovers (e.g. Elbe Estuary, Loire Estuary, Weser Estuary). This preference is supported by observations in the United Kingdom, where Avocets exclusively overwintered in estuaries, hosting around 5% of the whole EAF population, with a peak in abundance in November (Rehfisch et al. 2003). Post-breeding distributions in the Wadden Sea from August to November were also concentrated in estuarine areas like Jade Bay, Ems-Dollart, Leybucht, and the coastline of the Dutch province Friesland (Laursen et al. 2010). Estuaries are crucial due to their sediment structure, providing the high-mud-content areas necessary for Avocet’s specialised scything foraging technique (Glutz von Blotzheim et al. 1986; Moreira 1999); at low tide, Avocets sweep their beaks laterally through the remaining water or upper sediment layers, a technique that requires relatively low sediment resistance.

Estuaries are coastal wetlands, supporting the observed preference of Avocets for coastal rather than inland wetlands. Most stopover sites for Avocets migrating along the EAAF were also located at coastal wetlands, with few inland wetland sites (Wu et al. 2022; Wijethunge et al. 2024). Contrary to other populations where males preferred inland wetlands (at least for wintering; Wu et al. 2022), our data showed that all inland stopovers were made by females, except for one male. Most inland stopovers occurred at river shores, lagoons, or reservoirs where Avocets could also find shallow water zones for foraging, potentially reducing competition.

Our results also revealed that a high proportion of stopover sites were recognised as being worthy of protection (92.1% IBA) or were designated as nature conservation sites (63.5% Ramsar). Except for Baie de l’Aiguillon (a Natura 2000, Ramsar, and IBA site; protection status of Avocets under the Birds Directive: reproduction, wintering, concentration), all five most frequently used stopover sites were located in the Wadden Sea, part of a national park and a UNESCO World Heritage site where Avocets are specifically protected under the Natura 2000 Birds Directive (reproduction, concentration). Only five stopover sites used during autumn migration were neither recognised as worthy of protection by BirdLife International nor had any international conservation status: three sites in Spain, one in France, and one in Senegal. These sites were only approached once for a maximum of 2.6 days, indicating that they were not of high importance. Similarly, Catry et al. (2024) found that all Grey Plover stopover sites between the Wadden Sea and West Africa were classified as IBAs, with 60% being Ramsar sites. Avocet stopover sites thus had a slightly lower rate of IBA recognition (100% vs 92.1%) but a slightly higher rate of Ramsar designation (60% vs 63.5%).

The Avocet population has shown one of the most declining trends among migratory and wintering waterbirds in the Wadden Sea, with a 40% decrease from the 1990 s to the 2010 s (Kleefstra et al. 2022). Although the entire EAF breeding population trend has been relatively stable, there has been a significant decline in the Wadden Sea (Kleefstra et al. 2022; Koffijberg et al. 2022). Different pressures affect migrating waterbirds along the EAF, including sea level rise, coastal squeeze, and disturbances at high-tide roosts, mainly due to humans, predators, or air traffic in Europe, and litter/garbage and fishing in Africa (van der Kolk et al. 2020; Crowe et al. 2022; van Roomen et al. 2022; Kleefstra et al. 2023). Avocets are listed in Annex I of the EU Birds Directive and are thus subject to special conservation measures, establishing a ‘responsibility to protect’ (European Parliament 2019), which is taken seriously within the EU: Crowe et al. (2022) observed that 95% of assessed sites in Northern Europe and 100% in Iberia/North Africa were internationally designated as conservation areas. Unfortunately, the effectiveness of conservation measures in Africa is lower (ca 60%). Our study showed that tagging data were able to identify a high number of sites currently lacking a high conservation status, i.e. Ramsar vs IBA sites. While 92.1% of stopover sites are currently considered worthy of protection, it is essential to confer conservation statuses (e.g. Ramsar or Natura 2000) that mandate the implementation of conservation measures by the respective administrations to maintain the Avocet’s EAF population at a stable level.

Furthermore, Avocets appeared to mainly show highly aggregated distribution patterns close to the coastline at each of the most frequently used stopover sites. At Jade Bay for example, they almost exclusively used the western part of the bay in the south of Wilhelmshaven, whereas the main aggregation at the Dollart occurred in the eastern part. Maps of the distribution patterns at different stopovers sites are shown in Online Resource 1 (Figs. S6–S8). A similar situation was also observed at Banc d’Arguin (Mauritania), where Avocets consistently used the same parts of tidal flats with tidal pools and creeks (Blomert et al. 1990). These aggregated distributions suggest a reliance on habitat-specific conditions, such as prey availability and sediment structure. In Jade Bay, the western part has a higher mud content, essential for Avocets’ specialised foraging technique (Moreira 1999), compared to the eastern part (Schückel et al. 2013). In addition, the benthos community differs between these areas: Hydrobia ulvae dominated in the east, whilst Hediste diversicolor, an important prey species for Avocets (Enners et al. 2019), was abundant in the west (Schückel et al. 2013). This strong habitat dependency underscores the need for high-quality stopover sites along the EAF. Subsequent studies should focus in detail on the influence of benthos community and sediment structure on Avocets’ habitat choice at stopover sites, potentially explaining their aggregated distribution patterns.

Conclusions

Overall, this study emphasises the critical importance of stopover sites for Avocets, highlighting that the migratory route and the quality of approached sites along the EAF are integral to their annual cycle, with individuals spending one-third of the year at stopover sites during autumn migration. The large number of stopovers compared with other waders might suggest a sufficient availability of high-quality stopover habitats for Avocets in Europe, contradicting a stated lack of such habitats that forces other waders to adopt jumping or skipping migration strategies (Piersma 1987). However, the declining numbers of the Avocets in the Wadden Sea, both as breeders and migrants, observed over recent decades (Kleefstra et al. 2022; Koffijberg et al. 2022), raise concerns about whether these trends are linked to diminishing stopover site quality —an aspect that remains speculative. Comprehensive international studies are required to evaluate stopover site quality, particularly prey availability, to improve our understanding of high-quality stopover site availability along the Avocets’ migration routes. The Avocets’ reliance on specific sites, coupled with population declines, underscores the need for effective conservation measures to safeguard their preferred habitats.