The Svalbard breeding population of pink-footed geese migrates via Norway to autumn and wintering areas in Denmark, the Netherlands and Belgium (Madsen 1984; Madsen et al. 1999). The population currently totals 76,000–81,000 birds (2013–2014), and has been steadily increasing over the last decades (Madsen et al. 2014). During the non-breeding period, pink-footed geese forage most commonly on energy-rich agricultural plant matter such as waste grain on cereal stubble fields, newly sown winter cereals, pastures, and newly sown spring cereal grain fields (Fox et al. 2005). Like other goose species, they prefer open areas that allow sufficient look-out for potential dangers (Madsen 1985a; Rosin et al. 2012). Nighttime roosting takes place on large water bodies or isolated islets in coastal areas, and most inland foraging flights are within a few kilometres of the roost, although distances up to 30 km are occasionally covered.
Prior to the nineteenth century, Filsø was a large, 2800-ha coastal shallow lake, and an important staging and wintering site for numerous species of waterbirds in western Europe (Jepsen 2012). During the 1800s and 1900s, several rounds of land reclamation gradually reduced the lake’s surface area to a mere 70 ha (what is nowadays known as Fidde Lake, Fig. 1). From 1947 until 2012, the entire drained lowland area served as intensive agricultural land, primarily used for growing cereals and seed grass (Lorenzen and Madsen 1986), and the highly cultivated open landscapes on the former Filsø Lake bed became one of the most important autumn staging sites for pink-footed geese (Madsen 1984; Madsen et al. 1999). In the early 1980s, over 90 % of the entire population was counted on night roost in Fidde Lake within this area on a single day (J. Madsen, unpubl. data), and even after considerable population growth in recent decades, this number was still above 50 % in the 2000s. Habitat use at Filsø is confined to agricultural areas, and geese forage on waste grain on stubble fields and graze on newly sprouted winter cereals (Madsen 1985b; Lorenzen and Madsen 1986).
In 2010, 2320 ha of the Filsø area was bought by the Aage V. Jensen Naturfond (a private Danish nature conservation and wildlife protection fund) with the objective of restoring Filsø Lake to a size of 915 ha water surface and surrounding natural area. Agricultural practice continued until 2012, when drainage was stopped, pumping stations removed and dikes leveled to the ground. In two phases (2 July and 7 October 2012), water was released back into the lowland, efficiently flooding the former fields to a depth of 1–3 m (Fig. 1).
Changes in goose numbers
As a measure of the importance of the Filsø staging site, we relied on the maximum autumn count of pink-footed geese over a 4-year period prior to restoration (2008–2011), during the restoration year (2012) and 2 years post re-establishment of Filsø Lake (2013–2014). Maximum autumn counts from Filsø for all years were extracted from the DOFbasen (the online portal of the Danish Ornithological Society [Dansk Ornitologisk Forening]), and in order to clarify potential changes in the importance of neighboring autumn staging sites in response to the Filsø restoration, similar extractions were made for three other major staging areas (please see section "Delineation of staging sites"). Maximum counts were derived from summaries of roost counts from known roost sites in each staging area within the same week. In the case of pink-footed geese, the DOFbasen holds data from both systematic counts by trained observers, who specifically observe pink-footed geese 1–2 times per week at all major staging sites, and reporting by amateur volunteers. All maximum counts used in this study were reported by the trained observers.
Response of site-faithful geese
Pink-footed geese have been subject to a long-term ringing scheme that began in the late 1980s, and since 1990 geese have been fitted with individually recognizable neck collars enabling tracking of individual birds to study migration, dispersal and demographics (Madsen 2001; Madsen et al. 2002). Birds have been caught by either cannon-net captures in Denmark during the spring or roundups of moulting family groups on Svalbard during the summer, and the percentage of ringed birds in the population has generally been around 0.5–1.0 % (Ganter and Madsen 2001). The presence of neck collars seems to have no long-term effect on the geese (Clausen and Madsen 2014). Identification of birds faithful to the Filsø area was based on the criterion that individually marked geese had been observed and reported at the site in at least 3 of the 4 years prior to the restoration event. This criterion is rather strict, as geese may of course have visited the area without being reported. Although somewhat arbitrary, this approach (1) ensured that included birds indeed returned to this study area over a period of several years, and (2) provided a study period short enough to guarantee that significant numbers of birds survived the entire period. For birds fulfilling this criterion, we analysed annual site use (presence/absence: whether an individual was or was not observed in any given staging area) in September–October (time of peak migration in Denmark) for three periods corresponding to pre-restoration (2008–2011), during restoration (2012) and post-restoration (2013–2014). To guarantee a comprehensive and consistent re-sighting effort, systematic re-sighting campaigns have been carried out during autumn along the entire flyway of this population since 1990, ensuring several re-sightings of live birds during autumn migration (Madsen et al. 2002). In addition, the systematic sightings were supplemented by observations reported online by voluntary observers via the homepage www.geese.org. As a consequence of the topical issues and recent changes at Filsø, both the analyses of goose numbers and response of philopatric geese suffered from small post-restoration sample sizes. In both cases we rely on the t test to analyse potential differences, as it has proven reliable with small sample sizes (de Winter 2013). In the current case, we believe that changes between the pre- and post-restoration periods are quite dramatic, and that the comparisons do not suffer from this limitation.
Delineation of staging sites
The autumn migration of pink-footed geese through Denmark is more or less north–south along the Danish west coast (Madsen et al. 1999), and the main staging sites along this corridor can be distinguished by latitudinal position. To elucidate potential changes in site use, we delineated all major staging sites (holding observations of >5 % of the population total during Sept.–Oct.) as a geographical space bounded by an upper and lower latitude. This ensured that all reported sightings of marked birds (with known latitude and longitude coordinates) could be traced to the correct staging site along the migration corridor. Based on this reasoning, we defined four major stopover sites in Denmark and a final wintering site in Friesland, the Netherlands (Table 1, inset of Fig. 1), corresponding to the most important autumn staging areas used by pink-footed geese (for details, see Madsen et al. 1999). Each of these sites is associated with separate night roosts in the form of a large water body or isolated islets.
Effects on spring body condition
Each spring in Denmark, a monitoring program of pink-footed geese body condition is conducted, wherein the abdominal profile index (API) of individual geese is categorised on a scale of 1 to 7. This measure is related to both body mass and total energy content (Madsen and Klaassen 2006), and is a broadly accepted proxy of goose body condition. All APIs are assessed by a small number of intercalibrated observers with extensive experience of scoring abdominal profiles in the field (detailed description given in Madsen and Klaassen 2006).
In the years post-restoration (2013–2014) we compared April (the main API assessment period) abdominal profile indexes from Denmark of geese philopatric to the Filsø area with those of geese never observed at Filsø during the study period. This analysis should reveal whether any long-term effects on body condition persisted in geese that lost Filsø as a major stopover site during autumn migration. If a bird was scored more than once in the same period, the average was used.