Distribution and numbers of wintering waterbirds in Sweden in 2015 and changes during the last fifty years

Coordinated midwinter counts of waterbirds have been undertaken on a European scale since January 1967, when the midwinter counts of the International Waterfowl Census (IWC) were started in a number of countries including Sweden (Atkinson-Willes 1969, Nilsson 1975, Gillisen et al. 2002; see Nilsson 2008 for more references). Before that, Sweden and some other countries had been running specific national programs for a number of years (AtkinsonWilles 1963, Owen et al. 1986, Nilsson 1968, 1976). It was, however, soon realized that international cooperation was needed and accordingly the IWC was initiated by Wetlands International. Some of the major aims of this project are to get population estimates and trends both on a regional and global scale and to map the distribution of the different species. The IWC also aims to provide background information for international agreements relating to management of wetland birds such as the Ramsaar convention and AEWA (African Eurasian Waterfowl Agreement). For general information about IWC and the international results of the Midwinter Counts, see www.wetlands.org. Distribution and numbers of wintering waterbirds in Sweden in 2015 and changes during the last fifty years


Distribution and numbers of wintering waterbirds in Sweden in 2015 and changes during the last fifty years
Utbredning och antal av övervintrande sjöfåglar i Sverige 2015 och förändringar under de senaste femtio åren LEIF NILSSON & FREDRIK HAAS The International Waterfowl Census started in Europe in 1967, in Sweden already in 1966. In Sweden, country-wide surveys of the inshore coastal areas were undertaken in [1971][1972][1973]2004 and 2015 in addition to the annual sample counts. In 2015, the entire coast between the Norwegian border and the northern part of the Stockholm archipelago was covered. The total estimate was 616000 (excluding seaducks such as Long-tailed duck and scoters), a slight increase since 2004 and a large increase since 1971-1973. Long-term significant increase or little change was recorded for all species but the Longtailed Duck which decreased significantly. The distribution showed marked changes for several species, with larger proportions being found in the Baltic archipelagos in later years. In all EU the seaducks declined whereas most other species were stable or increased. In four diving ducks an increases in Sweden and other northern areas was balanced by decreasing numbers for countries to the south and southwest. Thus population changes in Sweden could be related to milder winters in the northern part of the wintering area.
Leif Nilsson & Fredrik Haas,Department of Biology,Biodiversity,Ecology Building,fredrik.haas@biol.lu.se In Sweden, midwinter counts started one year before the IWC, in January 1966, when a pilot study was carried out (Nilsson 1967). The results of the first forty years of the IWC in Sweden  have been reported by Nilsson (2008), where details of the early counts are to be found.
In short, the aim during the first years of the counts was to get as complete coverage of as many potential wintering areas of waterfowl as possible (Nilsson 1975). To reach this goal, aerial surveys were organized after a Danish model (Joensen 1968(Joensen , 1974(Joensen ) during 1971(Joensen -1973 eventually covering all inshore areas of south Sweden (more northern areas being ice-covered in winter in those years). However, it was realized that country-wide surveys could not be undertaken every year and the counts were therefore concentrated at covering a careful selection of important sites counted every year for the calculation of annual indices. A fullscale survey of the inshore areas was again organized in 2004, and in 1987-1989 some regions were fully covered (Nilsson 2008). The first countrywide survey of offshore waters was undertaken in 2007-2011 (Nilsson 2012) but extensive boat https://doi.org/10. 34080/os.v26.21854 surveys were made with the coast guard in some districts during the seventies (Nilsson 1980(Nilsson , 2012. In 2015, a third country-wide survey of all inshore coastal waters between the Norwegian border and the northern part of the Stockholm archipelago ( Figure 1 and Figure 2) was carried out. In this paper we analyse the results of this survey in comparison with the two former country-wide surveys to elucidate any changes in numbers and distribution of different species, especially in relation to climate change (Lehikoinen et al. 2013, Pavon-Jordan et al. 2015. Moreover, we present population trends for a set of species both on a national and regional scale. New analytical tools allow us to include more species than has previously been done (Nilsson 2008). We will also set the Swedish results in a larger perspective as the results from the IWC in the other Baltic countries are available for comparison (cf. Skov et al. 2011).
In the running text throughout this paper we use only English names of bird species. Scientific names will be found in the species accounts and in figures and tables. In Table 2 the species are listed with their English, scientific and Swedish names. All photos by L. Nilsson.

Organization
The International Midwinter Counts of Waterbirds in Sweden as in the other countries rely on voluntary counters, which cover one or more counting sectors at the specified dates. The counts are organized on the weekend with the Sunday closest to 15 January. Although the observers are encouraged to do the fieldwork during that weekend, it is allowed to do counts the week before and after to avoid bad counting conditions and to get a better coverage.
After the first few exploratory years, the entire Swedish coast was divided into counting sectors with established borders that have been kept the same since then. The strategy has been to have sectors with borders that can easily be recognized in the field. The size of the sectors was chosen to make it possible to cover the entire sector during  Table 1 and Nilsson (2008). In the analysis of distribution changes the regions are grouped into four larger units as follows: West = V and A-C, South = D-G, Large Islands = K and L, and Archipelago = H and M-O. Indelning av den svenska kusten i regioner. För beskrivning av regionerna se Tabell 1 och Nilsson (2008). I analyserna av förändringar i utbredningen har regionerna slagits samman i fyra storregioner enligt följande: Väst = V och A-C, Syd = D-G, Stora öarna = K och L, and Skärgård = H och M-O. one day. In practice it was normally possible for the counters to cover a number of sectors in a day. For inland Sweden, smaller lakes form one counting sector, whereas the larger lakes were split into several sectors in the same way as the coastal areas. Rivers were in many cases also split into smaller sectors.
During the first years of midwinter counts, the observers freely selected the sites to be covered but from the 1980s the counts for index calculations were standardized and a selection of larger coastal and inland areas (hereafter reference areas) was covered each year to form the basis for trend calculations (Nilsson 2008). These areas were subdivided into the same counting sectors as described above. In addition to the reference areas, counts from smaller sites spread over the country were also collected for the trend-calculations.
For the analysis of the country-wide surveys the coasts of southern Sweden were divided into fourteen regions with similar habitat features ( Figure  1, Table 1). These regions were described in detail in Nilsson (2008). For our analysis of changes in the distribution of more common species between the country-wide surveys we have grouped these regions into four larger units (Figure 1). One large unit is called West and includes the regions V, A, B and C, i.e. the coast between the Norwegian border and Falsterbo peninsula, thus including Öresund. A second large unit is called South and includes the regions D, E, F and G, i.e. the coast from Fal-sterbo to northern Kalmarsund (mainland coast). A third large unit is called Arhipelago and includes the regions H, M, N and O, i.e. the rest of the eastern coast of southern Sweden through Stockholm archipelago. The forth large unit, called Large Islands, includes the regions K and L, i.e. the islands of Öland and Gotland.
The Whooper Swan is included in the annual midwinter counts but in recent years it was realized that the species changed habits and was feeding to a higher extent in terrestrial habitats and could therefore not be accurately covered by the midwinter counts (Nilsson 2014). Therefore a special Whooper Swan survey was organized in 1995, and every fifth year since then, simultaneously with the midwinter count. The Whooper Swan survey in 2015, undertaken at the same time as the countrywide survey discussed here is reported separately in this same publication (Nilsson 2016).

Ground counts
During the ground counts the observer covered the counting sector(s) either by walking the shore line or from vantage points using telescope to cover the water areas, moving between counting points as fast as possible to minimize the effects of movements of the birds. Normally, one observer managed to survey one to several sectors within the same day.

Aerial surveys
For the aerial surveys in 2015 we used a Cessna 337 Skymaster, a twin-engined high-winged aircraft with good visibility. The flying altitude was about 70 m and the speed 150-180 km/h, i.e. the slowest possible. Aerial surveys were only carried out in good weather conditions. The flight path was recorded by a GPS taking positions every ten seconds. One to two observers covered each side of the aircraft. All observations and the time of each of the observations were recorded. By using the time the observations were later synchronized with the GPS positions and transferred to a database. During previous aerial surveys of inshore areas, and during surveys of some areas in this study, the aerial surveys covered the same counting sectors as used in the ground counts. Along open shores the flight path was laid out about 200 m from the shore, whereas in the archipelagos the different islands were circled so that the entire shallow water area within the sectors could be overlooked from the air. In the 2015 flight survey this method was used on the west coast, in Kalmarsund and along the open coasts of Öland. In contrast to previous aerial surveys there was no ice in the east coast archipelagos, which made the areas to cover too large for this method, so the remaining areas were covered by line transects (Figure 3).
The aerial line-transect method was developed for the offshore surveys (Nilsson 2012). In 2015 census lines were laid out so that all important water areas in the archipelagos could be covered. The distance between the survey lines here was 4 km. Fixed waypoints at the ends of each transect were established and navigation was undertaken with the aid of the GPS of the aircraft. We counted all waterbirds within a main belt extending 200 m on both sides of the aircraft. This means that we covered a zone 320 m wide as there was a dead angle below the aircraft. All observations of flocks outside the main belt were recorded as additional observation.
To estimate regional totals for the different species we used the counts within the main belts of the surveys and multiplied the number of observed birds with a factor of 12.5. Each flight transect represent a 320 m wide sample band through an area which was 2 km on each side of the aircraft, thus we extrapolated the number of birds by multiplying the number of observed birds with the quota between the width of the study area (4000 m) and the width of the sample band (320 m).
The efficiency of aerial surveys of different spe-cies was much studied during the seventies by air/ ground comparisons for different species in different habitats (Nilsson 1975, cf. also Joensen 1974. In these studies marked differences in the survey efficiency were found between species but also between different observers. During inshore surveys in the seventies, in 2004 and in some areas in 2015 the surveys were based on the counting sectors (see above), whereas the aerial surveys of the Swedish offshore waters in 2007-2011 (Nilsson 2012, Skov et al. 2011) as well as the east coast archipelago surveys in 2015 were performed as line transects. However, we do not believe that the use of two different survey methods severely influences the results. Thus, the results from 2015 are fully comparable with previous surveys.

Coverage of counts
The annual Midwinter Counts cover a sample of counting sectors in the southern part of Sweden ( Figure 2). During the last ten years between 644 and 944 counting sectors have been surveyed annually. For further details on the number of sectors covered at the counts see Appendix 1. For analytical reasons the majority of the counting sectors were grouped into 28 reference areas in the late 1980s. Within these reference areas a fixed selection of counting sectors were covered each year, whereas the counting frequency of other counting sectors has varied between years. In some of the reference areas all sectors could not be counted from the ground being hidden behind islands etc.
In the country-wide survey in January 2015, we used a combination of ground counts and aerial surveys to cover the inshore parts of the Swedish coast from the Norwegian border on the west coast to Norrtälje in the northern part of the Stockholm archipelago on the Baltic coast ( Figure 3). The areas with open coast, i.e. mainly southern Halland to southern Kalmarsund and Gotland and parts of Öland were covered by ground counts, whereas the rest was covered from the air. Details of the line transects in the east coast archipelagos can be seen from Figure 4. We could not fly the entire Stockholm archipelago, because of military activities in some areas. Moreover it was not possible to cover the archipelagos of the Södermanland county due to lack of flying time in the region. Thus, there are some gaps in the geographical coverage. In all we spent 28.5 flying hours on active counting of waterfowl during the survey.
The coverage of the country-wide survey in 2015 was similar to the country-wide surveys in 1971and 2004(Nilsson 2008). In 1972, 1973, 1987-1989 coverage was obtained in  some regions (see Tables 3-5, for details see Nilsson 2008). For details on the coverage of the counts in different years, see the annual reports that can be downloaded from the project webpage: http:// www.zoo.ekol.lu.se/waterfowl/index.htm .

Index calculations
We used the results from the land based counts to calculate population trends for a set of species both on national and regional scale. We based the regional analyses of the coastal regions on the four large areas presented in Figure 1. We did a separate analyses of inland waters. The inland waters were divided into two categories, Inland south and Inland north. We used freeze-up dates for small lakes, <10km 2, (SMHI 2015a) for the grouping. Inland south included lakes from the area where the freezeup dates normally were after 1 January (Skåne and parts of Bohuslän, Halland and Blekinge) and the other, Inland north, from areas where freeze-up dates were before 1 January (the rest of Sweden). Only a subset of the species analysed at a national level was included in the regional analysis due to limited sample size. At both national and regional level we calculated species population trends based on the maximal available time span for which data were available. This resulted in that the starting year of the time series differed between species, because some species were not surveyed during the first years or they were not abundant enough resulting in too much missing data. On national scale we also calculated species trends for the last ten years (2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015). The use of a time span of 10 years is arbitrary, what is important is that it allows us to contrast the long term trends with the population dynamics during recent years.
We used the freely available software package TRIM (http://www.ebcc.info/trim.html) to perform the analysis. TRIM is developed to analyse times series of count data and has the ability to handle missing data using a log-linear Poisson regression method (Pannekoek & van Strien 2001). The outcome of this procedure is yearly indices.
For each species we present an overall trend estimate, which is the slope of a linear regression where the dependent variable is the logarithm of the indices, and the independent is year (Pannekoek & van Strien 2001). We run the TRIM model under the assumption that data were serially correlated. Moreover, for 1971Moreover, for -2015Moreover, for and 2006Moreover, for -2015spectively, we produced multiple species indices based on the yearly indices that resulted from the TRIM-analysis of all but two of the species (n=27) included in the national TRIM runs. Two species (Gadwall and Black Guillemot) were excluded because of missing data in the early years of the first mentioned time series. The multiple species index for a given year was calculated by taking the geometric mean of all species' TRIM-indices that year (Gregory et al. 2005). This was then repeated for all years.

Weather and ice conditions
The wintering waterbird populations around the inshore parts of the Swedish coast are dependent on the availability of open water. In order to characterize the weather conditions of the winters during the study period we have used two different proxies obtained from the Swedish Meteorological and Hydrological Institute (SMHI), the means of monthly mean temperatures for January for ten stations spread over the southern part of Sweden (the areas covered by duck counts) and the annual maximum for the ice-distribution of the Baltic sea and Kattegat (total area: 420 000 km 2 ; SMHI 2015b).
Both these indicators clearly show that the winters in general have been milder in south Sweden during the years of midwinter counts. For the temperature there were several January months with low mean temperatures between 1966 and 1987, the latter the coldest January during the whole study period (Figure 5). After 1987, on the other hand, there was only one really cold January, in 2010.  1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 Figure 5. Annual means of monthly mean temperatures for January for ten meteorological stations spread over south Sweden over the area covered by Midwinter Counts. (From the monthly reports of SMHI, the Metrological and Hydrological Institute of Sweden). Årsmedeltal av de månatliga medeltemperaturerna för januari på tio väderstationer spridda over södra Sverige (Från SMIs månatliga rapporter).
The maximum ice coverage of the Baltic shows a similar picture. Up to 1987 there were six winters with more than 250 000 km 2 in maximum icecoverage, whereas there was only one such winter in the 28 years after 1987, namely 2011 ( Figure 6). The mean temperature for January 2011 was not especially low, but the cold period with much ice came after the midwinter count. Comparing the two periods 1966-1987 and 1988-2015, the mean maximum ice-coverage for the winters were 216 000 km 2 and 134 000 km 2 , respectively.
To evaluate the influence of ice coverage on bird abundance we correlated the multiple species index with the area of open water (420 000 km 2 minus area of ice coverage). We did this for three separate time periods; 1971-1987 (low numbers of wintering waterbirds), 1988-2000 (increasing numbers of wintering waterbirds), and 2001-2015 (high numbers of wintering waterbirds).

Abundance and distribution of wintering waterbird populations in 2015
In January 2015, the total wintering population of different waterbird species in inshore coastal waters of Sweden was estimated to be 616 000, excluding a number of seaduck species (Long-tailed Duck, Velvet Scoter and Common Scoter), which have their main over-wintering areas further out at sea. Their numbers are excluded in Table 2, which shows the estimated numbers of wintering waterfowls in Swedish coastal waters at the three country-wide surveys of Swedish inshore waters, but are included (when fully covered) for some regions in Table 3-5, which show totals for larger areas from the country-wide surveys and from regional surveys with full coverage. Regional totals from the 2015 are found in Appendix 2 and Appendix 3. The total number presented above for 2015 is likely to an underestimate. The area covered in 2015 was the same as in former country-wide coastal surveys but this winter was very mild. Thus, unlike the other years there were significant areas of open coastal water north of the surveyed area.
Full cover counts of Swedish inshore waters were also obtained in 1971 and 2004 (Table 2, Nilsson 2008). Total numbers of birds counted in 2004 was not very different from the totals counted in 2015; about 50 000 fewer were found in 2004. There are some differences for the different species, e.g. the Scaup was close to ten times more common in 2015 than in 2004. Other species that were markedly more common in 2015 were Goldeneye, Goosander and Mute Swan, whereas the total count for Tufted Duck was 43 000 higher in 2004 than in 2015.
Numbers counted at the coasts during the first country-wide survey in 1971 were appreciably lower for all species and the total was only 174 000.
In addition to the three country-wide coastal surveys discussed above, some regions were also covered fully in other years (see Table 3-5). The number of waterfowl in region V, the archipelago of Bohuslän (Table 3) shows the same general increase as was found in the national totals (Table 2). For the total waterbird fauna, numbers have increased dramatically between 1971 and 2015. The number of wintering waterfowls in this area is more than five times higher today compared to the beginning of the 1970s. The repeated surveys indicate that this increase has been a more or less continuous process, although a slight decline in numbers have occurred between 2004 and 2015, mainly so for Goldeneye, Eider, Mute Swan and Cormorant.
In the southern regions there were no marked changes in the overall numbers of wintering waterfowl between the early seventies and the late eighties, even if there was much variation between years for different species but also in the annual totals (Table 4). On the other hand overall totals more than doubled between the late eighties and the surveys in 2004 and 2015. The overall totals for the last two surveys were similar even if there were quite marked differences for some species.
Very marked changes in the number of wintering waterbirds were found for the archipelagos of the east coast when comparing the three surveys in [1971][1972][1973] 1966 1969 1972 1975 1978 1981 1984 1987 1990 1993 1996 1999 2002 2005 2008 2011 2014 Km 2 x 1000 Tufted Duck, Goosander and Mallard Mallard, Goosander and Tufted Duck   Table 4. Total numbers of wintering waterbirds in regions A-L (from northern Kattegat through northern Kalmarsund, Öland and Gotland; Figure 1) at surveys covering the inshore areas of this entire area. Species marked with a dash for 1971 were not counted during that year. For Clangula hyemalis, Melanitta fusca and Melanitta nigra (marked with grey in the table) only a small proportion of the total populations are found close to the shore, so the figures presented do not represent total population estimates for the regions but are the numbers actually counted in surveys of the inshore areas.  ure 1) during 2015), the totals for the other species in the inshore areas were four to six times higher in the two later surveys compared with the seventies. Very marked changes were found for most common species.

Species accounts
In this section the distribution and numbers of the more common wintering species in 2015 are discussed separately. For each species a map showing the national distribution according to the counts in 2015 is presented. Similar maps for the two previous country-wide surveys are found in Nilsson (2008), where details for the previous surveys can also be found. For some of the more common species their detailed distribution in three archipelago areas (the central part of the archipelago of Bohuslän (region V), Småland -Östergötland (regions H and M) and the Stockholm archipelago (region O)) is shown in maps. Changes in the distribution of the more common species between the three country-wide surveys over four larger regions (see division in Figure 1) are also presented in a series of graphs. Long-term and short-term trends for the different species will be discussed later (p 31).

Mallard Anas platyrhynchos
The most wide-spread species in winter is the Mallard which was found in good numbers over most of south Sweden, i.e. in the parts of the country where ice-free water was available. Along the coasts the species was found to be common in all areas except the east coast archipelagos north of Figure 7. National and regional midwinter distribution of Mallard Anas platyrhynchos in 2015.
Kalmarsund. The species occurs in most inland areas with open water (Figure 7). The detailed map over the west coast archipelagos shows that it is mostly concentrated to the inner waters even if some smaller groups were also found further out on the more exposed skerries. The relatively few flocks that were seen in the Stockholm archipelago occurred both in the inner and middle parts of the archipelago.
In the Mallard about 45% were found on the west coast including Öresund during the 1971 survey, this proportion decreased to about 20% in 2015 ( Figure 8). During this survey the islands Öland and Gotland hosted almost 45% of all Mallards counted. The proportion of Mallards found in the archipelagos showed only small changes between the three surveys even if the total number of individuals was much larger in 2004 and 2015 than in 1971.
In all, 87 000 Mallards were estimated to overwinter along the coasts in 2015, implying that the Mallard was the second commonest species in inshore coastal waters together with the Goldeneye. For the inland areas we do not have any estimate of the total wintering Mallard population for 2015 as it was a mild winter with much open water and the Mallard is wide-spread also in inland areas. During the survey in 2004 the total population of Mallards in Sweden was estimated to be about 150 000 compared to 68 000 in 1971. In 2004, the coastal total for the Mallard was 78 000 (Table 2), i.e. almost half of the national estimate. With this in mind it is most probable that the total population for Sweden in the winter of 2015 was more than 200 000.
Teal Anas crecca & Wigeon Anas penelope These two dabbling ducks have a restricted winter distribution in Sweden. When the counts started they were only seen in small numbers in the country during the winter but both species have estab-  lished a wintering habit in south Sweden in more recent years. They are mostly concentrated to Scania with some occurrences on Öland and Gotland ( Figure 9).

Tufted Duck Aythya fuligula
Among the diving ducks in inshore waters, the Tufted Duck was the commonest species followed by the Goldeneye and Mallard. In 2015 the total number of Tufted Ducks in the Swedish coastal waters was estimated to be 182 000 compared to 225 000 in 2004 and only 52 200 in 1971 ( Table 2). The Tufted Duck was distributed along the entire Baltic coast surveyed, with some larger concentrations also in the Öresund (Figure 10), whereas only smaller numbers were found on relatively few sites on the west coast north of Öresund. In the east coast archipelagos, the Tufted Ducks were markedly concentrated to the inner parts, where often large flocks were found. These sites are mainly used as daytime roosts from which the Tufted Ducks disperse over the feeding areas further out at sea (Nilsson 1972). Only low numbers were found on a small number of inland sites. Between the surveys in 1971 and especially 2015 there was a marked shift in the distribution from about 50% in the Öresund and the South region to about 25% here in 2015, the majority of the Tufted Ducks in Swedish coastal waters being found in the archipelagos and along the big islands in the Baltic, especially Gotland ( Figure 11).

Scaup Aythya marila
The Scaup was counted in unusually high numbers in 2015, 26 850 individuals were observed. Most of them (88%) were concentrated to the east coast of the island of Gotland ( Figure 12). Smaller numbers were found in the flocks of the numerous Tufted Duck in some other regions, but on Gotland several pure flocks with Scaup were found.
In all three surveys the majority of Scaups has been observed in the waters surrounding the Baltic islands and especially Gotland, but some flocks regularly occur also in the southern region and a few in Öresund ( Figure 13). The total number in 2015 was much higher than in the other two surveys and the dominance for Gotland was much more marked.
Pochard Aythya ferina Most Pochards occurred in a few areas along the coasts of south Sweden with a marked concentration in the province of Blekinge ( Figure 12). Totally 1208 individuals were observed in 2015, which can be compared with the 2660 Pochards that were counted in 2004. The total was much lower in 1971, with only 92 registered individuals. The Pochard will easily be underestimated during aerial surveys as they occur in the larger flocks of Tufted Ducks. In 2015 smaller numbers were also found inland.
Goldeneye Bucephala clangula The Goldeneye is well spread along the entire coast covered in the present survey (Figure 14), being equally common on the west and east coast. In all, 87 000 Goldeneyes were estimated to be present in the areas surveyed in 2015, but this does not fully represent the national total for the species. The Goldeneye is well spread in small numbers on several inland sites and there was only little ice in the inland of south Sweden in 2015. The same applies to the coasts north of the surveyed areas in the Baltic. During the earlier surveys these inshore areas were ice-covered but in 2015 many ice-free areas potentially used by the Goldeneye was not covered by counts. It is therefore highly probable that the wintering population of Goldeneyes in Sweden was between 90 000 and 95 000 or even higher. The coastal total for the inshore areas in 2004 was 72 000, whereas the total for 1971 was appreciably lower, 17 600 (Table 2).
On the west coast, the Goldeneyes were mainly found in the inner parts of the archipelagos, even if some flocks were found on more exposed sites ( Figure 14). In the narrow archipelago of Östergötland and Småland (regions H and M in Figure 1), they were also found in the innermost areas, whereas they occurred some distance from the mainland coast in the much wider Stockholm archipelago, but there were no Goldeneyes in the outer parts of this archipelago.
The Goldeneye is the species showing the most marked change in the distribution between the surveys ( Figure 15). In 1971, only about 5% of the total coastal number was from the archipelagos compared to about 50% in the survey in 2015. In the same time there were marked changes in the total counts of the species. 2004 were somewhat in between when it comes to the distribution over the areas and the counts were not markedly lower than in 2015.

Melanitta nigra
In the inshore coastal areas covered by the present survey, the Velvet Scoter and the Common Scoter were only found in larger numbers in Skälderviken and Laholmsbukten in southern Kattegat ( Figure  16). In the other areas only small numbers of these two species were recorded from the shoreline. Both species do however in most areas stay far out at sea, not being seen from the shore, they are regular in relatively large flocks in offshore waters south of Falsterbo and in Hanöbukten (Nilsson 2012, Green & Nilsson 2015. In Skälderviken and Laholmsbukten in southern Kattegat large proportion of the scoters can be counted from the shoreline in good weather condi-tions. In 2015, the total number of Common Scoters counted from the shore in south Swedish waters (Table 4) was 8162 and the number of Velvet Scoters was 5462.
Eider Somateria mollissima The Eider was concentrated to the west coast and was not found in any of the other archipelagos covered by the 2015 survey ( Figure 17). The only area not covered by the present survey where   large flocks are known to occur is the offshore waters south of Falsterbo (Green & Nilsson 2015). In other parts of the Baltic only small numbers were found, for example at the southern tip of Gotland.
In the archipelagos of Bohuslän (Figure17) the majority of Eiders were found in the outer zone with small skerries and sea shallows and only few individuals were found in the more sheltered inner water areas.
In the areas covered in 2015, 51 700 Eiders were counted, which is slightly more than in 2004, when 49 000 were present in the same areas. These two estimates were markedly higher than the total for 1971, when only 4500 were observed. Based on the offshore surveys south of Falsterbo (Green & Nilsson 2015, Nilsson 2012), there could be a further 10 000-13 000 Eiders in these waters, bringing the national total to little more than 60 000.

Red-breasted merganser Mergus serrator
The Red-breasted Merganser was distributed all around the coasts of south Sweden from the Norwegian Border to the northern part of Kalmarsund and around Öland and Gotland (Figure 18) but almost absent elsewhere. In the west coast archipelagos they occurred mostly in small groups spread over both the more sheltered and more exposed parts, but some flocks were found further to the sea. In 2015, the national total for inshore areas was 5200 compared to 5300 in 2004. As for most other species the wintering total was much lower in 1971, 2400. The totals do not represent the total wintering population of the species for Sweden as much larger numbers are sometimes found offshore  especially in SW Sweden (Nilsson 2012, Green & Nilsson 2015. The Red-breasted Merganser showed relatively small changes between the surveys compared to the other species, although the share for the Öresund and the west coast increased from about 30% to 50% ( Figure 19). As in 2015, the species was absent from the Baltic archipelagos in the two previous surveys. In the east coast archipelagos the Goosanders were mostly found in the innermost parts of the narrow archipelagos of Kalmar and Östergötland (H+M, Figure1), whereas they were more concentrated to areas a little further from the shore in the Stockholm archipelago.
The number of Goosanders in the coastal areas covered was about 35 000 in 2015, which is almost twice as many as was reported in 2004, when 18 000 were counted and much higher than the 7800 estimated for 1971 (Table 2). Especially during mild winters these figures do not represent national totals for the species. Like the Goldeneye, the Goosander is also found on a number of inland sites, and moreover, some Goosanders may have stayed in the ice-free areas north of the coastal areas covered by the present survey. The national total was probably a few thousand individuals higher than 35 000. The Goosander has a totally different distribution in south Swedish waters compared to the Redbreasted merganser. The numbers occurring in the Öresund and on the west coast are normally small and the vast majority of coastal Goosanders are found in the Baltic. Even in 1971, when the total numbers counted was much lower than in 2015, a large proportion was found in the archipelagos, this share increasing to about 80% in the mild winter of 2015 ( Figure 21).

Smew Mergellus albellus
In 2015, the total number of Smews wintering in Sweden was estimated to about 8000 individuals compared to 3700 in 2004 and 350 in 1971. Most Smews were noted along the south-eastern coast of Sweden with large flocks seen in the archipelago of Blekinge, in the Kalmarsund and on some sites on Gotland (Figure 22). Smaller numbers were also found inland.
At the first survey, the species was quite rare in Sweden and it was highly concentrated to the archipelago of Blekinge and to Öresund with some flocks and smaller groups on inland waters, especially in the southernmost province of Sweden,  Scania. At the next survey, in 2004, the situation had changed but the majority (60%) were still found in the south. However, some flocks had begun to appear in the Archipelago and the Large Islands ( Figure 23). This trend was accentuated in 2015. This winter the numbers were roughly equal in the three larger Baltic areas (3000 in South, 2600 in Large Islands and 2200 in Archipelago) but the proportional increase was larger towards northeast, 36%, 160% and 540%, respectively. The change in South was moderate and in one of the regions E, Blekinge, there was no change at all. Numbers were thus similar in the south, especially in Blekinge, most of the increase was found on Öland and Gotland and in the east coast archipelagos, especially in Kalmarsund and Östergötland.

Mute Swan Cygnus olor
The Mute Swan was distributed all over the Swedish coast, both in the archipelagos and on the open coasts. As the winter was mild, there were also large flocks on some inland lakes, especially in Scania, but also further north ( Figure 24). In the west coast archipelagos the Mute Swan was well spread in smaller groups over the inner parts and sheltered areas further to the sea. In the Baltic archipelagos the swans were mostly found in small groups and in many areas in pairs spread over the entire archipelago. In the Stockholm archipelago several smaller groups and pairs were found far out at the sea close to the outermost small skerries and island.
The number of wintering Mute Swans in Sweden show the same increase as most other species discussed here. In January 2015 somewhat more than 50 000 were estimated for the coast compared to 31 000 2004 and 7300 1971. The Mute Swan was found in relatively high numbers in all four regions. Simultaneous with the large increase in numbers between 1971 and 2004 there was a change in the proportion of the birds wintering in the different regions ( Figure 25). More than 50% of all Mute Swans were found on the west coast and Öresund in the 1971 survey, and this proportion continued to decline between 2004 and 2015. On the other hand, less than 20% of the total was found in the Archipelago area in 1971 and 2004 compared to more than 40% in 2015. Thus, the relative importance of the archipelago as wintering sites for Mute Swans has increased during past decades.

Whooper Swan Cygnus cygnus
Since the start of the IWC it was noted that the Whooper Swans gradually changed their habits to feed more and more on land, behaving like geese. This made it impossible to follow the population development of the species in the normal midwinter counts, so a series of special Whooper Swan surveys were started in Europe in 1995 (Laubek et al. 1999, Nilsson 1997, which has since then been repeated every five years. Thus the country-wide coastal survey in 2015 coincided with the special Whooper Swan survey in January. This survey will be discussed in a separate paper in this volume (Nilsson 2016).

Great crested Grebe Podiceps cristatus
The only common species among the grebes was the Great Crested Grebe with a total of 2700 in 2015 compared to 4300 in 2004; data are missing for 1971. Along the coasts it was distributed in flocks in the Öresund, the Hanöbukten and at the coasts of Gotland (Figure 26). In other regions the species was less abundant occurring only in smaller numbers. A large proportion of the wintering Great Crested Grebes in Sweden are known to stay in one large flock in Lundåkrabukten in Öresund. The flock often stays far out at sea and can be difficult to count from land especially in bad weather, which will cause quite marked variation in the counts between years.
Heron Ardea cinerea The Heron was counted in relatively small numbers. It was widespread in south Sweden both at the coasts and inland (Figure 26). Herons were includ-   Other species In addition to the species discussed above a number of other waterbird species have been counted in smaller numbers at the surveys in all three years but also during the annual ground based midwinter counts. For details of these species at the countrywide surveys see Tables 2-5 with a regional breakdown for 2015 in Tables A2 and A3 (see also Nilsson 2008 and the annual reports from the project, http://www.zoo.ekol.lu.se/waterfowl/index.htm, for older data).

Trends in the wintering waterbird populations 1966-2015
There is a remarkable difference between the longand short-term population trends ( ed in the country-wide surveys in 2004 and 2015, when about 500 and 600, respectively, were estimated for the coasts with smaller numbers inland. Cormorant Phalacrocorax carbo Cormorants were common along the west coast and the southern Baltic coasts of Sweden, whereas there were few occurrences of the species in the main Baltic archipelagos north of Kalmarsund ( Figure 27). The detailed map for central Bohuslän shows that this species occurs in all coastal waters, but most of the larger groups were seen on the outermost skerries and islands. The number of wintering Cormorants in Swedish coastal waters was estimated to be 14 300 in 2015 compared to 11 700 in 2004 and 1000 in 1971.
Coot Fulica atra The Coot was more or less concentrated to the southernmost coastal areas (76%) with some occurrences along the coast of Gotland (19%) (Figure 28). There were also some inland flocks, but no Coots were observed in the Baltic archipelagos. In 2015, 11 200 Coots were estimated for the inshore coastal areas compared to 16 000 in 2004 and 3900 in 1971.
a long series of mild winters, which may have influenced the short-term trends. Trend-diagrams for the different species are found in Appendix 5. The multispecies index for 1971-2015 ( Figure  29) shows a highly significant increase over time (r 2 =0.69, F 1,43 =97.7, p<0.001), which is not the case for the period 2006-2015 (r 2 =0.01, F 1,8 =0.08, p>0.05). From late 1980s until the millennium shift the multispecies index shows a more or less continuous increase, thereafter the increase has flattened out. Thus, the time series indicates a period with low numbers of wintering birds (roughly between 1971 and 1987), a period with increasing numbers (1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000), and a period with high numbers Table 6. Population trends, based on the national data set, until year 2015 according to the TRIM analyses for 29 seabird species in January for the entire (E) time series for each of the species and for last ten (10) years. Green: significant (p<0.05) increase; red: significant (p<0.05) decrease; white: no significant trend. Starting year of the long-term time series is shown in column E. Also shown are the slope of each species trend line and the corresponding standard error. (10)  The subdivision of the national data set allowed us to do regional trends of seven species in coastal areas and five in inland areas (Table 7, Appendix 4). Not surprisingly, the regional trends show the same general picture as the national ones. But opposed to the national trends a few regions show declining numbers of some species. In Kattegat and Öresund, Tufted Ducks as well as Red-breasted Mergansers have declined significantly in numbers between 1971 and 2015. The same is true for Mute Swans in Bohuslän and European Coots in Kattegat. The declining species are restricted to the Swedish west-coast; in the Baltic Sea all species analysed here are either increasing or stable. The latter is the case also for northerly as well as southerly inland waters.

Sweden as a wintering area for waterbirds
To evaluate the importance of Sweden as a winter area for waterbirds in the European perspective, the estimates based on the 2015 survey are compared with the population estimates published for the north-western European flyway (Table 8, Delany & Scott 2006, Wetlands International 2015. In some cases there have been changes in the populations that have not yet been accommodated in the published population estimates but anyhow these give a background for evaluating the Swedish winter areas from an international perspective. The country-wide survey in 2015 provides estimates of the wintering populations for most species occurring in inshore waters. To make the picture complete we have included estimates of the wintering populations of three seaducks from the offshore surveys in Swedish waters in 2009-2011 (Nilsson 2012).
For the Long-tailed Duck, two estimates are presented in Table 8, one from the mild winter of 2009 when all waters could be covered and the Swedish Table 7. Significance of regional trends for 1971-2015 based on the midwinter counts for seven species, all widespread in coastal areas, five of them widespread also inland. Regions consist of the following areas presented in Figure 1: V (Bohuslän), AB (Kattegat), C (Öresund), D (S and E Scania), E (Blekinge), FG (Kalmarsund), HMNO (northern Kalmar county, Östergötland and Stockholm arcipelagos), K (Öland), L (Gotland). For definition of inland waters (Inland, north and south, respectively, see method). Green cells denote significant increase, red significant decline, white no significant trend and grey missing data. Significant levels are indicated as: * = p<0.05, ** = p<0.01, *** = p<0.001. Slopes and s.e. are presented in appendix. Region V AB C D E FG HMNO K L Inland, N Inland, S Anas platyrhynchos *** *** *** *** *** *** *** *** *** Aythya fuligula *** ** * *** *** *** *** *** Bucephala clangula *** * *** *** *** *** *** *** *** *** Mergus serrator *** *** *** *** *** *** Mergus merganser ** *** ** *** *** *** *** Cygnus olor *** *** *** *** *** *** *** *** Fulica atra *** ** *** *** *** population was estimated to be 440 000 and one estimates from the cold winters of 2010 and 2011, when many areas of the Baltic were covered by ice and the wintering population of the Swedish areas covered was in the order of 700 000. The high counts in the two ice winters are most probably due to the freezing of the important wintering areas for the species in the outer parts of the Riga Bay and a subsequent move of some of these birds into Swedish waters. For the Velvet Scoter and the Common Scoter data from 2009 have been used, but much higher numbers of Common Scoter were found during a partial survey of the southernmost offshore areas in 2007 (cf. Nilsson 2012), the total number of Velvet Scoters and Common Scoters in the offshore waters of regions D, F and K (Figure 1) were then estimated to be 6150 and 39 900, respectively. It is however probable that these were temporary aggregations as such concentrations have never been seen here before. Numbers estimated from the surveys in 2009-2011 are likely to be more represent-ative for the wintering populations.
The Eider is also a typical seaduck, but the coverage of the main areas of the west coast was complete and the only area of importance for the species not covered in 2015 was the offshore waters around Falsterbo and southern Öresund with an estimated wintering population of about 10 000, a value included in the estimates in Table 8. The Red-breasted Merganser is also found in larger numbers in some offshore areas, especially around Falsterbo and in the southern Öresund (Green & Nilsson 2015, Nilsson 2012, the estimates having been adjusted accordingly. As presented above, some Goldeneyes and Goosanders are also staying on inland waters and smaller numbers winter along the coasts north of the areas covered in 2015 (being ice-covered during earlier surveys). The estimates from the 2015 survey (Table 8) have been adjusted to include also these birds.
Considering the Swedish coastal waters as a unit, the Long-tailed Duck is clearly the most common  (Nilsson 2012). For the inshore coastal waters the Tufted Duck is the most common species with a total estimated winter population of 190 000 in 2015, but more than 225 000 were counted in the previous country-wide survey in 2004. When comparing the Swedish estimates with the estimates of the winter populations for the entire north-western Europe (Table 8) it is clear that Swedish waters are of great international importance for some wintering species. The offshore banks in the Swedish Economic Zone are especially important for the Long-tailed Duck, where between 27% and 44% of the estimated total population for the Baltic and for north-western Europe (cf. Skov et al. 2011) were found in Swedish waters in 2009-2011. Other species for which Sweden has a high proportion of the total north-west European winter population are Tufted Duck, Goosander, Smew, Mute Swan and Whooper Swan. For these species between 15% and 20% of the estimated wintering population in north-west Europe was found wintering in Sweden.
Sites that regularly are used by more than 1% of the estimated population for the flyway are considered to be internationally important resting/wintering sites for the species according to the Ramsar convention (Wetlands International 2015). In the same way sites that regularly hold more than 20 000 waterbirds are also considered internationally important. Even if Sweden has important proportions of the wintering populations of some species, relatively few wintering sites qualify as Internationally Important Sites as most wintering waterbirds in Sweden occur quite dispersed when they are not concentrated to a few sites due to the ice conditions in cold winters.
For the most abundant waterbird in Swedish waters, the Long-tailed Duck, some offshore areas are clearly to be considered as internationally important sites. This especially applies to the Hoburg bank and the Midsjöbanks (Nilsson 2012), recent maximum counts being 426 000 and 213 000 Long-tailed Ducks respectively, i.e. 27% and 13% of the entire Baltic wintering population at some occasions.
Among the inshore species, internationally important wintering numbers were found on a few sites for the Tufted Duck, Smew and in some cases Goosander. For the Tufted Duck and Smew, internationally important numbers were regularly found in the Blekinge archipelago between Ronneby and Karlskrona. In mild winters such concentrations were also found in the inner parts of the archipelagos, e.g. in the Kalmarsund and Östergötland for the Smew and in the Stockholm archipelago for the Goosander and Tufted Duck, but the exact sites for the flocks varied between years due to ice conditions. However, most common inshore diving species are more dispersed. Even if flocks are sometimes surpassing the criteria it is difficult to fulfil it as it should be shown that the sites regularly hold large concentrations. This is normally not the case because the exact localization of the concentrations normally varies between times.

Development of the wintering populations during the last 50 years
The long-term trends for the wintering populations of waterbirds in Sweden over the past fifty years has been significantly increasing, with the exception of the Long-tailed Duck which was significantly decreasing, and three species that did not show any clear long-term trends: Shelduck Tadorna tadorna, Velvet Scoter and Red-throated Diver Gavia stellata. Of these three species, the Shelduck and the Red-throated Diver were only counted in small numbers. For the scoters, it is not clear if the indices presented are representative for the entire Sweden as they are based on ground counts from the shore and the majority of the Scoters are found offshore.
Among the seaducks a representative sample of the Swedish wintering population is covered by the annual counts for the Eider and the Red-breasted Merganser. The indices presented for the Longtailed Duck are based on a relatively large sample of sites where appreciable numbers of the species have been counted annually during the study period. For the Long-tailed Duck extensive data are available for the offshore areas around the mainland coast and Öland from the seventies showing a marked decrease in the wintering population of the species to the present day as was found in the indices from the ground counts (Nilsson 2012). The same pattern is indicated in restricted data available from the coasts of Gotland. For the Long-tailed Duck data from the offshore banks are available from a survey in 1992/1993 (Durinck et al. 1994) and the SOWBAS survey in 2007-2009(Skov et al. 2011, Nilsson 2012. Taken together these data indicate a major decrease in the wintering population of Long-tailed Duck in Swedish waters. For the seaducks an analysis of the trends for wintering waterbirds in the waters of the European Union  indicate a decreasing long-term trend. In 1992/1993, the Baltic population of Long-tailed Duck was estimated to be 4.2 million, whereas the estimate in 2007-2009 was only 1.48 million individuals, a dramatic decrease (Skov et al. 2011). Similarly, the wintering populations of the Velvet Scoter and the Common Scoter showed a marked decrease between the surveys. The surveys are made with different methods (ship-based and aerial surveys) and over long time and are therefore strictly not fully comparable but the marked decrease for the seaducks in the region is totally clear.
In the Eider the wintering population has also shown a marked decrease between the two all-Baltic surveys (Skov et al. 2011). This is also reflected in the breeding populations in the Baltic (Desholm et al. 2002, Ekroos et al. 2012. The time-series analysis presented here indicates an opposite pattern, both the long and short term trends are positive. However, this result is a bit misleading. A previous study, covering exactly the same time periods, showed that the number of over-wintering Eiders was declining on the Swedish east-coast, whereas the opposite was true for the west-coast (Nilsson & Haas 2015). Because most Eiders winter along the west-coast the national trend presented in this study is much influenced by the situation there. The total counts of Eiders along the west coast show no major changes between the two last surveys.
The Red-breasted Merganser is also considered as a seaduck but the proportion that occurs in inshore waters during winter makes the trend estimates relevant. The SOWBAS census in 2007-2009(Skov et al. 2011) indicates a major decrease of the Baltic population also in this species, which was not apparent in the national indices presented here. Moreover, the analysis in the SOWBAS project overlooked major concentrations of the species in SW Swedish waters (Green & Nilsson 2015).
The overall trend for the waterbirds in the European Union, considering all species covered by the IWC, is increasing , cf also van Roomen et al. 2012. A breakdown of the counts shows a decreasing trend for most seaducks as discussed above in a Baltic perspective. Species other than seaducks show increasing trends with the exception of the Pochard, where the European trend is decreasing. The Swedish long term trend shows an opposite direction, but during the past ten years numbers have declined For some species recent multi-country trend data that can be compared with the Swedish data presented here have been published, but there is a marked time-lag in the analysis and publication of trend data based on the IWC in different countries. For the Mallard, Dalby et al. (2013) analysed data from the Nordic countries. In general, the trends for the breeding populations of Mallard in the different countries were increasing, whereas the midwinter trends varied between countries.

Distribution of wintering waterbirds in Swedish waters in relation to climate change
For the three diving duck species Tufted Duck, Goldeneye and Goosander, Lehikoinen et al. (2013) analysed the IWC data from a number of countries in relation to climate change and found significant increases in the northern parts, whereas decreases were found in the southern areas. Similar results were also obtained in an international analysis of the IWC count data for the Smew (Pavon-Jordan et al. 2015). Van Roomen et.al. (2012) found decreasing numbers of wintering waterbirds in the Ijsselmeer area in the Netherlands in contrast to some of the international trends.
The northward shift in winter distributions is also indicated in the Swedish data. For several species the proportion of birds wintering in the Baltic archipelagos has increased, especially when comparing the most recent census with the census in 1971. The northward shift was most probably related to the better availability of open water in the archipelagos of the Baltic in recent years.
The regional trends also show a pattern in line with this. Of the seven species included in the regional analyses all were either increasing or stable on the east coast, whereas four species showed negative trends in at least one of the west coast regions.
Over the years there have not only been changes in distribution and numbers of the wintering waterbirds, but new wintering habits have also been established. When counts started, both the Teal and the Wigeon were only seen in very small numbers at the Midwinter counts, but new wintering habits were established during the latter part of the study period. An analysis of the European IWC data for the Wigeon (Fox et al. 2016) show increasing numbers wintering in the north and decreasing winter populations in the SW but the centre of gravity for the winter population is still in the Netherlands and the UK.
It is reasonable to believe that the general increase of waterbirds wintering along the Swedish coast to some extent can be explained by the gradual reduction of areas covered by ice. Until the mid-1980s the number of wintering waterbirds was on an obviously lower level than during the last 15 years. The transition between these two stages took place during a time period (1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999)(2000) with mild winters. In contrast to the low (1971)(1972)(1973)(1974)(1975)(1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987) and high abundance period (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012)(2013)(2014)(2015), respectively, the between year variation in multispecies index was not at all explained by the ice coverage during the transition period. This indicates that the abundance of waterbirds was at least partly limited by the availability of open water during the two periods when the wintering waterbird communities were relatively stable, but not when the wintering populations increased. Thus, it seems likely that mild winters promoted the increase of wintering birds. But this alone cannot explain the expansion of waterbirds. The number of birds has increased proportionally more than what can be explained if it was a linear relationship between bird abundance and area of open water (cf. Table 2 and Figure 29). However, the mild winters in the early seventies followed after a series of cold winters during the sixties Especially the winter of 1962/1963 was extremely cold in most parts of Europe with a high rate of mortality among waterbirds (Dobinson & Richards 1964, Owen et al. 1986). Thus, it is likely that our study took its start when the waterfowl populations were reduced to very low numbers. If that is the case it might be that the increase of wintering waterbirds was not limited by ice coverage until a certain point, i.e. around year 2000, when the number of birds had increased to a level where they saturated most areas with open water with suitable foraging conditions. What is troublesome in this context is that our results indicate that the number of waterbirds was influence by the ice conditions also before they started to expand. With data currently available we cannot find any reasonable explanation for this.
It is highly probable that the changes seen in the wintering populations of waterbirds in Sweden with the large increase during the 1990s is an effect both of a real increase in the size of breeding populations (cf. similar trends in a number of species in Europe ;Holt et.al. 2012, van Roomen et al. 2012 and changes in distribution due to the milder winters The changes in the distribution of waterbirds in relation to the milder winters in the north can over time cause problems for the management of the waterbird populations as the localization of important concentrations might change making the present network of protected areas less accurate. Pavon-Jordan et al. (2015) analysed the interna-tional IWC data for the Smew in relation to the EU Natura 2000 network and came to the conclusion that regular national and international assessments of the network was urgent to site-safeguard networks for this and other species of waterbirds.
In this study we use large scale data of ice coverage and we bundle data from the original counting sectors before performing the analyses. One possible way to go to further increase the understanding of the dynamics of wintering waterbirds, might be to make use of small scale data, but also to include factors like water depth and various measures indicating the quality of potential feeding areas.

Acknowledgements
The International Waterfowl Counts in Sweden have over the years been supported with grants from the Swedish Environmental Protection Agency, being part of the national Swedish monitoring program. The field work has been undertaken by a large number of voluntary observers that have counted the waterbirds of their sites in many cases over decades. We sincerely thank all participants for their large efforts over the years without which the project could not have been undertaken.
Regional trends for 1971-2015 based on the midwinter counts for seven species, all widespread in coastal areas and five also inland. See Figure 1 for Region. For definition of inland waters (Inland, north and south, respectively, see method). Population changes are represented with the slope of the trend line, the corresponding standard error (se) and significance level (* = p<0.05, ** = p<0.01, *** = p<0.001, ns = non-significant