SEARCH

SEARCH BY CITATION

Keywords:

  • shooting;
  • species diversity;
  • waterbirds

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Management implications
  8. Acknowledgements
  9. References

1. In most situations that have been studied, the creation of refuge areas for migratory waterfowl has generally increased the number of waterfowl using a site. However, experimental evidence to show that the increases were caused by reduced disturbance and not by confounding factors, e.g. improvement of habitat quality of the sites or general population increases, is generally lacking.

2. In two Danish coastal wetlands, waterfowl hunting was the most critical source of disturbance to autumn-staging waterfowl (Madsen 1998). To test whether hunting disturbance displaced birds from sites, experimental refuge areas were established, with hunting-free areas manipulated annually during a 4-year period, followed by permanent refuge establishments, monitored for a further 4 years.

3. In both areas, experiments showed that quarry geese and dabbling ducks redistributed according to the position of the hunting-free areas, whereas protected species did not. In quarry geese, there was a 6–8-fold increase in numbers from before to after the experiments; in quarry dabbling ducks there was a 4–50-fold increase. Dabbling ducks extended the staging period into the winter. Protected waterfowl as well as goldeneye and coot (both quarry species) did not redistribute; their numbers either fluctuated or increased (2 to > 50 fold).

4. In quarry geese and dabbling ducks, changes in food supplies (of the herbivorous species), intensities of recreational activities outside the refuge areas, weather or overall population sizes could not explain the increases. In protected species, as well as goldeneye and coot, trends in local numbers were in accordance with overall population trends for some species; in others, the rate of increase in local numbers was higher than in overall population size.

5. In both areas, species richness increased from before to after the experiments. This was most pronounced in one of the areas where the refuge included shallow-watered areas in association with salt marshes.

6. It is concluded that prior to experiments, waterfowl hunting caused a displacement of quarry species, resulting in a species-poor waterfowl community. Refuge creation is an efficient management tool to improve the conservation value and biodiversity of wetlands of importance to waterfowl.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Management implications
  8. Acknowledgements
  9. References

Throughout the Northern Hemisphere, wetlands used by migratory waterfowl are under increasing human pressure. While physical development in most cases causes an irreversible loss of wetland habitat, disturbance from recreational activities causing a functional habitat loss is reversible because it can be mitigated by access regulation. Although there is a large pool of circumstancial evidence of disturbance effects by various recreational activities on waterfowl behaviour and local distribution (see reviews by Davidson & Rothwell 1993; Madsen & Fox 1995; Hill et al. 1997), most studies have not clearly demonstrated that disturbance caused an under-exploitation of the sites.

In two Danish coastal wetlands, a 12-year experimental study was designed to test whether recreational activities caused disturbance of waterfowl which results in under-exploitation of the areas. During 4 baseline study years, waterfowl hunting was identified as the recreational activity causing most critical disturbance to migratory waterfowl during autumn and winter (Madsen 1998). This paper reviews the results of subsequent experiments, restricting waterfowl shooting, which aimed to test whether observed levels of disturbance lead to under-use of the two sites by waterfowl. It was hypothesized that if hunting disturbance caused an under-exploitation of the wetlands, the establishment of experimental refuges, shifted in position between years, would lead to an overall increase in waterfowl numbers, most pronounced on the refuge areas. Furthermore, it was predicted that, within the waterfowl community, the increases would be most pronounced in species more vulnerable to hunting disturbance, i.e. the quarry species.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Management implications
  8. Acknowledgements
  9. References

Study areas

Two coastal wetlands were selected for the study by the Reserve Committee of the National Wildlife Management Council, namely Nibe-Gjøl Bredning in the Limfjord and Ulvshale-Nyord on the island of Møn (Fig. 1). Both areas are designated as Ramsar sites and Special Protection Areas under the EU Directive for the Conservation of Wild Birds. In Nibe-Gjøl Bredning, there was a wildlife refuge nearby, Ulvedybet, prior to the study. Both areas comprise brackish fjords/bays with extensive shallow-waters, islets/sand bars, reed beds, swamps and salt marshes. There is little tidal amplitude (≈10 cm) in both areas, and changes in water levels are mainly wind induced.

image

Figure 1. The two study areas, Nibe-Gjøl Bredning in the Limfjord, north Jutland, and Ulvshale-Nyord on Møn, south-east Denmark. Transects where submerged vegetation was monitored are indicated by lines with attached labels. In Nibe-Gjøl Bredning, the shallow-watered area was divided into three experimental subunits: Nibe Bredning (south of the east–west channel); Gjøl Bredning and Attrup (north of the channel); and Attrup (extending from south of Attrup village to the spit west of transect line no. 3).

Download figure to PowerPoint

In Nibe-Gjøl Bredning, the shallow-water supports submerged Zostera marina L. with varying coverage, covering ≈45 km2 at a depth range of 0·5–2·5 m. Nibe-Gjøl Bredning is mainly used by herbivorous waterfowl: mute swan Cygnus olor Gmelin, wigeon Anas penelope L. and coot Fulica atra L.

At Ulvshale-Nyord, the shallow-watered areas range from 5 to 100 cm in depth, and are covered with a mixed community dominated by Ruppia spp. and Potamogeton pectinatus L.; in the depth range 1–2·5 m, Zostera marina dominates. Within the study area, the shallow-watered area with macrophytes covers ≈24 km2. Because of the combination of shallow-watered areas and adjacent salt marshes, there is a more varied community of waterfowl than in Nibe-Gjøl Bredning, with cormorants, swans, geese, dabbling ducks, diving ducks and waders.

Both areas are used for a variety of recreational activities: fishing (mostly recreational fishing from small boats using gill nets or traps, with few commercial fishermen), sailing/boating, windsurfing, beach walking and, in autumn, shooting. The open season for most quarry waterfowl ranges from 1 September to 31 December (until 1993, the season for greylag goose Anser anser L. opened on 1 August, but was postponed 1 month from 1994 onwards). Shooting of waterfowl by shoulder guns is undertaken over the marshes and reed beds mostly using cover and decoys, from anchored punts using decoys and mobile, non-motorized punts (without decoys) on the shallow waters. At Ulvshale-Nyord, shooting from motor boats and motorized punts also takes place.

Weather conditions

In late autumn and winter, frost and ice cover will have a pronounced effect on waterfowl usage of the study sites. The normal December temperature (monthly mean) for the eastern Limfjord area, and Møn was 1·4 and 3·0°C, respectively. During 1985–94, monthly means were close to the normal (range for the Limfjord: 1·3–3·4°C; for Møn: 3·1–4·7°C). However, December 1995 and 1996 were colder than normal, with mean temperatures of –3·3 and –1·7°C for the Limfjord and –1·1 and 0·1°C for Møn (source: Danish Meteorological Institute). In both Nibe-Gjøl Bredning and at Ulvshale-Nyord, the shallow-watered areas and ponds froze over during December both years.

At Ulvshale-Nyord, waterfowl use of salt marsh pans and creeks during August–September is dependent on water levels and, hence, affected by the balance of rainfall and evaporation. The normal cumulative rainfall during July–August for the county of Storstrøm, of which Ulvshale-Nyord is a part, is 131 mm. In 1985–89 1992 and 1993, rainfall was around or above normal. However, in 1990, 1991 and 1994–96, the amount of rainfall was 36 mm or more below normal (source: Danish Meteorological Institute). In most years, there was some water in the creeks and pans during August–September; however, in 1994–96, the creeks and pans were dry until mid September; in 1996 until early October.

Experimental design

The project was divided into three phases:

1. 4-year baseline study during 1985–88 (see Madsen 1998);

2. 4-year experiment during 1989–92 with changing restrictions on hunting;

3. 4-year follow-up study during 1993–96 on the longer-term development in waterfowl numbers after permanent refuge creation in the two areas.

Experimental refuge areas were designated by wildlife refuge orders from the Ministry of Environment. The administration, including wardening, of the refuges was carried out by the National Forest and Nature Agency of the Ministry of Environment. The orders described the refuge designs used during each hunting season throughout the experimental period. During the first three experimental years, changing refuge regulations were used. The experiments were originally designed to run for the 3-year period, but because a permanent refuge solution could not be enacted before 1993 for administrative reasons, the 1991 regulations were repeated in the interim year 1992 which was included as an experimental year (Figs 2 and 3).

image

Figure 2. Experimental regulations of hunting in Nibe-Gjøl Bredning 1989–92, and the permanent wildlife reserve regulation which was implemented from 1993 onwards.

Download figure to PowerPoint

image

Figure 3. Experimental regulations of hunting and public access at Ulvshale-Nyord 1989–92, and the permanent wildlife reserve regulation which was implemented from 1993 onwards.

Download figure to PowerPoint

Nibe-gjøl bredning

The purpose of the first experimental season (autumn 1989) was to test the effect of shooting from mobile punts, which was found to have the most serious disturbance effect on waterfowl behaviour and distribution (Madsen 1998). The purpose of the second, third and fourth experimental seasons was to test the disturbance effect of all shooting from punts (both mobile and stationary shooting punts). Only shallow-watered habitat was included in the experiments. The final permanent wildlife reserve which was designated following analysis of the results of the experiments and negotiations with all local user groups, included a slightly modified refuge compared to the third and fourth experimental seasons, as well as restrictions on windsurfing and speed limits of motor boating in the central Zostera beds.

Ulvshale-nyord

The aim of the first experimental season was to test the disturbance effect of shooting over shallow-watered areas by banning all shooting at egholm Sand and putting a ban on shooting from motorized punts and boats in the remaining area. In the second, third and fourth seasons, the purpose was to test the effect of the inclusion of salt marshes in the refuge.

To test the effect of shooting along the borders of the refuge, shooting was allowed in a 100-m zone outside most of the reed beds in the first and second seasons. In practice, hunters using punts or hunters coming from the landward side were shooting from the reed beds over decoys put out into the 100-m zone. In the other experimental years, when the border was defined by the daily high water line (which lies inside the reed zone) hunters were not allowed to shoot over the open water surface outside the reed zone, and were not allowed to place decoys on the open water surface.

The final permanent wildlife reserve was very similar to the refuge design used in the third and fourth experimental seasons, but also included a ban on shooting from mobile punts, windsurfing and motor boating at egholm Sand.

Constraints

The experiments were constrained by lack of suitable replicates and control plots on the study sites. Completely independent control areas within the area or in adjacent areas could not be established, because, theoretically, changes in waterfowl numbers in the experimental areas could result from a redistribution of birds from adjacent areas. To circumvent this problem, we operated with three internal categories of treatment: (A) no treatment; (B) treatment with a ban on certain types of hunting; and (C) treatment with a ban on all hunting ( Tables 1 and 2).

Table 1.  Experimental treatments applied in Nibe-Gjøl Bredning: A, no regulation of shooting; B, regulation of some types of shooting; C, ban on shooting
YearGjøl Bredning Nibe BredningAttrup
1985AAA
1986AAA
1987AAA
1988AAA
1989BBB
1990ACA
1991CAA
1992CAA
1993CAA
1994CAA
1995CAA
1996CAA
Table 2.  Experimental treatments applied at Ulvshale-Nyord: A, no regulation of shooting; B, regulation of some types of shooting; C, ban on shooting
Yearegholm SandHegnede BugtNyord south marshOther marshes
1987AAAA
1988AAAA
1989CAAA
1990CBCA
1991BCCA
1992BCCA
1993BCCA
1994BCCA
1995BCCA
1996BCCA

Furthermore, because of the nature of the experiment, external factors, such as overall waterfowl population development, changes in recreational activities outside the refuge areas and food availability could not be controlled, but only monitored.

During 1985–92, there was no major change in the Danish Hunting and Wildlife Management Act which could affect the experiments. From the hunting season of 1994 onwards, however, two changes may have had an influence: (i) hunting of greylag goose was postponed from 1 August to 1 September, and (ii) shooting of geese in general was only allowed from 1·5 h before sunrise until 10.00 hours (before 1994, goose shooting was allowed throughout the day).

Monitoring of waterfowl and human activities

During 1985–91, one monthly aerial survey (using a Cessna 172 or a Piper Cup aircraft) was performed in each of the two study areas and adjacent wetlands in order to monitor numbers and distributions of waterfowl and human activities. Observations were made by 1–2 observers and the study areas were subdivided into units relevant for the refuge experiments (for details, see Madsen et al. 1992a,b,c,d).

During August–December 1987–96, surveys were regularly performed from the ground to monitor in more detail numbers and distributions of waterfowl and human activities. During 1987–93, at least one weekly count was performed in each area (apart from Gjøl Bredning in 1987 when only one monthly count was carried out); from 1994 onwards counts have been made at 10-day intervals. Counts were made using telescopes from elevated points in the terrain or from observation towers. Flocks of birds and human activities were mapped on 1:5000 maps. To support the mapping of birds on the shallow-watered areas, poles were were put up on transects radiating from the observation towers and perpendicularly from the coast. Ground counts were performed during the morning and mostly on week-days and, hence, do not take into account diurnal or weekly variations in bird distributions and human activities, but rather give an index of numbers and distributions. In Nibe-Gjøl Bredning, it was often impossible to distinguish mute swans and whooper swans Cygnus cygnus L. roosting at long distance, and hence, the two species have been pooled. However, it is known from accurate counts that ≈ 90% of the swan-days were mute swans. On days with moderate visibility or with dense flocks of dabbling ducks at Ulvshale-Nyord, it was sometimes impossible to identify the species, and aggregations were recorded as dabbling duck spp.

The ground count units used matched the areas with different treatments. However, in Nibe-Gjøl Bredning there was a slight discrepancy between the final refuge borders in Gjøl Bredning (from 1993 onwards) and the established count units. Because the area east of the refuge was little used by waterfowl and for hunting (see below), the difference had negligible influence on the evaluation of the refuge effect.

Monitoring of submerged vegetation

In Nibe-Gjøl Bredning, the coverage and distribution of Zostera marina and Ruppia ssp. have been monitored annually in August along fixed transects since 1987 in Nibe Bredning and since 1988 in Gjøl Bredning (Fig. 1). At 100-m intervals, the coverage of Zostera and Ruppia was estimated from a boat in a radius of ≈8 m around the boat. The position of stations was measured by aid of a Decca navigator or (since 1993) a GPS navigator.

At Ulvshale-Nyord, the biomass of submerged vegetation in the Ruppia spp./Potamogeton pectinatus community has been sampled annually in August on a fixed transect since 1987 on Egholm Sand and since 1988 in Hegnede Bugt (Fig. 1). At stations at 300-m intervals, three samples were taken at 10-m intervals perpendicular to the main transect. On egholm Sand, samples were taken at 10 stations, and in Hegnede Bugt at five stations. Samples consisted of a probe with a diameter of 15 cm and included above-sediment green vegetation, as well as a 15-cm sediment core. Vegetation was washed, sorted into species, and Potamogeton was sorted into above and within sediment material. Samples were oven dried at 60°C for 48 h before weighing.

Data analysis

To express bird usage of units of area, the number of bird-days for August–December was calculated for each unit and species based on the ground count material. Data from Nibe-Gjøl Bredning has been presented in three units representing varying treatments: Nibe Bredning, Gjøl Bredning and Attrup (Fig. 1). Data from Ulvshale-Nyord has been presented in four units: egholm Sand, Hegnede Bugt, the Nyord south marsh (south of the east–west going road) and other marshes (the Nyord north marsh, the Ulvshale marshes north of the east–west going road, and marshes along the east coast of Hegnede Bugt). Peak numbers per year have been used to express the highest number of a given species which can be counted and are based on counts performed from the air, as well as from the ground. When aerial and ground counts performed at the same date or at 1 day's interval are compared, a high degree of similarity within count units was found for Nibe-Gjøl Bredning (range for swans and wigeon ± 15% and ± 12%, respectively, with no systematic error; n = 3 days × 4 count units), whereas at Ulvshale-Nyord, which is smaller and more complex than Nibe-Gjøl Bredning, there was generally an underestimation in numbers counted from the air compared to ground counts (average for swans and dabbling ducks 17 and 27%, respectively; n =3 days × 5 count units). For 1985–86, when only aerial counts were performed, the number of bird-days and peak numbers has been presented for Nibe-Gjøl Bredning. For Ulvshale-Nyord, aerial counts were not performed in August 1985 and October and December 1986. Therefore, only peak numbers are presented and only for those species which were covered by counts during the months when they are known to peak.

To visualize the distribution pattern of waterfowl, a grid map with 500 × 500-m squares was superimposed with flock distributions on the field maps. Assuming that there was an even distribution of birds within the flocks, numbers were apportioned between grid cells. The seasonal number of bird-days (August–December) was then calculated for each grid cell.

The distribution of waterfowl in relation to the varying treatments of restriction on shooting was analysed by a multivariate GLM procedure using area and treatment as main effects and year as a covariate. To reduce effects of seasonal fluctuations, the number of bird-days per unit area was expressed as the proportion of the total for the entire study area. To normalize data, percentages were arcsine transformed before analysis. In order to derive the most parsimonious species models when the overall model was significant, further analysis was performed with non-significant variables excluded where appropriate.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Management implications
  8. Acknowledgements
  9. References

Nibe-gjøl bredning

Waterfowl numbers and distribution

Five waterfowl species occurred regularly in numbers exceeding 500 individuals during autumn (Fig. 4).

image

Figure 4. Development in numbers of five species of waterfowl in Nibe-Gjøl Bredning 1985–96, expressed by the number of bird-days (August–December) in three subunits and the peak number recorded in the entire area within an autumn season. Swans comprise mute swans and some whooper swans, which could not always be distinguished.

Download figure to PowerPoint

1. Mute swan. There was a steady increase in peak numbers and the number of bird-days up to 1993 (tripling in number of bird-days), but since then, numbers have dropped again.

2. Light-bellied brent goose Branta bernicla hrota Müller. Brent geese were not observed before 1989, but numbers have steadily increased since then, reaching a peak of 1200 individuals in 1995.

3. Mallard Anas platyrhynchos L. Peak numbers fluctuated throughout the study period (excluding periods when adjacent lakes and Ulvedybet freeze over), but the number of bird-days increased by a factor of four from 1985 to 1991/92–1996.

4. Wigeon. Peak numbers and the number of bird-days increased by a factor of 4–5 from 1985–88 to 1992–93, but since then, numbers have dropped again.

5. Coot. Numbers fluctuated, reaching a peak during 1992–94 and a low-point in 1996.

Furthermore, annual peak numbers of shelduck Tadorna tadorna L. fluctuated between 150 and 295, and goldeneye Bucephala clangula L. peak numbers fluctuated between 120 and 430; both species did not show any trend during 1985–96.

During the baseline years, wigeon numbers peaked during October–November, and on average 1700 were observed by the end of December. In subsequent years, numbers peaked between October and December, and by the end of December, on average 9010 wigeon were observed.

The GLM analysis showed a significant area effect in all seven species tested (Table 3). In brent goose and mallard, there was a significant year effect; for brent goose this is explained by the fact that the species did not occur before 1989, and for mallard by an erratic occurrence before 1991, reflected by the discrepancy between peak numbers and number of bird-days. A significant treatment effect was found for mallard and wigeon (both quarry species), while there was no effect for goldeneye and coot (both quarry species), and for the three protected species. The changing wigeon distribution is shown for the two baseline years versus the first three experimental years in Fig. 5. Hence, in 1989, when there was a ban on shooting from mobile punts throughout the area, there was an increase in wigeon numbers all over Nibe-Gjøl Bredning; in 1990, when hunting was banned in Nibe Bredning, wigeon concentrated there, whilst in 1991, when the hunting-free area was moved to Gjøl Bredning, the centre of concentration shifted to there.

Table 3.  Outcome of a multivariate GLM analysis of the response by waterfowl to refuge experiments in Nibe-Gjøl Bredning 1985–96. Categories of treatment are given in Table 1. Quarry species are indicated by a q. When the overall model was significant, the outcome of the most parsimonious model is given
Speciesn (areas × years)Year effect (covariate)Area effectTreatment effect
  • *

    P < 0·05;

  • **

    P < 0·01;

  • ***

    P < 0·001; NS, not significant.

Mute swan36NS22·5***NS
Brent goose369·85**24·9***NS
Shelduck36NS16·3***NS
Mallardq364·91* 4·99*18·1***
Wigeonq36NS 7·38** 4·39*
Goldeneyeq36NS18·7***NS
Cootq36NS14·3***NS
image

Figure 5. Distribution of wigeon in Nibe-Gjøl Bredning before experiments (1987–88) and during 3 years with varying regulations of hunting (see Fig. 4), expressed by the number of bird-days (August–December) in 500 × 500-m grid squares.

Download figure to PowerPoint

Because of a similar trend in numbers of the key species, the species diversity, expressed in relative terms, did not change significantly during the study period. In absolute terms, the species richness increased. Thus, before experiments, three species occurred annually in peak numbers exceeding 1000 individuals (an arbitrarily chosen number), whereas during and after, between three and five species exceeded that number annually.

Submerged vegetation

During 1988–91, there was a high coverage of Zostera marina at water depths less than two metres in Nibe-Gjøl Bredning (Fig. 6; see also Madsen 1998). There was no systematic sampling of coverage before 1988, but from aerial photography and visual observations it is judged that there was a similarly high coverage during 1985–87. However, in both Nibe and Gjøl Bredning, the coverage decreased after 1991 and has remained low since. Similar decreases in Zostera coverage were observed in several areas in the Limfjord and was probably due to hypereutrophication and shading by algae (Anonymous 1993, 1994).

image

Figure 6. Average percentage coverage of (a) Zostera marina and (b) Ruppia spp. along transects in Nibe-Gjøl Bredning, August 1988–96. Each year, the same transects were monitored. Vertical bars show 95% confidence limits. Percentages were arcsine transformed before analysis.

Download figure to PowerPoint

Before 1994, Ruppia spp. was hardly observed in Nibe-Gjøl Bredning, except for the shallowest parts. However, since then it has become more widely dispersed, albeit at relatively low coverage, into deeper parts which were formerly covered with Zostera.

Shooting and other recreational activities

During 1987–96, there was no significant between-year difference in the number of punt hunters using the total area (Kruskal–Wallis test, H = 13·6, P > 0·05), implying that hunters redistributed within the area (Fig. 7). Before the experiments, punt shooting was equally distributed between Nibe and Gjøl Bredning, whereas there was little activity in the Attrup area. The ban on mobile punt shooting in 1989 did not have a significant effect on the overall number of hunters. In 1990, the ban on shooting in Nibe Bredning was compensated by an increased activity in Gjøl Bredning. During 1991–92, when there was a ban on hunting in Gjøl Bredning, shooting intensity increased in Nibe Bredning and in the Attrup area. From 1993 onwards, punt hunters have had access to the eastern part of Gjøl Bredning, and this has been used to a minor degree.

image

Figure 7. Daily mean number of punt hunters active on week-day mornings in Nibe-Gjøl Bredning, September–December 1987–96.

Download figure to PowerPoint

In 1992, a windsurfer training centre was established in Nibe. However, this increased activity has not affected the study because the permanent reserve regulations implemented from 1993 onwards included a ban on windsurfing in the central Zostera beds in Nibe-Gjøl Bredning outside the summer period (Fig. 2). In other recreational activities, no trends were observed.

Ulvshale-nyord

Waterfowl numbers and distribution

Sixteen waterfowl species occurred regularly in numbers exceeding 500 individuals (Fig. 8).

imageimageimage

Figure 8. Development in numbers of 16 species of waterfowl and dabbling ducks pooled at Ulvshale-Nyord 1985–96, expressed by the number of bird-days (August–December) in four subunits as well as the peak number recorded in the entire area within an autumn season. For 1985 & 1986, only the peak number is shown (see Materials and Methods).

1. Cormorant Phalacrocorax carbo sinensis Shaw. In 1985, a colony was established in the area and the number of breeding pairs increased from a few hundred to more than 2000; this development is reflected in the late summer and autumn roosting numbers in the area.

2. Mute swan. During 1985–93, numbers steadily increased (tripling in numbers), to decrease again during 1994–96.

3. Whooper swan. Numbers fluctuated widely; however, from 1993 onwards, numbers have been low.

4. Greylag goose. Peak numbers and the number of bird-days increased by a factor of six from 1987 to 1992; since then, the number of bird-days plateaued, while peak numbers decreased.

5. Dark-bellied brent goose Branta bernicla bernicla L. Numbers fluctuated greatly; however, both peak numbers and the number of bird-days tripled from 1985–90 to 1991–96.

6. Barnacle goose Branta leucopsis Bechstein. Before 1991, peak numbers of up to 564 individuals were observed for short periods; since 1991, the peak numbers and the number of bird-days has increased 12 fold.

7. Canada goose Branta canadensis L. From 1985–88 to 1993, numbers increased by a factor of eight; however, numbers have abruptly decreased since then. In 1996, ice covered much of the shallow-watered area in December and the number of Canada geese was reduced.

8. Mallard. Peak numbers and the number of bird-days increased by a factor of 4–5 from 1985–88 to 1993. Peak numbers have slightly decreased since 1993, while the number of bird-days has remained stable.

9. Teal Anas crecca L. From 1985–88 to 1993, numbers increased by a factor of 23; since then, peak numbers have decreased somewhat, while the number of bird-days has remained stable.

10. Pintail Anas acuta L. From 1985–88 to 1996, numbers have increased by a factor of 50.

11. Wigeon. From 1985–88 to 1993–96, numbers have increased by a factor of 30. From 1993 onwards, numbers have fluctuated dramatically.

12. Shoveler Anas clypeata L. From 1985–88 to 1995–96, numbers increased by a factor of 25.

13. Goldeneye. During 1985–96, both peak numbers and the number of bird-days fluctuated without any overall trend.

14. Coot. Numbers fluctuated, reaching a peak in 1992 and a low-point in 1996.

15. Lapwing Vanellus vanellus L. From 1985–88 to 1991–93, numbers increased by a factor of four, but then dropped again.

16. Golden plover Pluvialis apricaria L. From 1985–90 to 1991–93, numbers abruptly increased by a factor of six, but have dropped again since then.

The total number of dabbling ducks, including unidentified ducks, increased by a factor of seven from 1985–88 to 1993–96. To a large degree, the total number reflects the change in wigeon numbers.

Before the experiments, dabbling ducks, except mallard, peaked in numbers in September–October & hardly any stayed until December. From 1991 onwards, an average of 394 teal, 118 pintail, 1234 wigeon and 23 shovelers have been counted at the end of December (excluding 1996 when the shallow-watered areas were almost completely frozen).

The GLM analysis showed a significant area effect in all 16 species tested, except for whooper swan and teal (Table 4). Significant treatment effects were found for greylag goose, Canada goose, the five species of dabbling ducks and coot, i.e. in eight out of nine quarry species, while there was no effect in the seven protected species.

Table 4.  Outcome of a multivariate GLM analysis of the response by waterfowl to refuge experiments at Ulvshale-Nyord 1987–96. Categories of treatment are given in Table 2. For species which do not use terrestrial habitats, the sample size is reduced. Quarry species are indicated by a q. When the overall model was significant, the outcome of the most parsimonious model is given
SpeciesN (areas × years)Year effect (covariate)Area effectTreatment effect
  • *

    P < 0·05;

  • **

    P < 0·01;

  • ***

    P < 0·001; NS, not significant.

Cormorant20NS928·5***NS
Mute swan20NS 53·2***NS
Whooper swan20NSNSNS
Greylag gooseq40NS 6·26** 4·41*
Brent goose40NS 22·1***NS
Barnacle goose40NS 17·5***NS
Canada gooseq40NS135·4*** 4·69*
Mallardq40NS153·8***17·9***
Tealq40NSNS14·1***
Wigeonq40NS 63·4***14·3***
Pintailq40NS 24·1***13·2***
Shovelerq40NS 7·51** 5·88**
Goldeneyeq20NS241·2***NS
Cootq20NS 16·8**14·1***
Lapwing40NS 7·75***NS
Golden plover40NS 4·29*NS

The detailed changes in distribution of dabbling ducks (all species pooled) during the two baseline years versus the three first experimental years (Fig. 9) shows the importance of the undisturbed connection between the marshes and the adjacent shallow-water. To test the effect of the regulation in 1990, when shooting was allowed in some of the 100-m zones outside the reed beds in Hegnede Bugt (see Fig. 3), the dabbling duck density in grid squares which overlay the 100-m zone in Hegnede Bugt was compared for 1990 and 1991 (n = 27). To correct for the increase in total number of bird-days in Hegnede Bugt from 1990 to 1991, the density per square is expressed as the proportion of the total number of bird-days spent in Hegnede Bugt that year. On this basis, in 1990, 54% of all dabbling duck bird-days in Hegnede Bugt were spent in squares overlaying the 100-m zones with shooting. In 1991, when shooting was not allowed in the 100 m zone, 82% of the bird-days were spent in the same squares (two-tailed Wilcoxon paired-sample test: T = 54; P < 0·001).

image

Figure 9. Distribution of dabbling ducks (five species pooled) at Ulvshale-Nyord before experiments (1987–88) and during 3 years with varying regulations of hunting, expressed by the number of bird-days (August–December) in 500 × 500-m grid squares.

Download figure to PowerPoint

The number of waterfowl using the Nyord south marsh during August in the years after refuge creation has fluctuated greatly, which has been related to availability of water in the creeks and pans. Hence, in August 1991–93, when water was available, the number of bird-days spent by greylag geese, dabbling ducks and waders (lapwing and golden plover) was on average 13 850, 24 166 and 14 210, respectively. During 1994–96, when creeks and pans were dry, the number of bird-days spent by the three species groups was 61, 22 and 730, respectively. The lack of water was the main reason for the overall decreased use of the Nyord south marsh by waterfowl during 1994–96 and, in particular, in 1996, when surface water was first visible in early October. At least to some degree, the decreased use of the marshes was compensated by redistribution of birds (see greylag goose and dabbling ducks in Fig. 8); in lapwing and golden plover, there was a sharp decline in overall numbers from 1991–93 to 1994–96, which may be explained by the local lack of suitable wet marsh habitat.

Before the experiments, the community of waterfowl using Ulvshale-Nyord was dominated by mute swan and mallard, making up 60% of the total number of bird-days spent by the 16 species (Fig. 10). During the first 2 years of experiments, their dominance did not change. From 1991 onwards, when the refuge regulations have not altered importantly, species evenness has increased, with wigeon now as the most numerous species. In absolute terms, before the experiments, four or five species were annually recorded in numbers exceeding 1000 individuals; from 1989 onwards, 10–14 species were annually exceeding those numbers.

image

Figure 10. Waterfowl species diversity at Ulvshale-Nyord before experiments and during two phases with different refuge designs, expressed by the relative abundance of 16 species (percentage of total number of bird-days, August–December). For each of the three periods, the average between years is given.

Download figure to PowerPoint

Submerged vegetation

At egholm Sand during 1987–96, there was no significant between-year variation in the biomass of submerged vegetation in August (F9,29 = 2·45; P > 0·05). In Hegnede Bugt, there was a significant between-year variation, with 1990 and 1996 differing significantly from the other years (F8,14 = 7·13; P < 0·01, Tukey's test)(Fig. 11).

image

Figure 11. Average biomass of submerged vegetation (Ruppia spp./Potamogeton pectinatus community) at Ulvshale-Nyord, sampled at stations on fixed transects in August at egholm Sand 1987–96, and Hegnede Bugt 1988–96. Vertical bars show 95% confidence limits.

Download figure to PowerPoint

Shooting and other recreational activities

During 1987–96, the daily total number of hunters present in the area (mornings only) varied significantly between-years (Fig. 12; Kruskal–Wallis test, H = 35·7; P < 0·001), with 1996 significantly differing from the other years and 1987, 1988, 1992 differing from 1989–91 to 1993–95 [non-parametric multiple comparisons (Zar 1984)]. From 1993 to 1996, there was a decreasing trend in the number of hunters. In Hegnede Bugt, there was no shooting from punts from 1991 onwards, so the remaining hunters were shooting exclusively from the reed swamps. In other recreational activities, no trends were observed.

image

Figure 12. Daily mean number of hunters active on week-day mornings at Ulvshale-Nyord, September–December 1987–96.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Management implications
  8. Acknowledgements
  9. References

The experiments in both study areas showed that quarry waterfowl responded to the changing shooting regulations, both in terms of numbers and distribution. This is in accordance with the hypothesis that, if hunting was a disturbance source affecting the distribution and local numbers of waterfowl, it would be expected to affect most critically the quarry species. Some quarry species did not respond as predicted, namely goldeneye and coot (in Nibe-Gjøl Bredning). The possible explanation for goldeneye is that the species occurs in the deeper parts of the shallow-watered areas where it is not so affected by punt shooting; for coot, it has been shown that the species does not show as vigorous a behavioural response to shooting disturbance as wigeon (Madsen 1998) and, hence, are better able to cope within a site.

In the protected species which increased in numbers, i.e. cormorant, mute swan, light-bellied brent goose and barnacle goose, there was no evidence to suggest that the increases were due to restrictions on hunting. In lapwing and golden plover at Ulvshale-Nyord, the abrupt increases in numbers from 1989 to 1991 suggest an effect of the inclusion of the Nyord south marsh in the refuge; however, in the overall analysis of distribution, there was no significant treatment effect.

In neither of the two areas can the observed increases in numbers of herbivorous waterfowl be explained by increased food supplies. On the contrary, in Nibe-Gjøl Bredning, the decrease in Zostera marina was probably the reason behind the decrease in numbers of wigeon during 1994–96. The decrease in Zostera was partly compensated by an increase in Ruppia. However, Ruppia does not have the same shoot lengths and biomass as Zostera (Madsen et al. 1992b; P. Clausen, personal communication) and, in particular in situations with high water levels, wigeon probably encountered problems with food availability. Mute swan, light-bellied brent goose and coot may have partly encountered the same problems, but due to their longer necks and diving ability (coot), they are not as susceptible to reduced food supplies as wigeon.

Food stocks of species feeding on other items (seeds, zooplankton, invertebrates, fish and mussels) were not monitored. Therefore, it cannot be judged whether changes in these supplies influenced the experiments. However, the similarity in response by quarry geese and dabbling ducks, irrespective of diet, suggests that there was not a general trend in food supplies affecting the results.

Weather conditions cannot explain the increases in numbers. The droughts during late summer 1994–96 had a negative effect on the number of waterfowl using the marshes at Ulvshale-Nyord; however, except for lapwing and golden plover, there is no discernable effect on the overall use of the area. Ice cover played a significant role in 1996, but only for the numbers of Canada geese and, to a lesser extent, mallard, which arrive during late November to early December.

Changes in recreational activities, in particular shooting, did happen in both areas. In Nibe-Gjøl Bredning, the slightly increased shooting activity in the areas adjacent to the refuge could have reinforced the treatment effect; on most week-days, however, the density of hunters was well below the threshold at which wigeon numbers will be affected on a diurnal basis (see Madsen 1998). At Ulvshale-Nyord, the number of hunters decreased when the experiments started, which matches the decrease in available hunting areas. The upsurge in activity in 1992 was probably a result of the release of a report (Madsen et al. 1992a, of which a summary was published in the hunters′ magazine) describing the first results of the experiments, i.e. that waterfowl numbers had increased inside the refuge areas, as well as in adjacent areas still open for hunting. The decreased shooting activity during 1993–96 may have facilitated an increased waterfowl use of the areas adjacent to the refuge. It is, however, difficult to separate the effect of the decreased hunting activity from the drought effects during 1994–96. For example, the decreased use of Nyord south marsh by greylag geese during 1994–96 could be interpreted as a drought effect or the effect of the general ban on goose shooting during August implemented from 1994 onwards. Thus, due to the ban, the geese would disperse from the refuge to adjacent areas in August. The general ban on goose shooting after 10·00 hours may be the reason behind the decrease in numbers of Canada geese at Ulvshale-Nyord from 1994 onwards, because the geese have found suitable roosts and feeding areas which were hitherto unattractive due to hunting disturbance.

In neither of the two areas did substantial changes in habitat management other than the refuge creations take place during 1985–96.

Some of the increases in waterfowl numbers may potentially be an effect of an increase in overall population size. From the monitoring of wintering waterfowl in north-west Europe, coordinated internationally by Wetlands International and nationally by the National Environmental Research Institute, trends are known for most of the populations of waterfowl concerned (Rose 1995, giving trends for swans, ducks and coot up to 1993; the Goose Specialist Group of Wetlands International, unpublished, giving trends for geese up to 1995).

The Danish cormorant population has increased steadily during the 1980s, 1990s (Bregnballe & Gregersen 1995); and combined with the establishment and expansion of a colony at Ulvshale-Nyord, this has been the primary reason for the increase in numbers roosting during autumn. The Baltic population of mute swan almost doubled during 1987–93 following a series of mild winters, with a subsequent decrease following a series of cold winters in the 1990s. This development is mirrored in both study areas. The eastern Denmark breeding population of greylag geese, of which the autumn-staging geese at Ulvshale-Nyord are part, has been stable during the 1980s and 1990s (Jørgensen, Madsen & Clausen 1994); whereas numbers at Ulvshale-Nyord have increased. The population of dark-bellied brent goose has doubled in numbers from the mid-1980s to the mid-1990s; a similar development was observed at Ulvshale-Nyord. The Svalbard population of light-bellied brent goose has only slightly increased during the 1980s and 1990s, and the rapid increase in numbers using Nibe-Gjøl Bredning has to be explained as a redistribution of birds within the autumn-staging and wintering grounds of the population, driven by declines in food supplies in the traditional autumn-staging areas (Clausen et al. 1998). The Russian-Baltic breeding population of barnacle goose has doubled in size from the mid-1980s to the mid-1990s, compared to a 12-fold increase at Ulvshale-Nyord. The Baltic population of Canada goose has increased by ≈40% from the mid-1980s to the early 1990s, whereas at Ulvshale-Nyord, numbers increased by a factor of eight. The north-west European wintering populations of mallard, teal, pintail and shoveler have been stable since the mid 1980s, which is in contrast to 4–50-fold increases in the experimental areas. The north-west European wintering population of wigeon has doubled from 1986 to 1993, in contrast to a 4–30-fold increase in the two study areas. Finally, the European population of lapwing is regarded as stable, while the north-west European population of golden plover is decreasing (Rose & Scott 1994); both increased in numbers at Ulvshale-Nyord. For the remaining species which fluctuated in numbers (whooper swan, shelduck, goldeneye, coot); the population trends have been either stable (shelduck, coot) or slightly increasing (whooper swan, goldeneye).

To summarize:

1. the observed local trends in numbers of cormorant, mute swan, (whooper swan), dark-bellied brent goose, shelduck, (goldeneye) and coot largely reflect trends in population size;

2. in light-bellied brent goose, barnacle goose, Canada goose and wigeon, the overall population sizes have increased, but the rates of increase in the two study areas override the rate of increase in the whole population;

3. in greylag goose, mallard, teal, pintail, shoveler, lapwing and golden plover, overall population sizes have been stable or have decreased, while increasing in the two study areas;

4. changes in local food supplies (at least for the herbivorous species); weather conditions and recreational activities outside the refuge areas cannot explain the observed increases in waterfowl numbers;

5. for the quarry species, the most likely reason behind the increases was the creation of hunting-free areas, while for the protected species, it is not known whether the increases were due either to an attraction to the areas for reasons other than restrictions on shooting (e.g. in light-bellied brent, a discovery of a hitherto unexploited habitat) or to refuge creation.

In both study areas, but most pronounced at Ulvshale-Nyord, refuge creation resulted in a more species-rich waterfowl community, implying that prior to experiments, hunting disturbance caused an assymetry in response amongst different species. Specifically, the species hypothesized to be most vulnerable to hunting disturbance were displaced, i.e. the coastal quarry geese and dabbling ducks, while the more tolerant species remained, i.e. the protected species as well as quarry diving ducks and coot. This finding supports the disturbance ‘gradient’ hypothesis by Hockin et al. (1992), as well as the classification of species vulnerability to hunting disturbance (Bell & Owen 1990; Madsen, Pihl & Clausen 1998).

The reason that the species enrichment effect was most pronounced at Ulvshale-Nyord is probably because the refuge included marsh habitat in association with shallow-watered areas. The inclusion of the two habitats creates optimal conditions by which dabbling duck and waders can freely choose between feeding/roosting in the marsh or on the mudflats and shallow-waters during periods with low water levels in the bay. During high water periods, feeding opportunities in the shallow-watered habitat become restricted, but birds can still feed and roost in the marshes, provided that some water is available in creeks and pans.

The establishment of reserves in the two study areas has almost doubled the autumn and winter national totals for wigeon and shoveler. Ulvshale-Nyord has become one of the most important sites for coastal waterfowl in Denmark, and certainly the site with the highest species richness.

Where did the waterfowl, whose numbers have built up in the reserves so rapidly, come from? The speed by which the increases took place excludes the possibility that enhanced survival or fecundity of birds staying in the reserve areas played a major role. Two other explanations remain: (1) that the birds were attracted to the refuge areas from adjacent sites which have consequently lost their relative value, or (2) that individual birds have prolonged their length of stay in the refuges which cumulatively leads to an increase in overall numbers. The inference of (2) is that, before refuge creation, birds passed through the areas more rapidly on their way to more southern autumn-staging and wintering grounds; whereas now, birds are held back in the refuges and they are ‘missing’ further south in the flyway. Madsen et al. (1995) analysed counts of wigeon from key areas adjacent to the two refuges and found no negative trend in autumn numbers.

For some species, the increases cannot be accounted for by even a theoretically complete ‘drainage’ of all sites within the two regions. In the autumns of 1987 and 1988, national aerial surveys of waterfowl were performed (Laursen et al. 1997). Even though aerial surveys probably underestimate the true numbers present, they give an overview of total numbers, especially of species such as wigeon and pintail which congregate in relatively few coastal areas. In the entire Limfjord and northern Kattegat region, excluding Nibe-Gjøl Bredning, 6600 and 9900 wigeon were counted in autumn 1987 and 1988, respectively. In comparison, in Nibe-Gjøl Bredning, autumn peak numbers increased from 5000 to 26 000 within a few years. Similarly, in south-east Denmark outside Ulvshale-Nyord, 2300 and 1300 wigeon were counted in the autumns of 1987 and 1988, which is only around 10% of the peak numbers recorded at Ulvshale-Nyord after the experiments. In south-east Denmark, fewer than 100 pintails were observed in 1987 and 1988, compared to 2000–4000 solely at Ulvshale-Nyord after the experiments. Hence, for several species, e.g. Canada goose, teal, wigeon, pintail and shoveler, the most likely explanation is that the increases were primarily due to prolonged length of stay. This hypothesis will be explored in more detail elsewhere. In other species that occur in large numbers in south-east Denmark, e.g. greylag goose and mallard, it is likely that redistribution from nearby sites also played an important role.

In dabbling ducks, it took 2–6 years between the creation of the refuges and the time when peak numbers were reached, and for some species, numbers had not levelled off by 1996. One possible reason is that birds are sampling environmental conditions in several areas. In this way, an individual bird will gain experience over some years of the reduced predation risk associated with the newly-established refuge areas and will then gradually prolong its length of stay. In Nibe-Gjøl Bredning, the parallel decline in wigeon numbers and Zostera coverage after 1993 suggest that the carrying capacity of the food supplies may have been reached. Similarly, the stabilization in numbers of some species at Ulvshale-Nyord suggests that food supplies may set a limit on the numbers; however, for many species it is premature to judge, even 6 years after refuge creation.

Management implications

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Management implications
  8. Acknowledgements
  9. References

The study strongly suggests that waterfowl hunting caused a displacement of birds from important conservation areas and that the creation of hunting-free areas was an efficient management tool to improve waterfowl site usage, as well as biodiversity. The combination of shallow-watered areas and coastal marshes in a refuge provides an optimal design for dabbling ducks and waders. Shooting on the borders of a refuge affects distribution and probably habitat exploitation by waterfowl, and should be avoided, so that refuge borders follow natural borders in the landscape (see also Fox & Madsen 1997).

In areas designated to afford protection to migratory waterfowl, e.g. Ramsar sites and EU Special Protection Areas, sustainability is a management goal, which can be interpreted as enhancing conservation interests whilst maintaining some human recreational and commercial use. As shown by Madsen (1998) some activities are more disruptive to waterfowl behaviour and distribution than others, and activity levels and thresholds can be identified at which bird usage will not be affected. The finding was that mobile activities, e.g. mobile punt shooting, are more disruptive than activities at fixed positions. Such knowledge can be useful for the design of zoning of human activity in areas adjacent to a core refuge area.

In Denmark, as well as other countries which are either signatory to the Ramsar Convention or members of the EU, many conservation areas have been designated on the basis of the number of migratory waterfowl which were present, with no attention paid to the potential value of the sites. Certainly, the two experimental refuges have demonstrated that the potential for waterfowl was much greater than was suggested by the counts prior to experiments. The composition of the waterfowl community, especially the lack of species vulnerable to hunting disturbance, may give a clue whether hunting has a displacement effect, and this can be used as an indicator of the level of disturbance.

The results from the experimental reserves have served to form a basis for the designation and design of more than 50 new wildlife reserves within Danish coastal EU Special Protection Areas. The new reserves will be implemented during 1993–2000 (Madsen, Pihl & Clausen 1998). Because it is estimated that the number of migratory waterfowl passing through the country is an order of magnitude greater than what can be counted at any time in the autumn, it is anticipated that the new extended network of reserves has the potential to hold back significant proportions of the north-west European populations of dabbling ducks during autumn.

The wider implications of such refuge networks at population or flyway scale still have to be addressed. This question is, however, difficult to answer because, in migratory waterfowl, density dependence processes may operate thousands of kilometres away from the place where birds were disturbed (Madsen & Fox 1995). If refuge areas in the northern end of a flyway can hold back significant numbers of birds which would otherwise have been ‘pushed’ into wintering grounds with limited resources, then it is possible that the refuge creations will have a positive impact at a population level, achieved by increasing survival and/or fecundity.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Management implications
  8. Acknowledgements
  9. References

The experimental study was carried as a collaborative project with the National Forest and Nature Agency, and Palle Uhd Jepsen and Bjarne Søgaard are sincerely thanked for their support from the administrative side. I am very grateful to my colleagues who participated in the project: Ebbe Bøgebjerg, Jens Peder Hounisen, John Frikke, Jan Bolding Kristensen and Hans Erik Jørgensen. Peter Mikkelsen prepared the maps. Tony Fox is thanked for commenting on the manuscript.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Management implications
  8. Acknowledgements
  9. References
  • Anonymous 1993 Bundvegetation Limfjorden 1992. Report, The Limfjord Monitoring by the county councils of Ringkjøbing, Viborg and Nordjylland. (In Danish.).
  • Anonymous 1994 Bundvegetation Limfjorden 1993. Report, The Limfjord Monitoring by the county councils of Ringkjøbing, Viborg and Nordjylland. (In Danish.).
  • Bell, D.V. & Owen, M. 1990 Shooting disturbance—a review. Managing Waterfowl Populations (ed. G. V. T.Matthews), pp. 159171. International Waterfowl and Wetlands Research Bureau Special Publication no. 12. IWRB, Slimbridge, UK.
  • Bregnballe, T. & Gregersen, J. 1995 Development of the breeding population of cormorant Phalacrocorax carbo sinensis in Denmark 1938–94. Dansk Ornitologisk Forenings Tidsskrift, 89, 119134.
  • Clausen, P., Madsen, J., Percival, S.M., O'Connor, D. & Anderson, G.Q.A. 1998 Population development and changes in winter site use by the Svalbard light-bellied brent goose. 1980–94. Biological Conservation, 84, 157165.
  • Davidson, N. & Rothwell, P. 1993 Disturbance to waterfowl on estuaries. Wader Study Group Bulletin, 68, 1106.
  • Fox, A.D. & Madsen, J. 1997 Behavioural and distributional effects of hunting disturbance on waterbirds in Europe: implications for refuge design. Journal of Applied Ecology, 34, 113.
  • Hill, D., Hockin, D., Price, D., Tucker, G., Morris, R. & Treweek, J. 1997 Bird disturbance: improving the quality and utility of disturbance research. Journal of Applied Ecology, 34, 275288.
  • Hockin, D., Ounsted, M., Gorman, M., Hill, D., Keller, V. & Barker, M. 1992 Examination of the effects of disturbance on birds with reference to the role of environmental impact assessment. Journal of Environmental Management, 36, 253286.
  • Jørgensen, H.E., Madsen, J. & Clausen, P. 1994 Rastende bestande af grs i Danmark 1984–92. Danmarks Miljøundersøgelser Faglig Rapport, 97, 1112.
  • Laursen, K., Pihl, S., Durinck, J., Hansen, M., Skov, H., Frikke, J. & Danielsen, F. 1997 Numbers and distribution of waterbirds in Denmark 1987–89. Danish Review of Game Biology, 15, 1181.
  • Madsen, J. 1998 Experimental refuges for migratory waterfowl in Danish wetlands. I. Baseline assessment of disturbance effects of recreational activities. Journal of Applied Ecology, 35, 386397.
  • Madsen, J., Bøgebjerg, E., Hounisen, J.P., Kristensen, J.B. & Frikke, J. 1992a Forsøgsreservat Ulvshale-Nyord: Resultater af eksperimenter 1989–91. Danmarks Miljøundersøgelser Faglig Rapport, 55, 161.
  • Madsen, J., Bøgebjerg, E., Kristensen, J.B., Frikke, J. & Hounisen, J.P. 1992b Forsøgsreservat Ulvshale-Nyord: Baggrundsundersøgelser efteraÅret 1985 til foraÅret 1989. Danmarks Miljøundersøgelser Faglig Rapport, 46, 150.
  • Madsen, J. & Fox, A.D. 1995 Impact of hunting disturbance on waterbirds—a review. Wildlife Biology, 1, 193207.
  • Madsen, J., Frikke, J., Bøgebjerg, E., Kristensen, J.B. & Hounisen, J.P. 1992c Forsøgsreservat Nibe-Gjøl Bredning: Baggrundsundersøgelser efteraÅret 1985 til foraÅret 1989. Danmarks Miljøundersøgelser Faglig Rapport, 47, 157.
  • Madsen, J., Hounisen, J.P., Bøgebjerg, E. & Jørgensen, H.E. 1995 Rastende bestande af vandfugle i forsøgsreservaterne. 1985–93. Danmarks Miljøundersøgelser Faglig Rapport, 132, 140.
  • Madsen, J., Hounisen, J.P., Brgebjerg, E. & Frikke, J. 1992d Forsøgsreservat Nibe-Gjøl Bredning: Resultater af eksperimenter 1989–91. Danmarks Miljøundersøgelser Faglig Rapport, 53, 143.
  • Madsen, J., Pihl, S. & Clausen, P. 1998 Establishing a reserve network for waterfowl in Denmark: a biological evaluation of needs and consequences. Biological Conservation, in press.
  • Rose, P.M. 1995 Western Palearctic and South-West Asia Waterfowl Census 1994. IWRB Publication, 35, 1119.
  • Rose, P.M. & Scott, D.K. 1994 Waterfowl Population Estimates. IWRB Publication, 29, 1102.
  • Zar, J.H. 1984Biostatistical Analysis. Prentice Hall, New Jersey, USA.

Received 16 July 1997; revision received 7 February 1998