Data representability

Estimates of the annual spatial, temporal and sex and age distribution of the eider bag are based on extrapolations from the wing survey data and thus rely on the representativeness of the data. Noer et al. (1995) have previously shown a significant correlation (r > +0.79) between the number of eider wings received from separate counties and the county bag size in the 1980s. They also found that the seasonal distribution of wings over month corresponded to the seasonal distribution of ring recoveries of bagged birds confirming similarity in the temporal distribution. For the period 1992/93–1998/99, correlations between the number of wings and bag size in separate counties were highly significant in five of seven seasons (correlation coefficients between +0.85 and +0.98, t₁₂ > 5.6, P < 0.0005), whereas slightly weaker but still significant correlations were found in the seasons 1993/94 (r = 0.71, t₁₂ = 3.5, P < 0.005) and 1995/96 (r = 0.61, t₁₂ = 2.7, P < 0.05). Similarly, significant correlations were found between the average percentage of wings and bag size among individual counties (r = 0.93, t₁₂ = 8.9, P < 0.0001). Accordingly, the wing survey data are considered representative in relation to the spatial distribution of the eider bag among counties.

Statistical analyses

Long-term changes and trends in the eider bag were analysed using a program designed to dissect time-series index values into periods (segments) of more uniform development (Kirby & Bell 1996). This analysis, called Segment Analysis, allows changes to be identified and quantified within separate segments of the time period under consideration. Segments are contiguous when the underlying rate of change is effectively constant, and the index values of all years included in separate segments contribute to quantify the change. Changes are expressed as the annual rate of change, given in percent.

Dissection of the time series of eider bag index values into separate segments was done using a combination of random and directed searches. The overall strategy of these searches was to minimise a criterion measuring the lack-of-fit (residual mean square) of regressions of all segments. Kirby & Bell (1996) showed that to ensure detection of a global rather than a local minimum, a search strategy which first included a small number of random searches, then directed searches and ultimately a large number (> 1,000) of random searches was very successful. This strategy was applied in analyses of the full eider bag time series. Assessed from Akaike's Information Criterion (AIC; Lindsey 1997) in comparison of models including 1–8 segments, the most appropriate number of segments to fit the series was three (AIC = −134.8).

In order to identify and evaluate the importance of factors that potentially were responsible for the total annual bag size the following variables were included in a stepwise multiple regression analysis (Proc REG, model-selection = stepwise; SAS Institute 1988): the annual number of eider hunters, the annual number of days on which eiders were bagged (for separate months) and the annual ratio of juveniles to adult females (for separate month). The annual number of hunters and the number of hunting days were initially considered as independent variables as they were not correlated (r² = 0.14, t₁₆ = 1.54, P > 0.05). Since recordings of offshore hunting days, as defined above, started in 1983, the analysis was restricted to the period 1983/84–1999/2000. Values of P = 0.20 were used for variables to enter and P = 0.05 for variables to stay in the model. In all other analyses, the level of significance was set at 0.05.

As annual counts of the wintering population of eiders in Danish waters do not exist, the potential effect of changes in population size on bag size was not assessed using this analysis. Thus to assess the influence of the size of the winter population of eiders on the size of the eider bag, the changes in bag size were compared to the development in hunter numbers and to the number of eiders bagged per hunter in separate counties using standard linear regression analysis (Sokal & Rohlf 1981). Linear regression was accepted as residual variation in these time series was normally distributed assessed by both skewness and kurtosis values (ranging between −2.0 and 2.0) and by Shapiro-Wilks test of normality (Proc UNIVARIATE; SAS Institute 1988). Only in a few cases, mostly in counties with small bag sizes, were the criteria of normality not met (see Table 1). In all county-wise comparisons it was assumed that if a change in the bag size was influenced by eider numbers, then the number of eiders bagged per hunter would follow changes in bag size given that hunter numbers were stable, e.g. a decreasing bag size and a decreasing number of eiders bagged per hunter would be indicative of lower bird numbers present. Conversely, corresponding changes in bag size and in the number of hunters coupled with a stable number of eiders bagged per hunter, would suggest that bag size was not related to eider numbers.

Table 1.

Average proportion of total bag size of individual counties during 1980–1999, and results of regression statistics of the development in bag size, hunter numbers and bag size per hunter in separate counties and of pooled data during the hunting seasons 1980/81–1999/2000; df = 19 in all cases (see data in Appendix I). Cases where data deviated from a normal distribution (see Methods) are shown in italics.

Long-term trends in the eider bag

During the hunting seasons of 1958/59–2000/01, the total annual bag of eiders in Denmark increased from ca 100,000 in the late 1950s to ca 140,000 during the 1970s and 1980s and decreased during the 1990s to ca 83,000 (Fig. 2).

Figure 2.

Danish bag of eiders during 1958/59–2000/01 and the number of hunters that bagged eiders during 1980/81–1999/2000. Regression lines obtained from Segment Analysis of the bag size are fitted by eye (see text for details). ▴ indicates the occurrence of ice-winters.

Segment analysis separated the eider bag time series into three periods: a period with an annual rate of increase of 4.76% (95% confidence limits: 2.33%, 7.25%) from 1958/59 to 1969/70, a stable period (annual change: −0.38%; 95% c.l.: −1.18%, 0.43%) from 1970/71 to 1992/93, and a period with an annual decrease of −5.28% (95% c.l.: −9.74%, −0.61%) from 1993/94 to 2000/01 (see Fig. 2). The decrease during the period 1993/94–2000/01 is considered to be significant since the 95% c.l. do not encompass zero.

Eider hunting 1980–1999

The total size of the eider bag and the total number of active eider hunters have decreased markedly during 1980–1999 (bag size: r² = 0.58, t₁₉ = −4.97, P < 0.001; hunter numbers: r² = 0.84, t₁₉ = −9.83, P < 0.001; see Fig. 2 and data in Appendix I). The decrease in the number of eider hunters from ca 13,000 to ca 7,200 during this period was related to a specific reduction in the number of hunters that reported having shot eiders since the total number of hunter licenses in Denmark did not change during this period (r² = 0.18, t₁₉ = −2.02, P > 0.05). The proportion of successful eider hunters declined from ca 7.5% (N = 13,915) in the 1980/81 season to 4.4% (N = 7,359) in the 1999/2000 season.

Despite some year-to-year variation in hunter numbers, marked decreases were evident between the seasons 1985/86 and 1986/87 and between the seasons 1992/93 and 1993/94 (see Fig. 2). The decrease in hunter numbers between 1992/93 and 1993/94 coincides with a marked decline in bag size, but no corresponding relationship was evident for the decrease between 1985/86 and 1986/87.

The results of linear regression of national and county-wise development in bag size, in the number of hunters and in the average individual bag size per hunter during 1980/81–1999/2000 is shown in Table 1 (data in Appendix I). Most eiders were bagged in the central counties with long coastlines bordering southwest Kattegat (Vestsjælland, Fyn and Århus) and in counties adjacent to the southwest Baltic Sea and the Wadden Sea (Storstrøm, Sønderjylland and Ribe). The smallest numbers were bagged in the counties of Ringkøbing and Viborg in western Denmark adjacent to the North Sea and in the county of Bornholm in the Baltic Sea. In general, significant decreases in bag size were found in the eastern parts of Denmark, whereas those in western Denmark remained unchanged or even, in the county of Ringkøbing, significantly increased. Hunter numbers have significantly decreased everywhere, except in Ringkøbing and Viborg. In most counties the average number bagged per hunter has been stable (see Table 1).

Factors affecting bag size

Given that annual counts of the huntable wintering population of eiders in Danish waters do not exist, the potential influence of the size of the wintering population on the bag size was assessed from the development in bag size, number of hunters and average individual bag size of the different counties during 1980–1999 (see Table 1 and data in Appendix I). Comparisons of the trend in these variables between counties showed a consistent pattern in six of 14 counties (København, Frederiksborg, Roskilde, Storstrøm, Fyn and Vejle,) all having a significant decreasing bag size, significant decreasing number of hunters and stable individual bag sizes. A similar pattern was found in the counties of Sønderjylland, Århus and Nordjylland, although the tendency of a declining bag size was not significant. However, in these nine counties, representing on average ca 80% of the total annual eider bag, there was no indication that the total number of eiders available to hunters decreased as a result of a population decline. In the county of Vestsjælland a significant decrease in all variables suggests that the decline in bag size may not only be related to a decreasing number of eider hunters, but may also be affected by a decrease in the number of eiders during 1980–1999. In the county of Ribe bag size has remained stable, hunter numbers have declined whereas individual bag size has increased significantly. Thus, in this county fewer hunters have shot a similar number of eiders during 1980–1999 by improving individual bag size, which may indicate that the number of eiders available have increased or remained stable. Due to very low numbers shot (< 1% of the annual bag; see Table 1) in each of the last three counties, Bornholm, Ringkøbing and Viborg, these counties were not considered.

At a national level, no changes were found in the annual average number of eiders bagged per hunter during 1980–1999 (r² = 0.01, t₁₉ = 0.52, P = 0.612). Eider hunters bagged on average a stable number of 12.2 ± 0.29 (SE) birds per season, giving no indications that the decrease in bag size was related to a decrease in eider numbers, nor that the decrease in the number of eider hunters should be related to a decrease in individual bag size (see Table 1). As hunter numbers may decrease in response to lower bird numbers in the preceding season rather than to low bird numbers in the actual season, a positive correlation between changes in hunter numbers and changes in individual bag size in the preceding season(s) would be expected if hunter numbers decline due to lower numbers of eiders present. However, no significant relationship was found between changes in hunter numbers and preceding changes in individual bag size (one-year delay: r² = 0.007, t₁₇ = −0.34, P = 0.736; two-year delay: r² = 0.042, t₁₆ = 0.81, P = 0.431; three-year delay: r² = 0.064, t₁₅ = 0.98, P = 0.343). Likewise, the two marked departures of hunters recorded in 1986 and 1993 were not preceded by marked declines in hunting success. In the counties of Nordjylland and Århus hunter numbers have declined less markedly than in the counties of Fyn, Vejle, Sønderjylland and Ribe (see Table 1), even though the number of wintering eiders in the former counties have declined much more dramatically during the last 10-year period (see Pihl et al. 2001). Consequently, these analyses on both regional and national scales did not show any indications that bag size and hunter numbers were significantly affected by the size of the winter population of eiders.

The influence of the annual number of hunters, annual number of offshore hunting days and of the annual monthly ratios of juveniles to adult females in the hunted population on the annual size of the eider bag was assessed by stepwise multiple regression. This analysis showed that the number of hunters reporting to have shot eiders was the most important factor explaining 71.3% of the variation in bag size (see Model 1 in Table 2). The juvenile:adult female ratio in October significantly explained 10.6% of the variation. Thus more than 80% of the variation in the bag size was explained by these factors. The number of hunting days explained 4.2%, but did not significantly affect bag size. There was no significant contribution from the annual average juvenile:female ratios when included in the model instead of monthly ratios (see Model 2 in Table 2).

Table 2.

Results of stepwise multiple regression of the variability in annual bag size during 1983/84–1999/2000. Model 1 includes the juvenile ratio in separate months, and model 2 includes the annual average juvenile ratio.

No significant correlation existed between the juvenile ratio and the number of offshore hunting days in October for the period 1983/84–1999/2000 (r² = 0.07, t₁₆ = 1.09, P > 0.05). Thus the significant effect of the juvenile ratio on the bag size was considered to reflect variation in the reproductive success and/or juvenile survival in a given year and not differences in the number of hunting days during the period of high juvenile hunting kills. During the period 1983/84–1999/2000 there was no significant trend in the annual number of hunting days (76.4 ± 3.2 (SE); r² = 0.15, t₁₆ = −1.67, P > 0.05), nor in the number of hunting days in separate months (ANOVA: F = 2.27, df = 84, P > 0.05), averaging 15.3 ± 0.78 (SE) days per month. Although a few years with severe ice condition (i.e. 1981/82, 1984/85–1986/87 and 1995/96) had a lower number of hunting days, this result suggests that annual variation in hunting activity caused by weather generally is negligible.

Hunter numbers and activity

The strong explanatory power of hunter numbers on the size of the eider bag puts emphasis on the performance and activity of hunters in Denmark. Although the number of eider hunters has declined during the periods 1983/84–1985/86 and 1993/94–1999/2000, very marked declines were found between 1985/86 and 1986/87 and between 1992/93 and 1993/94 (see Fig. 2). The low number of hunters reporting having shot eiders in 1993/94 coincide with a ban against lead shot for sea duck hunting and a national ban on the sale of several game species including the eider. The lead shot ban which was implemented to encourage use of steel shot was probably the main cause for the decrease in the number of eider hunters since this matter was, and still is, a very sensitive and widely discussed subject among hunters and environmentalists in Denmark.

The reason for the gradual decline in the number of eider hunters after 1983/84 is unknown, but may relate to several factors. The three successive ice-winters, 1984/85–1986/87, may have lead to a reduction in hunting opportunities affecting the number of active hunters. Such an explanation seems, however, unlikely, since both Noer et al. (1995) and my study showed a decrease in hunter numbers in the counties of København, Roskilde and Storstrøm, where most eiders are bagged during the autumn (Noer et al. 1995), and thus are not susceptible to winter conditions. Likewise, no reduction in hunter numbers and bag size occurred after the severe winter of 1995/96 (see Fig. 2). A reduction in reporting frequency occurring from the season 1986/87 has been corrected for (see the section Methods), and thus should not result in lower numbers of hunters reporting to have shot eiders. However, an extensive public debate during the mid-1980s leading to prohibition of motorboat hunting in most Danish fjords in 1987, may have resulted in a marked decline in the number of active eider hunters during this period.

Eider hunting opportunities in the offshore environment are reduced during periods of strong wind and ice formation. Undoubtedly, wind speeds of > 5 m/second negatively affect the predominant motorboat hunting activity, but no correlation was found in the present data between the annual number of days with wind speeds of < 5 m/second and the annual number of days with > 4 eider wings (r = 0.27, t₁₁ = −0.90, P > 0.05) or compared to all days with reported eider wings (r = 0.23, t₁₁ = −0.77, P > 0.05), suggesting that weather had little influence on eider hunting activity. This may relate to alternative hunting methods, such as hunting from punts with decoys, which in certain parts of Denmark is practised at wind speeds of 5–10 m/second.

The significant decline in the number of eider hunters in all but two counties strongly suggests a general decline in eider hunting throughout Denmark since 1980. However, while the number of eiders retrieved per hunter was generally stable, hunter numbers dropped markedly two times during 1980–1999, with no known reports of dramatic population declines. Thus, it seems unlikely that declines in the number of hunters should be the result of sudden changes in the numbers of wintering eiders. Likewise, the similar development in hunter numbers among counties do not suggest that the winter distribution of eiders have changed markedly within Danish waters during the study period. This result contrasts somewhat with those of Noer et al. (1995), who found that changes in bag size between the 1970s and 1980s were mostly affected by changes in phenology of migrating and wintering eiders in Denmark, with increasing numbers taken in south and southwest Denmark.

Hunted populations

Recoveries of ringed eiders wintering in Denmark indicate that the majority of eiders bagged in Danish waters originate from the Baltic and Danish breeding populations (Noer 1991; NERI, unpubl. data). The seasonal distribution of these recoveries shows that Danish eiders are bagged mainly during October-November, while the frequency of Baltic eiders increases throughout the hunting season. Since the Baltic populations represent approximately 400,000 breeding pairs, while Danish populations account for ca 20,000–25,000 pairs (Noer et al. 1995, Lyngs 2000; NERI, unpubl. data.), the majority of Danish birds are bagged at the start of the hunting season, while the proportion of Baltic birds increases throughout the season.

Given the difference in both population size and in the temporal occurrence between Baltic and Danish eiders, the significant relationship between the high number of juvenile birds bagged per adult female in October and the annual bag size suggests that variation in reproductive success of Danish breeding colonies has a small, but significant effect on the annual bag size. However, since the highly variable annual reproductive output in eider populations (Mendenhall & Milne 1985, Swennen 1989, 1991a) will also have an effect on overall population size, the lack of a significant effect of the annual juvenile ratio, i.e. embracing both the Danish and Baltic populations, on the total bag size, is taken to suggest that overall population size has no marked effect on total bag size.

The significant contribution of the juvenile ratio in October to the total eider bag size most likely relates to a high proportional kill of juveniles in Danish waters during this month, as found by Noer et al. (1995). The large number of juvenile eiders bagged during the start of the hunting season, is probably also influenced by selective hunting close to breeding colonies where the majority of juvenile eiders are bagged at this time (NERI, unpubl. data on ringing recoveries) in combination with juvenile birds generally being more vulnerable to hunting than adults birds (see Krementz et al. 1987, Nichols et al. 1990, Caithness et al. 1991). Thus, despite the relatively small number of juvenile eiders compared to total population size (incl. adults and non-breeding immature birds; see Swennen 1991b), these relationships are believed to explain the observed significant contribution of juveniles in October to total annual bag size.