Introduction

Like other waterfowl (Anatidae), the female common eider (Somateria mollissima; hereafter eider), forms a brood patch prior to incubation (Afton and Paulus 1992). In waterfowl, brood patch skin is modified to enhance heat transfer from the parent to eggs, partly by moulting off down from the breast and belly (Jónsson et al. 2006a, b). The down that is shed from the brood patch area is salvaged by the female and used to line the inside of the nest. Nest down helps to maintain desirable heat and humidity for the eggs during the incubation period and also conceals the clutch while the nest is unattended (Afton and Paulus 1992; Hilton et al. 2004). However, a question is whether the salvaged brood patch down may not be a requirement for successful hatch, which ultimately depends on the nest attendance and individual properties of the incubating parent (Flint and Grand 1999; Bolduc et al. 2005; Jónsson et al. 2007).

Over the centuries, Icelandic farmers have collected down from eider nests during the incubation period and replaced with dry hay as insulating material (Snæbjörnsson 1998; Bédard et al. 2008). Down is considered the best insulation material for bird nests. Conversely, hay is considered the worst but the insulation abilities reverse with humidity i.e. wet down is considered a worse insulator than wet hay (Hilton et al. 2004). Nowadays eiderdown is collected, cleaned and used to make products such as duvets, sleeping bags and clothes (Bédard et al. 2008). In Iceland, the down is usually collected late in the incubation period rather than early (Snæbjörnsson 1998). Should down collection have a negative effect on incubation and hatching success of the eiders, differences will be observed in incubation behaviour and/or associated temperature characteristics between eiders nesting in down and hay-lined nests.

Different incubation temperatures have been observed in eider nests, depending on the location of the nest. In a comparative study in southwest Iceland, the fluctuations in incubation temperature were much higher in exposed areas than in sheltered ones (D’Alba et al. 2010). Similarly, incubation temperatures in a study in North Canada were lower at lower ambient temperatures, hardly never exceeding 38°C (Fast et al. 2007).

Female common eiders incubate without help from the males and thus rarely leave their nests during the incubation period, attending the nests for 90–95% of the time (Bolduc and Guillemette 2003a). Unattended nests are in risk of predation, and it costs the bird extra energy to reheat the eggs upon return from incubation recesses (Brummermann and Reinertsen 1991). Incubation recesses are more common during early incubation period compared to late incubation but even then the females attend the nests for 60–70% of the total incubation time (Bolduc and Guillemette 2003a). Eider females leave their nests to drink fresh water (Criscuolo et al. 2000) but have rarely been reported to leave the nests to feed (Swennen et al. 1993). Recess duration can vary greatly but studies have reported average recess duration of 4–17 min (Mehlum 1991; Criscuolo et al. 2000; Bottita et al. 2003; Bolduc and Guillemette 2003a).

While incubating the females rotate their eggs regularly, up to 20–24 times per day (Gerasimova and Baranova 1960). This behaviour is considered to maintain the same temperature at all sides of the eggs (Rahn et al. 1983). The side of the egg that is close to the brood patch is probably warmer than the side facing to the bottom of the nest. If hay is not as good a nesting material as down, one might expect that a female incubating on hay would have to turn her eggs more often than a female incubating on down.

Hatching success can vary greatly among eiders but it is usually connected to predation risk at the nesting site (Madsen et al. 1992; Jónsson 2001; Bolduc and Guillemette 2003b; Donehower and Bird 2008), breeding experience, body condition, timing of breeding and health status of the female (Bolduc et al. 2005; D’Alba et al. 2010) and environmental variables such as ambient temperature, shelter and human disturbance. In an investigation in Alaska, the hatching success ranged from 5 to 68% of total nests (Noel et al. 2005), in Scotland it was 63% (Milne 1972) and in west Iceland 70% of all laid eggs (Skarphédinsson 1996).

The aim of this study was to investigate whether down collection (replacing down by hay) affected the incubation temperature in the nests and the behaviour (incubation constancy, recess frequency, turning of the eggs) of the female eider. We used an experimental approach in which we removed down from half the nests and replaced it with hay (hay nests), whereas we left the down in the other half (control nests). We then compared incubation temperatures, incubation constancy, recess frequency and egg turning between hay nests and control nests.

Materials and methods

Study site

This study was conducted from 3 June to 9 July 2006, at Hvallátur islands located in Breidafjordur, west Iceland (65°25′79N, 22°46′05W) (Fig. 1). The islands, where the investigation was carried out, are small and the distance from an eider nest to the intertidal zone never exceeds 20 m. The predation risk for eiders is considered low, and only avian predators have been seen around the islands, which are too far away from the mainland (17 km) for arctic fox (Vulpes alopex) or American mink (Neovison vison) to reach there (Björnsson and Hersteinsson 1991).

Fig. 1
figure 1

A map of Breidafjordur west Iceland. The investigation area (Hvallátur islands) is inside the circle

Full size image

The mean ambient temperature and standard deviation (SD) during the study period was 8.9°C ± 2.3°C (min–max;−0.7–17°C), and the mean wind speed was 5.3 m/s ± 2.3 m/s (Icelandic Meteorological Office). In Breidafjordur, there are over 3,000 small islands with abundant food and good nesting sites for the eiders. Breidafjordur is a breeding, moulting and wintering area for the eiders and supports 20–25% of the large Icelandic population (Grimmett and Jones 1989; Jónsson et al. 2009), which is estimated as about 850,000 birds (Gardarsson 2009).

Data collection

In the present study, 28 temperature data loggers (DST centi 2010) were used to monitor temperature in nests and incubation behaviour of the female eiders and hatching success. The loggers (length = 5 cm, width = 1 cm, weight 15 g) were placed in white film boxes of about the same size as the eider eggs. The heat transfer from the eider female to data loggers in boxes is considered complete because data loggers without boxes showed the same results as those in boxes (Kristjánsson, unpublished data). The data loggers measured the nest temperature every 30 s, starting when they were put into the nests and stopped when removed following hatching or nest abandonment by the female. The clutch size in the nests ranged from 3 to 5 eggs, and incubation had lasted from 0 to 20 days in the nests when the study started.

When the investigation started, the incubation stage of the 28 nests ranged from 0 to 20 days. However, 4 females abandoned their clutch after the loggers had been put into the nests (incubation stage 0–17 days), leaving our sample size at 24 nests with data loggers. These 4 females were not included in the present results as the reason for nest abandonment is not related to the nesting material, but rather to the disturbance when inserting the data logger. In every other nest, all the down were replaced with hay (hereafter hay nests, n = 12), whereas the other half was not manipulated but used as controls (hereafter down nests, n = 12).

Incubation curve

Incubation stage was estimated both by (1) calculating the specific density of the eggs and using it to calculate the incubation day from an incubation curve and (2) by counting from the day eggs hatched backwards until the day the data logger was put into the nest. The average incubation period was 24 days both in hay nest and in down nest depending on the incubation curve (see Appendix). To estimate the laying dates of eggs, an incubation curve was calculated using a technique, which states that egg density (g/cm3) changes during incubation (Hoyt 1979; Furness and Furness 1981). The volume of the eggs was calculated by the rule that the shape of the eggs is the same (Furness and Furness 1981). It is therefore enough to know the length and width of the eggs to calculate the density. With this technique both laying and hatching dates of unknown eggs can be calculated with 95% accuracy (Hario 1983).

Interpretation of temperature data

The relationships between observed temperature changes, incubation recesses and rotation of eggs were confirmed by using a video camera (Pelco 35 mm) filming a single eider nest (n = 1) containing a temperature data logger, from 2 to 5 June 2007 (72 h). The insulation material for one and a half days was down (from 2 June at 12:00 to 3 June at 21:45) replaced by hay (from 3 June at 21:45 to 5 June at 12:00). The incubation stage was 15 days when filming started. The temperature fluctuations read from the data loggers were compared to the actual behaviour of the eider observed on the videotape. This comparison confirmed our interpretation of the temperature data with respect to timing, duration and frequency of incubation recesses. The eiders treated the data loggers as they were eggs, turning them through the whole incubation period.

The behaviour of the females was recorded based on the temperature data from the data loggers. A sudden temperature decrease and the duration of this decrease shows when, and for how long time, an incubation recess takes place. Rising temperature, followed by a steady temperature reading of 33–34°C, indicates when the female has sat on the nest again. Conversely, small temperature fluctuations indicated the rotation of the eggs as the females rotated the data logger as well as her eggs. We considered that a female took a recess when the temperature dropped more than 3°C and the slope was continuous and steep. At the study site, sun exposure was not considered a critical factor concerning nest temperature. The mean ambient temperature was only 8.9 ± 2.3°C during the incubation period, and data from abandoned nests showed that temperatures never exceeded 18°C. A rotation of the egg was considered when the temperature drop in the nest was 0.5–3°C and lasted for a shorter period (see results).

Data analysis

SeaStar model (Star-Oddi 2001) was used to read the temperature data from the DST data loggers. Our statistical tests were one-tailed, looking for a negative effect, because we expected down collection to lower incubation temperature and increase rates of temperature loss during incubation recesses. We only report nonparametric tests if assumptions of their parametric counterparts were not met.

Statistical comparisons were made using a Welch two sample t test when the data passed a normality test (temperature comparison for average and highest values, incubation stage comparison), otherwise a Mann–Whitney U-test when normality failed (frequency of incubation recesses and their duration) (Sokal and Rolf 2001). A Chi-square test (Sokal and Rolf 2001) was used for comparison of incubation recesses night vs. day (to examine for diurnal pattern) and for absolute hatching success. All these analyses were done using the SigmaStat 3.1 program, except for a linear mixed effect model (PROC MIXED, Littell et al. 1996, SAS Institute 2001), which was used for comparison of nest cooling rate in down and hay nests. We calculated the nest cooling rate as the temperature change that occurred during a recess divided by the length of that recess (min). The square root was considered to normalize the data when looking at relation between incubation days and the nest cooling rate. The nest cooling rate was used as a dependent variable, treatment (down vs. hay) and incubation day were fixed effects, and individual nests (which equal individual females) were a random effect.

The absolute hatching success in the two nest types (down and hay nests) was calculated by comparing the number of nests and eggs where hatching succeeded completely or partly (one or more egg lost) to the number of unsuccessful nests (abandoned). The relative hatching success was also calculated for relative proportion of hatched eggs in each nest type with a GLM model (Crawley 2007 (α < 0.05)), in which treatment (down or hay) and incubation stage (as a covariate) were the independent variables. We used these two methods to calculate hatching success, and both methods yielded very similar findings.

Results

Nest temperatures

In the incubation period, mean temperatures were similar in down (31.5°C, SE = 0.5) and hay nests (30.7°C, SE = 1.0) (t test; t = 0.72, P = 0.242, df = 14), where hatching succeeded, despite considerable temperature drops when the females left their nests (Figs. 2 and 3). The lowest mean temperature observed in a hay nest was 25.5°C, caused by one long recess, when the temperature went down to 9.8°C in a short period (Fig. 3).

Fig. 2
figure 2

Mean temperature, 95% confidence interval and extreme values in down nests where hatching succeeded for common eider (Somateria mollissima) in Hvallátur Islands summer 2006

Full size image
Fig. 3
figure 3

Mean temperature and 95% confidence interval and extreme values in hay nests where hatching succeeded for common eider (Somateria mollissima) in Hvallátur Islands summer 2006

Full size image

The mean highest temperatures (down: 38.2°C, SE = 0.7, hay: 37.4°C, SE = 0.6) and lowest temperatures (down: 16.8°C, SE = 1.4, hay: 14.2°C, SE = 1.2) were calculated. The mean of the highest temperature observations (t test; t = 0.81, P = 0.215, df = 14) and the lowest temperature observations (t test; t = 1.44, P = 0.086, df = 14) in the two nest types was similar.

When the females left their nests, the temperature decreased steadily until they returned to their nests. The mixed model indicated that nest cooling rate was not related to treatment (F = 1.601,14; P = 0.227) or to incubation day (2.491,207; P = 0.116). Average nest cooling rates were 0.34°C/min (SE = 0.06) and 0.44°C/min (SE = 0.06) in down and hay nests, respectively.

Recesses frequency

The average recess frequency and standard error (SE) per 24 h was similar in successful down nests (1.3 times, SE = 0.1) and successful hay nests (1.4 times, SE = 0.1) (Mann–Whitney; t = 7084.00, n = 80, n = 97, P = 0.459. The females left their nests 0–4 times per incubation day, both in down nest and in hay nest (Table 1). Similarly, the average recess frequency in unsuccessful down (1.8 times, SE = 0.2) and hay (1.7 times, SE = 0.1) nests was similar during this period (Table 1) (Mann–Whitney; t = 1297.00, n = 26, n = 66, P = 0.224).

Table 1 Number of incubation recesses per 24 h, the average frequency ± SE, range of recess duration and average duration ± SE of eiders nesting in successful and unsuccessful down and hay nests
Full size table

However, the average frequency of incubation recesses in all successful nests (down and hay) (1.4 times, SE = 0.1) was lower than the frequency in all unsuccessful nests (1.7 times, SE = 0.1) (Mann–Whitney; t = 14173.00, n = 92, n = 177, P = 0.002). Comparison between successful and unsuccessful nests within both hay (Mann–Whitney; t = 6021.50, n = 66, n = 97, P = 0.195) and down (Mann–Whitney; t = 1678.50 n = 26, n = 80, P = 0.018) showed a difference between the two groups.

The frequency of the incubation recess at day (08–20′ clock) and night (20–08′ clock) in nests where hatching succeeded was compared to investigate whether or not the eider showed a diurnal pattern. No significant difference was observed in frequency in day and night time (X 2 = 0.16, df = 1, P = 0.69).

Recess duration

Recess duration and standard error (SE) were similar in successful down and hay nests (Mann–Whitney; t = 17758.00, n = 130, n = 147, P = 0.320), (Table 1). The female left the nest from 10 min to 2 h with an average for 45 min (SE = 2.1) and 47 min (SE = 2.3) in down and hay nests, respectively (Table 1). The average incubation constancy was 97% in down nests and 96% in hay nests.

In the eight unsuccessful nests, recess duration was always longer than the longest duration in successful nests. In these nests, the maximum recess duration was from 100 to 1,440 min and these long recesses were in the middle of the laying period (investigation period). Therefore, the last absence was not counted as a recess (Table 1). In seven of the nests, the female left dead eggs, but in one nest the eggs were eaten by predators.

Egg rotation

Egg rotation was similar in successful down and hay nests. On average, a female in a down nest rotated her eggs 13.9 times per 24 h, (SE = 0.5), whereas a female incubating in hay 15.3 times (SE = 0.5) (t test; t = −1.56, df = 14, P = 0.07). The average egg rotation did differ between unsuccessful down and hay nests (Mann–Whitney; t = 888.00, n = 26, n = 61, P = 0.009) and was on average, 11.2 (SE = 1.2) in down but 14.4 (SE = 0.6) times in hay per 24 h, respectively.

Hatching success and incubation state of the females

In successful nests, incubation stages were similar in down (7–20 days) and hay nests (7–16 days) when the investigation started (t test; t = 0.07, df = 14, P = 0.473). No difference was observed between hatching success in down (0.60) and hay (0.67) when relative hatching success was investigated using a GLM model (P values < 0.462) nor when absolute hatching success was calculated, down (0.55) and hay (0.67) (Table 2).

Table 2 Hatching success (number of nests and number of eggs hatched) in down and hay nests in Hvallátur, Breidifjordur in 2006
Full size table

Incubation stage was similar in the unsuccessful down (0–16 days, mean = 5, SE = 1.8) and hay nests (1–9 days, mean = 9, SE = 3.4) when the investigation started (t test: t = 0.91, df = 6, P = 0.199). However, the incubation stage when data loggers were placed differed between successful (n = 18, mean = 12, SE = 1.3) and unsuccessful nests (n = 8, mean = 7.5, SE = 1.9) excluding difference in nesting material. The unsuccessful females had incubated for shorter period than the successful females when the investigation started (t test: t = 2.53, df = 22, P = 0.01).

Video filming

Temperature fluctuations in the eider nest, which was filmed for 72 h (2–5 June), are shown in Fig. 4. The incubation recesses seen on the video and the temperature drops from the data logger in the nest matched very well. The rotation of the eggs was also observed from the video as movements of the female on the nests and matched the smaller temperature fluctuations from the data loggers as well. In the first day of observation (2 June), the female left the nest for 45 min and the temperature drop was 5.8°C. The other eight observed fluctuations in these first 12 h (0.5–3°C) were caused by egg rotation. On June 3, the female left the nest 4 times, the temperature decrease was from 3.2 to 11.1°C and the duration of the recesses was 14–70 min. The longest duration and the largest temperature decrease occured when the down was replaced by hay. Twelve temperature fluctuations that were observed these 24 h were caused by rotation of the eggs. On June 4, there was only one recess, lasting 20 min and the temperature dropped 5.5°C. In these 24 h, the female rotated the eggs 20 times (0.8–4°C). From midnight until 12:00, the last day (5 June), the female did not leave the nest and turned the eggs only once.

Fig. 4
figure 4

Temperature fluctuations in an eider (Somateria mollissima) nest, which was video recorded in Hvallátur Islands summer 2006. The dots point where the female left the nest for recessing and the arrow when down was replaced by hay

Full size image

Discussion

Nest temperatures

In this study, average nest temperature during incubation was similar in successful nests with down (31.5°C) and hay (30.7°C). Likewise, the mean highest and lowest temperature recorded in the two nest types was similar. Average nest temperatures were similar to those reported from southwest Iceland, where the average temperature was 31.4°C in natural nests but 32.4°C where man-made shelters surrounded nesting sites (D’Alba et al. 2010). The average central egg temperature among eiders over the incubation period has been estimated as 33.6°C (Rahn et al. 1983). The observation that females can maintain similar average temperature in a hay nest, compared to a down nest, implies that they can compensate for poor insulating material with their body heat. The down is thus not a critical factor for incubation when the female is on the nest. Rather, the incubation constancy along with heat transfer from the female to the eggs is probably more important than the nest material. A future research question is whether females must expend extra energy to maintain nest temperatures in a hay nest. Such extra heat loss through the brood patch potentially can induce catabolism of energy reserves (Jónsson et al. 2006a).

Heat at the brood patch region of the eiders has been documented from 39.1°C up to 40°C (Rahn et al. 1983; Criscuolo et al. 2001). In this study, brood patch temperature was not measured but the average highest temperatures both in hay and in down nests were similar to brood patch temperatures reported by Rahn (1983). In the present study, the lowest temperature observed in successful nests was only 9.8°C but the highest was 41.8°C. These extreme values lasted for a short time of the total incubation period, or in 42 min the nest temperature was <15°C and for 20 min it was >40°C. In chickens (Gallus domesticus), embryos do not develop at egg temperature below 26°C (physiological zero temperature) or above 40.5°C (upper lethal temperature), these critical values are believed to apply to all bird species (White and Kinney 1974; Webb 1987). Our results suggest that the embryos were resistant to short-term temperature fluctuations. When the females in this study left their nests, temperatures decreased linearly but the nest cooling rates did not differ significantly between down and hay nests. However, the nest cooling rate was 29.4% faster in hay nests than in down nests. Thus, a higher sample size could have detected a significant difference between hay and down nests, although the variation within groups ranged 0.53 in down (min; 0.00, max; 0.53, median; 0.16, 25%; 0.10, 75%; 0.22) and 0.89 (min; 0.01, max; 0.90, median; 0.19, 25%; 0.13, 75%; 0.28) in hay nests.

Recess frequency

Replacing down by hay did not affect recess frequency of the female eiders in the present study. In 16 nests where hatching succeeded the females left the nests 0–4 times per 24 h (average 1.4 times). The 8 eiders, which did not hatch their clutch, took incubation recesses more frequently than those with successful nests (average 1.7 times). In these nests, the temperature fluctuations were higher and average recess duration was much longer. In a study conducted in Canada, the eiders only left their nests on average 0.5 times per 24 h and the high predation risk of the nesting site was considered the main reason for this low recess frequency (Bottita et al. 2003).

In waterfowl incubation, constancy is generally high and eiders are reported to have one of the highest average incubation constancies among ducks (Afton and Paulus 1992, Bolduc and Guillemette 2003a). The eider is the largest duck in the Northern Hemisphere; larger species generally have higher incubation constancies than smaller ones because they have higher fasting endurances than smaller species (Jónsson et al. 2006b; Jónsson et al. 2007). The average incubation constancy in the present study was 97% in down nests and 96% in hay nests. Bolduc and Guillemette (2003a) recorded 99.5% incubation constancy for eiders nesting in Denmark and related the relatively high incubation constancy to high predation rates from avian predators (Bolduc and Guillemette 2003a).

In the present study, there was no difference observed between the frequencies of incubation recesses between day and night hours. The incubation period of eiders in Iceland is May–June, when there is light for 24 h a day. The main predators in the present study, the gulls (Larus spp.), are active for 24 h a day during that time of the year. It has been demonstrated in Canada (Bolduc and Guillemette 2003a) and the Netherlands (Swennen et al. 1993) that eiders leave their nests more often during nights (during dark hours) than days to avoid predation from avian predators. However, in Alaska, most recesses of spectacled eiders (S. fischeri) were observed between 10:00 and 22:00 when predators were most active but the ambient temperature was warmer than that during night time (Flint and Grand 1999).

Recess duration

In the 16 successful nests, recess duration was similar in down and hay nests, females recessed on average 45 min, each recess ranging from 10 to 120 min. In a study conducted in Svalbard, the eiders left their nest for only 4–7 min each time (Mehlum 1991; Criscuolo et al. 2000), in Denmark for an average of 14 min (Bolduc and Guillemette 2003a) and in Canada for 17 min (Bottita et al. 2003). The longer recess durations observed in our study may be explained by different ambient temperature and/or different predation risk at the nesting sites. In Svalbard, snow covered the nesting grounds for half of the incubation period and the ambient temperature was most likely lower than in the present study. Eiders probably take shorter recesses in cooler atmosphere due to the risk of egg cooling and a high energy expenditure caused by reheating the eggs to optimal temperature (Mehlum 1991; Criscuolo et al. 2000). Birds incubating in northern hemisphere spend up to 50% more energy in the incubation period compared to individuals of the same species incubating at a lower latitude (Piersma et al. 2003; Eichhorn et al. 2010).

In our study, females left their nest for a longer period in each session as incubation progressed. Most eiders use their recess to drink fresh water and preen (Criscuolo et al. 2000) but the long recess observed here could indicate other activities as well. Eiders have been seen feeding during incubation (Swennen et al. 1993; Flint and Grand 1999) and as food is abundant at the study site and the nests always close to the sea; it is possible that some of the recess time was used to feed. The absence of mammalian predation and protective efforts against avian predators at our colony may explain to some extent longer incubation recesses giving the female an opportunity to feed or loaf. In our study, the eiders covered their nest when they left on an incubation recess, which is a widespread behaviour among waterfowl (Afton and Paulus 1992; Hilton et al. 2004; D’Alba et al. 2010).

We present our findings based on successful nests. However, in all the eight failed eider nests, one or more recess was longer than the longest recess observed in successful nests, causing larger temperature fluctuations. Such long recesses (up to 24 h) in failed nests occurred halfway through the incubation period. Unsuccessful females returned from these long recesses and incubated for some days, probably on dead eggs, before abandoning their nests. High temperature fluctuations in the nest while the eider takes incubation recess may lengthen the incubation period, even delay development and reducing probability survival of the embryo (Olsen et al. 2006). Long recess duration is considered to increase the risk of nest failure more than high recess frequency (Rauter and Reyer 1997).

The unsuccessful eiders abandoned their nest early in the incubation period. Inserting of the data logger could have had a stronger impact on the eider during early incubation (Bolduc and Guillemette 2003b), or this was an artefact of poorer quality females (younger and lighter) initiating nests later than those of better quality. It has been reported that fitness of the eider is strongly related to egg-laying dates (Spurr and Milne 1976; Bolduc et al. 2005), which points towards this conclusion.

Egg rotation

There was no difference observed in egg rotation between successful hay- and down-nesting females. Females rotated their eggs on an average of 14–15 times each 24 h in both nesting materials, even though the material it self might have different ability. It is considered that birds rotate their eggs in nests to maintain equal temperature and moisture at all sites of the egg and even out the temperature difference between the bottom of the egg, furthest from the brood patch, and the top, closest to the brood patch (Rahn et al. 1983). Another function for egg turning is to maximize utilization of the albumin by the embryo. Turning increases diffusion of protein and ions from the albumin to the yolk and prevents the embryo from adhering to the inner shell membrane (Deeming 2009). The egg rotation in this study was unrelated to nesting material indicating that the female could keep the same temperature at the bottom of the nest in both nest types.

Hatching success

Our findings indicate that replacing down by hay have no impact on relative hatching success in eiders nests as the success was similar in down (0.60) and hay nests (0.67) when the proportion of hatched eggs was investigated using a GLM model. However, these results point towards a better hatching success in hay than in down, which may indicate effects of a small sample size resulting in low statistical power.

Females with failed nests had incubated for shorter time and recessed more often and for a longer time than the females where hatching succeeded, which might indicate younger females as the older and heavier lay eggs earlier than young and light ones (D’Alba et al. 2010). Bolduc et al. (2005) found out that nesting success was principally related to female characteristics rather than to nest site characteristics, they suggested that eiders rely on nest attendance rather than on nest concealment to protect their nests. Nesting close to shore may shorten incubation recesses and improve hatchling survival when leaving the nest.

The methodology to evaluate the behaviour of nesting female eiders using temperature data loggers, as done in the present study, seems to work quite well. The data logger in the one nest, which was video recorded, showed that all behaviour of the eider matched the temperature data from the logger. Data loggers are useful in behaviour investigations where the ambient temperature is much lower than the average nest temperature, as in the present study. In this case, the nest attendance, incubation recess and its duration become obvious as temperature fluctuations. In a comparable study in the United States, the behaviour of ducks was monitored by using data loggers at the same time as video recorder and also here the results confirmed the temperature fluctuations well (Hoover et al. 2004).

Conclusion

In this study, we were looking for overall effects of down collection on productivity of the eider population by looking at some critical variables that might help understanding the mechanism driving any overall effects. It does not appear that down collection has any substantial effects on productivity, incubation temperature, recess frequency or recess duration, thus females can compensate for down loss while incubating on the nests. The strongest effect, however, occurred as increased rates of nest cooling during incubation recesses. This high cooling rate in hay-lined nests may increase embryonic mortality and reduce hatching success of eiders incubating in hay. However, because of our low sample size we may have failed to detect effects that actually exist. Nonetheless, even if we assumed that our effect sizes are real, the overall effects of down collection on population dynamics would be small and the practice of delaying the collection of down until late incubation would reduce the effects to less than what we estimated in this study.

Collection of eiderdown is believed to be non-harmful to the species (Snæbjörnsson 1998; Bédard et al. 2008), and down is considered the best insulation material for eiders but no study has been conducted on whether or not removal of the down and not replacing it at all has any impact on breeding success. A larger study sample in which eiders nesting in hay and down would be weighted at the beginning of the incubation and at hatching would be a good future study to compare nesting material and energy expenditure of the female. Likewise it would be informative to compare fitness of the eiders both early and late in the nesting period relating to breeding probability the following year.