Regular habitat switch as an important feeding strategy of an opportunistic seabird species at the interface between land and sea

Available online at www.sciencedirect.com

Estuarine, Coastal and Shelf Science 77 (2008) 12e22 www.elsevier.com/locate/ecss

Regular habitat switch as an important feeding strategy of an opportunistic seabird species at the interface between land and sea Philipp Schwemmer*, Stefan Garthe Research and Technology Center (FTZ), University of Kiel, Hafento¨rn 1, D-25761 Bu¨sum, Germany Received 3 May 2007; accepted 21 August 2007 Available online 20 September 2007

Abstract During deteriorated prey availability, purely pelagic, specialised seabird species have to alter their feeding strategy by extending foraging radii and/or time spent at sea or reducing feeding intervals of chicks. In contrast, more generalised species such as the opportunistic black-headed gull (Larus ridibundus) breeding at the German North Sea coast, can be assumed to react on prey shortages by switching foraging habitats. The coastal zone of the German North Sea provides a rich habitat mosaic consisting of the offshore zone, tidal flats and terrestrial habitats. Thus, we expected distinct temporal and spatial patterns of habitat switch in accordance with prey availability and environmental constraints. We carried out ship-based and aerial surveys as well as dietary analyses and observations on flight activity. We found a significant switch from terrestrial to marine feeding sites both on a daily basis (related to tidal cycle) and over the whole breeding season. Most likely, the latter switch is the result of lower prey availability in the terrestrial habitats and an increasing quality (in terms of prey abundance and energy intake) of the marine area. While there was only moderate variability in habitat use among different years, we revealed significant differences in the diet of birds from different colonies. The high dietary plasticity and flexible feeding strategy, switching between terrestrial and marine prey is certainly of major importance for the success of an opportunistic avian top predator in a complex coastal zone. It is suggested that e compared to situations elsewhere e the number of breeding pairs of black-headed gulls in the German North Sea coast are still stable due to the switch of foraging habitats performed by individuals in this region. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: habitat switch; foraging behaviour; tidal flats; abundance; diet; Larus ridibundus; Germany; North Sea; Wadden Sea; 8 Ee9 30 E; 55 Ne53 300 N

1. Introduction Coastal zones often provide highly important habitat structures for seabirds. In conditions of deteriorated prey availability, purely pelagic and specialised seabird species were found to alter their feeding strategy by extending foraging radii and/ or time spent at sea or reducing feeding intervals of chicks (e.g. Monaghan, 1996; Oro et al., 1997; Lewis et al., 2001). In contrast, more generalised species are able to shift their diet or to alter their foraging habitats (e.g. Spaans, 1971; Annett and Pierotti, 1989; Gonza´lez-Solı´s et al., 1997a). Opportunistic feeders might then use the whole habitat mosaic * Corresponding author. E-mail addresses: [email protected] (P. Schwemmer), [email protected] (S. Garthe). 0272-7714/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.ecss.2007.08.017

of the coastal zone for foraging including terrestrial habitats, tidal flats and the pelagic offshore zone. The black-headed gull (Larus ridibundus) in the German North Sea coast is one of the most generalist gull species. Former studies separately described either its utilisation of the marine habitat (e.g. Curtis et al., 1985; Gorke, 1990; Kubetzki and Garthe, 2003) or of the terrestrial nearshore area (Jones, 1985; Gloe, 2006). To our knowledge, Vernon (1970) and Mudge and Ferns (1982) present the only integrative approach to quantify habitat use of this species in the coastal zone of the UK as a whole. We consider such an integrative approach as highly useful to analyse the feeding ecology of an opportunistic avian predator. Switching between a variety of available foraging habitats might be an essential part of its feeding strategy. An integrative approach to analyse such a switch within different temporal and spatial scales has not been applied so far.

P. Schwemmer, S. Garthe / Estuarine, Coastal and Shelf Science 77 (2008) 12e22

We expected that the complex coastal habitat mosaic of the German North Sea coast (consisting of tidal flats, open-sea water areas as well as pastures and arable land) might provide excellent opportunities to switch feeding habitats according to food availability and environmental constraints. In detail, we derived the three following hypothesis: (1) Since the flight radius of black-headed gulls breeding at the North Sea coast was found to be comparatively low (i.e. ca. 5 km; Gorke, 1990), we assumed that colony location would reflect prey supply in the vicinity of the colony and that prey components should thus differ between different colonies, especially when these are located at different distances from the major habitat types (for detailed expected differences between colonies see chapter 2.3). (2) Superabundant prey should be used most intensively in order to reduce searching time, leading to a switch of foraging habitats according to prey availability and thus differences in gull nutrition in space and time (i.e. between colonies, seasons and years). (3) The hatching of chicks has been found to trigger diet switches (Annett and Pierotti, 1989) and additionally, energy demands of chicks increase with age (Spaans, 1971; Hartwig and Hu¨ppop, 1982). Thus, we expected dietary responses in adults leading to distinct differences in nutrition between incubation and chick-rearing period which might be again accompanied by a switch of foraging habitats. Black-headed gulls exhibited considerable increases in breeding pair numbers in the German North Sea coast since the second half of the 20th century (Garthe et al., 2000; Koffijberg et al., 2006). However, in some coastal parts, but particularly in the mainland, breeding pair numbers in Denmark, The Netherlands and Germany (here only mainland) showed a consistent decline since a few years (Stienen et al., 1998; Heldbjerg, 2001; Bellebaum, 2002; Koffijberg et al., 2006). By briefly reviewing historical studies on foraging ecology of black-headed gulls, possible implications of the quality of different coastal foraging habitats are discussed against the background of population trends of this species along the German coastal zone. 2. Methods 2.1. Study area The study was carried out at the North Sea coast and the adjacent mainland of Schleswig-Holstein, northern Germany (east of 8 ; west of 9 300 and south of 55 ; Fig. 1) during the years 2005e2006. The coastal mainland and most of the islands are characterised by a mix of intensive agriculture and pastoral farming. The coastal seas consist of tidal flats and shallow tidal streams belonging to the ‘‘National park Wadden Sea’’. Further offshore, the water depth increases slowly and tidal flats are followed by the offshore zone (Fig. 1). 2.2. Distribution over the annual cycle 2.2.1. Aerial surveys in the mainland Distribution of black-headed gulls in the coastal mainland was recorded using aerial surveys. Two flights were carried

13

out during the periods of incubation (1 Maye15 June) and chick-rearing (16 Junee31 July), covering a distance of 1145 km, respectively. Basically, we applied the common aerial census technique for recording seabirds at sea (Diederichs et al., 2002). Gulls were recorded within a counting transect of T ¼ 240 m width parallel to the route of the plane, flying at an altitude (Alt) of 150 m. The transect served to estimate an area sampled which allowed to calculate densities of gulls. Width of the transect was calculated with inclinometers measuring two angles against the horizon (25 ¼ distant boarder of the transect, 60 ¼ closer boarder of the transect; for further details see Diederichs et al., 2002): T ¼ Altðtanð90 e25 Þ  tanð90 e60 ÞÞz240 m

ð1Þ

Gull numbers and distance travelled (using GPS) as well as area surveyed (i.e. distance travelled  T ) were recorded at intervals of one second. Mean gull densities were computed for a grid of 1.5 latitude  2.5 longitude by dividing the total gull numbers within each grid cell by the total area surveyed within each grid cell. 2.2.2. Ship-based surveys Distribution of black-headed gulls sitting on or flying across open water (i.e. the offshore region and the tidal streams in the Wadden Sea area) were recorded for the same periods as in the mainland during the years 1996e2006 (German Seabirds at Sea database, version 5.07; October 2006). A total of 12,439 km were travelled during incubation, while 19,297 km were travelled during chick-rearing. We used ship-based surveys applying the transect method described by Tasker et al. (1984) and Garthe et al. (2002), following the same principle as stated for aerial surveys, despite that a transect of 300 m width was used and recording interval was 1 min (Garthe et al., 2002). All birds located on the water surface within the transect were counted. In order to avoid bias of flying individuals due to the comparably slow ship speed, we applied the ‘snapshot’ method; i.e. arbitrarily counting only those birds which flew into or across the transect at every full min to avoid counting any individual more than once (Tasker et al., 1984). Black-headed gulls foraging or resting on tidal flats were recorded during the years 2005e2006 (1.017 km travelled during incubation and 233 km during chick-rearing). Only parts of tidal areas that could be scanned easily from board of the vessel were observed (see hatched areas in Fig. 1). Since no transect method was applied, we did not correct sightings for observer effort and simply recorded individual numbers. Thus, distribution patterns of black-headed gulls on tidal flats should only be interpreted qualitatively. 2.3. Diet analysis A total of 648 pellets of black-headed gulls was collected in three different colonies (Fig. 1): (1) Friedrichskoog (FKoog; 8 490 12 E and 54 180 00 N; 1200 breeding pairs); (2) Hamburger Hallig (HHHallig; 8 480 36 E and 54 360 36 N; 600 breeding pairs); (3) Amrum (8 280 48 E and 54 400 48 N;

14

P. Schwemmer, S. Garthe / Estuarine, Coastal and Shelf Science 77 (2008) 12e22

Fig. 1. Location and details of the study area, with the location of the three black-headed gull colonies sampled for pellets (triangle: colony of Friedrichskoog, 1200 breeding pairs; square: colony of Hamburger Hallig, 600 breeding pairs; diamond: colony of Amrum, 80 breeding pairs; source: Ha¨lterlein unpubl. data; hatched area: tidal flats scanned during ship surveys).

80 breeding pairs). According to colony location and the principle of energy maximization (i.e. choosing foraging habitats close to the colony; Brandl and Gorke, 1988), compared to the other two colonies, the first colony was expected to consist of black-headed gulls which forage more on the mainland, while the second colony should provide better foraging opportunities in the Wadden Sea. The third colony was assumed to allow for both, offshore and Wadden Sea based foraging, but hardly for terrestrial foraging as the mainland is at a fair distance. Colonies were sampled once during incubation (i.e. 1 Maye15 June) and once during the chick rearing period (i.e. 16 Junee31 July) of both years 2005 and 2006. The colony of Amrum could not be sampled during the chick-rearing period in 2006 due to flooding. Number of pellets collected varied between colonies, seasons and years depending on number of breeding pairs and weather conditions, as rain might destroy pellets (see Table 1 for number of samples in each colony, season and year). The analysis of pellets was conducted following the method described by Duffy and Jackson (1986): we determined the

frequency of occurrence of each detectable food item by assessing its presence or absence in each pellet. Analysis of pellets may cause bias towards hard bodied prey species (Gonza´lezSolı´s et al., 1997b). However, in contrast to more quantitative analyses such as total mass or total number per prey species, presence and absence avoids potential bias caused by differences in digestibility to a large degree (Duffy and Jackson, 1986). Furthermore, analyses of pellets provide the advantage to determine many prey components to species level, which is particularly useful to reveal habitat use of the predator. In order to detect also small food remains (e.g. setae of polychaetes and oligochaetes), each pellet was scanned intensively using a 8  23 binocular. Food items found were determined to the lowest possible taxon. Only fresh pellets were collected. Remains of invertebrates and terrestrial mammals were determined following Schaefer (1994), while fishes were identified by otoliths, bones and scales following Harko¨nen (1986) and Leopold et al. (2001). For statistical analysis (see chapter 2.4), we grouped prey items into 11 categories according to foraging habitat of gulls and taxonomic considerations (see Table 1).

P. Schwemmer, S. Garthe / Estuarine, Coastal and Shelf Science 77 (2008) 12e22

15

Table 1 Frequency of occurrence [%] of 11 prey categories in 648 black-headed gull pellets from three colonies, two seasons and years (terr. arthrop. ¼ terrestrial arthropods) Friedrichskoog 2005

n pellets Insects/terr. arthrop. Earthworms Molluscs Polychaetes Crustaceans (tidal flats) Crustaceans (offshore) Fish Birds Mammals Plant material Waste

Hamburger Hallig 2006

2005

Amrum 2006

2005

2006

Incubation

Chick stage

Incubation

Chick stage

Incubation

Chick stage

Incubation

Chick stage

Incubation

Chick stage

Incubation

63 57 63 17 22 30 3 2 5 10 62 2

79 14 14 10 28 66 4 5 5 3 19 3

61 67 48 30 13 34 5 5 16 2 52 e

21 10 5 24 14 52 5 10 14 e 5 5

70 87 93 21 3 4 e 6 3 9 97 e

86 72 64 8 7 20 2 2 5 16 70 3

90 78 51 22 4 12 1 17 13 6 59 2

90 63 47 21 22 24 2 13 13 21 54 2

43 100 21 12 5 9 e 5 e 2 49 21

21 52 5 24 10 29 5 e 5 5 29 19

24 88 38 21 25 29 4 4 4 0 42 e

To document switch of feeding habitats, the food remains were classified as terrestrial (including freshwater) or marine according to their apparent origin. Classification was only doubtful in case of stickleback (Gasterosteus aculeatus) as this species occurs both in marine and freshwater. According to Vorberg and Breckling (1999) sticklebacks disappear nearly completely from the marine area during MayeJuly and migrate up small rivers towards the mainland. Thus, this species was classified as taken in terrestrial habitats. 2.4. Statistics on diet analysis We carried out Canonical Correspondence Analysis (CCA; Leps and Smilauer, 2003) on the basis of the 11 prey categories for each combination of colony, season and year to reveal spatial and temporal differences and possible habitat switch. Furthermore, we conducted Monte-Carlo permutation tests (Legendre and Legendre, 1998) using colony, season and year as predictor variables. We conducted 9999 unrestricted permutations with forward selection. Tests were performed on two different levels: (1) on the basis of all prey categories; and (2) with the 11 prey categories (see Table 1). The goal of this procedure was to test for differences in prey choice between colonies, seasons and years. To take into account rare food components we used the option ‘‘downweighting for rare species’’ (CANOCO 4.5). In order to focus on foraging habitat use, we determined whether the prey components in pellets originated from marine, terrestrial or both habitats (see chapter 2.3). We tested differences between colonies, years and seasons using a Linear Mixed Effect Model (LME) based on Restricted Maximum Likelihood (Venables and Ripley, 2002). We defined colony as fixed effect factor and season nested in year as random factor. To test for significant effects of the random factor we applied a Linear Model (LM) which was tested against the LME using ANOVA (Faraway, 2006). Permutation tests and CCA were performed with CANOCO 4.5, while all other tests were conducted with the open source

software package of R 2.4.1 (http://www.r-project.org) using the library MASS and NLME (Pinheiro and Bates, 2000; Venables and Ripley, 2002). All indicated p-values are twotailed. 2.5. Flight activity in relation to tide and time of day To assess short term habitat switch and the importance of the coastal mainland vs. the nearby marine area, flight activity patterns were observed in the colony of FKoog (see Fig. 1) on 6 d in early June and 4 d in mid July 2005. Individuals were counted during daylight hours when flying across two lines of 1 km length set arbitrarily along the colony edge. One line was set perpendicular to the mainland, the other line perpendicular to the marine area. Counts were conducted within 5 min intervals alternating both lines and covering different stages of the tidal cycle and time of day. Sixty-eight double 5 min counts were carried out during June and 42 during July. Total flight activity of birds heading towards/returning from each habitat was calculated and related to tide and time of day [UTC]. Flight activity patterns were analysed fitting a Generalised Linear Model (GLM) using backward selection (Venables and Ripley, 2002). For the total flight activity from/to the mainland and the marine area, respectively, the overall variance explained was determined by including tide, time of day and their interaction as independent variables. All GLMs were run under the software package R 2.4.1. 3. Results 3.1. Distribution On the mainland, densities were higher during incubation than during chick-rearing period, while numbers in the marine area changed complementarily (Fig. 2). On the mainland, distribution patterns during both periods were focused on the border to the marine area (particularly in the southern part of the

16

P. Schwemmer, S. Garthe / Estuarine, Coastal and Shelf Science 77 (2008) 12e22

Fig. 2. Distribution patterns of black-headed gulls along the western German North Sea coast and coastal mainland during: (a) incubation period; and (b) chickrearing period.

study area) indicating a fast access (to) and possible turnover with the marine area. During the chick-rearing period numbers in the marine area increased most dramatically in the southern part of the study area. 3.2. Diet A total of 117 different food components were found in black-headed gull pellets (Appendix I). Mean number of all food items per pellet was 3.64 (2.17), the maximum number found was 13. Pooling both years of observation, Kruskale Wallis tests showed that, in all three colonies, the number of food components per pellet was significantly higher

during incubation (inc) than during chick stage (cs) (FKoog: c2 ¼ 55.4; Ninc ¼ 124; Ncs ¼ 100; HHHallig: c2 ¼ 11.1; Ninc ¼ 160; Ncs ¼ 176; Amrum: c2 ¼ 21.1; Ninc ¼ 67; Ncs ¼ 21; df ¼ 1; P  0.001, respectively). This indicates a more diverse habitat use during the incubation period. Generally, in all colonies the most important terrestrial prey items were earthworms (Lumbricus spec.), followed by beetles (particularly Carabidae and Elateridae), while the most frequent marine prey were North Sea shrimps (Crangon crangon) and shore crabs (Carcinus maenas) (Appendix I). Considering all food items (Appendix I), permutation tests revealed significant differences between colonies (F ¼ 8.96), seasons (F ¼ 11.21) and years (F ¼ 7.22; P < 0.001,

P. Schwemmer, S. Garthe / Estuarine, Coastal and Shelf Science 77 (2008) 12e22

respectively). This was mainly caused by the preference of certain prey items during different years or seasons within each colony, indicating the use of superabundant prey species: e.g. diplopoda in FKoog during incubation in 2006 as well as the pine weevil (Hylobius abietis) or leatherjackets (Tipula paludosa) in Amrum during incubation in 2005 (Appendix I). Frequency of occurrence of the 11 categories of food types in pellets showed significant differences between colonies (F ¼ 19.92; P < 0.001), seasons (F ¼ 23.51, P < 0.001) and years (F ¼ 4.71; P < 0.005; Table 1). Insects/terrestrial arthropods as well as earthworms and plant material showed the highest presence values in all colonies and years, whereas crustaceans from the tidal flats (and partly polychaetes and molluscs) were important mainly for the colony of FKoog and particularly during chick rearing. Mammals (most commonly mice; compare Appendix I) appeared most frequent in the colony of HHHallig (Table 1). As revealed by CCA, the 11 food components differed most intensively between the three colonies, moderately between the two seasons and least between the two years (as indicated by the location of the combinations of colonies, seasons and years in the ordination diagram, Fig. 3). The strongest inter-colony difference was found between FKoog and Amrum. 3.3. Origin of food and seasonal effects For most combinations of colonies, seasons and years food originating from terrestrial habitats dominated in pellets (despite of FKoog during chick rearing in both years; Table 2). Unlike our assumptions, the colony of FKoog showed the highest proportion of marine prey, while HHHallig and Amrum mainly showed food components from terrestrial habitats. There was a consistent seasonal switch in feeding habitats at all three colonies and in both years, with more gulls apparently

0.10

0.15

Prey category Earthworms

Colonies FKoog HHHallig Amrum

Plant material

0.05

Incub, Y1 Incub, Y1

Chick, Y1

Incub Y2

Incub,Y1

ChickY1

0.00 -0.05

Axis 2

Crustaceans (tidal flat)

Polychaetes

Incub Y2

Crust. (offsh.)

Waste Chick Y1

Chick, Y2

Molluscs

Incub, Y2

-0.10

Chick, Y2 Insects / terr. arthropods

Mammals

17

foraging in marine compared to terrestrial habitats during the chick-rearing vs. the incubation stage (Table 2). This switch was most profound for the colony of FKoog. Furthermore, the proportion of pellets consisting of both marine and terrestrial prey decreased during the chick rearing stage (despite of Amrum in 2005 and HHHallig in 2006), indicating a more exclusive use of marine foraging habitats and a lower ‘‘mixed’’ habitat use (Table 2). The LME revealed significant differences in origin of food between colonies (effect of intercept: 2.13; P < 0.001; effect of colony: 0.19; P < 0.001; variance of random factor: <0.001; variance of residual: 0.85). ANOVA comparing LME and LM detected a significant improvement of the model adding season nested in year as random factor. This indicates a significant effect of both parameters on foraging habitat use. 3.4. Flight activity For flight activity of gulls commuting to and from the marine area the parameters tide and time of day explained 80.3% of the variance of the data, for gulls commuting to and from the mainland they explained 81.7% with no significant interaction of tide and time of day, respectively (partitioned variances for tide and time of day are given in the legend of Fig. 4). The correlation between flight activity and tide was strongest (Fig. 4a,b; for curve fit values see figure legend). Patterns of gulls that used the marine area and those that used the mainland were nearly complementary: The former individuals revealed peak numbers shortly before and after low tide (Fig. 4a), whereas the latter showed lowest flight activity during low and highest values during high tide (Fig. 4b). The correlation between flight activity and time of day was weaker then for tide (Fig. 4c,d, for curve fit values see figure legend). Gulls using the marine area showed two peaks of flight activity, both in the morning and in the evening hours, while flight activity slightly decreased during mid-day and was lowest during sunset and sunrise (Fig. 4c). Gulls using the mainland showed nearly complementary patterns: Flight activity showed a slight peak around mid-day (Fig. 4d). Generally, total flight activity of gulls using the mainland was significantly lower compared to those using the marine area (ManneWhitney U-test: U ¼ 1841.5; Ncounts mainland ¼ 110; Ncounts marine area ¼ 110; P < 0.001). Total flight activity of individuals using the mainland differed during the two months of observation, with significantly more individuals using the mainland during June compared to July (ManneWhitney U-test: U ¼ 801; Ncounts June ¼ Ncounts July ¼ 68; P < 0.001), while total flight activity of individuals using the marine area increased slightly but not significantly during July. 4. Discussion

Birds Fish

-0.10

0.00

0.10

0.20

4.1. Habitat switching

Axis 1 Fig. 3. Canonical Correspondence Analysis of 11 prey categories in pellets sampled in different colonies, seasons and years (Incub ¼ incubation; Chick ¼ chick-rearing; Y1 ¼ year 2005; Y2 ¼ year 2006).

Our study clearly showed that habitat switches are an important feature in the use of the coastal zone by black-headed gulls. In general, we found two temporal levels of habitat

P. Schwemmer, S. Garthe / Estuarine, Coastal and Shelf Science 77 (2008) 12e22

18

Table 2 Proportion of pellets with food items originating from terrestrial, marine or both habitats 2005 Incubation

Fkoog HHHallig Amrum

2005 Chick stage

2006 Incubation

2006 Chick stage

Terrestrial

Marine

Terrestrial & marine

Terrestrial

Marine

Terrestrial & marine

Terrestrial

Marine

Terrestrial & marine

Terrestrial

Marine

Terrestrial & marine

38% 70% 81%

22% 0% 0%

40% 30% 19%

18% 71% 48%

70% 15% 29%

13% 14% 24%

36% 63% 33%

20% 4% 8%

44% 32% 58%

24% 46% e

71% 17% e

5% 38% e

switch: (1) A regular short term switch according to the tidal cycle (and to a minor extent also due to time of day) as revealed by flight activity patterns; and (2) a seasonal habitat switch as revealed by a stronger utilisation of marine vs. terrestrial feeding habitats: 4.1.1. Short-term habitat switch and role of the tidal cycle Regular daily switches of foraging habitats in relation to time of day and tide have been described before in herring gulls

b

200

Number of individuals

Number of individuals

a

(Larus argentatus) (e.g. Verbeek, 1977; Sibly and McCleery, 1983), a species highly associated with tidal flats. For blackheaded gulls (at least at the colony of FKoog) tide seems to be the predominant factor, as indicated by the strong correlation of flight activity with tide. In contrast, for other gull species such as lesser black-backed gull (Larus fuscus), a diurnal rhythm was found to be more important (Schwemmer and Garthe, 2005). It is not known if a tide-related habitat switch also took place in the other two colonies. However, the high

150

100

200

150

100

50

50

0

0 0

200

400

0

600

Minutes after high tide

400

600

Minutes after high tide

d

200

Number of individuals

Number of individuals

c

200

150

100

200

150

100

50

50

0

0 10

15

Time [h]

20

10

15

20

Time [h]

Fig. 4. Flight activity of gulls in relation to tide (a) commuting to/from the marine area; (b) commuting to/from the mainland and flight activity of gulls in relation to time of day [UTC]; (c) commuting to/from the marine area; and (d) commuting to/from the mainland. Bold line: Generalised linear model of flight activity. Dashed lines: 95% confidence interval of the model. Curve fit values: (a) logit y ¼ 0.02x  9.1  105 x2 þ 1.7  107x3  1.1  1010x4; R2 [ 0.59; (b) logit y ¼  5.5  105x2 þ 1.728  107x3  1.34  1010x4; R2 [ 0.6; (c) logit y ¼ 5.5x  0e6x2 þ 3.1  102x3  5.4  104x4; R [ 0.22; (d) logit y ¼  20x þ 3.15x2  0.23x3 þ 8.1  103x4  1.1  104x5, R2 [ 0.21.

P. Schwemmer, S. Garthe / Estuarine, Coastal and Shelf Science 77 (2008) 12e22

number of pellets with mixed food suggests a similar pattern as in Fkoog. 4.1.2. Seasonal habitat switching As the distribution patterns indicate, the use of marine and terrestrial foraging habitats apparently changed between the seasons (Fig. 2), indicating a habitat switch from the mainland to the marine area between the incubation and chick-rearing period. This switch was reflected by a higher proportion of marine food in all three sampled colonies (Table 2). A more exclusive use of the marine habitat is indicated by a reduced number of pellets containing mixed food (i.e. samples with both terrestrial and marine prey) and also by lower numbers of gulls heading towards the mainland compared to the marine area. As we used two different methods for revealing distribution patterns at land and sea (i.e. aerial and ship-based surveys), our data are not directly comparable and were consequently not tested statistically. However, the general distribution pattern revealed by both methods indicates a basic seasonal change in habitat use (Fig. 2). For seasonal changes in habitat use there appear to be two main reasons: (1) prey depletion and density-dependent effects causing competition (Ashmole, 1963; Gorke and Brandl, 1986; Lewis et al., 2001); and (2) changing quality of different foraging habitats as determined by abundance, availability and quality of prey species. The quality of prey might be determined by energetic or nutritional demands of growing chicks (Spaans, 1971; Hartwig and Hu¨ppop, 1982; Pierotti and Annett, 1990, 1991). We can assume that either one of these factors or a combination might explain the habitat switch from terrestrial to marine foraging habitats in the late breeding season. In contrast to inland colonies where black-headed gulls have to increase their foraging radius during the breeding season (Gorke and Brandl, 1986), black-headed gulls of the North Sea coast showed constant foraging radii (Gorke, 1990). Thus, prey depletion caused by the birds themselves should not be considered as a main effect. Switch of food as a consequence of dietary demands of hatched or growing chicks has been described before in gulls (Spaans, 1971; Annett and Pierotti, 1989). However, prey originating from marine habitats has not generally higher energetic values than terrestrial food. In our study, the strongest increase of abundance values in pellets was found in shore crabs (Table 1). This food type does not have a higher energy value compared to many terrestrial prey species (Cummins and Wuycheck, 1971). Spaans (1971) as well as Annett and Pierotti (1989) described a high importance of highly energy-rich fish in gull diets during the chick-rearing period. In contrast, our results do not show a higher proportion of fish or other food of high energy in the diet (except a higher proportion of terrestrial mammals in HHHallig during chick rearing; Table 1). Presumably, food originating from terrestrial habitats would be sufficient to rear gull chicks despite of increasing energy demands of growing young. It is important to consider, that energy content of prey is not the only issue affecting prey choice of seabirds. Pierotti and Annett (1990, 1991) explained the preference for

19

food of lower energetic value by gulls with other nutritional components, such as calcium, which is essential for the development of bones. The increasing utilisation of shore crabs (Table 1) might be a suitable source for such nutrients. Moreover, plant material consisting of seeds, herbs and grass (although in some cases perhaps ingested while taking up other prey) should be regarded as source for essential vitamins and amino acids as well (e.g. Bezzel and Prinzinger, 1990). Thus, we can conclude that the quality of feeding habitats (i.e. sites showing high abundance, availability and quality of prey species) might play the most important role to cause habitat switch. Indeed, there seems to be a general increase in the quality of the marine area as foraging habitat for black-headed gulls from spring to early autumn: (1) total biomass of the macro-benthos community on tidal flats reaches peak values during the late breeding season/early autumn (Beukema, 1974; Curtis et al., 1985; Stolpmann, 2004). (2) Individual benthic organisms reach highest qualities for predators during that time as a consequence of mass gain (e.g. Beukema, 1974). This would equal a higher energy intake rate per unit foraging time. Oppositely, the quality of the terrestrial feeding habitat seems to decrease during the late breeding period, which is reflected by the lower numbers of black-headed gulls recorded during aerial surveys, lower numbers of gulls flying towards the mainland and a lower proportion of food originating from terrestrial habitats. Most probably the main reasons for decreasing foraging site quality are reduced farming activities (Schwemmer and Garthe, unpubl. data) and high vegetation that reduces access to soil invertebrates (e.g. Morris, 2000; Butler and Gillings, 2004). Moreover, the availability of different prey species can be directly associated with seasonal weather effects, as for instance earthworms are less available during dry conditions, which might dominate during chick-rearing, leading to a switch to marine foraging sites (Sibly and McCleery, 1983; Thyen and Becker, 2006). Other seasonal effects like the swarming of ants or other insects might cause high short-term food availability leading to a habitat switch. 4.2. Inter-colony diet variation We could not confirm our hypothesis of intensive area utilisation in the vicinity of the breeding colonies. In contrast, our results revealed the opposite of what was expected: In FKoog we found the highest proportion of food originating from marine habitats of all colonies (although a high proportion of terrestrial food was expected). In addition, we found significantly more individuals heading towards the marine area than towards the mainland. At HHHallig and Amrum we found highest proportions of terrestrial food (although a higher proportion of marine food was expected). Thus, we may conclude that location of foraging habitats does not seem to influence foraging behaviour at first hand. Moreover, it can be assumed that foraging habitat quality might again play a more crucial role, rather than moderate distances to foraging sites. The quality of the marine area near FKoog is potentially higher compared to the other two colonies for a number of

P. Schwemmer, S. Garthe / Estuarine, Coastal and Shelf Science 77 (2008) 12e22

20

reasons: (1) this area holds extensive fine grained tidal flats that are particularly rich of invertebrates (Stolpmann, 2004). (2) The intensity of Brown Shrimp fishery in this area is higher compared to the other regions (Ehrich et al., 2006; our own unpubl. data). The high number of pellets from FKoog with undersized shrimps that probably originate from fishing vessels as discards, support this conclusion (Appendix I). (3) High breeding pair numbers of particularly herring and common gulls (Larus canus), that have the closest foraging niche overlap with black-headed gulls (Kubetzki and Garthe, 2003), might cause interspecific competition at least in the marine zone. Such colonies are located closest to Amrum and HHHallig. The high proportion of terrestrial prey in HHHallig and Amrum, thus, might also be caused by intensive foraging on islands and mainland to avoid competition in the marine area. The excellent foraging conditions near FKoog are confirmed by a consistent increase of breeding pair numbers during recent years, while the colonies of Amrum and HHHallig showed a stagnating trend (Ha¨lterlein unpubl. data). 4.3. Long-term change of feeding site quality Although we found increasing use of the marine feeding habitats during the late breeding period in all three colonies and both years of sampling, the overall proportion of food originating from the marine area can be considered low when compared to findings of other studies conducted during the last 20 years (Table 3). A few years ago, the proportion of marine-littoral prey in the diet of black-headed gulls was far higher compared to the findings of our study. Of course it is important to note that the reviewed studies were conducted in different sites and might be methodologically inconsistent. However, the general minor use of terrestrial feeding sites in the past years compared to our study is obvious. There are

two possible reasons for this: (1) increased interspecific competition with other gull and shorebird species; and (2) a decrease of feeding site quality. Breeding pair counts showed a long-term decrease of herring gulls during the last few years along the whole German North Sea coast (Garthe et al., 2000; Koffijberg et al., 2006). This species mainly exploits tidal flats (e.g. Kubetzki and Garthe, 2003). Additionally, a high number of migrating and breeding shorebird species (mainly mussel eaters) showed considerable declines in the study area as well (Blew et al., 2005; Koffijberg et al., 2006). Thus, a decrease of feeding site quality in the marine habitat seems to be likely. Breeding pair numbers of black-headed gulls also showed a wide ranging decline in many countries of northern Europe, which have been explained by food shortages (e.g. Stienen et al., 1998; Heldbjerg, 2001; Bellebaum, 2002). However, this decline does not yet seem to take place in the vicinity of the Wadden Sea (Stienen et al., 1998; Bellebaum, 2002; Thyen and Becker, 2006). Thus, the opportunity to switch between two major feeding habitats (i.e. marine and terrestrial) as well as the flexible foraging strategy of black-headed gulls might be able to explain the difference in breeding pair trends in coastal and mainland colonies. 4.4. Conclusions In contrast to purely pelagic, specialized seabird species, as well as in contrast to wading birds that are forced to utilise the tidal flats, the black-headed gull of the North Sea coast switches intensively between foraging habitats (i.e. the offshore zone, tidal flats and the coastal mainland) during different tidal stages, times of day and seasons/breeding stages. Poor prey availability during certain times in the breeding period will thus most likely be avoided. Moreover, this feeding strategy

Table 3 Historical data taken from studies on black-headed gull food, originating from marine and terrestrial foraging habitats as presence of prey items in pellets/faeces (%) Study Period Study area

Gorke (1990) Incubation 1986e1988a Island of Norderoog

Dernedde (1993) Post breeding 1991b Island of Sylt

Kubetzki and Garthe (2003) Incubation 1997 Island of Juist

Kubetzki and Garthe (2003) Chick rearing 1997 Island of Juist

Marine prey Bivalves Gastropods Polychaetes Crustaceans Fish

59 <1 43 10 19

17 11 44 18d 43

86 20 28c 8 8

42 16 11 18 11

Terrestrial prey Terrestrial arthropods Oligochaetes Birds Eggs Mammals Plant material Refuse

11 ? <1 <1 <1 40 <1

17 13 ? ? ? 19 ?

13 6 0 2 1 14 0

26 8 0 0 0 45 3

a b c d

Mean values of 327 pellets collected during the three years. Mean values of 27 faeces samples from September and October; only selected food components available. Polychaetes: Mean values of 83 pellets and 25 faeces. Crustaceans: Mean values of presence of Crangon crangon, Gammarus spec., Carcinus maenas.

P. Schwemmer, S. Garthe / Estuarine, Coastal and Shelf Science 77 (2008) 12e22

is generally an example of superior competitive performance of a flexible and opportunistic predator that is able to exploit the whole variety of the habitat mosaic of a complex coastal zone. Other less opportunistic gull species have much smaller foraging niches and consequently mainly use distinct habitat types (Kubetzki and Garthe, 2003). The flexible feeding strategy, switching between foraging habitats will e in contrast to situations elsewhere e most probably sustain breeding pair numbers at the German Wadden Sea for the near future, despite of possible substantial changes in one of the feeding habitats. Acknowledgements Parts of the study were carried out as an element of the project MINOSplus financed by the Federal Environmental Ministry. S. Adler gave important statistical support. J. Garzke, N. Guse, L. Gieschen and T. Tischler assisted in diet analysis. V. Dierschke, R. Rehm and S. Weiel helped with field work. B. Ha¨lterlein provided data of breeding pair numbers. F. Colijn and N. Markones gave useful comments on an earlier version of this manuscript. D. Oro and one anonymous reviewer made very valuable suggestions to improve the manuscript.

Appendix I. Supplementary information Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.ecss.2007.08.017. References Annett, C., Pierotti, R., 1989. Chick hatching as a trigger for dietary switching in the Western Gull. Colonial Waterbirds 12, 4e11. Ashmole, N.P., 1963. The regulation of numbers of tropical oceanic birds. Ibis 103, 458e473. Bellebaum, J., 2002. Ein ‘‘Problemvogel’’ bekommt Probleme: Bestandsentwicklung der Lachmo¨we Larus ridibundus in Deutschland 1963e1999. Vogelwelt 123, 189e201. Beukema, J.J., 1974. Seasonal changes in the biomass of the macro-benthos of a tidal flat area in the Dutch Wadden Sea. Netherlands Journal of Sea Research 8, 94e107. Bezzel, E., Prinzinger, R., 1990. Ornithologie. Ulmer, Stuttgart, 552 pp. Blew, J., Eskildsen, K., Gu¨nther, K., Koffijberg, K., Laursen, K., Potel, P., Ro¨sner, H.U., van Roomen, M., Su¨dbeck, P., 2005. Migratory birds. In: Essink, K., Dettmann, C., Farke, H., Laursen, K., Lu¨erßen, G., Marencic, H., Wiersinga, W. (Eds.), Wadden Sea Quality Status Report 2004. Wadden Sea Ecosystem No. 19. Trilateral Monitoring and Assessment Group, Common Wadden Sea Secretariat, Wilhelmshaven, Germany, pp. 287e304. Brandl, R., Gorke, M., 1988. How to live in colonies: foraging range and patterns of density around a colony of black-headed gulls Larus ridibundus in relation to the gulls’ energy budget. Ornis Scandinavica 19, 305e308. Butler, S.J., Gillings, S., 2004. Quantifying the effects of habitat structure on prey detectability and accessibility to farmland birds. Ibis 146, 123e130. Cummins, K.W., Wuycheck, J.C., 1971. Caloric equivalents for investigations in ecological energetics. Mitteilungen der Internationalen Vereinigung fu¨r Theoretische und Angewandte Limnologie 18, 1e158. Curtis, D.J., Galbraith, C.G., Smyth, J.C., Thompson, D.B.A., 1985. Seasonal variations in prey selection by estuarine black-headed gulls (Larus ridibundus). Estuarine, Coastal and Shelf Science 21, 75e89.

21

Dernedde, T., 1993. Vergleichende Untersuchungen zur Nahrungszusammensetzung von Silbermo¨we (Larus argentatus), Sturmmo¨we (L. canus) und Lachmo¨we (L. ridibundus) im Ko¨nigshafen/Sylt. Corax 15, 222e240. Diederichs, A., Nehls, G., Petersen, I.K., 2002. Flugzeugza¨hlungen zur großfla¨chigen Erfassung von Seevo¨geln und marinen Sa¨ugern als Grundlage fu¨r Umweltvertra¨glichkeitsstudien im Offshorebereich. Seevo¨gel 23, 38e46. Duffy, D.C., Jackson, S., 1986. Diet studies of seabirds: a review of methods. Waterbirds 9, 1e17. Ehrich, S., Ra¨tz, H.J., Wilhelms, I., 2006. Raumordnung in der AWZ: Mittlere Anlandungen der deutschen und internationalen Fischereiflotten aus deutschen Nordseegewa¨ssern im Zeitraum von 2000e2004. Informationen aus der Fischereiforschung 53, 1e5. Faraway, J.J., 2006. Extending the Linear Model with R. Chapman & Hall, London, 301 pp. Garthe, S., Flore, B.O., Ha¨lterlein, B., Hu¨ppop, O., Kubetzki, U., Su¨dbeck, P., 2000. Brutbestandsentwicklung der Mo¨wen Laridae an der deutschen Nordseeku¨ste in der zweiten Ha¨lfte des 20. Jahrhunderts. Vogelwelt 121, 1e13. Garthe, S., Hu¨ppop, O., Weichler, T., 2002. Anleitung zur Erfassung von Seevo¨geln auf See von Schiffen. Seevo¨gel 23, 47e55. Gloe, P., 2006. Zum Auftreten von Mo¨wen Laridae als Ga¨sten im Binnenland von Dithmarschen (westliches Schleswig-Holstein). Corax 20, 129e137. Gonza´lez-Solı´s, J., Oro, D., Jover, L., Ruiz, X., Pedrocchi, V., 1997a. Trophic niche width and overlap of two sympatric gulls in the southwestern Mediterranean. Oecologia 112, 75e80. Gonza´lez-Solı´s, J., Oro, D., Pedrocchi, V., 1997b. Bias associated with diet samples in audouin’s gulls. The Condor 99, 773e779. Gorke, M., 1990. Die Lachmo¨we (Larus ridibundus) in Wattenmeer und Binnenland. Seevo¨gel 11, 5e48. Gorke, M., Brandl, R., 1986. How to live in colonies: spatial foraging strategies of the black-headed gull. Oecologia 70, 288e290. Harko¨nen, T., 1986. Guide to the Otoliths of the Bony Fishes of the Northeast Atlantic. Danbiu Aps, Hellerup, 256 pp. Hartwig, E., Hu¨ppop, O., 1982. Zum Energie- und Nahrungsbedarf einer Kolonie der Lachmo¨we (Larus ridibundus L.) an der Schlei. Seevo¨gel spec. issue, 93e105. Heldbjerg, H., 2001. The recent decline in the population of black-headed gulls Larus ridibundus in Denmark and its plausible causes. Dansk Ornithologiske Forenings Tidsskrift 95, 19e27. Jones, M.J., 1985. Distribution and activity of common gulls and black-headed gulls on grassland in Manchester. Bird Study 32, 104e112. Koffijberg, K., Dijksen, L., Ha¨lterlein, B., Laursen, K., Potel, P., Su¨dbeck, P., 2006. Breeding birds in the Wadden Sea in 2001. Results of the total survey in 2001 and trends in numbers between 1991e2001. In: Wadden Sea Ecosystem Rep. No. 22. Wilhelmshaven, 132 pp. Kubetzki, U., Garthe, S., 2003. Distribution, diet and habitat selection by four sympatrical gull species in the southeastern North Sea. Marine Biology 143, 199e207. Legendre, P., Legendre, L., 1998. Numerical Ecology, second English ed. Elsevier Science BV, Amsterdam, 870 pp. Leopold, M., van Damme, C.J.G., Philippart, C.J.M., Winter, C.J.N., 2001. Otoliths of North Sea fish: Interactive Guide of Identification of Fish from the SE North Sea. Wadden Sea and Adjacent Fresh Waters by Means of Otoliths and Other Hard Parts. CD-Rom. ETI, Amsterdam. Leps, J., Smilauer, P., 2003. Multivariate Analysis of Ecological Data using CANOCO. Cambridge University Press, Cambridge, 269 pp. Lewis, S., Sherratt, T.N., Hamer, K.C., Wanless, S., 2001. Evidence of intraspecific competition for food in a pelagic seabird. Nature 412, 816e819. Monaghan, P., 1996. Relevance of the behaviour of seabirds to the conservation of marine environments. Oikos 77, 227e237. Morris, M.G., 2000. The effects of structure and its dynamics on the ecology and conservation of arthropods in British grasslands. Biological Conservation 95, 129e142. Mudge, G.P., Ferns, P.N., 1982. The feeding ecology of five species of gulls (Aves: Larini) in the inner Bristol Channel. Journal of Zoology London 197, 497e510. Oro, D., Ruiz, X., Jover, L., Pedrocchi, V., Gonza´lez-Solı´s, J., 1997. Diet and adult time budgets of audouin’s gull Larus audouinii in response to changes in commercial fisheries. Ibis 139, 631e637.

22

P. Schwemmer, S. Garthe / Estuarine, Coastal and Shelf Science 77 (2008) 12e22

Pierotti, R., Annett, C.A., 1990. Diet and reproductive output in seabirds. BioScience 40, 568e574. Pierotti, R., Annett, C.A., 1991. Diet choice in the herring gull: constraints imposed by reproductive and ecological factors. Ecology 72, 319e328. Pinheiro, J.C., Bates, D.M., 2000. Mixed-Effects Models in S and S-PLUS. Springer, Berlin, 528 pp. Schaefer, M., 1994. Brohmer, Fauna von Deutschland, 19th ed. Quelle & Meyer Verlag, Heidelberg, Wiesbaden, 809 pp. Schwemmer, P., Garthe, S., 2005. At-sea distribution and behaviour of a surfacefeeding seabird, the lesser black-backed gull (Larus fuscus), and its association with different prey. Marine Ecology Progress Series 285, 245e258. Sibly, R.M., McCleery, R.H., 1983. The distribution between feeding sites of herring gulls breeding at Walney Island, U.K. Journal of Animal Ecology 52, 51e68. Spaans, A.L., 1971. On the feeding ecology of the herring gull Larus argentatus Pont. in the northern part of the Netherlands. Ardea 59, 75e188. Stienen, E.W.M., Arts, F.A., de Boer, P., Beeren, W.J., Majoor, F., 1998. Broedresultaten van Kokmeeuwen in Nederland in 1997. Sula 12, 1e11.

Stolpmann, M., 2004. Watt im Wandel. Vera¨nderungen der Abundanz und Diversita¨t der Makrobenthos-Organismen im schleswig-holsteinischen Wattenmeer zwischen 1993 und 2003. Diploma Thesis, University of Bremen, 168 pp. Tasker, M.L., Jones, H.P., Dixon, T., Blake, B.F., 1984. Counting seabirds at sea from ships: a review of methods employed and a suggestion for a standardized approach. The Auk 101, 567e577. Thyen, S., Becker, P.H., 2006. Effects of individual live-history traits and weather on reproductive output of black-headed gulls Larus ridibundus breeding in the Wadden Sea. Bird Study 53, 132e141. Venables, W.N., Ripley, B.D., 2002. Modern Applied Statistics with S. Springer, New York, 495 pp. Verbeek, N.A.M., 1977. Comparative feeding ecology of herring gulls Laurs argentatus and lesser black-backed gulls Larus fuscus. Ardea 65, 25e42. Vernon, J.D.R., 1970. Feeding habitats and food of the black-headed and common gulls. Part 1 e Feeding Habitats. Bird Study 17, 287e296. Vorberg, R., Breckling, P., 1999. Atlas der Fische im schleswig-holsteinischen Wattenmeer. Boyens, Heide, 178 pp.