Functional Ecology 2004 18, 793– 801

Linking the foraging performance of a marine predator to local prey abundance

Blackwell Publishing, Ltd.

D. GRÉMILLET,*† G. KUNTZ,*¶ F. DELBART,‡ M. MELLET,‡ A. KATO,§ J.-P. ROBIN,* P.-E. CHAILLON,¶ J.-P. GENDNER,* S.-H. LORENTSEN** and Y. LE MAHO* *Centre d’Ecologie et Physiologie Energétiques, Centre National de la Recherche Scientifique. 23 rue Becquerel, 67087 Strasbourg Cedex 02, France, ¶Université du Québec à Rimouski, 300 allée des Ursulines, Rimouski (Québec) Canada G5L 3A1, ‡Institut Polaire Français Paul-Emile Victor, Technopôle Brest-Iroise, BP 75–29280 Plouzané, France, §National Institute of Polar Research 1-9-10 Kaga, Itabashi-ku, Tokyo 173–8515, Japan, and **Norwegian Institute for Nature Research, Tungasletta 2, NO-7485 Trondheim, Norway

Summary 1. Knowledge of the functional response of predators to prey densities conditions our understanding of food webs. Such links are still poorly understood within the higher trophic levels of marine ecosystems. 2. We present the first field study recording the foraging effort and foraging yield of a seabird (the Great Cormorant, Phalacrocorax carbo) as well as the abundance and quality of prey within its foraging area. 3. We confirm that Great Cormorants foraging off West-Greenland show the highest foraging performance recorded for a marine predator (between 17 and 41 g fish caught per minute underwater). Former work suggests that such high foraging yield should be based upon the exploitation of extremely profitable prey patches. 4. Contrary to this hypothesis, average prey abundances estimated within the foraging areas of the cormorants were low (0·03–0·09 prey m−2, depending on methods), as was the average calorific value of the prey items (4·2 kJ g−1). 5. Our study suggests that Great Cormorants remain highly successful predators even when exploiting modest prey resources. These findings have implications for our understanding of predator–prey relationships, and for the management of Great Cormorant populations. Keywords: Catch per unit effort, Great Cormorant, Phalacrocorax carbo, scuba transects, wildlife telemetry Functional Ecology (2004) 18, 793– 801

Introduction How do prey resources influence predator populations? Predator–prey interactions underpinning this key question have been extensively studied both theoretically and in controlled environments (Stephens & Krebs 1986; Cuddington & Yodzis 2002; Shertzer et al. 2002; Yoshida et al. 2003; Johnson & Agrawal 2003). Field studies linking prey abundance and predator foraging efficiency nonetheless remain challenging (Pulido & Diaz 1997; O’Donoghue et al. 1998; Naef-Daenzer, Naef-Daenzer & Nager 2000; Vucetich, Peterson & Shaefer 2002). This is due to technical difficulties in collecting sufficient and concurrent information about predators and prey, particularly about the prey intake rates of free-ranging predators. Such constraints are enhanced in the © 2004 British Ecological Society

†Author to whom correspondence should be addressed. E-mail: [email protected]

marine environment where, besides laboratory studies (Wellenreuther & Connell 2002; Shertzer et al. 2002) and studies confined to benthic organisms (Tokeshi & Romero 1995; Menge et al. 1997; Whitlatch et al. 1997), little is known about the functional links between prey abundance and predatory performance. This is particularly true for the higher trophic levels of marine food webs, as top predators (predatory fish, sea birds and mammals) typically operate underwater. Seabirds are partly an exception to this, as many of them travel in the air, and feed at the sea surface where they can be observed. Seabirds also raise their offspring on islands where they are accessible during the breeding season. Long-term studies conducted at these breeding sites have shown strong links between the availability and the quality of food resources and seabird reproductive output (Annett & Pierotti 1999; Bertram, Mackas & McKinnell 2001; Durant, Anker-Nilssen & Stenseth 2003). Furthermore, at-sea observations and measurements 793

794 D. Gremillet et al.

have unravelled the complex associations between seabirds populations and their prey at different spatial and temporal scales (Hunt, Heinemann & Everson 1992; Balance, Pitman & Reilly 1997; Mehlum et al. 1999; Fauchald, Erikstad & Skarsfjord 2000; Davoren, Montevecchi & Anderson 2003). At the individual level, breeding and interbreeding seabirds have been equipped with miniaturized electronic devices. These tools have provided detailed information about the position of birds at sea and their foraging behaviour (see Croll et al. 1998; Wilson et al. 2002 for reviews). The insights gathered through these complementary approaches have greatly enhanced our current knowledge of seabirds as marine predators. However, the link between seabird foraging performance and prey abundance has never been studied in the field. We investigated this aspect by recording synoptically the foraging effort and foraging yield of a diving seabird, as well as the abundance and quality of prey targeted within its foraging area. The study was conducted on Great Cormorants (Phalacrocorax carbo) breeding in West-Greenland. Great Cormorants are a particularly suitable model species for two reasons. First, they are coastal diving seabirds, which mainly feed on demersal and bottomdwelling fish (Johnsgard 1993). Their food resources can thus be studied within well-defined coastal areas. Second, Great Cormorants have a partly wettable plumage, very limited thermal insulation underwater

(Grémillet et al. 2005) and high foraging costs when diving (Schmid, Grémillet & Culik 1995). This is particularly marked in West-Greenland, where heat losses from the bird’s body to cold waters are amplified (Grémillet & Wilson 1999; Grémillet et al. 2003). Great Cormorants nonetheless manage to breed and winter in the high Arctic (Grémillet et al. 2001) where their daily food intake is not higher than that of conspecifics from temperate Western Europe (Grémillet et al. 1999a). Birds are thought to compensate for high foraging costs via an increase of their foraging success, i.e. the colder the water, the faster they acquire food. Consequently, it has been proposed that Great Cormorants foraging in cold Greenlandic waters target prey patches with very high density (Grémillet et al. 2001). This link between prey abundance and Great Cormorant foraging performance is thought to be stronger than in any other species of diving bird (Grémillet & Wilson 1999). We tested the hypothesis that high foraging performance in Greenland Great Cormorants is linked to high-density/ quality food resources within their foraging areas.

Materials and methods We studied Great Cormorants (Phalacrocorax carbo) breeding on Qeqertaq (69°30′ N, 54°05′ W) in the Diskofjord area, Disko, West-Greenland (Fig. 1) in June and July 1998, 2001 and 2002. All experiments were performed under permits from the Greenland Homerule Government and Danish Polar Center. We used a noose mounted on a telescopic pole to catch Great Cormorants at the nest site to equip them with electronic devices. Care was taken to minimize stress, and handling of individual birds lasted less than 10 min in all cases. Great Cormorants breeding in Greenland are sexually dimorphic, males being significantly heavier than females (3·2 kg vs 2·6 kg on average). We used this criterion to distinguish sexes.

 

© 2004 British Ecological Society, Functional Ecology, Fig. 1. Location of the study area in West-Greenland. 18 , 793–801

In June–July 1998, adult cormorants were radio-tracked for at least 10 foraging trips chosen at random during a period of 10 days (see Grémillet et al. 1999b for technical details) The study area consists of deep (>200 m) fjords with very steep shores. Great Cormorants usually forage along the seabed in c. 10 m water depth, with maximum dive depths of 35 m (Grémillet et al. 1999b). Their foraging zone is thus probably restricted to a very narrow band (max. 300 m) along these shores. Knowing the bearing of the diving zone and the position of the coast line, we determined the approximate location of each foraging area. We visited these sites, and made direct observations of the birds which confirmed our radio-tracking assessments. We observed that foraging Great Cormorants rested after each diving session on the shore next to their feeding area (this was also determined via radio-tracking). They build roosts on rocks adjacent to their foraging zones, which are covered with

795 High foraging performance at low prey densities

a conspicuous white layer of faeces. In 2001 and 2002 we consequently used these roosts to determine the areas used by the foraging cormorants. Direct observations of birds flying between the breeding site and the feeding zones, as well as observations at the feeding zones and their adjacent roosts confirmed that the areas used were the same as in 1998.

  We used three indices of foraging effort: (1) the total time spent underwater per foraging trip (i.e. the time available for the bird to search and target prey), (2) the total duration of each foraging trip ( i.e. the time spent away from the nest), and (3) the number of foraging trips per day. Foraging trip duration and number of foraging trips per day were determined using data from automatic nest balances (see ‘foraging yield’). The total time underwater per trip was determined using two different methods. In 1998 we used radio-tracking data. Our recordings were analysed to determine the duration of each dive (accuracy 1 s). These durations were summed to calculate total time underwater per foraging trip. In 2002 total time spent underwater per foraging trip was determined using recordings of time-depth data loggers (M190D2GT, relative accuracy 0·1 m, 15 mm diameter, 60 mm length, 20 g, Little Leonardo, Tokyo, Japan). The recorders were programmed to record depth every second, and were attached underneath the tail of the birds with waterproof Tesa® tape.

  Nest balances were used to determine the amount of food caught during each foraging trip (see Grémillet et al. 1996). In 1998 birds were studied using the equipment described in Grémillet et al. (1999a). In 2001 we used the equipment described in Grémillet et al. (1996). A new system was used in 2002 (J.-P. Gendner, CNRS, Strasbourg), consisting of a rounded fibreglass podium (50 cm diameter, 20 cm height) linked to the base of the balance by a strain gauge which acts as a scale sensor. This sensor (accuracy: ±10 g under field conditions) is connected to a multipurpose data logger set to record every 10 s.



© 2004 British Ecological Society, Functional Ecology, 18, 793–801

The prey spectrum of Great Cormorants breeding on Qeqertaq was determined in 1998, 2001 and 2002 via pellet analysis following Harris & Wanless (1993). Fish species were identified to the lowest possible taxonomic level from otoliths using Härkönen (1986) and our own reference collection. We also gathered fish regurgitated by the birds when being handled or disturbed at the nest. This sample was primarily used to determine the average mass of the major prey types, but also to crosscheck the diet spectrum derived from pellet analysis (see Grémillet & Argentin 1998).

    Specimens from the different fish species were sampled to match the sizes taken by the birds. Once collected the fish were weighed (±0·1 g) in the field and stored frozen (−18 °C) until analysis. Each fish was freeze-dried to constant mass and ground under liquid nitrogen to obtain a fine homogeneous powder. Energy content was determined on dry aliquots (0·7–1·4 g) by using an adiabatic bomb calorimeter Parr 1241 with benzoic acid as standard (Parr Instrument Co., Moline, Illinois, USA). Reproducibility between two measurements of fish samples was 1·32 ± 0·47%. Just before being analysed the samples were freeze-dried again to eliminate potential traces of water.

  Previous studies showed that Great Cormorants from Greenland feed exclusively on fish, and primarily on bottom-dwelling sculpin Myoxocephalus scorpioides and Gymnacanthus tricuspis (Grémillet et al. 2001). These species are sedentary, and do not form schools (FishBase 2004). Spatial censuses of their abundance are therefore not affected by patchiness. We studied the abundance of demersal and benthic fish within the feeding area used by the cormorants (as determined by radio-tracking and visual observations, see above) using four independent methods: underwater video, fish trap, gill net and scuba transects. The trials were conducted during the breeding season of the cormorants and within their foraging areas as determined via radio-tracking and direct observations (see above). Great Cormorants breeding in Greenland are mainly active between 06.00 and 21.00 (D. Grémillet et al., unpublished data). Trials were therefore conducted during this time period. In 2001 we used an underwater camera system (Mariscope, Kiel, Germany). This equipment consists of a waterproof camera (20 cm length, 10 cm diameter, operational depth: 300 m, type: 1/3″ CCD B & W, picture elements: 512 (H) × 582 (V) pixel, resolution: 380 TV lines, lens: 4·48 mm, F1·8, sensitivity: 0·1–0·01 lux, viewing angle: 43° in water) connected to a standard video recorder via a 50 m reinforced coaxial cable RG174 (Mariscope, Kiel). We deployed the camera from an inflatable boat, lowering it slowly to 1 m above the ground. In this position the unit records pictures from approximately 1 m2. For each station the camera was set to record for 1 min once stabilized above the sea floor. Video tapes were analysed to record the number and the species of fish present at each station. In 2001 fish abundance within the cormorants’ foraging areas was also studied using a baited fish trap (lobster trap design; 50 cm diameter, 120 cm long, mesh size 5mm, funnel diameter 8 cm). The trap was baited with gadid offal and set for between 1 and 3days. In 2002 we used a three-layer, fine-meshed gill net (50 m × 1·7 m; central layer: mesh knot to knot = 23 mm; lateral layers: mesh knot to knot = 250 mm) to assess

796 D. Gremillet et al.

Results    In 1998 we counted 179 occupied nests, and an average of 2·5 ± 0·94 chicks per nest (n = 170). In 2001, 298 nests were counted, with an average of 3·0 ± 1·58 chicks per nest (n = 71). In 2002, the colony had grown to 401 nests, with an average 3·5 ± 1·29 chicks per nest (n = 50). Interviews of the local Inuit hunters indicated that this colony has been present in the Diskofjord area ‘with several hundred birds’ as far back as they could recall.

 

Fig. 2. Foraging zones of Great Cormorants breeding on Qeqertaq, Diskofjord, as determined via radio-tracking within the study area.

prey abundance. The net was set vertically from the bottom in 3 – 8 m depth for 3 –12 h. We remained within 500 m of the net throughout the deployments to make sure that no diving bird would approach it. In 2002 we also conducted underwater transects by scuba diving. This approach followed the line-transect method (Buckland et al. 1993). We used transects of 10 m length conducted in 2, 4, 6, 8, 10, 12, 14, 16 and 18 m water depth. This depth zone corresponds to the zone used by Great Cormorants (Grémillet et al. 1999b; D. Grémillet et al., unpublished data). All transects were conducted by the same observer, who recorded the abundance of the different fish species within 0·5 m on each side of the transect line while swimming between 0·5 and 1 m above the ground at approx. 0·15 m s−1. The surveyed area per transect was thus assumed to be approximately 10 m2.

 

© 2004 British Ecological Society, Functional Ecology, 18, 793–801

Summary statistics of cormorant foraging effort and foraging yield were computed for the entire sample using individual birds as the sampling unit. Averages are provided ± SD. The effect of sex and year on foraging trip duration, time underwater per foraging trip and food intake per foraging trip were analysed using residual maximum likelihood analyses ( REML, Patterson & Thompson 1971). Logarithmic transformations were carried out to fulfil the criterion of normality. The number of trips per day was compared between sexes and years in a Generalized Linear Mixed Model (GLMM) with Poisson errors and log link function (Schall 1991). In all analyses, sex and year were entered as fixed factors and bird as a random factor. In all REML and GLMM models, the effect of sex and year was determined by comparing Wald statistics with F-distributions.

We radio-tracked six female and five male Great Cormorants in 1998 during a total of 163 foraging trips. Birds mainly foraged along the South shore of Qeqertaq (55% of trips, Fig. 2, zone A) and along the Northern shores of the fjord (22% of trips, Fig. 2, zone B). On fewer occasions they also visited the Southern shores of the fjord entrance (12% of trips, Fig. 2, zone C), and the inner part of the fjord (2% of trips, Fig. 2, zone D). In 9% of the trips we could not radio-track the bird continuously, and therefore could not determine its exact feeding location. All feeding locations were within 15km of the breeding colony, and the average foraging range was 6·7 ± 2·5 km. Visual observations (75%).

     We recorded foraging trip duration, the number of foraging trips per day, and food intake per foraging trip for six male and five female Great Cormorants in 1998, five males and five females in 2001, and four males and four females in 2002. The time spent underwater per foraging trip was recorded for six males and five females in 1998, and six males and six females in 2002. Table 1 provides an overview of the results. Individual birds conducted between one and four foraging trips lasting taking 30 min to 14 h per day. During each foraging trip they spent between 0·3 and 54 min underwater, catching between 20 and 1376 g of fish. Foraging trip duration and food intake per foraging trip were significantly different between years (F2,28 = 11·1, P = 0·004, and F2,28 = 30·5, P < 0·001, respectively), but not between sexes (F1,28 = 2·0, P = 0·16, and F1,28 = 3·1, P = 0·078, respectively). Time underwater per foraging trip was significantly different for the different sexes (F1,28 = 6·6, P = 0·01), but not for the different years (F2,28 = 0·7, P = 0·42). The number of foraging trips per day was not significantly different between years (F1,23 = 3·8, P = 0·15), or between sexes (F1,23 = 0·01, P = 0·92). Using the ratio of the amount of food caught per foraging trip to the time spent underwater per foraging

797 High foraging performance at low prey densities

Table 1. Foraging investment and foraging yield in Great Cormorants breeding in West Greenland. Mean values are given ± standard deviation. Median values are given with the range 1998

Trip duration (min)

Trips per day Time under-water per trip (min)

Food intake per trip (g)

2001

Males

Females

Males

Females

Males

Females

143 (31– 495) n=6 2·9 ± 0·9 n=6 8·2 (0·3 –54·1) n=6 334 (100 –787) n=6

147 (31– 440) n=5 3·0 ± 1·1 n=5 12·1 (0·8 –51·5) n=5 410 (50 – 926) n=5

174 (40 – 853) n=5 2·6 ± 0·5 n=5

189 (31– 462) n=5 2·7 ± 0·6 n=5

391 (20 –1376) n=5

300 (32– 691) n=5

150 (30 – 414) n=4 2·8 ± 0·6 n=4 8·1 (1·9–30·7) n=6 281 (80 –735) n=4

216 (36–626) n=4 2·4 ± 0·9 n=4 13·0 (2·9–50·2) n=6 215 (30–807) n=4

trip we calculated an average catch per unit effort (CPUE g min−1) of 40·7 g min−1 and 33·9 g min−1 for males and females in 1998 and of 34·7 g min−1 and 16·5 g min−1 for males and females in 2002, respectively. Similarly, the product of the time spent underwater per foraging trip and of the number of foraging trips per day provides an estimate of the total time spent underwater per day (23·8 and 36·3 min for males and females in 1998 and 22·8 and 31·2 min for males and females in 2002, respectively). Despite some intersexual and interannual variations (see also tests above) both features indicate high levels of bird foraging success throughout the measurements.

 We extracted 1288 otoliths from 52 pellets in 1998, 1654 otoliths from 51 pellets in 2001, and 660 otoliths from 51 pellets in 2002. These samples showed that Great Cormorants breeding in the Diskofjord area feed on at least 13 different species of fish. Sculpins (Gymnacanthus tricuspis and Myoxocephalus scorpioides) were largely predominant both by number and by mass in all years (Table 2), followed by Capelin (Mallotus villosus), and gadids (Gadus sp.), although the proportion of these two prey types varied among years (Table 2). Further species such as Ammodytes marinus, Pholis fasciatus, Eumesogrammus praeciosus, Hippoglossoides platessoides, Liparis sp., Lycodes sp., Salmo sp. and Sebastes sp. were also present in small numbers. Regurgitations collected at the breeding site confirmed that 60–75% of food loads consisted of sculpins in all years.

   

© 2004 British Ecological Society, Functional Ecology, 18, 793–801

2002

We could gather samples for six prey species/groups. Their average energy contents (expressed as kJ per gram fresh mass) were as follows: Mallotus villosus X = 3·9 ± 0·3 kJ g−1 (n = 11), Gymnacanthus tricuspis X = 4·5 ± 0·4 kJ g−1 (n = 9), Myoxocephalus scorpioides X = 4·1 ± 0·4 kJ g−1 (n = 13), Gadus sp. X = 4·4 ± 0·03 kJ g−1 (n = 2),

Ammodytes marinus X = 6·8 ± 0·2 kJ g−1 (n = 9) and Pholis fasciatus X = 5·7 ± 0·7 kJ g−1 (n = 10). Capelins were sampled after spawning.

  In 2001 we sampled a total of 144 video stations, 87 in zone A and 57 in zone B. The average sampling depth was 6·1 ± 3·9 m (range 1–16 m), with 63% of the sampling effort between 2 and 10 m, i.e. within the preferred foraging depth of Great Cormorants (Grémillet et al. 1999b; D. Grémillet et al., unpublished data). The benthic zones consisted mainly of hard substrata (83%), with an average kelp coverage of 44% (range 0–100%, 18% with 100% kelp cover) showing substantial iceberg scouring. The remaining 17% of the sampled zones were soft bottom situated at the mouth of glacial streams. Potential prey items were present at only 6·9% of all stations. At these 10 stations, an average of 1·4 ± 0·7 potential prey items were present. The fish sighted consisted of 46% sculpins, 23% flounders, 23% Pholis fasciatus and 8% gadids. The recorded densities correspond to an estimated average 1·4 prey-items per square metre considering the stations where fish were sighted, and an average of 0·09 prey-items per square metre for all 144 stations. In 2001 the fish trap was deployed on three occasions within zone A for periods of 1, 2 and 3 days, respectively, at approximately 6m depth on hard, kelpcovered substrata. On all three occasions the yield was nil, suggesting that this method was not particularly appropriate in this area. In 2002 we deployed a gill net at 13 stations situated in zone A (7 stations), zone B (5 stations), and zone C (1 station). The net was set at an average depth of 5·2 ± 0·9 m (range 3–8 m) for periods of between 3 and 12 h (average 4·1 ± 2·6 h). For 23% of the trials the yield was nil. For the remaining 77% an average of 4·3 ± 7·9 potential cormorant prey were caught. These 56 prey items consisted of 57% Capelin, 32% sculpins, 7% gadids, 2% flounder and 2% Lycodes reticulates. Finally in 2002 a total of 115 underwater transects were conducted,

49·7 30·8 3·8 9·4 1·7 4·7 82·4 13·4 2·8 1·4

65·2 28·8 5·1 0·9

Proportion in diet (% occurrence) Proportion in diet (% mass) Proportion in diet (% occurrence)

58·8 13·4 2·0 14·6 7·3 4·0 61·5 16·9 20·4 1·2 39·4 15·2 12·8 10·8 15·7 6·1 41·2 ± 20·7 29·4 ± 5·5 42·0 ± 6·5 2·8 ± 0·2 Sculpins Capelin Gadids Sandeels Others Undetermined

Proportion in diet (% occurrence) Prey species

corresponding to approximately 1150 m2 surveyed (assuming a surface of 10 m2 per transect, see Materials and methods). The sampling effort was divided between zone A (42 transects), zone B (59 transects) and zone C (14 transects). The depth zone investigated stretched between 2 and 18 m, yet 72% of the sampling occurred between 2 and 10 m, as this is the preferred foraging depth of Great Cormorants (D. Grémillet et al., unpublished data). The sampled zones consisted of 82% hard and 18% soft substrata, respectively. For 68% of the transects no potential cormorant prey was recorded. For the remaining 32% we found 73% Capelin, 23% sculpins and 4% further bottom-dwelling fish. The average estimated prey abundance for the transects within which fish were counted was 15·2 ± 15·2 (range 1– 80), corresponding to an estimated 1·52 prey items per square meter. Taking all transects into account, the average prey abundance per transect was 0·29 ± 0·24. This corresponds to approximately 0·03 prey items m−2. Such results are within the same range as estimates provided by the video survey conducted in 2001 (see above). There were no significant differences in prey abundance between zone A and zone B (t = 0·55, P = 0·59), yet prey abundance decreased significantly with increasing depth (F1,7 = 7·05, P = 0·03).

Discussion

Average mass (g ± SD)

Proportion in diet (% mass)

2002 2001 1998

Table 2. Diet of Great Cormorants breeding in Greenland in 1998, 2001 and 2002. See Materials and methods and Results sections for details on the calculations and sample sizes

© 2004 British Ecological Society, Functional Ecology, 18, 793–801

Proportion in diet (% mass)

798 D. Gremillet et al.

To our knowledge, this is the first field study that has recorded the foraging effort and the prey intake of a seabird or a marine mammal while simultaneously assessing the abundance and the quality of its prey within its foraging area. Our results highlight important facts about the ecology of marine predators such as Great Cormorants. We confirm that Great Cormorants breeding in Greenland are extremely efficient predators. This is underlined by the short time periods spent underwater and the very high prey capture rates (Table 1). Despite some interannual and intersexual variability, all average catch per unit effort (CPUE) values were higher than those recorded for conspecifics breeding in Europe (i.e. 12 g min−1; Grémillet 1997), reaching levels 30 times higher than those recorded for other seabird species (typically around 1 g min−1 in penguins (Grémillet 1997), but see Wilson 2004). Furthermore, birds from the Diskofjord study site were raising on average over three chicks during the field trials, indicating a substantial reproductive output. Finally, the study colony increased by 100% during the study period. One would expect these predatory and reproductive performances to be based upon vast numbers of highly profitable prey within the vicinity of the breeding site (Grémillet et al. 2001). Contrary to this expectation the estimated abundance of Great Cormorant prey within their foraging area was low. Regardless of the technique used, the majority of the sampling stations considered did not reveal any potential bird prey. The video stations and the scuba transects provided estimates of the local prey abundance with average densities of between 0·03 and 0·09 prey m−2.

799 High foraging performance at low prey densities

© 2004 British Ecological Society, Functional Ecology, 18, 793–801

These densities of demersal fish are typical for Arctic rocky shores (Hoff 2000; Born & Böcher 2001; FishBase 2004). In the Bering Strait area, for example, Hoff (2000) estimated a sculpin abundance of less than 0·01 fish m−2. Our estimates of demersal fish densities off WestGreenland are about 10 times lower than in comparable habitats from boreal coastal zones. For example the abundance of bottom-dwelling labrids averages several individuals per square meter of rocky shores off West Scotland (data recorded using scuba transects; Magill & Sayer 2002). Furthermore, an average of 1·5 bottomdwelling fish m−2 (mainly labrids) was recorded via scuba transects in Brittany (Siorat 2003). Such labrid species offer a good comparison, as they occupy the same niche in boreal coastal zones that Arctic sculpins occupy in Arctic coastal zones. In these regions, they are the major prey of local Great Cormorants (Grémillet & Argentin 1998). There is no fully accurate way to assess fish abundance (Willis, Miller & Babcock 2000), even in a fairly simple arctic benthic ecosystem. The fish abundances provided in our study therefore probably underestimate the true prey density in the study area. However, similar biases apply to studies of labrid densities conducted in the boreal zone (Magill & Sayer 2002 and Siorat 2003; see above), because they used exactly the same methodology (scuba transects). These studies indicate abundances of potential cormorant prey about 10 times higher than at our study site. Hence, even when taking into account a certain level of underestimation, this comparison shows that cormorant prey densities are markedly lower in Greenland than they are along European rocky shores. This is confirmed by recent studies from central Norway (Lorentsen et al., 2004) which provide concurrent estimates of cormorant predatory performance and fish abundance in an area where the birds forage in warmer water than in Greenland (12 °C in Norway vs 5 °C in Greenland). In this zone, average cormorant CPUE was assessed as 9·76 ± 4·49 g min−1 spent underwater for 10 breeding individuals, using exactly the same methodology as in our study (recording of dive patterns using VHF transmitters and automatic weighing at the nest, S.-H. Lorentsen, unpublished data). These CPUE values are similar to those for Great Cormorants from temperate Normandy (Grémillet 1997), and markedly lower than for Great Cormorants studied in Greenland. Average prey abundances (labrids and gadids) of between 0·5 and 2·3 fish per m2 were recorded using scuba transects along the west coast of Norway (for a total transect length of 12 750 m, K. Sjøtun, unpublished data). These abundances are similar to those recorded along other European rocky shores (Magill & Sayer 2002 and Siorat 2003; see above). Taken together, these findings show that Great Cormorant foraging performance is higher in Greenland than it is in Europe, and that estimated prey abundances are lower in Greenland than they are in Europe. We therefore cannot support the hypothesis that the foraging performance of Greenland Great Cormorants is based upon super-abundant food resources (Grémillet

et al. 2001). Food resources off West-Greenland are not only scarce, they are also energetically unattractive. The birds studied mainly fed on small bottom-dwelling sculpins (Table 2). The energy content of this prey is low compared with that of a wide range of fish exploited by seabirds (Hislop, Harris & Smith 1991). Moreover, this food resource occurs in a marine area where Capelin populations reach large numbers (Friis-Rødel & Kanneworff 2002; Carscadden & Vilhjalmsson 2002). Great Cormorants also take some Capelin (Table 2), yet, unlike most seabirds and marine mammals in this zone, they make limited use of this bounty. Capelin stocks show substantial spatio-temporal variability (Carscadden & Vilhjalmsson 2002), whereas sculpins feature a stable, sedentary population. We propose that cormorants favour an apparently less numerous, yet highly predictable food resource over a temporarily super-abundant one. This predictability could explain the high consistency of the breeding chronology of Great Cormorants in Greenland. This potential strategy nonetheless does not explain how Greenland Great Cormorants sustain high levels of foraging efficiency on such limited food quality and quantity. This can only be due to selection pressure favouring individuals with extraordinary foraging capabilities. This hypothesis remains to be tested.

  It has been proposed that food availability principally limits the growth of tropical seabird populations (Ashmole 1963), and possibly of seabird populations in general. This paradigm is largely based upon the assumption that seabirds have a costly way of life, and therefore require vast amounts of food to balance their energy budgets. As alluded to in the Introduction, this assertion is particularly difficult to test because the population dynamics of a seabird species, its foraging behaviour and the abundance of prey within its foraging zone are very difficult to record simultaneously. Nevertheless, indirect evidence has been gathered which currently supports Ashmole’s theory (Furness & Birkhead 1984; Birt-Friesen et al. 1989; Cairns 1989; Lewis et al. 2001). In this context, it is rather surprising to find that seabirds such as cormorants can sustain themselves on modest food resources and raise large numbers of chicks, even when foraging in a challenging arctic environment. This suggests that the foraging success of a diving endotherm is not necessarily a good indicator of environmental quality (Houston et al. 2003). Our findings should also be considered in the context of the perceived conflict between European Great Cormorant populations and freshwater fisheries. Great Cormorant numbers increased sharply throughout Europe between the mid-1980s and the late 1990s. Since then claims have risen that Great Cormorants deplete freshwater fish stocks (review in Carss et al. 2003). The rate of population growth decreased slightly in recent years (Bregnballe et al. 2003). This has been attributed to growing intraspecific competition for food, which

800 D. Gremillet et al.

might induce a self-regulation of European Great Cormorant populations ( Frederiksen & Bregnballe 2000). Our study questions this interpretation, as it indicates that Great Cormorants might show a substantial predatory and breeding performance, even when exploiting modest prey resources.

Acknowledgements This study was funded by the Institut Polaire Français Paul-Emile Victor and by the Centre National de la Recherche Scientifique. J.-P. Gendner was funded by the Mission des Resources et Compétences Technologiques du CNRS. The Arctic Station in Godhavn provided logistic support throughout four field seasons on Disko. The Greenland Homerule Government, the Danish Polar Center, and the Greenlandic Radio Administration gave permission to work within areas under their control. R. Dey, J. M. Jensen, N. Pekruhl and M. Krause survived a difficult first field season in 1998. C. Gilbert and E. Pettex assisted D.G. during the 2003 field season. Patient support was also provided by B. Jessen-Graae, F. Nielsen, A. Mølgaard, the crews of the vessels Porsild & Maya, the people of Kangerluk, and the staff of the Arctic Station board, the French Polar Institute and the CEPE-CNRS. J.-M. Canonville performed the dietary analysis for 2001 and 2002. D. Boertmann helped during the planning and the writing-up phases. A. Myers and M. Enstipp corrected the English. K. Sjøtun kindly provided unpublished data. We warmly thank all of them.

References

© 2004 British Ecological Society, Functional Ecology, 18, 793–801

Annett, C.A. & Pierotti, R. (1999) Long-term reproductive output in Western Gulls: consequences of alternate tactics in diet choice. Ecology 80, 288 –297. Ashmole, N.P. (1963) The regulation of numbers of tropical oceanic birds. Ibis 103, 458 – 473. Balance, L.T., Pitman, R.L. & Reilly, S.B. (1997) Seabird community structure along a productivity gradient: importance of competition and energetic constraint. Ecology 78, 1502–1518. Bertram, D.F., Mackas, D.L. & McKinnell, S.M. (2001) The seasonal cycle revisited: interannual variation and ecosystem consequences. Progress in Oceanography 49, 283 –307. Birt-Friesen, V.L., Montevecchi, W.A., Cairns, D.K. & Macko, S.A. (1989) Activity-specific metabolic rates of free-living northern gannets and other seabirds. Ecology 70, 357–367. Born, E.W. & Böcher, J. (2001) The Ecology of Greenland. Atuakkiorfik Education, Nuuk. Bregnballe, T., Ensgtröm, H., Knief, W., Van Eerden, M., Van Rijn, S., Kieckbusch, J.J. & Eskildsen, J. (2003) Development of the breeding population of great cormorants Phalacrocorax carbo sinensis in the Netherlands, Germany, Denmark, and Sweden during the 1990s. Vogelwelt 124, 15 –26. Buckland, S.T., Anderson, D.R., Burnham, K.P. & Laake, J.L. (1993) Distance Sampling. Estimating Abundance of Biological Populations. Chapman and Hall, London. Cairns, D.K. (1989) The regulation of seabird colony size – a Hinterland model. American Naturalist 134, 141–146. Carscadden, J.E. & Vilhjalmsson, H. (2002) Capelin – what are they good for? ICES Journal of Marine Science 59, 863–869.

Carss, D.N., Bregnballe, T., Keller, T.M. & Van Eerden, M. (2003) Reducing the conflict between cormorants Phalacrocorax carbo and fisheries on a pan-European scale: REDCAFE opens for business. Vogelwelt 124, 299–307. Croll, D.A., Tershy, B.R., Hewitt, R.P., Demer, D.A., Fiedler, P.C., Smith, S.E., Armstrong, W., Popp, J.M., Kiekhefer, T., Lopez, V.R., Urban, J. & Gendron, D. (1998) An integrated approach to the foraging ecology of marine birds and mammals. Deep-Sea Research II (45), 1353–1371. Cuddington, K. & Yodzis, P. (2002) Predator–prey dynamics and movement in fractal environments. American Naturalist 160, 119 –134. Davoren, G.K., Montevecchi, W.A. & Anderson, J.T. (2003) Search strategies of a pursuit-diving marine bird and the persistence of prey patches. Ecological Monographs 73, 463 – 481. Durant, J.M., Anker-Nilssen, T. & Stenseth, N.C. (2003) Trophic interactions under climate fluctuations: the Atlantic puffin as an example. Proceedings of the Royal Society of London B 270, 1461–1466. Fauchald, P., Erikstad, K.E. & Skarsfjord, H. (2000) Scaledependent predator–prey interactions: the hierarchical spatial distribution of seabirds and prey. Ecology 81, 773 –783. FishBase (2004) http://filaman.uni-kiel.de. Frederiksen, M. & Bregnballe, T. (2000) Evidence for densitydependent survival in adult cormorants from a combined analysis of recoveries and resightings. Journal of Animal Ecology 69, 737–752. Friis-Rødel, E. & Kanneworff, P. (2002) A review of capelin (Mallotus villosus) in Greenland waters. ICES Journal of Marine Science 59, 890 – 896. Furness, R.W. & Birkhead, T.R. (1984) Seabird colony distributions suggest competition for food during the breeding season. Nature 311, 655 – 656. Grémillet, D., Dey, R., Wanless, S., Harris, M.P. & Regel, J. (1996) Determining food intake by great cormorants (Phalacrocorax carbo) and European shags (Phalacrocorax aristotelis) with electronic weighing units. Journal of Field Ornithology 67, 637– 648. Grémillet, D. (1997) Catch per unit effort, foraging efficiency and parental investment in breeding great cormorants (Phalacrocorax carbo carbo). ICES Journal of Marine Science 54, 635 – 644. Grémillet, D. & Argentin, G. (1998) Cormorants, shags and fisheries in the Chausey Islands area. Le Cormoran 10, 196 –202. Grémillet, D. & Wilson, R.P. (1999) A life in the fast lane: energetics and foraging strategies of the great cormorant. Behavioral Ecology 10, 516 –524. Grémillet, D., Wilson, R.P., Wanless, S. & Peters, G. (1999a) A tropical bird in the Arctic (the cormorant paradox). Marine Ecology Progress Series 188, 305 –309. Grémillet, D., Wilson, R.P., Gary, Y. & Storch, S. (1999b) Three-dimensional space utilization by a marine predator. Marine Ecology Progress Series 183, 263 –273. Grémillet, D., Wanless, S., Carss, D.N., Linton, D., Harris, M.P., Speakman, J.R. & Le Maho, Y. (2001) Foraging energetics of arctic cormorants and the evolution of diving birds. Ecology Letters 4, 180 –184. Grémillet, D., Wright, G.A., Lauder, A., Carss, D.N. & Wanless, S. (2003) Modelling the daily food requirements of wintering great cormorants: a bioenergetics tool for wildlife management. Journal of Applied Ecology 40, 266–277. Grémillet, D., Chauvin, C., Wilson, R.P., Le Maho, Y. & Wanless, S. (2005) Unusual feather structure allows partial plumage wettability in diving great cormorants. Journal of Avian Biology 36, 1 – 7. Härkönen, T. (1986) Guide to the Otoliths of the Bony Fishes of the North East Atlantic. Danbiu ApS, Hellrup. Harris, M.P. & S.Wanless (1993) The diet of shags Phalacrocorax aristotelis during the chick-rearing period assessed by three methods. Bird Study 40, 135 –139.

801 High foraging performance at low prey densities

© 2004 British Ecological Society, Functional Ecology, 18, 793–801

Hislop, J.R.G., Harris, M.P. & Smith, J.G.M. (1991) Variation in the calorific value and total energy content of the lesser sandeel (Ammodytes marinus) and other fish preyed on by seabirds. Journal of Zoology 224, 501–517. Hoff, G.R. (2000) Biology and ecology of threaded sculpin, Gymnocanthus pistilliger, in the eastern Bering Sea. Fishery Bulletin 98, 711–722. Houston, A.I., McNamara, J.M., Heron, J.E. & Barta, Z. (2003) The effect of foraging parameters on the probability that a dive is successful. Proceedings of the Royal Society of London B 270, 2451–2455. Hunt, G.L., Heinemann, D. & Everson, I. (1992) Distributions and predator–prey interactions of macaroni penguins, Antarctic fur seals, and Antarctic krill near Bird-Island, South Georgia. Marine Ecology Progress Series 86, 15 –30. Johnsgard, P.A. (1993) Cormorants, Darters, and Pelicans of the World. Smithsonian Institution Press, Washington, DC. Johnson, M.T.J. & Agrawal, A.A. (2003) The ecological play of predator–prey dynamics in an evolutionary theatre. Trends in Ecology and Evolution 18, 549 –551. Lewis, S., Sherratt, T.N., Hamer, K.C. & Wanless, S. (2001) Evidence of intra-specific competition for food in a pelagic seabird. Nature 412, 816 – 819. Lorentsen, S.-H., Grémillet, D. & Nymoen, G.-H. (2004) Does annual variation in breeding Great Cormorant (Phalacrocorax carbo carbo) diet reflect varying stock recruitment of gadoids? Waterbirds 27, 161–169. Magill, S.H. & Sayer, M.D.J. (2002) Seasonal and interannual variation in fish assemblages of northern temperate rocky subtidal habitats. Journal of Fish Biology 61, 1198 – 1216. Mehlum, F., Hunt, G.L., Klusek, Z. & Decker, M.B. (1999) Scale-dependent correlations between the abundance of Brünnich’s guillemots and their prey. Journal of Animal Ecology 68, 60 –72. Menge, B.A., Daley, B.A., Wheeler, P.A., Dahlhoff, E., Sanford, E. & Strub, P.T. (1997) Benthic–pelagic links and rocky intertidal communities: bottom-up effects on top-down control? Proceedings of the National Academy of Sciences of the USA 94, 14530 –14535. Naef-Daenzer, L., Naef-Daenzer, B. & Nager, R.G. (2000) Prey selection and foraging performance of breeding great tits Parus major in relation to food availability. Journal of Avian Biology 31, 206 –214. O’Donoghue, M., Boutin, S., Krebs, C.J., Zuleta, G., Murray, D.L. & Hofer, E.J. (1998) Functional responses of coyotes and lynx to the snowshoe hare cycle. Ecology 79, 1193 –1208. Patterson, H.D. & Thompson, R. (1971) Recovery of interblock information when block sizes are unequal. Biometrika 58, 545 –554. Pulido, F.J. & Diaz, M. (1997) Linking individual foraging behavior and population spatial distribution in patchy

environments: a field example with Mediterranean blue tits. Oecologia 111, 434 – 442. Schall, R. (1991) Estimation in generalised linear models with random effects. Biometrika 78, 719 –727. Schmid, D., Grémillet, D. & Culik, B. (1995) The energetic costs of under-water swimming in cormorants. Marine Biology 123, 875 – 881. Shertzer, K.W., Ellner, S.P., Fussmann, G.F. & Hairston, N.G. (2002) Predator–prey cycles in an aquatic microcosm: testing hypotheses of mechanism. Journal of Animal Ecology 71, 802–815. Siorat, F. (2003) Faisabilité de la mise en place d’un protocole de suivi à long terme des populations de poissons littoraux subtidaux de l’archipel des Sept-Iles, Côtes d’Armor. Ligue Pour le Protection Des Oiseaux LN-0203 –06, unpublished report. Stephens, D.W. & Krebs, J.R. (1986) Foraging Theory. Princeton University Press, Princeton, NJ. Tokeshi, M. & Romero, L. (1995) Quantitative analysis of foraging behavior in a field population of the South American sun star Heliaster helianthus. Marine Biology 122, 297–303. Vucetich, J.A., Peterson, R.O. & Schaefer, C.L. (2002) The effect of prey and predator densities on wolf predation. Ecology 83, 3003 –3013. Wellenreuther, M. & Connell, S.D. (2002) Response of predators to prey abundance: separating the effects of prey density and patch size. Journal of Experimental Marine Biology and Ecology 273, 61–71. Whitlatch, R.B., Hines, A.H., Thrush, S.F., Hewitt, J.E. & Cummings, V. (1997) Benthic faunal responses to variations in patch density and patch size of a suspension-feeding bivalve. Journal of Experimental Marine Biology and Ecology 216, 171–189. Willis, T.J., Millar, R.B. & Babcock, R.C. (2000) Detection of spatial variability in relative density of fishes: comparison of visual census, angling, and baited underwater video. Marine Ecology Progress Series 198, 249 –260. Wilson, R.P., Grémillet, D., Syder, J., Kierspel, M.A.M., Garthe, S., Weimerskirch, H., Schafer-Neth, C., Scolaro, J.A., Bost, C.A., Plötz, J. & Nel, D. (2002) Remote-sensing systems and seabirds: their use, abuse and potential for measuring marine environmental variables. Marine Ecology Progress Series 228, 241–261. Wilson, R.P. (2004) Reconstructing the past using futuristic developments. Trends and perspectives using logger technology on penguins. Memoires of the National Institute of Polar Research Tokyo 58, 34 – 49. Yoshida, T., Jones, L.E., Ellner, S.P., Fussmann, G.F. & Hairston, N.G. (2003) Rapid evolution drives ecological dynamics in a predator–prey system. Nature 424, 303–306. Received 20 January 2004; revised 27 May 2004; accepted 1 June 2004