Introduction

Territoriality is a common strategy in animals to ensure access to resources. Individuals or groups of individuals have exclusive access to food, favourable sites or mates, and banish others from resources (Maher and Lott 2000). The occupation of defended territories should guarantee sufficient food supply and lead to successful reproduction. In polar regions, feeding conditions for birds are highly variable; the scarcity or absence of the main food species can lead to complete reproductive failure, e.g. krill abundance and reproductive success of Wilson's storm-petrels (Oceanites oceanicus) (Quillfeldt 2001), lemming cycles and breeding success of rough-legged buzzard (Buteo lagopus) (Potapov 1997).

Seabirds as top predators in the marine food web defend breeding territories in the close vicinity of the nest but the defence of feeding territories at sea is unknown, perhaps because of ephemeral food patches. However, skuas, gull-like predatory seabirds, are unlike most other seabirds because, during their reproductive period, they may feed on terrestrial resources (Furness 1987). Skuas of the genus Stercorarius in the Arctic are known to occupy huge territories, defending them against intra- and interspecific competitors (e.g. territory size of ca. 200 km2 in Alaskan pomarine skuas Stercorarius pomarinus, Maher 1974). Southern hemisphere skuas (Catharacta sp.) prey on petrels, shearwaters and penguins nesting in large colonies on the sub-Antarctic islands and continental Antarctica (review in Reinhardt et al. 2000). In some Antarctic populations, some breeding pairs establish territories in these seabird colonies (Trivelpiece et al. 1980; Pietz 1987; Young 1994; Mougeot et al. 1998) whereas others do not forage in feeding territories (Parmelee 1992). This simultaneous occurrence of two foraging types, i.e. foraging with and without a feeding territory, in a population offers the unique opportunity for investigating their consequences on reproductive performance.

We define territoriality only in terms of feeding: breeding pairs were considered as territorial when they defended a feeding territory in one or more of the penguin sub-colonies, irrespective of whether this territory was in close vicinity to the nest or clearly separated from it (Trivelpiece et al. 1980). To identify the consequences of each foraging strategy on reproductive performance, we carried out an analysis in two stages. In the first stage, we characterised the feeding territories and quantified attributes of individual birds with regard to their feeding strategy. We acted on the assumption that territory ownership should be related to individual characters like body size, because the establishment and defence of (nesting) territories in skuas involve fighting (Higgins and Davies 1996). The quality of a feeding territory is a function of food availability and, thus, of the number of penguin nests within a territory. This number increases with territory size, but larger territories are more difficult to defend (Davies 1978). Therefore, territory size and the distance between nest and feeding site should be restricted. To ensure the advantage of occupied penguin nests in territories, territorial skuas should defend them vigorously throughout their reproductive season. In the second phase, we elucidate the consequences of each strategy on reproductive performance. As the establishment and maintenance of feeding territories is likely to be more costly than foraging without territory maintenance because of greater risk of injury and expenditure of time and energy (Calow 1998), we hypothesised that territory holders should achieve reproductive benefits. Therefore, we compared several aspects of reproductive biology between pairs with and without feeding territories: time of reproduction, parental nest attendance during incubation and chick rearing, chick survival, and breeding success. In birds, the onset of reproduction very often reflects food availability during the beginning of the reproductive season (Kelly and VanHorne 1997). Female skuas are stimulated to breed by courtship feeding of their males (Phillips and Furness 1998). Together with guaranteed food supply, females of pairs owning feeding territories should lay earlier and, consequently, their chicks should hatch earlier than in pairs without territories (Ens et al. 1992).

Skuas normally lay clutches of two eggs and are not able to increase the annual number of offspring by increasing egg number. Therefore, differences in reproductive output are caused by differential survival probabilities of eggs and chicks. During the incubation and chick-rearing period, parents have to guard their offspring against conspecifics, because intraspecific predation is a common cause of offspring mortality in Catharacta skuas (Lamey 1995). High parental nest-attendance rates are therefore crucial for successful offspring defence (Furness 1987), and high attendance rates reflect good foraging conditions or efficient foraging by the adults (Bukacinska et al. 1996; Catry and Furness 1999; Caldow and Furness 2000). If territoriality increases food availability for territorial owners, we expect higher nest-attendance rates and higher survival probabilities of their offspring in comparison to pairs without feeding territories. Therefore, we expected benefits in terms of the reproductive output in pairs with feeding territories compared to skua pairs without feeding territories.

Materials and methods

The study was carried out on Potter Peninsula, King George Island, South Shetland Islands between mid-December and the end of February in the Antarctic summers 1998/1999 and 2000/2001, including annual skua reproductive periods from late incubation to chick fledging. The ice-free area of the Potter Peninsula contained a population of brown skuas with 26–32 breeding pairs and up to 50 non-breeders (Hahn et al. 2003).

All breeding birds and a large number of non-breeders were captured by noosing their legs and marking by plastic bands with a letter code, which allowed for long-distance identification. On each bird captured, we measured total head length (including bill and cranium ±0.1 mm), tarsus length (±0.1 mm), wing length (±1mm) and body mass (±10 g). Individual body size was calculated as the PCA score derived from tarsus, head and wing length for each sex separately (Rising and Somers 1989; Phillips and Furness 1997). The first axis of the PCA explained 55.8% of total variance in females (n=67) and 58.7% in males (n=60). For body-size related comparison of body mass, we calculated body condition as the residual from a linear regression of mass against body size expressed as a percentage of the predicted value (Hamer and Hill 1993). The sex of marked birds was determined by DNA-analysis of blood samples (Fridolfsson and Ellegren 1999), taken by puncture of the brachial vein. To determine annual changes in status (breeder, non-breeder, mates or new pair establishment), we checked all nest sites at the beginning of every field season.

Characteristics of feeding territories and territory defence

The food of brown skuas on the Potter Peninsula contained more than 80% penguin components (Reinhardt 1997). The closest feeding ground was the mixed-species penguin colony at Stranger Point with 17,000–20,000 breeding penguin pairs (Adelie Pygoscelis adelie, gentoo P. papua, and a very small number of chinstrap penguins P. antarctica, reviewed in Hahn et al. 1998). The colony consists of several small sub-colonies (separate groups of nesting penguins) and one unitary colony with approximately 12,000 nests. The numbers of penguin nests in the sub-colonies were counted annually during the 3rd week of December.

We defined territorial skuas as breeding birds defending a particular feeding territory (FT) inside a penguin colony beside the nesting site. Birds that did not occupy such a feeding territory were defined as non-territorial (NFT no feeding territory), but they could defend nest sites. To determine the foraging strategy of each bird, i.e. FT or NFT, the locations of all skuas in the area of the penguin colony were mapped (i.e. look-outs, resting sites, feeding places). The territory borders were determined on the basis of 214–262 spot observations per field season, referring to 14–17 data points per territorial breeding pair and year. The number of penguin nests in a feeding territory was considered as an appropriate measure of territory quality. The distance between nest and feeding site was derived from GPS location data, for FT pairs, from nest site to the central point of the feeding territory, and for NFT pairs, from nest site to the centre of the main penguin colony.

Territory defence against intruders was experimentally measured using discovery time and intensity of aggression against a potential intruder—a free-standing museum specimen of a female brown skua placed in the central part of the feeding territory beside a penguin carcass. Discovery time was defined as the time a territory holder needed to locate the intruder. The power of aggression was scored as follows: 0 no attack, only threat, 1 hesitant plumage picking, 2 heavy push away, 3 strong physical aggression (head pecking), and 4 dive-bombing. The sex of the defender was also recorded. The experiments were repeated in eight territories at intervals of 3 weeks on 26.12.1999 (early chick stage, mean chick age 3 days), 14.01.2000 (late chick stage, 21 days old), and 05.02.2000 (fledgling stage, 42 days).

Time of reproduction

Laying dates in skuas reflect arrival times in the breeding area and nutritional status of females (Phillips and Furness 1998). We recorded the onset of reproduction by nest checks at 6-day intervals. The hatching date for 84% of the chicks was known exactly. For the remaining chicks, the hatching date was recalculated via the relationship between the density of the egg and hatching date (Furness and Furness 1981). The mean egg density on hatching was 0.89 g/cm3 and egg density (y) declined until hatching with y=(−8.9*10−5*x 2)−(9.8*10−3*x)+0.887 with x=days before hatching (R 2=0.78, df=215, P<0.001). To compare the timing of reproduction, all data were standardised according to within 1 year: y i=x i,real−x mean−i and between years: y iii=(x real−x mean−i)+x mean−iii (i=year, iii=3 years, x real=real date, x mean=mean of particular year) (Catry and Furness 1999).

Nest attendance and offspring survival

Parental nest attendance is considered to be a valuable measure for nest and offspring defence and reflected the trade-off between nest defence and foraging (Catry and Furness 1999). During the incubation and chick stage, parental nest attendance was recorded by regular nest checks at 6-day intervals. To compare the number of adults present with different foraging strategies, we calculated an annual average value for nest attendance during the chick stage (mean number of birds present per check). Chick survival was determined within 3 days (mean date between last positive and the negative nest check) and survival probabilities were calculated by Kaplan-Meier-survival analysis (SPSS 10.0) with territoriality as fixed factor and chick number as stratum.

Reproductive success

All eggs and chicks were marked individually. For a detailed quantification of reproductive performance, egg volume, clutch size, hatching and reproductive success were recorded. All eggs were measured (length and breadth) and their volumes calculated (Coulson 1963). Laying order was determined by hatching order or by volume due to egg loss before hatching (the larger egg assigned as the first egg in a clutch, Catry and Furness 1997). Chicks were assumed to have fledged if they survived until the end of the study period (age >50 days). For statistical comparison, data were analysed with GLIM (Crawley 1996).

Results

Approximately one-third of all breeding skua pairs at the Potter Peninsula occupied feeding territories (10 pairs in 1998, 9 in 1999, 9 in 2000), whereas 16–23 pairs did not defend feeding territories. One territorial pair failed to breed in 1999 and 2000, but nevertheless defended its feeding territory. Changes in status, e.g. NFT to FT and vice versa, were never recorded during the 3 years of study.

Females that occupied feeding territories were significantly larger than females of NFT pairs or non-breeding females (GLM: F=3.41, df=2.64, P=0.04), but in males no differences in body size were found (GLM: F=0.60, df=2.57, P=0.55, Fig. 1). Additionally, there was no correlation of body size of sexes in pairs (partial correlation controlled for territoriality: R=−0.20, df=1.36, P=0.23). Body condition differed between breeders (FT and NFT pairs) and NFT non-breeders (GLM: F=12.6, df=2.125, P=0.001); however, no difference was found between skuas of FT and NFT pairs (Bonferoni: t-test: P=0.74).

Fig. 1.
figure 1

Body size of females and males of brown skuas of different status (FT—territorial pairs defending a feeding territory in the penguin colony, NFT—pairs without a feeding territory, NB—non-breeders without feeding territories). Body size was calculated for each sex separately by a PCA of tarsus, wing and total head length. Size differences were significant in females (GLM: F=3.41, df=2.64, P=0.04), but not in males (GLM: F=0.60, df=2.57, P=0.55)

Full size image

Characteristics of feeding territory and territory defence

The majority of territories were all-purpose territories, including nest and feeding sites (n=6). Four territorial pairs defended feeding territories with nest and feeding sites separated. The distance between nest and feeding site differed greatly for owners of all-purpose territories (median: 252 m), pure feeding territories (median: 457 m) and pairs without feeding territories (median: 2,142 m; H=13.51, df=2.32, P=0.001). The territory size ranged from 48 to more than 3,000 penguin nests in 1–21 sub-colonies (Fig. 2). The only >3,000 nest-territory included 1 sub-colony and the northeastern margin of the main penguin colony. Seventy-six percent of territories contained fewer than 1,000 nests (median 421 nests). Thus, territory holders occupied 93% of all penguin sub-colonies and 1 side of the main colony, whereas NFT skuas (breeders and non-breeders) had free access to only 7% of the sub-colonies (620 penguin nests) and to the unoccupied main colony margin.

Fig. 2.
figure 2

Distribution (% of occurrence) of feeding territories with different size at Potter Peninsula. Territory size is given in categories of 500 penguin pairs. Territory size and number of occupied penguin sub-colonies were correlated (r 2=0.85, df=25, P<0.01; small graph)

Full size image

Territory owners defended the feeding territories during the whole field season (from incubation till fledgling stage), but the intensity differed considerably. An intruder was discovered after 13.2±14.6 min, irrespective of reproductive stage (RM ANOVA: F=0.65, df=2.23, P=0.54), but aggression intensity varied with reproductive stage (RM ANOVA: F=5.04, df=2.23, P=0.02) and averaged 1.9±1.6 and 2.0±1.8 in early and late chick stage, respectively. During the fledgling stage, territory holders attacked the intruder in none of the experiments, but observed it continuously (n=8). The initial defence was sex-specific, with males starting (77%), and females joining later (binomial-test: P=0.01, n=24).

Time of reproduction

Chicks from FT pairs hatched on average 5.5 days (range: 3–7 days) earlier than chicks from NFT pairs (GLM: F=9.50, df=1.124, P=0.003), irrespective of laying sequence (GLM: F=0.03, df=1.124, P=0.86) or year (GLM: F=0.19, df=2.121, P=0.82). Mean hatching date for chicks of FT and NFT parents was 27 December±6.2 (n=37) and 2 January±10.2 (n=88), respectively. Furthermore, the hatching of chicks from FT pairs was highly synchronised (19.7 days) compared to chicks from NFT pairs (35.7 days, comparison of annual standard deviations: t=−3.81, df=5, P=0.02). Laying dates were recalculated to the last week in November for pairs with feeding territory and to the first week in December for pairs without feeding territories (calculation with an average incubation period of 30 days (30.5 days Burton 1968; 30 days Williams 1980).

Nest attendance and chick survival

During incubation, both adults of FT and NFT pairs were present at the nest in 92% (n=51) and 78% (n=162) of all nest checks, respectively (Fisher exact: χ 2=4.98, P=0.02). During the chick stage, FT skuas guarded their offspring more often in pairs (83.3%, n=174 nest checks) than NFT pairs (65.5%, n=385, Fisher exact: χ 2=18.6, P<0.001). Therefore, the average nest attendance during the chick stage was 1.84±0.15 birds and 1.67±0.26 birds for FT and NFT pairs, respectively (GLM for territoriality: F=6.42, df=1.73, P=0.01, for year: F=0.33, df=2.68, P=0.77). Parental nest attendance was negatively correlated with the distance between nest and feeding area (R s=−0.35, df=73, P=0.003).

Chicks from pairs with feeding territories survived with higher probabilities than chicks from pairs without feeding territories in 2 of 3 years (1998: log-rank=4.84, df=1.32, P=0.03, 1999: log-rank=6.95, df=1.30, P=0.008, 2000: log-rank=0.01, df=1.42, P=0.96). The overall survival probability of chicks from FT and NFT parents was 71.4% (n=35) and 45.3% (n=75, log-rank=5.59, df=1.108, P=0.018). The difference was even higher in clutches of two because, in FT pairs, the second chick survived with the same probability as the first chick, whereas the survival probability of the second chick from NFT pairs decreased continuously at least to the age of 50 days (Fig. 3).

Fig. 3.
figure 3

Kaplan-Meier-survival probabilities of skua chicks from parents with different foraging strategies. Filled circles (±SE) are chicks from pairs with feeding territories; unfilled circles (±SE) are chicks from pairs without feeding territories. First chick refers to chicks of one-egg clutches and first chicks from two-egg clutches, second chicks are last-hatched chicks from two-egg clutches

Full size image

Reproductive success

Females with feeding territories generally laid larger eggs than females without feeding territories (ANCOVA with female size as covariate: F=7.07, df=1.150, P=0.009, irrespective of laying sequence: F=2.64, df=1.151, P=0.11). The mean difference in egg volume of FT and NFT females was 3.6 cm3 and 3.9 cm3 for the first and second egg, respectively. Egg volume and female body size were correlated (partial correlation controlled for laying sequence: R=0.17, df=149, P=0.02). There was no difference in clutch size in conjunction with foraging strategy (χ 2=0.14, P=0.91, Table 1). FT pairs suffered from high egg mortality/failure, resulting in an average hatching success of only 72.0% (n=50) compared to 81.3% (n=107) of NFT pairs (χ 2=8.16, P=0.02). Consequently, the reproductive success of birds owning feeding territories was not significantly greater than that of pairs without feeding territories (GLIM, all P>0.05, Table 1). Likewise, the pooled data of breeding success in 3 years of FT pairs (0.89 chicks per pair) did not differ significantly from NFT pairs (0.75 chicks per pair, GLIM: χ 2=0.44, df=86, P>0.05).

Table 1. Reproductive performance of brown skua pairs with feeding territories (FT) and without feeding territories (NFT) at Potter Peninsula/King George Island from 1998/1999 to 2000/2001. Data for clutch size and breeding success are means±SD; in 1998 one territorial pair had a clutch size of three. Hatching success is given as the percentage of hatched chicks per eggs laid; breeding success refers to number of chicks fledged per occupied nest
Full size table

Discussion

The harsh and highly variable environment of polar regions requires special adaptations in birds to survive and reproduce. Flying seabirds can ensure their food demand for maintenance and breeding by large feeding areas and long-distance foraging, i.e. albatrosses (Weimerskirch and Robertson 1994; Waugh et al. 2002). Another possibility to ensure sufficient food is to establish feeding territories in areas with terrestrial food. In the Antarctic, both skua species (brown skuas and south polar skuas, Catharacta maccormicki) have been observed to occupy feeding territories inside penguin colonies and defend them against competitors (e.g. Trivelpiece et al. 1980; Young 1994). We argue that territory establishment and maintenance depend on at least two aspects: the local form of penguin colony and the effective defence against conspecific competitors. The latter was quantified in this study: females owning feeding territories were larger than females without feeding territories, but no such differences were found in males. The result was not influenced by body weight, because body condition as a standardised measure of body weight by body size did not differ between both groups. Therefore, only females from pairs with feeding territories had an advantage during conflicts, because body size is a good predictor of fighting ability (Piper 1997). Consequently, these females should play a major role in territory maintenance and/or more successfully banish intruders from resources than males (S. Hahn, unpublished work).

The second assumption of a correlation between the type of penguin colony and the occurrence of feeding territories is obvious: spatially separated penguin sub-colonies are more accessible for predation than nests in the main colony, because predators may attack from many sides (Davis and McCaffrey 1986; Emslie et al. 1995). Large and unpartitioned penguin colonies are not effectively defendable by territory holders and, hence, act as a generally available resource for other (NFT) individuals (Stonehouse 1956; Sladen 1958; Burton 1968; Young and Millar 1999). Explicit recording of southern-skua foraging behaviour showed that feeding territories were only established in the presence of penguin sub-colonies (Table 2). The majority of feeding territories in our study contained fewer than 1,000 penguin nests with a maximum value of approximately 3,000 nests. Earlier studies reported skua feeding territories not to exceed 3,190 penguin nests (Trillmich 1978). The remarkable range for an upper threshold of 1,715–3,190 nests per territory (Table 2) might be ascribed to local differences in number and size of occupied sub-colonies. However, territorial skuas formed 28–67% of a breeding population, and in conjunction with a high degree of resource occupation (93% in our study), a substantial part of the population had to exploit alternative resources (i.e. fish, Parmelee 1992; Young 1994; krill or seal carcasses, Stonehouse 1956; Reinhardt et al. 1998).

Table 2. Occurrence of feeding territories in brown (Catharacta antarctica lonnbergi) and south polar skuas (C. maccormicki) and conjunction with penguin colony type used as feeding ground. Sub-colony: penguin colony divided into several groups of nesting penguins; type of feeding territory: APT all-purpose territory, FT feeding territory separated from nesting territory; territory size: number of occupied penguin nests, % within a population: percentage of territorial pairs within a particular breeding population (AP Antarctic Peninsula, KGI King George Island, RI Ross Island, SG South Georgia, VL Victoria Land, Antarctica, A Antarctica)
Full size table

Territory holders have to be able to defend their feeding territories as well as their offspring at all times. Territorial pairs with nests in the vicinity of the penguin colony (Burton 1968) saved the expense of flight in comparison to pairs without feeding territories and, additionally, they were able to permanently control their feeding territory. Consequently, individuals holding feeding territories, mainly the males, located intruders after almost constant discovery times throughout the whole chick stage. However, attack intensity declined towards the end of the reproductive period. During incubation and early chick stage, territory owners did not tolerate any foreign bird inside their feeding territories. In the middle of February, penguin fledglings are no longer available and territorial skuas decreased their effort in territorial defence and tolerated non-breeding skuas in abandoned penguin sub-colonies (Young 1963a). We therefore interpret the decreasing attack intensity rather as an adjustment in defence readiness due to lower territory profitability than as a seasonal decline in (nest) territory defence (Brunton 1990).

Reproductive performance

Abundant food supply during the pre-incubation period leads to an early breeding stimulation of females, and is reflected in early laying and larger eggs (Hiom et al. 1991; Ens et al. 1992). In our study, FT females laid on average 5.5 days earlier and produced larger eggs than NFT females, suggesting a clear advantage of FT pairs regarding food availability (see Catry and Furness 1997; Phillips and Furness 1998). Furthermore, the early onset of reproduction in skuas owning feeding territories was well synchronised with the breeding cycle of their major food source, penguin (Parmelee 1992; Young 1994).

FT pairs foraged in the vicinity of their nests and could, therefore, almost permanently guard their offspring, confirming the benefit of a short resource-nest distance and better food availability (Hamer et al. 1991; Bukacinska et al. 1996; Catry and Furness 1999; Caldow and Furness 2000). Consequently, chicks from FT parents survived with a higher probability, irrespective of laying order, in comparison to chicks of NFT pairs. However, the reproductive success of FT pairs was only slightly higher (difference of 0.14 chicks per pair per year) than the reproductive output of NFT pairs. This output would lead to a significant reproductive advantage of feeding territoriality only after 26 reproductive periods (estimation in GLIM, Poisson-distribution for χ 2-statistics, P<0.05). The reason for equal reproductive output was attributed to lower hatching success in pairs with feeding territories. Unfortunately, we had no exact information about their egg failures (e.g. addled eggs or egg predation by roaming non-territorial skuas).

Little is known about the lifetime reproductive success of skuas in conjunction with individual foraging strategies. The reproductive output of skuas with and without feeding territories is considerably different in different investigations: in within-year comparisons, FT pairs produced more fledglings (Young 1963b; Trillmich 1978; Trivelpiece et al. 1980), but the difference decreased in longer studies (Neilson 1983; this study). The equality in annual or even multi-annual reproductive success does not necessarily lead to equal lifetime reproductive success. If, for example, the annual chick provisioning by NFT parents is more costly, and NFT birds operate at their limit, the mortality in these birds should be higher than in adults with feeding territories. Hence, the reduced life span of NFT individuals would result in a lower lifetime reproduction success. In addition, Catry et al. 1998) found in great skuas (C. skua), that early hatching increased survival probability of fledglings at least in the 1st year of life, and early-laying birds had higher over-winter survival than late layers. If this applies to brown skuas as well, then we should expect birds owning feeding territories to produce more recruits on a lifetime scale and benefit on a broader scale from their predictable and stable food resource in the Antarctic.