The study was carried out in 1998 on the Isle of May, south-east Scotland (56°11′ N, 02°33′ W). Age and experience are tightly coupled in European shags. Changes in breeding performance in relation to age are manifest largely in differences between pairs containing a 2-year-old male breeding for the first time, and pairs containing a male older than 2 with at least 1 year's breeding experience (Potts et al. 1980; Aebischer 1993). Foraging performance was measured in males from these two experience classes (hereafter ‘naive’ and ‘experienced’). The age of 2-year-olds was known because they had been ringed as chicks and/or from their distinctive plumage characteristics (Potts 1971), and all were breeding for the first time (European shags never attempt to breed at 1 year). For the experienced group, individuals had been ringed either as chicks, in which case their exact age was known, or as breeding adults, thus their minimum age was known; all had at least 1 year's previous breeding experience. Although no independent effect of female age has been recorded in European shags, female and male age are highly correlated within pairs (rs = 0·56, P < 0·001; Daunt et al. 1999).
Naive pairs laid 12 days later than experienced pairs on average (experienced, 6 May ± 7 days, n = 38; naive, 18 May ± 6 days, n = 38; t74 = 7·54, P < 0·001). We manipulated the timing of hatching using a cross-fostering protocol (for full details see Daunt et al. 1999). A four-way swap was carried out, comprising two experienced and two naive pairs matched for clutch size and, within each experience class, for laying date (n = 17 swaps with a clutch size of three; n = 2 swaps with a clutch size of two). No pairs reared their own young, with control pairs receiving a clutch from the same experience class, and experimental pairs receiving a clutch from the other experience class. As a consequence, experimental naive and control experienced pairs hatched chicks (from eggs laid by experienced pairs) early in the season, while experimental experienced and control naive pairs hatched chicks (from eggs laid by naive pairs) late in the season.
Indirect evidence from the behaviour of European shags indicated that environmental conditions differed markedly early and late in the season, enabling us to examine whether the effect of experience on foraging performance varied with conditions. In the Isle of May breeding population as a whole (consisting of 621 pairs), some broods were left unattended by adults late in the season, an unusual event indicating that both parents had to forage simultaneously to obtain sufficient food for their offspring (28% of nests not part of the experiment left broods unattended late in the season; no records early in the season; n = 43).
Of the 76 nests in the experiment, foraging performance of 18 experienced males (10 early in the season, eight late) and 20 naive males (10 early in the season, 10 late) was recorded during the guard phase, approximately mid-way through chick rearing, using VHF telemetry. By alternating between the experience classes on a daily basis, the date that radio tracking took place did not differ between the experience classes (experienced early: 30 June ± 3 days; experienced late: 15 July ± 4; naive early, 30 June ± 3; naive late, 17 July ± 3; general linear model (GLM): experience, F1,35 = 0·21, P = 0·65; time of season, F1,36 = 30·56, P < 0·001; interaction term, F1,34 = 0·06, P = 0·81). Brood age at radio tracking (range 14–31 days) did not differ with parental experience, but was older late in the season because the spread of laying dates was narrower than that of radio tracking dates (GLM: experience, F1,35 = 0·63, P = 0·43; time of season, F1,36= 23·70, P < 0·001; interaction term, F1,34 = 0·59, P = 0·45). The two experience classes were matched for brood size, which declined late in the season (Generalized Linear Model (GLM) with binomial errors and a logit link function: experience, χ2 = 0·45, P = 0·50; time of season, χ2 = 3·96, P < 0·05; interaction term, χ2 = 0·03, P = 0·86). The combination of an increase in brood age and decline in brood size with time of season resulted in no difference in brood biomass, a determinant of foraging effort and food load in shags (Wanless, Harris & Russell 1993b; GLM: experience, F1,36 = 1·96, ns; time of season, F1,35= 1·43, ns; interaction term, F1,34 = 0·03, ns).
Males were caught and a VHF radio transmitter (Biotrack Ltd, Wareham, UK; mass 20 g, ≈1·5% of body mass) was attached to two central tail feathers with Tesa tape. Each bird was weighed and wing length (maximum flattened chord), tarsus length, and head and bill length were measured. We recorded no adverse effects resulting from handling. Tag attachment took place at dusk each day, and the first trip after dawn was tracked. We targeted the first foraging trip since it followed a long fast (European shags on the Isle of May do not feed at night during the breeding season) and is therefore likely to be more revealing of foraging performance, as both naive and experienced adults, and their brood, will have fasted for a similar period. Birds were radio tracked from a station at 73 m a.s.l., using a system consisting of two parallel eight-element Yagi aerials joined by a 2-m crosspiece, attached to a 5-m mast that allowed the aerials to rotate through 360°. The aerials were connected to an ATS R4000 scanning receiver, operating in the 173 MHz band. A typical foraging trip consisted of a flight out to the feeding site, a series of dives with periods between dives on the sea surface, and a return flight to the colony. From the strength and consistency of the signal, it is possible to determine a precise time–activity budget, namely whether the bird is flying (strong, continuous signal), swimming on the sea surface (unsteady, continuous signal), or diving (signal disappears; Wanless, Harris & Morris 1991; Wanless et al. 1993a). An estimate of foraging location is obtained from dead-reckoning, using the bird's bearing and flight duration (Wanless et al. 1991). Birds at this colony typically favour three main foraging areas, two areas inshore of the island along the mainland coast north and west to north-west, respectively, and close to the island in any direction (Wanless et al. 1991).
Where possible, birds were caught immediately on their return after the foraging trip (experienced early, n = 7; experienced late, n = 6; naive early, n = 6; naive late, n = 4) and the stomach contents obtained by flushing the stomach with water (under licence; for full details see Wanless et al. 1993b). No birds showed any adverse effects from this procedure.
Foraging efficiency was calculated as the ratio of energy gained (in kJ) to energy expended (in kJ) multiplied by the assimilation efficiency (77%; Grémillet, Schmid & Culik 1995).
Energy gained was obtained by calculating energetic content of food loads. For those containing lesser sandeels (A. marinus), the energy content was derived from the equation: energy (kJ) = 0·0031 length3·745 (Hislop, Harris & Smith 1991). For the remaining species, we multiplied the mass by an energy density of 5 kJ g−1 (Hislop et al. 1991). The energy content of the load was the sum of the prey species’ energy values.
We converted the time–activity budget to energy expended using activity-specific energy costs taken from the literature. Flight costs were calculated from Pennycuick (1989, updated: http://www.bio.bris.ac.uk/people/staff.cfm?key = 95). Wing span was calculated from wing length, using the equation wing span = (2·752 × wing length) + 0·360 (derived from breeding adult males in 1999, n = 18, r2 = 0·71). Aspect ratio was set at 6·85 (Pennycuick 1997), and air pressure at 1·23 kg m−3 (0·5 m a.s.l., the approximate flying height). The chemical power expended at a speed of 15·4 m s−1 (Pennycuick 1997) was used. Data on double-crested cormorants (Phalacrocorax auritus, a similarly sized species) taken from Enstipp, Grémillet & Jones (2006) were used for the costs of diving (26·22 W kg−1 for deep diving at 11 °C, the sea temperature during the study), incorporating the dive : pause cost ratio from great cormorants (Phalacrocorax carbo; Grémillet et al. 2003). The costs of swimming on the sea surface (12·73 W kg−1 in water at 11 °C) were taken from Enstipp et al. (2006).