Temporal and intrapopulation variation in prey choice of wintering geese determined by stable isotope analysis



    1. Division of Environmental and Evolutionary Biology, Institute of Biomedical & Life Sciences, Graham Kerr Building, University of Glasgow, Glasgow, G12 8QQ, UK; Queen's University Belfast, School of Biology and Biochemistry, Medical Biology Centre, 97 Lisburn Road, Belfast BT9 7BL UK;
    Search for more papers by this author

    1. Division of Environmental and Evolutionary Biology, Institute of Biomedical & Life Sciences, Graham Kerr Building, University of Glasgow, Glasgow, G12 8QQ, UK; Queen's University Belfast, School of Biology and Biochemistry, Medical Biology Centre, 97 Lisburn Road, Belfast BT9 7BL UK;
    Search for more papers by this author

    1. NERC Life Sciences Mass Spectrometry Facility, Scottish Universities Environmental Research Centre, Rankine Avenue, East Kilbride, G75 0QF, UK;
    Search for more papers by this author

    1. Wildfowl and Wetlands Trust, Castle Espie Ballydrain Road, Comber, County Down, BT23 6EA, Northern Ireland, UK; and
    Search for more papers by this author

    1. R.S.P.B. Northern Ireland Headquarters, Belvoir Park Forest, Belfast, BT8 4QT, Northern Ireland, UK
    Search for more papers by this author

    1. Division of Environmental and Evolutionary Biology, Institute of Biomedical & Life Sciences, Graham Kerr Building, University of Glasgow, Glasgow, G12 8QQ, UK; Queen's University Belfast, School of Biology and Biochemistry, Medical Biology Centre, 97 Lisburn Road, Belfast BT9 7BL UK;
    Search for more papers by this author

    1. Division of Environmental and Evolutionary Biology, Institute of Biomedical & Life Sciences, Graham Kerr Building, University of Glasgow, Glasgow, G12 8QQ, UK; Queen's University Belfast, School of Biology and Biochemistry, Medical Biology Centre, 97 Lisburn Road, Belfast BT9 7BL UK;
    Search for more papers by this author

R. Inger, Division of Environmental and Evolutionary Biology, Institute of Biomedical & Life Sciences, Graham Kerr Building, University of Glasgow, Glasgow, G12 8QQ, UK. Tel.: +44(0)141 3302430. Fax: +44(0)141 3305971. E-mail: inger@udcf.gla.ac.uk


  • 1Individual variability in prey preferences can have marked effects on many demographic parameters from individual survival and fecundity to the vital rates of entire populations. A population level response is ultimately determined by individual prey choices; however, the effect of individual dietary choice is often overlooked.
  • 2We determined prey choice by individual consumers, light-bellied Brent geese Branta bernicla, during the overwintering period. Two hundred and eighty-one individuals were sampled at distinct temporal points over two winters. Stable isotopic ratios of carbon and nitrogen for blood cells and blood plasma, from each sampled individual were measured. Isotopic ratios for potential prey items were also measured.
  • 3δ15N and δ13C for blood samples were both significantly different between sample months. Generally we found a decrease in both isotopic ratios during the course of the winter. All potential prey items were also isotopically distinct. Multisource mixing models (isosource) were used to determine the range of possible contribution to the diet of individuals.
  • 4During early winter, diet consisted almost exclusively of sea grass Zostera spp. The level of Zostera spp. in the diet dropped until mid-winter, and was supplemented by the utilization of green algae Ulva lactuca, and Enteromorpha spp., and terrestrial grasses. Terrestrial grass comprised an increasing proportion of the diet in late winter, representing virtually the exclusive food source by April.
  • 5By examining intrapopulation variability in resource utilization we highlight a number of ecologically important factors not addressed by previous population level studies.


Individual variability in diet choices is a common feature among animal populations and may explain a significant amount of variation in a suite of life-history and demographic parameters (Sutherland 1996; Krebs & Davies 1997; Schmitz, Beckerman & Litman 1997). Optimal foraging theory suggests consumers will forage in such a way as to maximize net rate of energy intake (Krebs & McCleery 1984). Hence, when faced with the choice between prey species (all other things being equal), consumers should select the one that will deliver highest energy intake and therefore maximum fitness (Charnov 1976). Although initially thought to be of trivial importance (Pyke, Pulliam & Charnov 1977), recent empirical studies have shown that factors such as nutrient content (McKay, Bishop & Ennis 1994; Hassall, Riddington & Helden 2001), digestibility (Prop & Vulink 1992), or a combination of these factors (Amano et al. 2004; Durant, Fritz & Duncan 2004) may substantially affect resource selection. In addition to intrinsic properties of prey items, other nondietary factors including, predation risk (Inger et al. 2006) disturbance (Madsen 1988; Béchet, Giroux & Gauthier 2004) and local prey density, in terms of biomass in a patch (Vickery et al. 1995) may have a strong effect on dietary choice. Indeed changes in biomass by depletion of preferred prey items can cause animals to switch habitats in search of alternate resources (Rowcliffe et al. 2001), and may be the only factor required to explain habitat switches (Vickery et al. 1995).

Generally, studies of resource usage consider conspecifics, within the same population, as ecological equivalents (Bolnick et al. 2003), and thus ignore any between individual variation. Hence patterns of resource utilization often describe dietary choice only at the population level (Durell 2000; Bolnick et al. 2003). However, consideration of average levels of resource utilization within a population may conceal underlying individual variation, which may have important ecological and evolutionary implications (Smith & Skulason 1996; Bolnick et al. 2002; Bolnick et al. 2003; Urton & Hobson 2005; Bearhop et al. 2006). Indeed many populations that appear to be composed of generalists actually comprise a range of individual specialists (Bolnick et al. 2002). Furthermore, demographic segregation in resource utilization, and the subsequent effects of population dynamics (Newton 1998) will be overlooked by population level studies, but may be quantified by examining intrapopulation variation (Bolnick et al. 2002). Decisions made by individuals ultimately determine the population level response; therefore, it is highly desirable that diet selection studies measure the diet of individuals.

Until recently, methodological constraints often rendered it impractical to determine directly the diet of individuals. Direct observation of individuals’ habitat utilization provides limited information on actual diet, as the community composition of a habitat may not reflect prey selection (Rowcliffe, Sutherland & Watkinson 1999; Stephens et al. 2003). The analysis of gut content, regurgitates or faeces have serious limitations (Bearhop et al. 1999; Votier et al. 2003), and may be biased to particular prey types (Hobson, Piatt & Pitocechelli 1994; Russell et al. 1996). In addition these methods only provide data on resource selection during the study period, or immediately prior to analysis and are typically hard to quantify in a truly objective manner.

However, the measurement of stable isotopic ratios of consumer tissues and food offers an alternative approach and has advanced our knowledge of feeding ecology (e.g. Gannes, Martínez del Rio & Koch 1998; Bearhop et al. 1999; Podlesak, McWilliams & Hatch 2005). Consumer tissues are ultimately derived and dietary sources, and are integrated into tissues in a predictable manner, providing a direct record of dietary choice (Hobson & Clark 1992; Matthews & Mazumder 2004; Bearhop et al. 2002 Bearhop et al. 2004). In particular, there are marked differences in the ratios of the stable isotopes of both carbon (13C and 12C) and nitrogen (15N and 14N) between marine and terrestrial biomes. Thus measuring these two isotope ratios (expressed as δ13C and δ15N) in the tissues of consumers can provide an accurate assessment of marine vs. terrestrial prey in the diet (Bearhop et al. 1999; Ben-David, Titus & Beier 2004). Temporally, incorporation of dietary isotopic ratios is a function of the metabolic turnover rate of the tissue. Hence tissues with different turnover times record diet over different temporal periods.

Recent advances in stable isotope mixing models (Phillips & Gregg 2003) allow us to ascertain the range of contributions of multiple food sources to a consumer's diet. These methods provide a powerful tool in nutritional ecology and have recently been utilized to provide unique insights into the foraging ecologies of a range of animals from crustaceans to mammals (Ben-David et al. 2004; Newsome et al. 2004; Abed-Navandi & Dworschak 2005; Urton & Hobson 2005).

Here we apply these techniques to evaluate the feeding ecology of wintering East Canadian light-bellied Brent geese Branta bernicla hrota (O.F. Müller), wintering at Strangford Lough, Northern Ireland. The population, along with other Brent goose populations, has dramatically increased in size over the last few decades (Robinson et al. 2004). This rise has in turn led to increased depletion of the preferred prey, a marine angiosperm, Zostera spp. (R. Inger, in press), and has led to the birds seeking alternative food sources in different habitats (Mathers & Montgomery 1997) including agricultural land (Merne et al. 1999). This pattern of habitat usage has occurred in other populations of Brent geese in the UK, often causing conflict with agriculture (Charman 1979; Tubbs & Tubbs 1982; Summer & Critchley 1990; McKay et al. 1993; Vickery et al. 1995). Reduced persecution and the ability to adapt to novel resources are probably the key factors underpinning the recent rises in population sizes (Inger et al. 2006). However, the underlying ecological mechanisms driving variability/changes in habitat utilization remain unclear. Models predict that resource guarding has a strong effect on goose aggregations (Rowcliffe et al. 1999), and hence we suspect that the presence of social dominance hierarchies within goose populations may have a powerful influence on the timing and extent of resource utilization. Furthermore, we predict dominant individuals and groups will have greater access to preferred resources in support of the competition hypotheses proposed to explain differential foraging behaviour among individuals (Monaghan 1980; Goss-Custard et al. 1982; Ekman & Askenmo 1984; Gustafsson 1988; Koivula et al. 1994).

In this study we use stable isotopic ratios coupled with multisource mixing models to quantify the prey choices of individuals throughout the overwintering period. We propose that, in order to explain resource utilization, a number factors in addition to maximizing energy intake must be evoked.


study area

Six sites around Strangford Lough, County Down, Northern Ireland, UK, were used to sample both Brent geese and potential food sources. This 150 km2 marine inlet is an internationally important site for a number of wintering wildfowl species, including the light-bellied Brent goose. The intertidal zone of the northern Lough consists of extensive mudflats, which contain large beds of the marine angiosperms Zostera marina L. and Zostera noltii (Hornemann). The green algae Enteromorpha spp. and Ulva lactuca L. are also abundant around the shoreline of the Lough. Much of the surrounding county is agricultural pastureland, principally perennial rye-grass Lolium perenne L., and Timothy Phleum pratense L., collectively referred to in this paper as terrestrial grasses.

study population

The East Canadian High Arctic population of light-bellied Brent geese Branta bernicla hrota, spend the winter in western Europe, with the majority of the 30 000 individuals distributed around the coast of Ireland. In the early winter about 75% of the population pass through Strangford Lough, with numbers peaking about mid-October (O’Brian & Healy 1991; Merne et al. 1999). Here they utilize the large stocks of Zostera spp., depleting it rapidly between October and December. The birds remaining at Strangford Lough increasingly utilize Enteromorpha spp. and Ulva lactuca (Mathers & Montgomery 1997), and since the mid-1970s Brent geese have also been found feeding inland on agricultural land (Merne et al. 1999), mainly on, terrestrial grasses. In addition birds occasionally feed on discarded potatoes from local agriculture. All data were collected between October 2003 and October 2005.

sample collection

Birds were captured in cannon nets, individually marked with coloured leg rings, sexed by cloacal examination, and morphometric measurements taken. In addition we took blood samples from the caudal tibial vein, which were subsequently separated into plasma and cells by centrifugation. Brent goose plant food sources, Zostera spp., Enteromorpha spp., Ulva lactuca and terrestrial grasses were collected from a total of 14 sites around the Lough, and stored at −20 °C until used for stable isotope analysis.

stable isotope analysis

All samples were freeze-dried and ground into a homogenous powder, before being weighed into tin cups for analysis. Analysis was carried out at the East Kilbride node of the Natural Environment Research Council Life Sciences Mass Spectrometry Facility. Stable carbon and nitrogen isotope measurements were carried out using continuous flow isotope ratio mass spectrometry (CF-IRMS), using a Costech (Milan, Italy) ECS 4010 elemental analyser interfaced with a Thermo Electron (Bremen, Germany) Delta Plus XP mass spectrometer. Stable isotope ratios are reported as δ-values and expressed in , according to the following: δX = [(Rsample/Rstandard) − 1] × 1000, where X is 13C or 15N and R is the corresponding ratio 13C/12C or 15N/14N, and Rstandard is the ratio of the international references PDB for carbon and AIR for nitrogen. Replicate analyses of internal lab standard gelatine and alanine during measurements reported in this paper yielded standard deviations better then 0·24 for δ15N and 0·04 for δ13C. Actual mass, and hence percentage carbon and nitrogen were also recorded.

isosource modelling

Mixing models can be used to determine the isotopic contribution of food sources in a consumer's tissues and thus determine the relative contributions to the consumer's diet. For n isotopes, a unique solution can be calculated for up to n + 1 sources. Where this constraint is not met, as in our example where we have two isotopes and up to five sources, a most-likely solution must be estimated. The program isosource (Phillips & Gregg 2003) allows the possible contribution of additional sources using an iterative approach to calculate all possible feasible solutions for any give sources and isotopic mixtures, generating a distribution of feasible solutions. We used the mean δ15N and δ13C values for Brent goose blood plasma and blood cells sampled in different months of the winter in the analysis. Values were first adjusted for trophic fractionation (3·2 for δ15N and 1·45 for δ13C) using fractionation values determined from literature sources (Bearhop et al. 1999; Hobson & Bairlein 2003; Evans Ogden, Hobson & Lank 2004) based on captive birds studies before input to isosource. The δ15N and δ13C values for plant samples were input as sources. A sensitivity analysis was carried out on the fractionation factors. We varied the value used for δ15N between 3 and 5, and the value used for δ13C between 1 and 2. Where a feasible solution could be found model outputs varied by a maximum of 5% for the mean contributions to the diet.

The results from isosource enable us to quantify the proportion of different food sources in the Brent goose diet at different times of the year. As the two tissue types, blood cells and plasma, used for stable isotope analysis have different turnover rates, they provide information over two temporal periods. Plasma generally has a half life of 3 days, while avian blood cells may have a half life of up to 4 weeks (Hobson 2005). Therefore, the plasma records dietary intake over the previous few days, while the cells provide information over the previous few weeks. Each sampling event therefore provides dietary information over two temporal periods. In addition differences between isotopic ratio of plasma and cells from an individual provides a measure of any recent changes in diet.

statistical analysis

All tests were carried out in SPSS Version 12·0 (SPSS Inc., Chicago, IL). All data were tested for departures from normality using Kolmogorov–Smirnov tests, and for homoscedasticity using F-tests or Levene's tests. Non-normal data were either log or rank transformed (Iman 1974; Conover & Iman 1981; Iman, Hora & Conover 1984), as outlined in Zar (1999) before analysis. Stable isotope ratios were analysed using general linear models (type III SS). Model selection was achieved by fitting the maximal model and subsequent hierarchical backwards elimination of nonsignificant factors until a model containing only significant terms was created. Scheffe tests were used for post-hoc analysis. In order to better visualize the stable isotope data by month we assign 95% confidence ellipses to the data and to the mean (Fig. 1a,b) of the bivariate data comprising δ15N and δ13C using the methods of Sokal & Rolf (1995).

Figure 1.

δ15N and δ13C for all individuals sampled separated by month, for results from cells (a) and plasma (b). Solid lines indicate 95% confidence interval of the mean, dashed lines indicate 95% confidence interval of the data.


sampling of birds

We captured and marked 406 birds in nine winter months during the 2-year study period, at five sites around Strangford Lough and one site south of Strangford Lough (January 2005) (Table 1).

Table 1.  Birds caught between October 2003 and April 2005
Oct 0326  9  5  4  5  4  9 5644
Oct 05 7–9 30 29  1 15 15 30 9750
Nov 0417–19 10  6  4  5  5 10 6050
Dec 0415–16 43 43  0 20 22 4110051
Jan 0412–14 30  9 21 12 18 29 3060
Jan 0528 90 67 23 37 53 20 7459
Feb 0417–19125 64 61 63 62 74 5150
Feb 0518 32 22 10 18 14 32 6944
Apr 05 8–10 37 31  6 17 16 36 8443
Total 406276130192209281  
Mean        6950

While we would have ideally sampled birds during the same period in each of the study years the stochastic behaviour of birds and logistic consideration made this impossible. Results are presented by date during the course of the overwintering period, in order to clearly demonstrate dietary changes throughout the winter, which is broadly consistent across years. Where birds have been caught twice in the same month, the birds caught earliest in that month are displayed first.

In total 68% of birds captured were adult and the sex ratio was 52·7% females/47·3% males. However, the age composition differed between catches, from 30% adult in January 2004 to 97% adult in October 2005, and 100% adult in December 2004.

stable isotope analysis

Brent geese

Of all the captured individuals, 281 were blood sampled and blood separated into plasma and cells and analysed for stable isotopic ratios, producing 562 data points each for δ15N and δ13C. We found highly significant differences between months for blood cells δ15N (F[8,272] = 54·5, P < 0·001) and δ13C (F[8,272] = 422, P < 0·001) (Fig. 1a) and blood plasma δ15N (F[8,272] = 93·4, P < 0·001) and δ13C (F[8,272] = 467, P < 0·001) (Fig. 1b). For plasma δ15N we found an initial increase from October to December followed by a decrease from December to April A similar pattern was found in blood cells, which peaked in February, then decrease until April In the case of δ13C there was a general decrease throughout the year from October to April for both blood plasma and cells. Sample year was removed from models due to nonsignificance.

The age of individuals, either adult or juvenile, had a significant effect on δ13C in both blood plasma (F[1,271] = 13·1, P < 0·001) (Fig. 2a) and blood cells (F[1,271] = 9·63, P = 0·002) with δ13C consistently higher in adults than juveniles. δ15N was also higher in adults than juvenile blood cells (F[1,271] = 10·7, P = 0·001) and blood plasma (F[1,271] = 5·78, P = 0·017) (Fig. 2b). We found no significant differences between the sexes for δ15N plasma (F[1,240] = 0·36, P = 0·72) and cells (F[1,240] = 0·24, P = 0·79), or for δ13C plasma (F[1,240] = 0·35, P = 0·71) and cells (F[1,240] = 0·225, P = 0·799).

Figure 2.

δ13C (a) and δ15N (b) values form blood plasma. Filled circles represent adults, empty circles represent juveniles. Error bars represent 95% confidence intervals. No juveniles sampled in December 2004.

food sources

Mean values of δ15N were lowest for terrestrial plants. Terrestrial grasses (4·43, SD = 2·27) had the lowest values of all plants, follow by potatoes (6·32, SD = 0·04), Zostera spp. (6·49, SD = 1·46), Enteromorpha spp. (9·82, SD = 0·83) and Ulva lactuca (11·20, SD = 1·11). With regards to δ13C, terrestrial plants were considerably lower than marine plants. Terrestrial grasses were the lowest (−30·88, SD = 0·64), followed by potatoes (−25·96, SD = 0·01), Enteromorpha spp. (−14·06, SD = 1·17), Zostera spp. (−11·17, SD = 1·21) and Ulva lactuca (−11·17, SD = 1·96). Significant differences were found between the plants samples for δ15N (F[4,40] = 12·2, P < 0·01), and δ13C (F[4,39] = 546, P < 0·01).

isosource modelling

Mean isotopic data from each month was used in isosource models. The results are summarized in Fig. 3. isosource models were highly conclusive for the two main dietary components, Zostera spp. and terrestrial grasses, producing a narrow distribution of possible source contributions for each month of the winter, providing a highly specific description of dietary preference during the study period.

Figure 3.

Box plot representing the range of possible proportions of Zostera ssp. (a), terrestrial grasses (b), Ulva lactuca (c), and Enteromorpha spp. (d) in the diet of Brent geese, during the winter months. Boxes indicate interquartile range, bars indicate the range of feasible solutions as determined by program isosource (Phillips & Gregg 2003). Filled boxes indicate results from cells, empty boxes indicate results from plasma.

Over the annual cycle, the results represent a decrease in the use of Zostera spp. throughout the winter with a concurrent increase in the utilization of terrestrial grass, with the green algae comprising the remainder of the diet, particularly in mid-winter. The decrease in the utilization of Zostera spp. is particularly abrupt in December and January

isosource results for birds sampled in October show that Zostera spp. constitutes the majority of the birds’ diet. It was impossible to calculate any feasible solutions for blood cells form birds sampled in early October. This is likely to be due to the fact that at least some of the period of synthesis was in Iceland, hence the cells will reflect the isotopic signatures of Icelandic food sources. In contrast, isosource results for blood plasma probably represent purely Irish intertidal feeding, with Zostera spp. representing between 93 and 95% of the diet. Results from birds caught in late October show Zostera spp. comprises 76–87% of the diet in late September and early October, while plasma results show that by late October Zostera spp. represents 73–75% of the birds’ diet. The remainder of the diet consisted of terrestrial grasses (cells 14–19%, plasma 8–9%), U. lactuca (cells 4–13%, plasma 8–18%) and Enteromorpha spp. (cells 6–17%, plasma 11–24%).

The blood cells from birds sampled in November represent diet during the end of October and early November, and so temporally overlap the results from October plasma. Consequently, the Zostera spp. composition in the diet determined from November cells is virtually identical to that of October plasma, representing 72–75% of the diet, with terrestrial grasses making up the majority of the rest of the diet. That the results are so similar, even though birds’ were sampled in separate years gives us confidence in our methodology and that there is no major seasonal effect at this time of the year. In contrast, the results for November plasma show a marked drop in the levels of Zostera spp. by mid-November, this component is replaced in fairly equal parts by terrestrial grasses, U. lactuca, and Enteromorpha spp.

Blood cells collected in December, representing diet during late November and early December, show that Zostera spp. (64–71%) is still the major dietary component; however, by mid-December there is a marked dietary change, with levels of Zostera spp. falling to 3–13%, with concurrent marked increases in the other dietary components. The range of possible contributions for terrestrial grasses is well constrained (20–26%); however, the ranges for both U. lactuca (0–63%) and Enteromorpha spp. (0–82%) are less conclusive.

Results for January (2004) and December (2004), showing a large decrease in the proportion of Zostera spp. in the diet between later December and early January, with a consequent increase in terrestrial grasses, and to a greater extent green algae. It would appear that birds sampled in January 2004 were undergoing a similar dietary shift to birds sampled in December 2004, but later in the season. In addition to the four main food sources, geese in January 2004 were also utilizing discarded potatoes from local agricultural sites, accounting for 14–31% of the diet.

Birds caught is January 2005 had different dietary preferences to those caught in January 2004, with Zostera spp. only forming a minor part of the diet, and a much higher grass content. This is possibly a flock level effect due to localized Zostera stocks and subsequent depletion, resulting in an earlier move to terrestrial foraging.

By February of both years terrestrial grasses were the major diet component. In 2004, grasses represented 56–59% of the diet in early February, and 58% in mid-February; while in 2005, grasses were by far the largest constituent of the birds diet, represented between 77 and 83% in both early and mid-February.

Results from geese sampled in April show a similar, but slight increase in reliance on terrestrial grass, which formed between 89 and 95% of the diet in late March and early April, and between 90 and 91% by mid-April.

carbon and nitrogen content of plants

Standard methods for measuring protein content rely on determining nitrogen content by the Kjeldahl method (AOAC 1975). Here we use the percentage nitrogen obtained by mass spectrometry as a measure of relative protein content. The percentage nitrogen was significantly different between plants (F[4,40] = 7·31, P < 0·001). The lowest protein content as determined by percentage nitrogen was found in potatoes (0·57%, SD = 0·01), followed by Enteromorpha (1·39%, SD = 0·57), Ulva lactuca (1·92%, SD = 0·53), Zostera spp. (2·97%, SD = 0·97) and terrestrial grasses (3·55%, SD = 0·63). Post-hoc analysis revealed no significant differences between Zostera spp. and terrestrial grasses.

We found a strong relationship with percentage carbon values and published energetic values for Mathers & Montgomery (1997) (r2 = 0·96), and Charman (1979) (r2 = 0·93), therefore we used percentage carbon as a measure of energetic content. Percentage carbon was also significantly different between plants (F[4,40] = 12·2, P < 0·001). Enteromorpha spp. (18·4%, SD = 11·31) had the lowest carbon followed by Ulva lactuca (21%, SD = 3·8), potatoes (33%, SD = 3·75), Zostera spp. (36%, SD = 5·61), and terrestrial grasses (40·3%, SD = 3·8). No significant differences were found between Zostera spp., terrestrial grasses and potatoes.


A population level response is ultimately determined by the choices made by individuals. However, surprisingly few studies consider that individuals are not typically equivalent. Here we provide a rare demonstration of how determining dietary preferences of individuals and quantifying intrapopulation variations reveals patterns of resource utilization that are not apparent from population level results.

Habitat utilization and habitat switching have been well characterized in herbivorous wildfowl (Charman 1979; Tubbs & Tubbs 1982; Summer & Critchley 1990; Mckay et al. 1994; Vickery et al. 1995; Mathers & Montgomery 1997; Nolet et al. 2002). Previous work has concentrated on the habitat preferences of populations; however, the direct quantification of the dietary preference of individuals has not been previously attempted.

Here we apply recent advances in multisource stable isotope mixing models to define the dietary preference of individual animals. This approach has a number of advantages over traditional approaches. The stable isotope ratios of a consumer's tissue reflect its diet in a predictable manner, and represent actual assimilation rather than ingestion. The use of two blood components, red blood cells and plasma, with different turnover rates allows us to quantify the diet over two temporal periods. Blood plasma reflect diet over the previous hours and days, while blood cells represent the diet over the previous weeks, hence we have a continuous record of assimilation, rather than a snapshot of ingestion as provided by traditional methodology (Hobson 2005). Furthermore, by sampling individuals we are able to assign data to the demographic or social level. Consequently, we can quantify dietary choice and differential foraging patterns of groups within the population, rather than an overall population level response. While our methodology offers a number of advantages over traditional techniques it may not to suitable in all circumstances. In order to use stable isotope analysis the individuals must first be captured, which is not always possible. In addition in order to quantify diet, the components must be isotopically distinct.

At the population level, our study was in good agreement with previous studies recording a general movement from feeding in marine habitats early in the winter to terrestrial habitats as the winter progresses. However, in contrast to other studies, we found no evidence for an abrupt habitat switch (between marine and terrestrial habitats), but rather a gradual decrease in the utilization of prey type in one habitat, with a concurrent increase in prey from another habitat. When birds began to utilize a new habitat it is not necessarily coincidental with abandonment of previous habitats. Rather, birds utilized multiple prey types, and multiple habitats within the same temporal period. In addition, by quantifying intrapopulation variability, we identified demographic differences in timing and extent of resource use. Generally, the data describe a strong preference for Zostera spp. in early winter, which declines throughout the winter. By mid-winter, the green algae Enteromorpha spp. and Ulva lactuca became more important, although the results from isosource give a wide distribution of possible contributions to the diet, so that although the quantitative importance of these food sources remains unclear, it is likely to be variable. For the remainder of the winter terrestrial grasses became the most important food source in the diet, accounting for over 80% of the diet by April.

Perhaps of most interest is that not all individuals changed their foraging habitats in the same way. Juvenile individuals had significantly lower (more terrestrial) isotopic ratios than adults, suggesting that throughout the winter juveniles were utilizing terrestrial habitats earlier, and to a greater extent than adult birds. It may be argued that lower isotopic ratios of juveniles may, to some extent, be due to differential fractionation between adults and juveniles, which are still growing. However, this seems unlikely as when birds are purely feeding on Zostera spp. (October 2005) there are no significant differences in either isotopic ratio between adults and juveniles, indicating the absence of any age-related fractionation effect.

That juveniles appear to be utilizing terrestrial prey types earlier than adults, suggest that competitive ability is one factor that may produce patterns of variability in habitat use among individuals, or demographic groups. This is consistent with the finding of Tubbs & Tubbs (1982) who first reported terrestrial feeding in juvenile dark-bellied Brent geese. In this study it was found that terrestrial feeding was stimulated by the presence of juvenile birds in the flock.

Habitat segregation based on dominance status, often age related, is most frequently explained by the competition hypothesis (Monaghan 1980; Goss-Custard et al. 1982; Ekman & Askenmo 1984; Gustafsson 1988; Koivula et al. 1994) with subordinates unable to compete with dominant social groups for preferred resources. However, until now there was little empirical evidence in support of this hypothesis (Wiens 1991; Sol, Santos & Cuadrado 2000). Our results clearly show differential patterns of habitat usage and prey choice between age classes, and represent compelling support for the competition hypothesis.

The movement of individuals or groups of individuals between habitats and prey choice decisions are likely to be influenced by a number of factors, including the ability to maximize energetic intake, nutritional value of food, protein content (particularly for herbivores), and the suitability of the habitat with regards to predation risk and tradition. Previous studies have suggested a variety of mechanisms influencing diet selection in overwintering Brent geese. Probably the most influential factor is resource depletion (Vickery et al. 1995), which may cause birds to move habitats, although diet selection within habitats may be due to nutritional requirements (Summers et al. 1993; McKay et al. 1994; McKay et al. 2001).

At Strangford Lough, of the four major food types, our results suggest that, in terms of nutritional content, Zostera spp. and terrestrial grass should be the most desirable prey items. Our use of nitrogen and carbon content as proxies for protein and carbohydrate content are consistent with the values of Mathers & Montgomery (1997), who assessed the nutritional value of Brent goose food sources using standard techniques. Indeed, if geese were selecting prey items purely on nutritional content they would select terrestrial grasses over Zostera spp., with terrestrial grasses having a higher carbon and nitrogen content that Zostera spp. In addition, if diet choice were based on biomass we would expect terrestrial grasses to be the preferred dietary item. However, terrestrial grasses have a much higher fibre content, compared with marine food sources (Mathers & Montgomery 1997) decreasing digestibility, which (considering the poor digestive efficiency of geese), may be a major factor in prey selection. Factors other than the intrinsic properties of prey items are also likely to influence prey choice. Zostera spp. is the traditional prey choice for Brent geese, which have historically fed on Zostera spp. beds in preference to other food sources, with terrestrial feeding unreported until the 1970s (Charman 1979; Tubbs & Tubbs 1982). The risk of predation also has a strong influence on the distribution of Brent geese, with intertidal areas that harbour Zostera spp. beds being considered a lower risk habitat than terrestrial habitats (Inger et al. 2006).

That Zostera spp. is the preferred prey choice appears clear, as it constitutes the majority of the diet after migration from their Canadian breeding grounds. The transition from Zostera spp. to alternate food sources can be explained in terms of depletion. Empirical and modelling studies have found a strong relationship between bird numbers and Zostera spp. depletion (Inger et al. in press). In agreement with Vickery et al. (1995), depletion of biomass to levels where foraging becomes unprofitable is likely to be the main cause of changes in prey exploitation. Indeed density dependence appears to take effect very early in the winter, with levels of Zostera spp. falling significantly during October. However, we found no clear switch between habitats or prey items, more a gradual change in diet composition during the winter, the timing of which was strongly influenced demographic factors.

Previous studies have suggested that Brent geese move down a profitability gradient based on food preference or maximizing energy intake (Charman 1979; Mathers & Montgomery 1997). However, since geese started to utilize terrestrial grasses in the 1970s this resource has become increasingly important, representing the main food source during late winter and spring, and the main resource prior to migration and during spring staging in Iceland. Indeed the birds show substantial weight gains in spring due almost exclusively due to terrestrial foraging. Brent geese have clearly adapted well to terrestrial feeding and this is probably an important factor in recent increases in the populations of all races of Brent geese in western Europe. If we are to effectively conserve these populations and reduce the conflict with agriculture it is important that we make decisions based not only on conserving traditional intertidal feeding areas, but also areas of terrestrial grassland utilized by geese.

In conclusion we have demonstrated how resource utilization by individuals can be determined using stable isotope ratios coupled with multisource mixing models. These techniques require some a priori knowledge of the study system, and should therefore build on, not replace traditional methods for determining dietary choice. However, by accurately determining the dietary choice of individuals over predictable time-scales we are able to further our understanding of the complex relationships between consumers and resources, and how the choices made by individuals determine a population level response.


We thank Robin Ward, Kerry Mackie, Alex Portig, Graham McElwaine, and Hugh Thurgate for assistance in catching geese. We also thank, Seamus Magouran, Mark Ruddock, George Henderson, Lynne Tinkler, Gillian Robb and Kerry Crawford for additional field support. Birds were caught and ringed under licences from the Environment and Heritage Service, the British Trust for Ornithology. Isotope analyses were run at the NERC LSMSF via a grant-in-kind no. RK65-02/04. This work was partially funded by the Environment and Heritage Service (Northern Ireland), and the National Parks & Wildlife Service (Republic of Ireland). R.I. was funded by a N.E.R.C. studentship with additional C.A.S.E. award from the Wildfowl and Wetlands Trust. S.B. was funded by a N.E.R.C. postdoctoral fellowship.