Diet studies of seabirds: a review and recommendations Robert T. Barrett, Kees (C. J.) Camphuysen, Tycho Anker-Nilssen, John W. Chardine, Robert W. Furness, Stefan Garthe, Ommo Huppop, �� Mardik F. Leopold, William A. Montevecchi, and Richard R. Veit Barrett, R. T., Camphuysen, C. J., Anker-Nilssen, T., Chardine, J. W., Furness, R. W., Garthe, S., Huppop, �� O., Leopold, M. F., Montevecchi, W. A., and Veit, R. R. 2007. Diet studies of seabirds: a review and recommendations. ��� ICES Journal of Marine Science, 64: 1675���1691. We review the different methods that are used to collect dietary data from marine birds. We consider their limitations and practi- calities and emphasize critical data gaps in our knowledge of the feeding ecology of seabirds (na mely diets outside breeding seasons). To enhance comparability of findings among studies, species, and oceanographic regions, we make recommendations on standards for the reporting of results in the literature. Keywords: foodwebs, large predators, oceanographic comparisons, seabird diet sampling. Received 10 May 2007 accepted 15 September 2007 advance access publication 26 October 2007. R. T. Barrett: Department of Natural Science, Troms�� University Museum, NO-9037 Troms��, Norway. C. J. Camphuysen: Royal Netherlands Institute for Sea Research, PO Box 59, NL-1790 AB Den Burg, Texel, The Netherlands. T. Anker-Nilssen: Norwegian Institute for Nature Research, NO-7485, Trondheim, Norway. J. W. Chardine: Wildlife and Landscape Science Directorate, Environment Canada, PO Box 6227, Sackville, NB, Canada E4L 1G6. R. W. Furness: Institute of Biomedical and Life Sciences, Graham Kerr Building, University of Glasgow, Glasgow G12 8QQ, UK. S. Garthe: FTZ, University of Kiel, Hafento ��rn 1, D-25761 Bu ��sum, Germany. O. Hu ��ppop: Institute of Avian Research ���Vogelwarte Helgoland���, Field Station Helgoland, PO Box 1220, D-27494 Helgoland, Germany. M. F. Leopold: Wageningen-IMARES, PO Box 167, NL-1790 AD Den Burg (Texel), The Netherlands. W. A. Montevecchi: Cognitive and Behavioural Ecology Programme, Departments of Biology and Psychology, Memorial University of Newfoundland, St John���s, NL, Canada A1B 3X9. R. R. Veit: Biology Department, College of Staten Island, 2800 Victory Boulevard, Staten Island, NY 10314, USA. Correspondence to R. T. Barrett: tel: ��47 77645013 fax: ��47 77645520 e-mail: rob.barrett@tmu.uit.no Introduction Seabirds are integral components of marine ecosystems and, besides being the subject of general scientific interest, are excellent indicators of changes in the marine environment (Furness and Monaghan, 1987 Furness and Camphuysen, 1997). For example, seabird data give early indications of fluctuations in fish stocks and oceanographic conditions (Montevecchi, 1993 Frederiksen et al., 2004), and monitoring programmes for seabirds have been implemented in many parts of the world to investigate these relationships. In its broad sense, monitoring can be defined as ���the process of gathering information about system state variables at different points in time for the purpose of assessing system states and drawing inferences about change in state over time��� (Yoccoz et al., 2001). The systems of interest here are typically seabird populations, and the state variables include breeding population size, reproductive success, adult survival, and seabird diets, but can also include broader foodweb and ecosystem extrapolations. Many methods are used to study seabird diet. Some are based on opportunism whereby samples are collected ad hoc, e.g. from watching food uptake directly or by collecting dropped fish, regur- gitated food, or faeces. Others take a more systematic approach through regular collections or sightings made within a specified time. Techniques vary greatly and range from the direct killing of birds to inspect their stomach contents through to totally non- invasive and repeatable observations of fish-carrying birds. Indirect methods include observations of feeding flocks, analyses of faeces or regurgitated food remains, or tissue collection for stable isotope or fatty-acid analyses. All methods have, however, biases of one kind or other (Duffy and Jackson, 1986 Rodway and Montevecchi, 1996 Carss et al., 1997 Gonzalez-Sol��s �� �� et al., 1997 Andersen et al., 2004), and almost all methods and studies refer to the short breeding season when birds are readily accessible on or near land. When seabirds are not breeding and are dispersed along the coasts and over the open seas, there is no completely sat- isfactory non-destructive method for sampling their diets. Consequently, far too little is known about what and how much seabirds eat when they are at sea, or how the diets of immature birds and non-breeding birds compare with those of breeding adults or chicks. The variable approaches to diet sampling and the different formats of data presentation often make it difficult to assess shifts in diets over time or spatial patterns in the exploitation of particular prey. Consistency is required to allow comparisons of the size and energetic content of prey items to be made, and a detailed reporting of techniques (methods) used to calculate, for example, prey body size and weight from prey fragments is very important. There is also a need to be as clear and informative as possible with respect to taxonomy, a subject that is constantly being revised and refined. This review of diet sampling methods and our recommen- dations on how to report results in a standard manner are an out- growth of work conducted by the ICES Working Group on Seabird Ecology meetings in 2006 and 2007. We describe the methods used # 2007 International Council for the Exploration of the Sea. Published by Oxford Journals. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org 1675

to sample seabird diets and trophic relationships, and provide rec- ommendations for standardizing and enhancing the comparability of data collections and reporting. Dietary sampling methods Stomach sampling/regurgitations To assess the diets of seabirds directly, it is necessary to obtain or extract items from the digestive tract where they may be found in the oesophagus, crop, proventriculus, gizzard, or small intestine. Generally, the only items retained in the gizzard are hard parts such as bones, shells, exoskeletons, polychaete jaws, and squid beaks. Everything from the proventriculus to the oral cavity can often be sampled by lavage (see below) without harming the bird, whereas sampling the gizzard or intestine is only possible from dead, dissected specimens. Once the food samples are obtained, they can be sorted and identified, and measurements, e.g. weight, linear dimensions, volume, taken. Size of digested prey can often be estimated from measurements of undigested hard parts such as otoliths, bullae, bones, shells, and polychaete or squid beaks, but the accuracy depends greatly on the amount of digestion and wear of these items (see section on Pellets below). Dead birds Shooting birds at sea is one way of obtaining dietary data. Shooting has, however, the obvious limitation of killing the birds, which raises ethical issues, especially in relation to species of conservation concern. Moreover, a substantial fraction (often 30% or more of birds shot at sea) do not contain any food items other than bony fragments in the gizzard (RRV and G. L. Hunt, unpublished data from South Georgia and the Bering Sea). Therefore substantial numbers need to be shot to obtain an adequate sample. For these reasons, shooting is becom- ing increasingly unacceptable as a sampling tool. In addition, because many seabirds feed socially, shooting a sample of birds at a single location may give a misleading indication of diet that may vary between aggregations. Consequently, shooting as a sampling tool can rarely achieve a representative picture of the spatial and temporal variation in diet. However, birds shot for other reasons, e.g. for pollutant analyses, harvesting (such as the Newfoundland murre, Uria spp., hunt), or shot as pests (although those killed at, for instance, aquaculture sites may provide very biased data), have been used for diet studies (e.g. Rowe et al., 2000). Other sources of dead birds are oil spills, bycatches in fishing gear, and beached carcasses of oiled or wrecked birds (e.g. Blake, 1983 Lorentsen and Anker-Nilssen, 1999 Ouwehand et al., 2004), although birds from the last group have very often starved to death and yield few or biased data. Dead birds often arrive on beaches in a trickle, e.g. as a result of chronic oil pol- lution, or may hit a coastline en masse after an oiling incident, or a wreck following extended extreme weather (e.g. Stenhouse and Montevecchi, 1996). Such large-scale events should be seized for diet studies whenever possible, because they often provide large samples across a range of species from the same time and location (Ouwehand et al., 2004). In some oil spills, specimens may be sequestered for litigation purposes (Carter et al., 2003), and, unless sufficient excess material is available, they become impossible to access for years after the incident. Efforts should, however, be made to ensure that they can be used after litigation, because they are often then discarded. Once the birds are retrieved, the stomach or entire digestive track should be removed as soon as possible, and preferably frozen. Preservation in ethanol is a poor option because it leads to tissue discolouration, which can be problematic in identifi- cation of some small prey. The use of formaldehyde, even when buffered, is strongly discouraged owing to health concerns and because otoliths quickly dissolve in it. During the subsequent analysis, allowance must be made for the differential digestion of food items in different portions of the digestive track. Items in the crop can be near intact, but the further an item progresses through the system, the more it is digested and consequently the more difficult it may be to identify and measure accurately. Items in the gizzard may be retained for a considerable time some- times until they are forcibly regurgitated as a pellet (see below). Squid beaks or polychaete jaws, for example, may be retained for a month or longer (Jackson and Ryan, 1986 van Heezik and Seddon, 1989 Putz, �� 1995), so such retention needs to be taken into account when estimating dietary composition based on dis- sected dead birds and/or regurgitated samples. As the soft parts of squids or polychaetes are digested quickly, the beaks or jaws in the gizzard are often the only evidence of their presence in the diet. Using the number of these items in the gizzard will, never- theless, likely overestimate their proportional contribution because of their long retention times, so such counts need to be weighted (e.g. Duffy and Jackson, 1986). There is also need to con- sider interspecific variation in this. Some seabirds, such as gulls (Laridae) and skuas (Stercoraridae), empty the gizzard from time to time by regurgitating pellets of indigestible material, whereas others, such as procellariiforms, rarely do so, so may retain hard parts of prey in the gizzard for many months. Regurgitations Some birds, especially nocturnal petrels and shearwaters (Procellariidae and Hydrobatidae, respectively), when attracted to lights at night become disorientated and land on a ship���s deck or the ground. To lessen weight or as a panic response, they often vomit the contents of the upper digestive tract. At breeding colonies, stormpetrels can also be caught in mist-nests where they will regurgitate or can be induced to regurgitate (Montevecchi et al., 1992 Hedd and Montevecchi, 2006). Sampling this way can be especially valuable because it may be the only way to obtain dietary information from birds at sea and/or outside the breeding season. The problem with this technique of sampling outside the breeding season is that it is entirely opportunistic and dependent on weather conditions, because birds are much more likely to be attracted to lights during foggy, overcast, and/ or rainy weather. Nevertheless, such sampling can produce valu- able information on the food types available at prey patches at sea. Other species such as gannets (Sulidae), cormorants (Phalacrocoracidae), gulls, and terns (Sternidae) at or near the nest or on their way to feed chicks often regurgitate food held in the proventriculus if disturbed. Chicks may also spontaneously regurgitate in response to disturbance, or can be easily stimulated to regurgitate. Such samples are often only partly digested material and readily identifiable in the field (e.g. from gannets, cormorants) or on return to the laboratory (gulls, kittiwakes, Rissa spp.). Another advantage is that this type of sampling can be repeated (using the same or different birds) through the breeding season. Hard body parts (otoliths, bones, etc.) are also often not worn by digestion (although there are different digestion rates among opaque and hyaline otoliths Jobling and Breiby, 1986), so allowing reliable determinations of prey size. Note, however, that the pro- portion of ingested items in the regurgitations varies, so the 1676 R. T. Barrett et al.

amount regurgitated cannot be used as an estimate of meal size. Another limitation of this method is that the disturbance involved in some breeding colonies reduces the numbers of visits possible. There may also be biases where some food types are easy to regur- gitate whereas others are not. Stomach lavage, emetics If a bird does not regurgitate voluntarily, the upper digestive tract can be sampled without harming the bird by flushing the contents with water. This process, referred to as lavage, stomach flushing, or water offloading, involves pumping water through a tube inserted in the oesophagus of a bird and catching the regurgitated contents in a bag, sieve, or bucket (Wilson, 1984 Ryan and Jackson, 1986). A latex tube is inserted deep into the bird���s oesophagus, and water (preferably salt water) pumped (using a syringe) in the other end of the tube. If working in cold regions, the water should be slightly heated to avoid cold stress. The bird is then inverted over a suitable receptacle into which the water and stomach contents are emptied. The process may be repeated to ensure as complete an emptying of the gastric system as possible (Neves et al., 2006). Note that in some countries, the use of this method may require a licence. One limitation of lavage relates to how the birds are captured, because many birds vomit immediately upon being captured, so appear to be empty upon having their stomachs flushed. It has also proved difficult to use in some groups of seabird that do not regurgitate food to offspring, e.g. auks, though see Wilson et al. (2004). Birds do not always eject all the contents of the upper gut tract during lavage, and can sometimes be induced to do so using an emetic (Ryan and Jackson, 1986). Emetics can, however, increase the risk of harming birds, especially if used by inexperienced researchers, so their use is not recommended. Excrement Bird excrement has been used in various ways to reconstruct diets. Hard parts from prey, such as bones, scales, eggs or otoliths of fish, parts of the exoskeletons of crustaceans, squid beaks and jaws and setae of nereid worms, calcite plates and spines of echinoderms, or shell hinges of molluscs may all survive digestion and are often excreted. If such parts are recognizable and still bear a relationship with original prey size, they may be used to identify prey and reconstruct prey size. This method has been applied to many different mammalian piscivores, most notably pinnipeds and otters, Lutra lutra (Pierce et al., 1991 Tollit et al., 1996, 2004 Kingston et al., 1999 Andersen et al., 2004). Seabirds that excrete such remains are also candidates for similar studies and many have been carried out on omnivorous gulls and skuas (Andersson and Gotmark, �� 1980 Ambrose, 1986 Kubetzki et al., 1999 Kubetzki and Garthe, 2003), piscivorous ducks (Anatidae Rodway and Cooke, 2002), mollusc-eating seaduck (Swennen, 1976 Nehls, 1989 Nehls and Ketzenberg, 2002 Leopold et al., 2007), benthos-feeding waders (Scolopacidae Dekinga and Piersma, 1993 Scheiffarth, 2001), and other birds (e.g. Ormerod and Tyler, 1991 Taylor and O���Halloran, 1997). Relatively few such studies have, however, been carried out on other seabird taxa (e.g. terns, Veen et al., 2003 Stienen et al., 2007). Advantages of the method are that it is non-invasive and simple. Furthermore, large sample sizes can be processed and time-series built by repeated sampling schemes. Given that differ- ent methods often reveal different prey types, studying remains in excrement could reveal prey species previously unknown, e.g. Nereis jaws in sandwich tern (Sterna sandvicensis) excrement (Stienen et al., 2007). Genetic analyses of faecal or scat samples may also be used to identify the sex of the predator (Reed et al., 1997). Being widely used and with samples readily available, the method has also been extensively tested against other methods of diet study (Prime and Hammond, 1987 Dellinger and Trillmich 1988 Cottrell et al., 1996 Taylor and O���Halloran, 1997). Such tests have demonstrated that studies of excrement, like many other indirect methods covered here, are unlikely to reveal all prey taken by the predator. Some prey are easily fully digested, and some birds also use other means to rid themselves of prey hard parts, e.g. through regurgitation of pellets (see below). Moreover, some parts survive better than others and some prey may be completely overlooked or greatly underesti- mated. For example, sandeel (Ammodytidae) otoliths appear in the faeces of great black-backed gulls (Larus marinus), but otoliths of gadoid fish too large to pass through the intestine are voided in pellets (RWF, unpublished data). Another disadvantage is that excrement is unlikely to be collected at sea, unless a suitable plat- form on which they are deposited is available for sampling (Camphuysen and de Vreeze, 2005). Also, processing faecal samples can be unpleasant, although several washing methods have been developed (Bigg and Olesiuk, 1990), and estimating prey size from their remains is also time-consuming compared with measuring whole fish in a bird���s oesophagus. The identifi- cation of prey from their remains, be they faecal or regurgitates, requires good identification guides (Harkonen, �� �� 1986 Watt et al., 1997 Leopold et al., 2001) and/or reference collections. Pellets Several seabirds regurgitate indigestible prey remains in discrete pellets. These may be collected and the remains sorted, using methods similar to those described earlier. Pellet analysis has been used widely on cormorants and shags (Phalacrocorax spp. Kennedy and Greer, 1988 Barrett et al., 1990 Hald-Mortensen, 1995 Derby and Lovvorn, 1997 Gremillet �� and Argentin, 1998 Leopold et al., 1998 Olmos et al., 2000), gulls (Meijering, 1954 Spaans, 1971 Wietfeld, 1977 Garthe et al., 1999b Kubetzki et al., 1999 Kubetzki and Garthe, 2003), skuas (Votier et al., 2004, 2006, 2007), terns (Favero et al., 2000 Granadeiro et al., 2002 Veen et al., 2003 Bugoni and Vooren, 2004 Mauco and Favero, 2005), black skimmers (Rhynchops niger Naves and Vooren, 2006), and other birds such as waders, kingfishers (Alcedinidae), and dippers (Cinclus cinclus Swennen, 1971 Jost, 1975 Cairns, 1998). Being widely used and with samples readily available, particularly from cormorants, the method has been tested extensively both with captive birds fed known diets (Votier et al., 2001), and against other diet study methods (Brugger, 1993 Harris and Wanless, 1993 Russell et al., 1995 Trauttmansdorff and Wassermann, 1995 Zijlstra and van Eerden, 1995 Suter and Morel, 1996 Casaux et al., 1997, 1999 Votier et al., 2003). Like faeces collection, the method is non-invasive and simple and can provide large samples over time, although finding pellets is often restricted to breeding colonies or roosts. However, species such as gulls aggregate in mixed groups, especially at roosts and resting sites, so that pellets can sometimes not be allocated to a specific species. Pellets can be collected from any dry surface such as offshore lighthouses and platforms, or even specially designed floating pellet-collecting devices (Gagliardi et al., 2003), and are Seabird diet study review and recommendations 1677