A Review of the Diet and At-sea Distribution of Penguins Breeding within the Camlr Convention Area

Managing fisheries in an ecologically sensitive manner demands that catches do not depress stocks within the foraging areas used by predators to levels that reduce their reproductive success or survival. Spatially–explicit bioenergetics models that estimate the amount of prey consumed by predators are required to inform such policy. These models require information on the number of predators in a population, their nutritional demands, their diet composition and their seasonal distribution in the marine environment. This paper reviews all published information on the diet and at-sea distribution of the six penguin species that breed in the CAMLR Convention Area, the methods used to collect these data and the uncertainties inherent in them. The review will be of utility to modellers as a source of parameters, and to penguin biologists by providing comparative information for their own findings and by highlighting significant gaps in existing knowledge and methods that could be used to fill these. Ratcliffe and Trathan 76 The target audience of the review are penguin biologists and ecological modellers. Penguin biologists will find the information in the review useful for comparison with the findings of their own studies, to identify important gaps in current knowledge and to determine the most suitable methods that could be used to fill such gaps. Ecological modellers will be able to extract values of interest from the review and gain an improved understanding of the uncertainties that may be associated with these. Those readers wishing to extract large numbers of diet values may find it easier to do so from the online spreadsheet compiled by Raymond et al. (2011). Estimating diet composition Quantification of prey remains in stomach contents Most diet studies of penguins are based on examination of the contents of the oesophagus, proventriculus and gizzard, typically referred to collectively (though anatomically inaccurately) as stomach sampling. The contents can be accessed by dissection, administration of emetics or by stomach flushing (also termed as water offloading, stomach pumping or gastric lavage). Dissection involves killing birds and cutting open the stomach to extract its contents; this is the most thorough method as all contents can be extracted (Croxall and Furse, 1980; Croxall and Prince, 1980; Volkman et al., 1980). This was only used in early studies of diet during the 1970s, and most researchers have subsequently considered it unjustifiable on ethical grounds. Emetics were also used in early studies (Jablonski, 1985; Volkman et al., 1980) but were ineffective and their use has been discontinued. Almost all studies since the mid-1980s have used stomach flushing, in which a tube is inserted down the oesophagus, down which warm water is poured until the stomach is full. The bird is then inverted over a bucket in which the regurgitated stomach contents are collected (Gales, 1987; Wilson, 1984). This procedure is often repeated for a fixed number of times or until only clear water is returned, depending on the study. The method does cause birds distress, and in exceptional cases results in death of a study animal, but it is less intrusive than killing birds for dissection. Food recovered from stomachs is typically filtered through sieves with increasingly fine meshes to recover solid constituents, and prey remains are identified, counted and weighed. The main source of uncertainty is that prey are often heavily digested when recovered, which may cause problems with identification and estimation of their original mass upon ingestion (Gales, 1987). However, some parts of prey types resist digestion better than others and have sufficiently unique characters to allow identification to the level of genus or species. The beaks of squid, eye-capsules of crustaceans and sagittal otoliths of fish are particularly widely used in this respect. Counting these allows numbers consumed to be estimated, and original size of prey at digestion can be estimated from equations describing relationships between the size of the body parts and the whole body size, obtained from measurements of whole specimens captured in trawls (Hill et al., 1996). The resulting corrections from such analyses are usually referred to as reconstituted mass, as opposed to wet mass measured from raw stomach contents. Problems with this approach still remain, since erosion of these parts may result in underestimates of their numbers and size, and because these parts are digested at differing rates. For example, fish components are digested particularly rapidly, while squid beaks are extremely resistant to digestion and can accumulate in stomachs for several days or even weeks, resulting in overestimates of the squid component (Pütz, 1995; van Heezik and Seddon, 1989; Wilson et al., 1985). Most of the recent studies overcome the latter problem by only counting those squid beaks still attached to the buccal mass. Data derived from stomach content studies can be presented in three ways. The simplest method is percentage occurrence, which is the percentage of stomachs in which at least one prey item of a given type is present. This tends to exaggerate the importance of small food items that tend to occur in most stomachs but are never significant in terms of numbers or mass. Percentage by number is the percentage of food items of a given type in a stomach, and is usually averaged across stomach samples. This tends to exaggerate the importance of numerous but small items and underestimates the importance of large items, which are inevitably less frequent owing to limitations of stomach volume. Percentage by wet mass or reconstituted mass are the percentage of the total mass of stomach contents comprised by a given prey type, and is the only measure that is of use in prey consumption