which represents a small mass of digestible flesh, and a few large beetle mandibles,
each of which is equivalent to much more food. By searching the whole of the
channel for beetle mandibles, but only one in every six of the 10 degree sections
for chaetae and scaling up, it was possible to increase the efficiency of sample
processing (Green and Tyler 1989).
Some plant food items can be identified macroscopically, for example, from
seed coats, but many other items cannot. The plant epidermis is reasonably resis-
tant to herbivore digestion and so provides a more reliable measure. For example,
it has been used to quantify the fruit, leaf, and seed diet of Madeira Laurel
Pigeons Columba trocaz(Oliveira et al. 2002). The droppings can be dispersed
in water (a couple of drops of sodium hypochlorite helps clear the sample) and
examined under 10or higher magnification. Phase contrast illumination
microscopes are also useful for identifying plant epidermis because they allow
cell walls to be seen without having to stain the sample first.
A reference collection of potential prey is essential. Ideally watch the
species feeding and then catch prey within that habitat. Slides are prepared by
dissecting out the part, such as a mandible and placing within a drop of a moun-
tant such as Faure-Berlese solution (or Canada Balsam if the reference collection
is not permanent) on a slide and then placing a cover slip on top. It may be desir-
able to remove soft tissue with potassium hydroxide or a proteolytic enzyme
solution.
Nestlings typically have less efficient digestive efficiency, which makes the
identification of fragments in droppings easier. Nestling diet often differs from
adult diet, and if both are being studied, it is easiest to start with nestling droppings
to gain experience in identifying fragments.
A common mistake is to round up individuals per dropping and record one
mandible as one individual. This will greatly increase the estimated abundance of
rare prey. Five mandibles should thus be recorded as 2.5 individuals rather than 3.
A considerable problem is differential digestibility between prey species (e.g.
Tigar and Osborne 2000). Few remains of a species may be found either because
few were eaten or because their identifiable remains are usually digested. This is
best overcome by conducting calibration trials with captive birds. Feed a captive
animal on an identifiable food (e.g. poultry pellets, mealworms), then give a
known number of prey items (perhaps a range of species), record the number
uneaten, return to feeding with identifiable food, collect all droppings, and exam-
ine for prey remains. After a gap the experiment can be repeated. Stone Curlew
voided all prey within 24 h (Green and Tyler 1989), geese voided all food within
2 h (Marrion and Forbes 1970), and Knot Calidris canutusvoided all within 4 h
(Dekinga and Piersma 1993). The recovery rate is then calculated as the number
238 |Diet and foraging behavior