USGS Home
Contact USGS
Search USGS

A. Methods Common to Most Studies of Feeding Ecology

Analysis of esophageal food samples is currently the preferred method of studying the diets of prebreeding and breeding waterfowl. The advantages of esophageal samples have been demonstrated with field (Bartonek and Hickey, 1969; Sedinger, 1986) and experimental (Swanson and Bartonek, 1970) studies. Alternatively, analysis of fecal samples can provide valuable data on diets of grazing species when the collection of specimens is not possible (Owen, 1975a).

Once food samples have been obtained, most researchers record the frequency of occurrence and weight or volume of each type of food in each sample. Dry weight is preferred for the latter measurement because wet weight and volumetric data complicate the interpretation of nutrient and energy intake (Sugden, 1973; Reinecke and Owen, 1980).

Frequency data from food samples generally are used to analyze food selection, and weight or volumetric data are used to estimate diet composition and interpret nutritional relationships. Frequency data can be used to assess food selection only if food resources are sampled when birds are collected. If this is done, food selection can be inferred from differences in the frequencies of occurrence of various food types between samples from the birds and samples from the feeding sites. Readers should consult Thomas and Taylor (1990 and references therein) for guidance on experimental design. It is probably best to interpret the results of these tests conservatively, because strong evidence for food selection requires unbiased samples of the food available to foraging birds. Obtaining unbiased samples is difficult, especially in studies where ducks feed on active organisms in complex aquatic habitats.

Two methods currently are used to estimate diet composition (Swanson et al., 1974). In the aggregate weight or volume method, the percentage of each food type in the diet is estimated as the cumulative weight or volume of that food in all samples expressed as a percentage of the cumulative weight or volume of all foods in all samples. In the aggregate percent method, the percentage of each food in the diet is estimated as the average over all birds of the weight or volume of that food in each sample expressed as a percentage of the total weight or volume of the corresponding sample. If the aggregate percent method is used and each bird represents an independent observation, then standard parametric (e.g., Reinecke and Owen, 1980) and nonparametric (e.g., Drobney and Fredrickson, 1979) statistics can be used to test for differences in diet between sexes or among groups differing in reproductive status.

The aggregate percent method of estimating diet composition has several advantages: (1) it prevents large food samples with rare food types from unduly influencing estimates of diet composition (Swanson et al., 1974); (2) it avoids the restrictive assumption that all foods are digested at similar rates (Swanson and Bartonek, 1970); and (3) it facilitates statistical testing. However, the aggregate percent method also makes an assumption: that composition of the diet is unrelated to the size distribution of food samples. Generally, the aggregate percent method seems more appropriate when food-processing rates limit food intake, and the aggregate weight or volume method seems more appropriate when time spent searching for prey limits food intake.

B. Methods Specific to Studies of Geese and Other Herbivorous Waterfowl

Researchers studying the feeding ecology of geese and other herbivorous waterfowl encounter special problems but also have unique opportunities. One problem concerns the interpretation of laboratory analyses of the protein content of plant tissues. Traditional proximate analyses estimate the percentage of crude protein in a sample as the percentage of nitrogen multiplied by 6.25 to account for the average nitrogen content of protein (Robbins, 1983). Crude protein overestimates the potential availability of plant protein to geese and other herbivores, however, because stems and leaves of herbaceous plants contain significant quantities of nonprotein nitrogen, mostly in refractory structural compounds. Sedinger (1984) made independent estimates of nitrogen and protein in foods of the Cackling Canada Goose and found that crude protein overestimated true protein by 22-52%. Future studies should adopt similar procedures or at least consider the implications of biased estimates when crude protein data are used to assess the availability of protein in plant tissues.

Another problem involving the chemical characterization of plant foods is the analysis of cell structural compounds. Traditional proximate analyses provide an estimate of crude or acid detergent fiber, which consists of cellulose and lignin (Robbins, 1983). In the past, most researchers (e.g., Reinecke and Owen, 1980) accepted the conclusion (Mattocks, 1971) that cellulose is indigestible, and assumed that the energy in crude fiber was unavailable to waterfowl. However, recent studies have shown that herbivorous waterfowl can metabolize 25-74% of the hemicellulose in plant foods (Buchsbaum et al., 1986; Dawson et al., 1989; Sedinger et al., 1989). In contrast, data regarding digestibility of cellulose or acid detergent fiber are inconclusive (cf. Buchsbaum et al., 1986; Sedinger et al., 1989). Clearly, the ability of geese to utilize cell structural compounds appears to be an important area for future research.

Although cell structural compounds present problems, they also provide a unique opportunity to study the nutrition of free-living geese. If it is assumed that crude fiber, lignin, or another indigestible substance can be used as a tracer in the food and feces, traditional methods of estimating nutrient retention rates for captive birds can be used in the field. This approach, which has been explored more by European (Ebbinge et al., 1975) than by North American researchers, requires that the birds being studied: (1) feed on land and produce feces that are easily collected; (2) eat one or only a few plant species; and (3) forage in a uniform habitat patch long enough to produce feces representing the local food source. If these conditions are met, food and feces can be collected and analyzed, and digestibility estimated, with standard methods (Robbins, 1983). Using this approach, Madsen (1985a) estimated that Pinkfooted Geese in Denmark were able to digest 24-29% and 53-64% of the dry matter in pasture grasses and barley grain, respectively.