Northern Prairie Wildlife Research Center
We found that 2 independent indices to suites of life history characteristics (these can be equally considered as indices of r and K-selection) varied together. Average values indicated that pintail, mallard, blue-winged teal, and shoveler have the most features associated with unstable or unpredictable environments. Gadwall American wigeon, and green-winged teal were intermediate, and attributes of the diving ducks were associated with the use of stable or predictable environments.
The conclustions of 3 other studies are consistent with ours. For one, Patterson (1979) suggested that high correlation coefficients between duck numbers and pond numbers in the Canadian prairie and parklands indicated an opportunistic response to habitat conditions associated with r-selection. He also determined correlations between changes in duck numbers and the previous population count, large positive correlations interpreted as evidence for competitive self-regulation associated with K-selection. He concluded that, among these 10 duck species, the mallard, blue-winged teal, pintail, and shoveler were r-selected dabblers. American wigeon gadwall, and green-winged teal were K-selected dabblers, and redhead, canvasback, and lesser scaup were further along the spectrum toward K-selection.
Bailey (1981) compared mallard and canvasback life histories with data gathered in Saskatchewan by Stoudt (1971). He concluded that the mallard was more of an r-selected species than the canvasback in that the mallard responded more opportunistically to improving habitat and accepted a greater risk of reproductive failure.
Finally, Vickery and Nudds (1984) tested for density-dependent relations in duck populations on study areas in Alberta and Saskatchewan. They found that gadwall, northern shoveler, redhead, canvasback, and lesser scaup showed some density dependence, characteristic of K-selection. The remaining 5 species did not.
It would be worthwhile also to array the 10 species according to other associates of r- and K-selection, notably longevity and reproductive effort. Species considered more r-selected tend to have shorter lifespans but higher fecundity; K-selected species typically live longer but produce fewer young (e.g., Pianka 1974).
A comparison of ecological longevity among duck species is best made in terms of annual survival rates, which can be estimated from long-term banding studies. D. H. Johnson, J. D. Nichols, and M. D. Schwartz (Breeding dynamics of ducks, unpubl. manuscript presented at Ecol. and Manage. of Breeding Waterfowl Symp., Winnipeg, Manit., Canada, 18-22 Aug 1987) reviewed studies that used modern statistical methods of estimating survival rate. They found some variation among species, but not all of the 10 species had been studied, and estimates available pertained to different geographic areas, time periods, and banding seasons, making them difficult to compare. Moreover, all studied populations had been subjected to recreational hunting, so that estimates may no longer reflect survival rates under conditions to which the species had adapted.
Comparisons of reproductive effort among species are often made on the basis of age at first reproduction and clutch (or litter) size. For waterfowl, total number of eggs laid during a breeding season is more relevant than clutch size because it takes differing renesting probabilities into account.
We reviewed Palmer (1976) and Bellrose (1980) for information on age at first reproduction. The 4 species we indicated were most r-selected (pintail, mallard, blue-winged teal, and shoveler) typically breed as yearlings, as does American wigeon, an intermediate species. Of the other 2 intermediate species, the gadwall often defers breeding until its second year, and no information was presented on the green-winged teal. Large numbers of the 3 most K-selected speciesócanvasback, redhead and lesser scaupódo not breed until they are 2 years old. Thus, age at first reproduction parallels the other attributes of rand K-selection that we examined.
Total eggs laid in a breeding season is more difficult to address because it involves clutch size and the number of nesting attempts made. Further, clutch size typically declines with later nesting attempts, and both clutch size and probability of making a nesting attempt depend on the condition of the breeding bird and its habitat (D. H. Johnson, J. D. Nichols, and M. D. Schwartz, Breeding dynamics of ducks, unpubl. manuscript presented at Ecol. and Manage. of Breeding Waterfowl Symp., Winnipeg, Manit., Canada, 18-22 Aug 1987). Keith (1961) estimated for an Alberta study area both clutch size and numbers of nests per female. We used the product of those 2 values as an estimate of total eggs laid (except for the canvasback, which had limited sample size, and pintail and American wigeon, which Keith thought were not adequately represented). For the 7 remaining species, total estimated egg production was correlated with our index of r- and K-selection (r = 0.705, P = 0.077; Table 5).
The analyses we present, the other evidence we review, and the reports of Patterson (1979), Bailey (1981), and Vickery and Nudds (1984) do not justify any causal implications, but suggest an association among certain life history features, including the predictability of favored habitats.
Effective management of waterfowl populations is enhanced by a basic understanding of the distributions of the birds and the factors that affect those distributions. On a continental basis, for example, much effort is expended in attempts to estimate the size of each year's breeding population of key species of ducks. Waterfowl surveys to obtain such estimates are labor-intensive and costly. Conversely, extensive habitat information may be gathered inexpensively with satellite imagery, as a by-product of other activities. By knowing how ducks distribute themselves in response to habitat conditions, especially those of wetlands, surveys could be designed in a more optimal fashion. Sampling intensity could vary annually to place more effort in strata where the habitat information predicts the presence of more ducks, for example. Results presented in this report will be useful in planning extensive waterfowl surveys, but they also warn that the 10 species vary, so that an optimal design must be a compromise among the species of interest.
Even without new information, such as satellite imagery providing habitat conditions, an understanding of the factors that determine duck distributions can improve estimates of duck numbers. Johnson (1981, 1986) showed how auxiliary information could be employed to develop estimators, known as empirical Bayes estimators, that averaged 20% more accurate than customary estimators. Auxiliary information took 1 of the following 3 forms: counts of a given species in a particular stratum during previous years, the pond count in that stratum during the year of the survey, and the total number of ponds with water within the pond-surveyed area. Although the previous count was generally an effective form of auxiliary information, estimates for some species in some strata were better with either the local or continental pond count. These differences among species and strata corresponded to findings of this report.
One of the main purposes of estimating breeding populations of ducks is to predict the fall flight so that hunting regulations can be established consistent with anticipated numbers of available ducks. For any species, the fall flight depends on the size of the breeding population and its productivity. Productivity in turn depends on several factors and is variable from area to area and from year to year. Again, an understanding of the distribution of breeding ducks is important in predicting their productivity. For example, greater productivity may be expected if most mallards are dispersed across breeding habitats in the prairies and parklands, and if pond conditions there are good, than if they are displaced to more northern strata. Results in this report can be used to refine predictions beyond such obvious generalities.
On a long-term basis, there is much interest in knowing whether populations of ducks are increasing, fluctuating about some long-term mean, or declining. Determining trends is not as straightforward as it would seem because of the variable distributions of the species and the fact that numbers of ducks are related to numbers of ponds with water. Analyses like those included here can help clarify the situation. A study of mallard trends was presented by Johnson and Shaffer (1987).
Considering more local situations, such as a wetland manager might deal with, it is equally important to understand how ducks distribute themselves. This information will help assess the effects of on-the-ground management practices. For example, it may be feasible to attempt to develop a large breeding population of diving ducks or some other species with high rates of homing by attracting pairs and ensuring high production. Surviving adult females and many of their female offspring would be likely to return to the area, if the breeding habitat were stable from year to year. This effort would be less likely to succeed if directed toward northern pintails or blue-winged teal, however, because of their reduced homing tendencies. The development of attractive wetlands in southern parts of the breeding range would more likely attract these species.
Beyond their efforts to improve duck habitats and populations, managers can fill a vital scientific role. Most of their management activities can be viewed, not only as a positive action for the ducks, but as a scientific test. Theories such as presented here generate predictions of responses to specific changes. They can be applied to the manager's actions and results used not only to evaluate the impact on the duck populations, but also to evaluate the scientific theory. We hope that this report not only provides some guidance for managers, but stimulates actions that will test its predictions.