Sedinger and Rexstad (1994) objected to the model selection procedures
in our analysis of mallard (Anas platyrhynchos) survival (Smith and R
eynolds 1992), especially in our analysis of the additive and compensa
tory nature of hunting mortality. Sedinger and Rexstad (1994) disagree
d with pooling data over age-sex classes and the averaging over bandin
g reference areas that was part of our ultrastructural analysis. We be
lieve these objections are unfounded and are based on a misinterpretat
ion of the principle of parsimony. Sedinger and Rexstad (1994) propose
d the addition of a density-dependent term to the ultrastructural mode
l to explain observed changes in survival patterns. Their analysis sug
gested, as we also believe, that density dependence may be a viable al
ternative. Certainly density-dependent mortality should be investigate
d further using appropriate mechanistic models. However, a comparison
of the 2 models indicates that the hunting hypothesis fits historical
data better, and therefore is the more appropriate model. We disagree
with Sedinger and Rexstad's (1994) call for implementing differing man
agement actions in the absence of supporting information.