Modeling Predator–Prey Dynamics
279
Recommendations
My motivation for writing this review is to encourage wildlife ecologists to
begin focusing their research on testing ecological principles. When inconsistencies emerge between the models and field observations, the solution is not
to reject modeling and modelers, but to fix the models. Improved models will
emerge if field ecologists work with modelers to change models to accommodate the peculiarities of their study organisms.
An obvious area needing attention is the development of models for functional response. Although we know that functional response has important
consequences for population dynamics (Sherratt and MacDougall 1995), seldom do carnivore food habit studies relate results to the predictions of alternative functional response models. This seems to be particularly rich ground
for multispecies functional response models that are poorly understood in theory or practice. I am unaware of any models that go beyond linear substitutability among prey (Abrams 1987), yet I suspect that this assumption is
often violated.
Likewise, improved understanding of numerical response is essential to our
understanding of population dynamics. For equilibrium systems, Messier
(1994) presents a convincing illustration of how the combination of numerical and functional responses allow us to anticipate the role of predators in the
regulation of prey numbers. The outcome is largely a consequence of the
shapes of the numerical and functional response curves. For dynamic systems,
obtaining reliable estimates of the numerical response will be more complex
and should receive careful thought.
Fitting complex models to data is an area needing careful attention. Even
the simplest predator–prey model requires the estimation of several parameters, and estimating these parameters directly can be difficult without detailed
laboratory studies. Yet Bayesian procedures hold promise for accommodating
models with moderate numbers of parameters (McAllister et al. 1994).
REMEMBER THE AUDIENCE
Ecologists usually approach models as tools for understanding ecological
processes. But models are becoming essential tools in natural resource conservation and management. In this context, targeting the audience carefully is
crucial to the success of a model. Dueling models have dominated courtroom
debates over the management of striped bass in eastern United States (Bart-
279
Recommendations
My motivation for writing this review is to encourage wildlife ecologists to
begin focusing their research on testing ecological principles. When inconsistencies emerge between the models and field observations, the solution is not
to reject modeling and modelers, but to fix the models. Improved models will
emerge if field ecologists work with modelers to change models to accommodate the peculiarities of their study organisms.
An obvious area needing attention is the development of models for functional response. Although we know that functional response has important
consequences for population dynamics (Sherratt and MacDougall 1995), seldom do carnivore food habit studies relate results to the predictions of alternative functional response models. This seems to be particularly rich ground
for multispecies functional response models that are poorly understood in theory or practice. I am unaware of any models that go beyond linear substitutability among prey (Abrams 1987), yet I suspect that this assumption is
often violated.
Likewise, improved understanding of numerical response is essential to our
understanding of population dynamics. For equilibrium systems, Messier
(1994) presents a convincing illustration of how the combination of numerical and functional responses allow us to anticipate the role of predators in the
regulation of prey numbers. The outcome is largely a consequence of the
shapes of the numerical and functional response curves. For dynamic systems,
obtaining reliable estimates of the numerical response will be more complex
and should receive careful thought.
Fitting complex models to data is an area needing careful attention. Even
the simplest predator–prey model requires the estimation of several parameters, and estimating these parameters directly can be difficult without detailed
laboratory studies. Yet Bayesian procedures hold promise for accommodating
models with moderate numbers of parameters (McAllister et al. 1994).
REMEMBER THE AUDIENCE
Ecologists usually approach models as tools for understanding ecological
processes. But models are becoming essential tools in natural resource conservation and management. In this context, targeting the audience carefully is
crucial to the success of a model. Dueling models have dominated courtroom
debates over the management of striped bass in eastern United States (Bart-