16 Experimental management
In practical terms, there are two different modes of wildlife management: that in which
management decisions flow from personal experience and received wisdom, and that
in which they are based upon data and analysis. For want of better names we will
call these the “traditional” and “experimental” modes of wildlife management,
respectively. The value of the traditional mode should not be underestimated. Its
dominant characteristic is conservatism, a lack of interest in trying out new ideas.
That is sometimes a strength rather than a weakness because most new ideas turn
out to be wrong (Caughley 1985). However, some new ideas are useful and these
are best identified by the experimental approach. In this chapter we explain how a
technical judgment can be evaluated, by posing it as a question (hypothesis).
One way to evaluate a hypothesis is through a standard statistical test of inference.
We provide guidelines for designing effective experiments and outline some
standard statistical methods of analysis of such experiments. We describe the use
of replication to sample the natural range of variability and the use of controls to
render the conclusions unambiguous. However, standard statistical tests are some-
times inadequate for identifying which of many possible hypotheses provide the “best”
explanation for the observations at hand, and for deciding which course of manage-
ment would be most effective. In order to pursue these issues, we need a different
approach, namely model evaluation and inference. In Chapter 15, we outlined some
of the principles of model evaluation and showed how this method can provide a
powerful tool in the resource-manager’s arsenal.Wildlife management is not like civil engineering. The theory and practice of civil
engineering is placed on public display every time a bridge is built. No expertise is
needed to interpret that test. If the bridge remains operational for the length of its design
life, the engineers got it right. If it collapses they got it wrong, and we look forward
to hearing the details of how and why they got it wrong at the subsequent court case.
Wildlife management differs from civil engineering in a number of respects. First,
the managers are not erecting something new but acting as custodians of something
already there. They are not responsible for the initial conditions but these often con-
strain their options.
Second, in civil engineering the question is usually obvious. In ecology the
appropriate question is seldom obvious. Choosing the appropriate question is the most
difficult task, much more difficult than answering that question. Good design does
not correct an inadequate grasp of the problem.
Third, criteria for success and failure are seldom tight and often are not available
to the public. Compare these two statements:26816.1 Introduction
16.2 Differentiating success from failure
WECC16 18/08/2005 14:47 Page 268