Errors in indices can be estimated by comparing results with a known population
similar to the way we estimate bias errors in counts (see Section 13.4.2) (Eberhardt
and Simmons 1987). For example, the number of sightings of fallow deer (Dama dama)
in France along a transect (the index) was calibrated against a known population.
The sighting index was found to be an effective standardized method to detect trends
in the population (Vincent et al. 1996).Although known-to-be-alive is sometimes used as a one-off estimate of population
size, it is more often used to track trends in population size. The operating rules
governing these two uses are quite different. In the first exercise we seek the most
accurate estimate we can get. In the second we seek consistency of method among
several estimates such that their bias is held constant. In the first case we put in
as much work as possible. In the second we put in precisely the same amount of
sampling on each surveying or capturing occasion. Otherwise the trend in the
estimates may reflect no more than variation in capturing effort.
A variant of this aberration of effort, very common in ecological research, is to
boost the number known to be alive (because they were caught) on a given occa-
sion by the number of individuals not caught on that occasion but which must have
been there because they were caught on both previous and subsequent occasions.
Although the accuracy of the estimate of absolute numbers is thereby enhanced, the
consistency of the string of estimates is thereby lowered. Estimates for the earlier
occasions are inflated relative to those of later occasions, the rate of increase being
underestimated if density is rising and the rate of decrease being overestimated if
density is falling.Animal numbers can be estimated by total counts, sampled counts, mark–recapture,
or various indirect methods. In each case the usefulness of the method is determined
by how closely its underlying assumptions are matched by the realities of what the
animals do and how difficult they are to see, trap, or detect. The range of methods
provided should allow wildlife managers to choose one that will be adequate in any
given circumstance.COUNTING ANIMALS 243
13.7.1Known-to-
be-alive used as
an index