Geoscience Reference
In-Depth Information
Another misguided example is Tomiuk and Loeschcke (1994). Their
mathematics cover up the real problem of obtaining realistic estimates of the
parameter values to use in the models. Their model emphasizes demographic
variation and ignores the bigger issues of temporal variation and individual
heterogeneity.
A common problem with PVA is that the sampling variation of the param-
eter estimates is ignored (e.g., Stacey and Taper 1993; Dennis et al. 1991). In
both cases, estimates of persistence are too pessimistic because the sampling
variation of the population parameters is included in the population model as
if it were temporal variation. Furthermore, individual heterogeneity was left
out of the model, further biasing the estimates of persistence too low.
“Most PVAs have ignored fundamentals of ecology such as habitat, focus-
ing instead on genetics or stochastic demography” (Boyce 1992:491). For
small populations (less than 50) of endangered species, such a strategy may be
justified, particularly for short-term predictions. But incorporating only
demographic variation results in overestimates of persistence because temporal
variation has been ignored. On the other hand, the remaining survivors of an
endangered species may be the individuals with high survival and reproductive
rates, so the lack of individual heterogeneity may underestimate persistence.
The studies just mentioned should not lead the reader to believe that use-
ful attempts to estimate persistence do not exist. Schneider and Yodzis (1994)
developed a model of marten (
Martes americana
) population dynamics that
incorporated the behavior and physiology of individual martens, spatial
dynamics, and demographic and environmental stochasticity. Undoubtedly
some readers would quibble with some of the assumptions and data used to
build the model, but I contend that a realistic model with some of the inputs
guessed (and clearly stated to be such) is a much more reasonable approach
than a simplistic model that ignores important processes affecting persistence.
Furthermore, such realistic models identify data needs that can be addressed
with time, even though the actual estimate of persistence is of questionable
value. The alternative of using simplistic and naive models leads to invalid esti-
mates and little progress in improving the situation, with a rapid loss of credi-
bility by the field of conservation biology.
Murphy et al. (1990) proposed two different types of PVA. For organisms
with low population densities that are restricted to small geographic ranges
(typical vertebrate endangered species), genetic and demographic factors
should be stressed. For smaller organisms such as most endangered inverte-
brates, environmental uncertainty and catastrophic factors should be stressed
because these organisms are generally restricted to a few small habitat patches,
