This application performs power analyses to determine the probability
of statistically detecting ecological relationships between a species
and its required resources.
We affectionately refer to this application as Sample Simon
based on its key role of executing numerous sample simulations.
Publication:
Smith, B.D., C. Goodinson and K.M. Martin 2004. The challenge
of statistically detecting species-habitat relationships on an uncooperative
landscape. Ecological Applications (in preparation).
Abstract:
Point-counts in combination with vegetation plots have been used
to collect data in bird studies exploring species-habitat relationships.
In an ideal world, analysis of the resulting data would yield
correct and significant correlations between point-count detections
and various putative habitat attributes.
However, in practice the data are invariably noisy and their interpretation
inconclusive, perhaps even misleading.
In order to explore possible sources of variance, we constructed
a spatially explicit model, which simulates the sampling of a territorial
species within a landscape characterized by a number of habitat
resources.
The spatial distribution of each resource and the distribution
of territories across the landscape, given species-habitat relationships,
were defined a priori.
The point-count and vegetation plot regime was then superimposed
on the landscape and sample data generated.
For a given set of simulation parameters, sample data were generated
and analyzed repeatedly, in order to determine the minimum sampling
effort required to correctly detect the true species-habitat relationships.
The results of our simulations showed that only under highly simplified
conditions were the true species-habitat relationships detectable
for a realistic sampling effort.
In addition to previously acknowledged sources of variance in
the data, such as species detectability and density, we determined
that the synchrony of the spatial periodicity of the habitat resources,
the average territory size, and the radii of both point-counts and
vegetation plots, was critical to a successful sampling effort.
|