We conducted these tests only for species present at four or
more sites (minimal number of occurrence needed to calculate
D⁄) in either two or three forest types; for these species, the results
of the D⁄ tests were used to assign a specialization score to the species
(see below).
For species that did not meet this criterion, we
based our assignment on occurrence patterns alone.
We calculated
the occurrence rate for a bird species in a forest type by dividing
the number of sites at which it was recorded by the total number
of sampled sites, rounding the occurrence measures to the nearest
tenth (e.g., to 0.1, 0.2, etc.).
A species was considered to have its
habitat use associated with a single forest type when the difference
in the occurrence measures between the forest types exceeded 0.3.
Typically, associations were much stronger than described by this
cut-off. We feel it would be unreasonable to make inferences about
forest type specialization for very scarce or localized species, which
we deemed to be those that occurred in six or fewer sites. Thus
these species are excluded from our quantitative analyses. We
present data on these species (see Appendix) and in the Discussion
reflect on potential biases that not including these rare species
might have for our conclusions