We review the evidence for changes in species composition and population vi
ability caused by forest fragmentation in one of North America's most biolo
gically diverse regions, the Klamath region of northwestern California and
southwestern Oregon. Our studies on small mammals and understory plant comm
unities suggest that a number of species are restricted to, or closely asso
ciated with, remnant patches of late-successional or old-growth forest. Lik
ewise, we found evidence for changes in population sizes in species that we
studied in-depth: California red-backed vole (Clethrionomys californicus M
erriam) and western trillium (Trillium ovatum Pursh.). We also identified m
any species that do not fit the island-ocean analogy that is often applied
to habitat fragmentation. For example, some species were found mostly in re
mnants, while other species were restricted to clearcuts, or were ubiquitou
s across sites. Evidence from these studies suggests that the landscape mat
rix can provide important habitat for many species. The conservation of bio
logical diversity in this region will require not only knowing which specie
s are adversely affected by fragmentation and what role remnant habitat pla
ys in the viability of populations; it will also require a better understan
ding of the role of matrix in providing both habitat and opportunities for
dispersal of species between remnants. Studies of matrix conducted in the C
ascade and Coast Ranges of Oregon and Washington provide relevant informati
on, but there is a striking paucity of data concerning matrix in the Klamat
h region, where we predict that the drier, warmer growing season is likely
to result in different, species-specific responses to fragmentation. Resear
ch on these responses would provide a vital contribution toward understandi
ng how to conserve biological diversity in the Klamath region.