Ja. Howell et Rh. Barrett, CALIFORNIA WILDLIFE HABITAT RELATIONSHIPS SYSTEM - A TEST IN COASTAL SCRUB AND ANNUAL GRASSLAND HABITATS, California fish and game, 84(2), 1998, pp. 74-87
We tested predictions of the California Wildlife Habitat Relationships
(CWHR) System in coastal scrub and annual grassland. We detected a to
tal of 28 species of terrestrial vertebrates: 18 mammals, 9 reptiles,
and 1 amphibian. The CWHR System prediction omitted 4 of these species
: 3 domestic mammals and 1 reptile. For the 2 habitats combined, CWHR
predicted a total of 38 species: 23 mammals, 13 reptiles, and 2 amphib
ians. We detected 64% of these predicted grassland species and 71% of
predicted coastal scrub species. For the habitats combined, we detecte
d 65% of the species predicted to be present by the CWHR System. We de
tected 68% of the mammals, 62% of the reptiles, and 50% of the amphibi
ans predicted for these habitats. The CWHR System theoretically predic
ts absence rather than presence, since it is assumed that all 288 regu
larly occurring mammals, reptiles, and amphibians occur anywhere unles
s one can argue that a specific habitat, location, or habitat element
is not available. By including predictions of species absence in the a
ssessment of model performance, observed accuracy of the CWHR model pr
edictions increased to 96% for both habitats.