Premium
USE OF A SMOOTHER TO FORECAST OCCURRENCE OF EPIPHYTIC LICHENS UNDER ALTERNATIVE FOREST MANAGEMENT PLANS
Author(s) -
McCune Bruce,
Berryman Shanti D.,
Cissel John H.,
Gitelman Alix I.
Publication year - 2003
Publication title -
ecological applications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.864
H-Index - 213
eISSN - 1939-5582
pISSN - 1051-0761
DOI - 10.1890/1051-0761(2003)13[1110:uoastf]2.0.co;2
Subject(s) - lichen , ecology , geography , habitat , elevation (ballistics) , indicator species , indicator value , forestry , environmental science , biology , mathematics , geometry
We used habitat models to forecast the frequency of occurrence of epiphytic lichen species in a forested landscape under two alternative plans: a literal application of standard prescriptions in the Northwest Forest Plan and a plan patterned in part after natural disturbance regimes. The plans were evaluated for the Blue River watershed in the Cascade Range of Oregon, USA. We used two model types: logistic regression and an ecological neighborhood model using a form of nonparametric regression (SpOcc; Species Occurrence Modeler). Both logistic regression and SpOcc successfully estimated the occurrence of the lichen Lobaria oregana in the current landscape, based on elevation and structural classes of forests. Structural classes were defined by combinations of (1) overstory retention of remnant trees, (2) young cohort age, and (3) stream position (upland vs. riparian). Lobaria oregana rapidly diminished in frequency in the landscape at elevations above 900–1000 m. Young, even‐aged stands had little or no Lobaria oregana . Most of the other structural classes, however, were very similar in frequency of Lobaria . The differences in lichens between retention levels and between mature and old growth stands were very small compared to the differences with respect to elevation and clearcutting. Similar results, with varying importances of elevation and forest structure, were obtained for most of the 20 species with the strongest models. Many of these species were associated with old growth, and many were forecasted to be more frequent under either management plan than in the current landscape. This is a likely response to the reduction of even‐aged management in this landscape. We recommend using SpOcc for flexible nonparametric fitting of species relationships to a multivariate habitat. SpOcc avoids the need to specify response functions and, because it is multiplicative rather than additive, automatically models responses to interactions among predictors.