Re-evaluation of stand indicators for the assessment of the representativity status of the Natura 2000 habitat type forests

Abstract

Habitat inventories and monitoring require the evaluation of the status of various charac- teristics of these habitats - indicator traits. In the case study of Estonian forests in Viru County (Natura 2000 Annex I forest habitat types *9010, *9080, and *91D0), the survey data of old-growth and natural forests were used to test for the efficiency of the indicator set that has been suggested for the evaluation of the habitat's representativity status. The data from expert-graded forests (121 stands) were alternatively clustered with k-means clustering, their characteristics were tested for indicator power with discriminant analysis, and the resulting efficient set of characteristics was clustered again for an updated classification. In studying the differences between analyses of expert grading and cluster system, we found that different characteristics had different weights in forest classification. In addition to the standard structural characteristics, signs of anthropogenic activity and landscape pattern proved to be of importance. From the testing for various precision scales of classification, we concluded that different indicator traits of structure and composition are required, and the three-grade system appears to be practical for the purpose of avoiding over-interpretation. We found that additional studies are needed to define reasonable indicator traits for wet and swampy forests, and also for forests on unproductive soils.