INTRODUCTION

Habitat loss is considered as one of the greatest threats to biodiversity worldwide (Wilson 1988; Pimm and Raven 2000). Both theoretical and empirical studies indicate that there are critical limits to the amount of habitat that can be lost without reducing the viability of populations (e.g. Fahrig 2001) or disrupting important ecosystem processes (Hobbs 1993). In analogy to the concept of critical load of airborne pollution (Nilsson and Grennfelt 1988), which may lead to the loss of ecological integrity in forest ecosystems, critical loss and fragmentation of habitat may lead to dysfunctional habitat networks (Angelstam et al. 2004a,b).

Knowledge about the necessary characteristics, quantity, spatial configuration, and dynamics of different ecosystem types required to maintain forest biodiversity is a prerequisite for effective conservation and restoration planning. Gaining such knowledge requires studies of dose–response, where the dose is the parameter value of a given ecological variable (e.g. amount of a given resource or rate of a key process) and the response can be measured as the status of biodiversity components (e.g. presence of a species, or even better, fitness of individuals in a population). However, the present level of empirical knowledge on that topic is limited, both with respect to which variables are critical at different spatial scales and the parameter values associated with the presence of viable populations or functioning ecosystems (Tear et al. 2005).