Knowledge of habitat requirements of saproxylic insects and their response to habitat changes is critical for assessing the ecological impacts of forest management. Several studies have demonstrated a positive relationship of tree-species richness, deadwood volume, or structural diversity with saproxylic species diversity, while the relationship with the abundance of potential pest species have often been negative. A better understanding of which factors drive saproxylic insects’ occurrence is therefore essential for deriving urgently needed thresholds for key habitat conditions. We tested a rapid assessment method applicable at large scale based on recorded feeding galleries and boreholes assessed during the Austrian National Forest Inventory to investigate the drivers and habitat thresholds of different saproxylic insect families; i.e. Buprestidae, Cerambycidae, Curculionidae, Siricidae, at multiple spatial scales; i.e. at the object, forest stand and landscape level. We modelled the relative abundance of all insects and these families considering nineteen explanatory variables using ordinal logistic regression models. Key habitat characteristics were identified using recursive partitioning. Our results revealed complex interactions among influencing factors at different spatial scales. We showed that deadwood volume was of surprisingly little importance. Instead, individual tree characteristics were of major importance, demonstrating the value of resource quality and variability. The abundance of all saproxylic insect families increased with advancing decomposition, on trees taller than 18 m, and above a living stand volume of 41 m3ha−1. Aiming to guide forest management, not only forest type-specific, but tree species specific deadwood management is needed, taking into account site-specific conditions, including temperature and precipitation. For assessing temporal trends in insect colonization and habitat dynamics as well as the effects of forest management, we propose a continuous monitoring of insect traces, including living but weakened trees. This will allow for further thresholds that are urgently needed for maintaining biological diversity in forest ecosystems in the face of climate change. Austria; Climate change; Coleoptera; Conditional Inference Tree; Deadwood; Logistic Regression; National Forest Inventory; Threshold
