Natural disturbances are increasing around the globe, also impacting protected areas. Although previous studies have indicated that natural disturbances result in mainly positive effects on biodiversity, these analyses mostly focused on a few well established taxonomic groups, and thus uncertainty remains regarding the comprehensive impact of natural disturbances on biodiversity. Using Malaise traps and meta-barcoding, we studied a broad range of arthropod taxa, including dark and cryptic taxa, along a gradient of bark beetle disturbance severities in five European national parks. We identified order-level community thresholds of disturbance severity and classified barcode index numbers (BINs; a cluster system for DNA sequences, where each cluster corresponds to a species) as negative or positive disturbance indicators. Negative indicator BINs decreased above thresholds of low to medium disturbance severity (20%–30% of trees killed), whereas positive indicator BINs benefited from high disturbance severity (76%–98%). BINs allocated to a species name contained nearly as many positive as negative disturbance indicators, but dark and cryptic taxa, particularly Diptera and Hymenoptera in our data, contained higher numbers of negative disturbance indicator BINs. Analyses of changes in the richness of BINs showed variable responses of arthropods to disturbance severity at lower taxonomic levels, whereas no significant signal was detected at the order level due to the compensatory responses of the underlying taxa. We conclude that the analyses of dark taxa can offer new insights into biodiversity responses to disturbances. Our results suggest considerable potential for forest management to foster arthropod diversity, for example by maintaining both closed-canopy forests (>70% cover) and open forests (<30% cover) on the landscape.