Biodiversity loss due to intensive timber production is a ubiquitous conservation issue across temperate and boreal forest ecosystems. Retention forestry, the retention of deadwood and old-growth features within production forest, is one management strategy that has been implemented in various countries around the world to conserve a wide range of taxa within managed forests. The success and ecological implications of retention forestry are currently subject to intensive investigation and while some taxa like birds and insects have already been studied frequently, larger mammals have obtained less attention. Pine martens are one of the few larger mammals in central Europe preferring older forest and potentially profiting directly from deadwood retention as a consequence of implemented retention forestry. The goal of our study was to assess the response of European marten species to deadwood retention in montane mixed forests. Using marten detection rates from camera traps on 135 research plots we assessed the response of martens to deadwood on three different spatial scales using generalized linear mixed models. We found no effect of lying deadwood on marten detections on the plot scale (one-hectare) or in a ten-meter radius around the camera traps. However, we found a significant increase in marten detections if logs (>10 cm) were directly in front and in view of the camera trap. Our results show that deadwood retention as a measure of retention forestry does affect microhabitat use of martens, but not stand selection during the growing season. Logs directly in view of the camera trap increase marten detection rates as martens choose to move and forage along fallen trees when they are available. When using camera trapping to collect data on martens, trap positioning in front of logs can heavily bias trapping results when unaccounted for.

We set up camera traps on 135 predetermined 1-ha research plots. We used camera traps (Bushnell Trophy Cam HD Aggressor Low Glow) over five sampling rounds in spring (April-early July) and autumn (late August-November) from 2019 to 2021. The exact positions of the camera traps in the first sampling round were assigned randomly to one of three fixed points within the plot. Afterwards, cameras were shifted systematically among those positions. We aggregated marten detections in events, which included a sequence of detection of the same species with less than five minutes between pictures. The sum of marten events is used as an index of relative abundance. Additionally, we assessed the presence/volume of deadwood at three different scales (view of the camera, 10 m radius of the camera, 1-ha plot).