Development over time of the tree-related microhabitat profile: the case of lowland beech–oak coppice-with-standards set-aside stands in France

European forest managers are implementing set-aside measures in managed forests to restore key structures for forest biodiversity such as tree-related microhabitats (TreMs). However, the time required to regenerate these structures is little known. We assessed the patterns of thirteen TreM types on 282 plots in 24 lowland forests in southwestern France. We applied a synchronic sequence ranging from 1 to 80 years after the last harvest, a time frame which is considered long enough to observe significant changes in the TreM profile. We sampled lowland beech–oak coppice-with-standards stands representative of the different forest ownerships and management regimes occurring in France; our assessment included both public and private forests with or without formal management plans. We found that both TreM density and diversity just after harvesting were lower, though not significantly so, in private stands without any management plan than in the other management regimes. We observed both significantly higher TreM density and diversity in plots harvested 10–15 years ago than in plots harvested 1–5 years ago. The next marked difference did not occur until stands had been harvested 70–80 years ago. Globally, time since last harvest was the best explanatory variable for variations in both TreM density and diversity. We therefore recommend: (1) conserving more habitat trees in harvested areas, particularly cavity trees, since the densities we observed were much lower than the densities required by cavity-dwelling species, and (2) letting set-aside patches freely complete several full silvigenetic cycles. This latter practice would avoid the inefficacy (or possibly even negative effects) of a temporary conservation network, and would also simplify management of the network over time. Further research should assess TreM occurrences in a permanent reserve network. Diachronic observations would make it possible to highlight the drivers of TreM profile development.