1. Saproxylic community assembly is structured both by deadwood and forest habitat gradients, as well as biotic interactions such as competition, predation, and parasitism. However, covariation between abiotic and biotic conditions in natural systems have limited our ability to disentangle these mechanisms. Furthermore, focus towards beetles and fungi in temperate or boreal forests has led to important taxonomic and geographic knowledge gaps. 2. Here, we tested how experimentally-manipulated tree diversity, deadwood position (lying vs. standing), and biotic interactions with a dominant antagonist (ant exclusion) structure the community assembly of deadwood-cavity-nesting bees, wasps, and their parasitoids in a subtropical forest. 3. Our findings reveal that lying deadwood supports a nested subset of the communities occurring in standing deadwood, with less diversity and abundance of hosts and parasitoids. We found that increased moisture, rather than ant activity, was the primary mechanism filtering Hymenoptera communities, as deadwood in contact with the forest floor retained twice as much moisture as standing substrate. Moreover, moisture gradients within each substrate type further reduced host abundance – likely due to reduced brood cell production and survival. In contrast, forest habitat (tree species richness, canopy cover, and coarse woody debris) had comparatively minor roles in shaping cavity-nesting community assembly. 4. Our results provide a mechanism for the positive association between cavity-nesting Hymenoptera and standing deadwood in forests. Because standing deadwood is typically scarce in many managed forests, these findings support the retention and enhancement of such substrates to promote these ecologically-important insects.

See full manuscript archived in bioRxiv for details Two sites, A and B, 64 plots in total, 192 deadwood traps in total. Deadwood blocks of Schima superba were pre-drilled to offer cavities to cavity-nesting Hymenoptera. In each plot we had three deadwood traps, one lying on the floor, two suspended by a tree. Of the two suspended traps, one included insect glue to exclude ants from accessing. We explores forest stand and deadwood-level environmental drivers of cavity-nesting Hymenoptera community assembly and relationship with parasitoids.

VIP plots, sites A and B

deadwood was set out in the field for 11 months

cavity-nesting Hymenoptera and ants

Datasets and meta-data are also available in Fighsare at: https://doi.org/10.6084/m9.figshare.31979481 R code for data analysis is available on Github via Zenodo at: https://doi.org/10.5281/zenodo.19493070