On each site, ten pitfall traps were evenly spaced along two 50 m transects, with a distance of 10 m between individual traps and 20 m between transects. Pitfall traps were filled with 100–200 ml of a mixture of ethylene glycol and water (1:1 vol/vol) with a drop of liquid soap to break surface tension. Traps were exposed for 7 days each during two to five sampling events in both the dry and wet seasons between May 2011 and October 2012. As the number of individuals collected in ten traps was very high, we had to confine the sorting and subsequent analysis to sub-sets of at least three traps per sampling site and sampling event. The number of sampling events and considered samples is given in Table S2 (Röder et al. 2017). To address imbalances in the data set, we first calculated the mean abundances per trap per sampling event for each sampling site. We then calculated mean abundance per sampling site from the mean abundances per sampling event such that the number of replicates for statistical testing summed up to a total of N = 59 sampling sites. Unfortunately, we had to find out later that the ethylen glycol procured locally was actually a mixture of ethylen glycol and 2-ethoxyethanol, which is a strong oxidizing chemical. Therefore, any sequencing of specimen caught in pitfall traps was impossible.

Further descriptions:1) For method: Pitfall trap counts = pitfall traps, exposed for 7 days, containing 100-200ml sampling liquid - ethylenglycol : water (50:50) plus a drop of liquid soap to break the surface tension, trap level samples were sorted to order level and counted.Mean per trap calculated from all analyzed trap samples per plot and sampling round, then means over all sampling rounds per plot.2) For Coleoptera = Beetles assigned to morphospecies by Peter Schüle, Carabidae identified at species level, Tenebrionidae checked by Wolfgang Schawaller (Museum Stuttgart).