Abstract. 1. Cellulose baits are commonly used for semi-quantitative sampling of subterranean wood-feeding termites, with a single food choice sampling programme implemented most often. In most situations, however, the composition and feeding preferences of the subterranean termite assemblage remain unknown.
2. The diversity, frequency, foraging activity, and intensity of attack of termites were assessed regularly at 144 baits representing 12 different bait choices over 8.5 months, in two northern Australian tropical savanna sites that differed in vegetation structure (closed vs open). Baits differed in type (paper rolls, cardboard, wooden stakes), position (surface, buried), and moisture status at installation (wet, dry).
3. Sixteen species were recorded, including 11 wood-feeders. Average species diversity, foraging activity, and bait consumption were greater at buried baits than at surface baits. Wooden baits were most attractive early in the experiment, and paper baits more attractive later. Mean species diversity was greatest at wooden stakes in the closed site. Species frequency of occurrence varied across bait choices.
4. A composite bait sampling protocol of stakes and paper rolls installed above and below ground gave an accurate assessment of the activity, diversity, and structure of the termite guild sampled across all baits over 8.5 months.
5. The choice of bait, its presentation, and time of examination are critical to the success of a termite baiting programme. If the aim is to characterise the structure and foraging activity of the subterranean termite assemblages that are attracted to baits, composite baiting protocols should be implemented.