Tree species sampling and wood decomposition experiment
From October to November 2017, we selected 32 important and representative woody species in TT and PT, and 2 island endemic woody species in PT (Pittosporum tobira, Eurya emarginata ). The species set included 7 evergreen trees and 11 evergreen shrubs, 7 deciduous trees and 4 deciduous shrubs, 4 conifer trees and 1 monocot tree (bamboo) (full species names see Supp. Mat. Table S1). Here “shrubs” included both short stature, single-stem understorey trees and multi-stemmed shrubs from exposed habitats. In total, 147 individual healthy adult trees/shrubs were selected, i.e., 34 species × 3 replications (but 6 replications for shrubs). We chain-sawed those trees/shrubs and collected wood samples as 20-cm long stem sections of 5 ± 0.5 cm diameter so as to control for size effects on decomposition. Adjacent to each end a 2-cm thick disk was sawn out for analyses of initial wood traits. Here we used the healthy living trees standardizing the initial, undegraded phase for all samples, thereby avoiding complications of prior pathogen attack having altered wood quality and providing a direct link to the functional traits of living trees (Cornelissen et al. 2012). Moreover, in the forests studied typhoon-induced wind-throw as well as logging is very common, so a large fraction of the trees is still alive when downed. We therefore consider these wood samples as representative of deadwood going into the decomposition process.
We used 25×15 cm litter-bags differing in mesh size to assess the termite contribution to litter decomposition: (1) termite exclusion treatment, enclosed stainless steel litter-bags with 0.05 mm mesh; (2) termite access treatment, nylon litter-bags with 4 mm mesh. Each fresh wood sample was weighed immediately after cutting (see below for measurement of water content), and sealed into a litter-bag. In total, 1224 litter-bag samples were produced, i.e., 34 tree species × 2 treatments × 3 replications (plots) × 2 incubation sites × 3 harvest times.
In December 2017, we established three 20 m × 30 m replicate plots in PT and TT, respectively. Plots within each site had broadly similar slope, altitude, soil type, litter substrate and forest structure and composition and plots within a site were 10 m apart to minimize interference between them. We started litter-bag incubation in December 2017, when termite activity and wood consumption was low because of low temperatures. The litter-bags of each species were pinned onto the forest floor in their respective subplots within each of the three replicate plots randomly, and each species’ replicate had three litter-bags per treatment in each subplot, i.e., one for each harvest. The distances between subplots were approx. 2 m. Litter-bags were harvested after 6 months (July 2018), 12 months (Dec. 2018) and 18 months (July 2019).
We transported the litter-bags to the laboratory where we carefully removed extraneous materials and soil from the deadwood remains within a few days. Each wood sample was put in a large tray with tall enough edges for the termites not to crawl out. We cut the sample into small pieces and cleaned mud and sand (brought in by termites) with a brush carefully. We then collected the termites from the tray and counted them. All pieces of the wood sample were then oven-dried at 75 °C to constant mass and weighed (dry mass).