Relationships between wood quality, termite activity and decomposition over time
Multiple traits, notably wood N, P, cellulose and lignin content and wood density, together explained 55.9% of all variance explained by PC1, interpreted here as the WES (see Methods and Fig. S1). Wood litter of more resource acquisitive strategy species (i.e., high WES values), generally decompoesed faster than that of conservative strategy species; in the treatment without termite access (Fig. S1, Fig. 2) there were significant positive linear relationships between WES score and mass loss in the periods 0-6 and 12-18 months (but non-significant ones at 6-12 months), and cumulatively over 18 months, in both sites.
Our termite access treatment revealed that termites not only significantly accelerated wood mass loss rate overall, but there was a clear time pattern within this acceleration (Fig. 3), as hypothesized (Fig. 1). In the first year (periods 0-6, 6-12 months) decomposition rates scaled positively and linearly with WES in both sites (Fig. 2a,b,c,d), indicating that termites preferentially consumed the species of resource acquisitive strategy (high nutrient content, less lignin, less dense structure). Thereby, the termites’ consumption amplified the initial WES effect on decomposability, i.e., it increased its positive linear regression slope.
In contrast, partly owing to termite activity, the wood of the acquisitive species had been considerably depleted after 12 months. In the subsequent period of 12-18 months, the initially medium quality species were consumed more by the termites, which modulated the tree species’ decomposition trajectory on the WES. Now, in contrast to the positive linear relation in the treatment without termite access, there was a hump-back relationship between WES and period mass loss in the treatment with termite access, at least in the more termite-rich PT site (Fig. 2e, Table 1). Termite abundance showed a corresponding humpback pattern with mass loss (equation in Fig. 2e). In the less termite-rich site, TT, such a humpback relation was not apparent, but here the positive relationship between WES and mass loss was less steep in the latter (Fig. 2d,f) than in the initial period (Fig. 2b). This pattern also suggests negative termite feedback on the positive relation between WES and mass loss. These changing patterns of WES over period-specific mass loss were confirmed by a significant interaction among WES (covariate), harvest time and termite presence/absence on period mass loss in PT and TT, respectively (Table 2). Together, as hypothesized (Fig. 1), these deviating relationships over time caused overall convergence of cumulative mass loss along the WES between the termite treatments in both sites (Fig. 2g,h), as indicated by a lack of interaction of WES (covariate) x termite access treatment on cumulative mass loss over 18 months (Table 2).
Termite abundance patterns in the wood samples were consistent with the above changing patterns of period-specific wood mass loss (Fig. 3) and the termite contribution to decomposition (Fig. S3) along the WES through time. At 6 months, there was a significant, exponential increase from the conservative end to the acquisitive end of the WES in both sites (Fig. 3a,b). At 12 months, the termite abundance peak occurred at slightly higher than medium initial wood quality (Fig. 3c,d). By 18 months in PT, the termite abundance peak had moved further towards the conservative end of the WES (Fig. 3e), i.e., to the centre of the range, matching the humpback for mass loss (Fig. 2e). In TT both the height and the width of the peak (i.e., the range) increased at 12-18 months (Fig. 3f). All in all, the peaks of termite abundance and termite contribution to mass loss shifted from high to medium initial wood quality during the decomposition process, broadly matching that of mass loss itself.