Changes in the composition of plant functional traits may affect ecosystem processes through influencing trophic interactions. Bottom‐up control by plant species through food availability to animals may vary with time. However, such dynamics and their consequences for deadwood turnover are poorly known for detrital food webs. We introduce a dynamic conceptual model of the feedback of tree functional traits, (deadwood‐feeding) termite populations and deadwood decomposition. We hypothesized that tree functional diversity (in terms of a wood resource economic spectrum [WES]) supports the sustenance of termite populations via complementary food supplied through time, as deadwood varies in traits both initially across species and because of different decomposition rates. Simultaneously, driven by this temporal dynamics of food quality, the consumption of deadwood by termites should hypothetically sustain deadwood turnover in a functionally diverse forest over time. We tested our hypothesis through an 18‐month termite‐exclusion decomposition experiment by incubating coarse (i.e. 5 cm diameter) deadwood of 34 woody species in two subtropical forests in East China. One site still sustained a healthy population of pangolins as the keystone termite predator, whereas another had lost its pangolins due to hunting and illegal wildlife trade. The results supported our hypothesis: in the first 12 months, termites amplified the positive linear relationship between % wood mass loss and initial wood quality (WES). In contrast, between 12 and 18 months, termite‐mediated consumption, and associated wood mass loss, showed a humpback relation with the initial WES. This shift in termite preference of deadwood species along the WES reflects complementary food availability to termites through time. Synthesis. Our findings imply that tree functional composition, with variation in deadwood quality through decomposition time, can help to sustain termite populations and thereby forest carbon turnover. Future studies need to test whether and how our conceptual model may apply to other detrital systems and food webs. In general, food web research would benefit from a stronger focus on temporal patterns for better understanding the interactions of basal resource functional traits and consumers on ecosystem functions.

Litter quality and decomposers are critical to carbon and nutrient cycling through litter decomposition. However, how relationships between litter quality and invertebrate detritivores change litter mass loss through time is poorly known. Species’ initial leaf litter quality, as a legacy of their position on the ‘leaf economics spectrum’ (LES), may determine the invertebrate contribution to litter mass loss. This contribution may change through time, as both population peaks of invertebrate detritivores and litter quality of given species will change through time. Here we introduce invertebrate phenology into a conceptual model of drivers of litter mass loss. We hypothesized that in the early decomposition period, LES can predict litter decomposability with or without a strong invertebrate contribution, that is, litter with higher nutrient content would decompose faster. But in the later decomposition period, when higher quality litter will already have decomposed too much and lower quality litters have still been less degraded, a strong invertebrate peak would coincide with relatively more consumption of initially lower quality litters; this would lead to a humpback relationship between leaf litter mass loss and initial LES position in this period. We tested our hypothesis through a 1‐year field decomposition experiment using leaf litter of 41 woody species in each of two sites in subtropical forest in China; only one of these sites had a strong late peak of leaf litter‐feeding moth larvae in the litter layer. LES score of litter species had a positive linear relationship with litter mass loss before the key invertebrate consumer peaks in the litter layer. However, with the invertebrates peaking later into the decomposition process, the invertebrate consumption peaked at initially lower quality litters, which altered the species’ decomposability trajectory on the LES, consistent with the hypothesized humpback relationship between leaf litter mass loss and LES. This phenomenon resulted in a strongly reduced slope of cumulative mass loss on initial LES score across species. Our finding highlights the importance of considering interactions between the timing of detritivore activities and the timing of litter quality for better understanding the relationships between soil animals and ecosystem carbon and nutrient cycling. A free Plain Language Summary can be found within the Supporting Information of this article.

The plant economics spectrum integrates trade‐offs and covariation in resource economic traits of different plant organs and their consequences for pivotal ecosystem processes, such as decomposition. However, in this concept stems are often considered as one unit ignoring the important functional differences between wood (xylem) and bark. These differences may not only affect the performance of woody plants during their lifetime, but may also have important “afterlife effects.” Specifically, bark quality may strongly affect deadwood decomposition of different woody species. We hypothesized that (1) bark quality strongly influences bark decomposability to microbial decomposers, and possibly amplifies the interspecific variation in decomposition by invertebrate consumption, especially termites; and (2) bark decomposition has secondary effects on xylem mass loss by providing access to decomposers including invertebrates such as termites. We tested these hypotheses across 34 subtropical woody species representing five common plant functional types, by conducting an in situ deadwood decomposition experiment over 12‐month in two sites in subtropical evergreen broad‐leaved forest in China. We employed visual examination and surface density measurement to quantify termite consumption to both bark and the underlying xylem, respectively. Using principal component analysis, we synthesized seven bark traits to provide the first empirical evidence for a bark economics spectrum (BES), with high BES values (i.e., bark thickness, nitrogen, phosphorus, and cellulose contents) indicating a resource acquisitive strategy and low BES values (i.e., carbon, lignin, and dry matter contents) indicating a resource conservative strategy. The BES affected interspecific variation in bark mass loss and this relationship was strongly amplified by termites. The BES also explained nearly half of the interspecific variation in termite consumption to xylem, making it an important contributor to deadwood decomposition overall. Moreover, the above across‐species relationships manifested also within plant functional types, highlighting the value of using continuous variation in bark traits rather than categorical plant functional types in carbon cycle modeling. Our findings demonstrate the potent role of the BES in influencing deadwood decomposition including positive invertebrate feedback thereon in warm‐climate forests, with implications for the role of bark quality in carbon cycling in other woody biomes.

1. Subordinate species have relatively low abundance compared to dominant species; however, they may contribute significantly to functional diversity and ecosystem functionality, particularly if they differ strongly from the dominants in key traits. Here we investigated whether this phenomenon might be applied to litter