**1. Introduction**

Understanding the mechanisms that control the biomass and density of trees in forests is important for ecological, economic, and conservation motivations [1]. For instance, growing forests sequester substantial amounts of carbon [2], and as global atmospheric carbon dioxide (CO2) continues to rise, understanding tree growth and the mechanisms that drive carbon accumulation and storage has become a pressing issue. Abiotic factors like climate, soil type, and disturbance frequency are known to be among the major determinants of the productivity of plant communities [3]. However, in recent decades, interest has shifted towards how community properties (e.g., number and relative abundance of species), and functional traits (e.g., [4]), influence productivity [5]. This is particularly important in studies of community structure, because these features are amenable to experimentation and as a managemen<sup>t</sup> tool. Additionally, since di fferent plant species vary in their nutrient requirements [6], it is possible that nutrient cycling is a fundamental feature contributing to the observed link between diversity and productivity.

The availability of nitrogen (N) limits productivity in many terrestrial ecosystems, and its increase typically promotes plant-community productivity [7,8]. This e ffect is further facilitated through complementarity, because di fferences in N-use e fficiency lead to positive feedbacks in N-cycling rates [9]. Nitrogen cycling is heavily dependent on the input and output of various forms of N [10,11], which is often supplied in various quantities by di fferent plant species (and their litter). The plant litter can comprise a mixture of fine detritus (e.g., leaf litter, bark, twigs) and coarse material (branches and whole stems of fallen trees), which are then decomposed and transformed into available nutrients, such as nitrogen [12], via ammonification, nitrification, and N mineralisation. Nitrification is the dominant process in most soils, where nitrifying organisms are often constrained by, and are dependent on, the rate of ammonification [13]. Because of the scarcity of N in soils, soil-N transformations have a strong influence on the potential productivity of an ecosystem [14,15].

It is well known that decomposition rates can vary due to di fferences in leaf quality (C:N ratio) among plant species, and often increase when litter of di fferent species is combined [16,17] due to complementary stimulation among microbial groups [18]. For example, sclerophyllous eucalypts produce litter that has low nutrient concentrations and low rates of decomposition [19] whereas, by contrast, the litter of N-fixing acacias has comparatively high N concentrations and decomposition rates [20,21]. This has led to speculation that Australian forest communities containing species with more readily decomposable and nutrient-rich litter (such as *Acacia* spp.) might have higher litter decomposition rates than those with *Eucalyptus* litter alone [22,23]. However, this idea has ye<sup>t</sup> to be explored in natural forest systems or linked to overall tree size and biomass.

Here we use field and laboratory data to determine the local-scale e ffect of an N-fixing tree species (*Acacia dealbata* Link) on the size and total biomass of the dominant eucalypt species (*Eucalyptus obliqua* L'Hér) in a tall hardwood forest community in Tasmania, southern Australia. We used a combination of:

