**4. Discussion**

We have shown that experimental mixtures of plant species from Australian eucalypt forests that contain a higher proportion of nitrogen-fixing *A. dealbata* litter have higher decomposition rates (Figures 4 and 5). This was driven primarily by the three-fold increase in nitrogen (N) content in *A. dealbata* litter compared to, for example, *B. salicina* (another understorey tree) and the canopy-forming *E. obliqua*. This result is supported mechanistically by previous studies (e.g., [36,37]) showing that decomposition rates of litter mixtures are driven predominantly by substrate quality, particularly the amount of C and N in the litter mixtures. Plant functional traits, such as nitrogen fixation, are key drivers of higher-quality substrates in forest ecosystems, with the potential to mediate and control plant abundance, biomass, and composition via inherent differences between species in quality (C:N, lignin:N) and quantity (%C, %N) of plant litter [38]. Indeed, the quantity of carbon (C) in the soil at our Tasmanian forest sites was the best predictor of average eucalypt tree size, followed by the average size of *A. dealbata* (Figure 1). Studies that have examined this phenomenon in other forests have demonstrated that the identity and functional traits of litter-creating species can surpass species diversity as the main driver of decomposition [39–42].

Given that *A. dealbata* presence and abundance indirectly controls litter decomposition via nitrogen availability and was consequently a useful predictor of average *E. obliqua* tree size, this research supports the hypothesis that N availability and cycling would be higher in sites with a greater *A. dealbata* biomass. Our results align with those of [23], who demonstrated that the presence of *Acacia mearnsii* (another N-fixer) influenced the quantity and rates of N and P cycling in *Eucalyptus globulus*

mixed forests compared to monocultures, in an 11-year experimental trial. Similar relationships have also been found in boreal forests, grasslands, and tropical forests, through field N fertilisation trials [42–44] and in laboratory experiments [45,46]. However, in our study, there was no observable relationship between *A. dealbata* biomass and the total amount of organic matter (C and N) or the concentrations of NH4 +, NO3 − or with total mineral N and P in the soil, despite high heterogeneity among sites in N availability and cycling (Figure 3). Further, we found no correlation between total available soil-mineral N and average *A. dealbata* tree size. As *A. dealbata* was the sub-dominant canopy species in our forest system, and there was 30% more biomass of *A. dealbata* relative to other sites, the potential for N-rich soil from N-fixers with high N litter should have been realised. As such, this null result demands an explanation.

There are several plausible reasons as to why a higher N content was not observed at sites with high *A. dealbata* biomass (Figure 1), with these inferences potentially useful for forest managemen<sup>t</sup> during early succession and development. Two leading candidates are: (1) *A. dealbata* is a nitrogen fixer and therefore is not reliant on high soil N to grow, whereas this might be limiting for other species (such as *E. obliqua*); and (2) due to the relatively short lifespan of *A. dealbata* (~40 years) compared to *E. obliqua* (>100 years) [47], a higher N litter might have been present during the early stages of forest succession, when growth was faster and intraspecific competition more intense, but that early signal has since dissipated (e.g., [48,49]). The development of a forest is characterised by both changes in structure and succession, with concomitant consequences for productivity of standing forests and regrowth [50]. For example, restoration of forest ecosystems with multiple forest species helps rebuild multiple functions, such as species structure, litter composition, and soil nutrients of an area [9,51]. Selective logging of natural forests is a form of disturbance, and managemen<sup>t</sup> plans require some older trees to be left in a logged area for animal habitat and seedling regeneration. Our results sugges<sup>t</sup> that complimentary tree species, such as *A. dealbata* and *E. obliqua*, should be left in such areas, to maintain the overall functionality, especially during early successional development [52]. However, our experimental results only show that complementarity and nitrogen facilitation can occur, whereas determining this relationship explicitly within a natural setting (such as [53,54] is challenging to demonstrate, especially in the face of inevitable inter-site heterogeneity).

In terms of confounding e ffects, there are also many factors that can influence environmental and topographic heterogeneity within eucalypt forests, which might act to mask any relationship and thus cause the field results to deviate from the expectations of the experimental incubations. For example, the realised distribution of *A. dealbata* litter at the field sites might be influenced by localised erosion or mineral dilution of litter and topographic variation, causing the build-up of litter in soil depressions or exclusions from mounds, but which were not reflected in controlled soil collections. The few studies that have examined patterns of litter fall dispersal and collection in field conditions support the notion of high heterogeneity in litter-sample collection, finding large disparities between the location of the litter source (local tree canopy) and the actual litterfall sample that was collected [55]. Similarly, [56] found that under-representation of certain components of vegetation, such as litter, can have important implications for the estimation of forest productivity and nutrient cycling, leading to a scale mismatch between sample points and the character of the soil across the forest.

While this study focussed in detail on N transformations and cycling within a Tasmanian eucalypt forest system, N losses could not be determined explicitly or assessed in the field. Leaching of N from the soil at field sites might contribute to the low N availability; this suggests that future studies should focus on determining how much N is below the 15 cm of topsoil that was collected in our sampling protocol. Stable isotope tracing is one method that could be used to track N pathways through the soil, providing information of N availability and movement at various depth stratifications. If leaching does occur, then it is likely that N will accumulate further down the soil profile. This e ffect might have been exacerbated by the overwhelming dominance of the deep-rooted *E. obliqua* (numerically and in terms of biomass), whose leaves, due to variations in litter characteristics between the two species such as litter size, mass, decomposability and chemical composition [57,58], dominate the surface litter composition, and ye<sup>t</sup> it is able to access historically deposited nutrients that are now only found deep in the soil profile.
