*2.1. Study Site*

The field sites were located in a tall eucalypt forest at an altitudinal range of 310–440 m above sea level, on the foothills of Mount Wellington, in south-eastern Tasmania (42.895◦ S–147.268◦ E). Vegetation in the region is dominated by *E. obliqua*, a widespread and abundant tall tree found across south-eastern Australia, with the occasional occurrence of *Eucalyptus regnans* F.Muell and *Eucalyptus delegatensis* R.T.Baker. The community also contains two acacia species as sub-dominants: *A. dealbata* and *Acacia melanoxylon* R.Br. The understorey species included *Bedfordia salicina* D.C., *Coprosma quadrifida* (Labill.) B.L.Rob., *Pittosporum bicolor* Hook., and *Pomaderris apetala* Labill. The site is predominately covered by grey-brown podzolic soils on mudstone, classified as Kurosol under the Australian Soil Classification system [24]. Kurosol soil is known as being acidic and nutrient poor [25] with vegetation structure strongly dependent on rainfall [24]. The climate has warm, dry summers and cool, moist winters, with a mean annual precipitation of 720 mm and a mean annual temperature of 17.5 ◦C. We used the basal area of trees per unit area as a proxy for long-term integrated 'productivity' (>four decades of canopy-tree-biomass accumulation). The forest stand is even-aged, being a natural regeneration following a 1967 stand-replacing fire. The few trees that had obviously lived through the fire (e.g., diameter at breast height >100 cm were avoided during field measurements or excluded from analysis). For this study, we were interested in local heterogeneity across space, rather than temporal comparisons across di fferent site histories. Landscape features were homogeneous across sites (e.g., little variation in topography, canopy, understory, and rock cover).

### *2.2. Site Selection and Survey Technique*

To collect data on community structure and to examine the relationship between tree size and total basal area we surveyed 24 sites using a modified point-quarter method [26], where each site was chosen using a random bearing and distance greater than 50 m from a randomly selected starting point located within the forest (Table S2). The tree closest in proximity to this distance and bearing was designated the starting tree. Each starting tree was identified to species level, and diameter at breast height (DBH in cm) was measured. From the starting tree, the nearest neighbour in each of four compass-bearing quadrants (northeast, northwest, southeast and southwest) was selected. These four trees were also identified to species level and their DBH was also measured, with distance from the starting tree determined using a laser meter. This nearest-neighbour selection method was repeated for each of the four initial trees, giving a total of 17 trees (including the starting tree) at each of the 24 sites.

### *2.3. Soil Sampling and Processing*

Soil samples were collected during early autumn from within the study community. Nine of the 24 sites were selected to represent variation in the proportion of *Acacia* (nitrogen fixers) present, relative to total basal area of all trees (Table S1). At each of these *Acacia* sites, five sampling locations were selected using a random bearing (in NW, SE, SW, NE directions) and a random distance from the starting tree. Five soil cores (2 × 15 cm) were randomly placed in a 2 × 2 m quadrant within each location for a total of 25 soil core samples per site. A site's soil samples were then mixed and homogenised by passing through a 2 mm sieve, which removed large roots, litter, and rocks. To determine relative soil water content, a 10 g subsample from each site was dried in an oven at 110 ◦C for 48 h. A 5 g soil subsample from each site was also ground in a ball mill (MM200 Mixer Mill, Retsch GmbH, Haan, Germany) for total C and N determination via a controlled combustion process in a PerkinElmer 2400 Series II Elemental Analyser (PerkinElmer, Australia [27]).

### *2.4. Assessing Field Nutrient Availability*

To determine nutrient availability in the forest community, a 10 g subsample of each sieved soil sample was extracted with 40 mL of 2 M KCl to determine available NH4+ and NO3− concentrations (Table S3). The tubes containing the soil and solution were shaken for 1 h, then centrifuged for 3 min at 4000 rpm. A total of 10 mL of supernatant was removed with a syringe and filtered through a Whatman No. 42, 2 μm syringe filter. The filtered extracts were analysed for NH4+ and NO3− concentrations in a SmartChem Discrete Analyser (Westco Scientific Instruments, Brookfield, CT, USA). Additionally, a portion of the soil was also assessed to determine phosphorus (P) concentrations in the soil at each site using the extraction methods described in [28]: 2 g of soil was mixed with 40 mL 0.005 M NaCl2 solution and shaken for 17 h, then the mixed solution was extracted and analysed using the syringe method described above, with available P concentration determined using a SmartChem Discrete Analyser.

### *2.5. Determining Nitrogen Mineralisation Rate*

The N mineralisation rate of the soil from each of the nine subsampled sites was determined using published protocols [29]. Briefly, 20 g of fresh soil from each composite sample was incubated in a sealed 500 mL glass jar placed in a darkened incubator at 25 ◦C for 34 days. Jars were ventilated regularly to prevent soils becoming anoxic. Net N mineralisation, nitrification, and ammonification were calculated as the difference between initial total available N, NO3<sup>−</sup>, and NH4+ concentrations and those measured following the incubation period (Table S4). The rates of nitrogen mineralisation (mg N g<sup>−</sup><sup>1</sup> soil) in relation to the proportion of *A. dealbata* in the community was analysed from incubation measurements from soil taken from nine sites, as the sum of NH4+ and NO3<sup>−</sup>. Each of the nine sites had a different proportion of *A. dealbata*, and this natural cross-site variation was used to determine whether: (a) There were local-scale differences between nitrogen mineralisation rates and (b) there was a relationship between the proportion of *A. dealbata* and mineralisation rates of nitrogen. Results were analysed with a linear regression, using the resampling technique described below in Section 2.6.
