**3. Results**

### *3.1. Relationship between Size of Trees and Total Basal Area across All 24 Plots*

The average size of *A. dealbata* (biomass) was the best predictor for total basal area across the 24 plots (*df* = 3, *<sup>w</sup>*AIC*c* = 0.32), explaining 12% of the total deviance. There was little information-theoretic support for the saturated model (containing all three variables), nor for the two predictor variables (i) variance of distances between trees (*df* = 3, *<sup>w</sup>*AIC*c* = 0.13) and (ii) proportion of *A. dealbata* (biomass) (*df* = 3, *<sup>w</sup>*AIC*c* = 0.10); these were no better supported than the null expectation. Average tree size of *A. dealbata* (biomass) in additive combination with the variance of distances between trees was the best model for predicting average size of eucalypts (*df* = 4, *<sup>w</sup>*AIC*c* = 0.46, %DE = 29.1), followed by the saturated model (*df* = 5, *<sup>w</sup>*AIC*c* = 0.35, %DE = 36.5). A post hoc analysis suggested that of the two best predictors for average size of eucalypts, variance of distances between trees was 1.6 times better supported than the average size of *A. dealbata* (biomass). See Supplementary Materials for all raw data associated with the linear models and associated analyses.

### *3.2. Local Soil Nutrient Availability*

There was a positive relationship between soil mineral N availability and soil C content (%) (*R*<sup>2</sup> = 0.51, slope = 1.680, 95% CI = 0.4–2.90; Figure 1a), indicating that nitrogen availability was dependent upon the amount of organic matter in the soil. However, neither soil total N (%) content nor available mineral nitrogen concentrations were related to the basal area of *A. dealbata* across the nine sites sampled ( *R*<sup>2</sup> = 0.03, slope = −0.01, 95% CI = −0.03–0.02 and *R*<sup>2</sup> = 0.04, slope = −0.29, 95% CI = −1.40–0.81, respectively; Figure 1). Conversely, *A. dealbata* average size did show a strong negative relationship to soil mineral N availability ( *R*<sup>2</sup> = 0.47, slope = −0.61, 95% CI = −1.10–0.12), but not to soil N content (%) ( *R*<sup>2</sup> = 0.19, slope = −0.01, 95% CI = −0.03–0.01). This result shows that none of the predictors for nutrient availability in the soil were predictive of total basal area of *A. dealbata*.

**Figure 1.** Relationship between total basal area of *A. dealbata* and the amount of nitrogen at each soil site in two different forms: (**a**) Relationship between total soil carbon (%) and available mineral nitrogen (N). Concentrations are determined from soil collected from each of the nine sites in a eucalypt forest. (**b**) Total soil organic matter quantity in terms of total soil nitrogen (%) and the total basal area at each soil site and (**c**) Available mineral N present at each site and the relationship to total basal area of *A. dealbata*.

### *3.3. Local C and N Dynamics*

Since C mineralisation is an enzymatic process, its rate is dependent upon the availability of substrate; as expected, there was a clear relationship in our analysis between C mineralisation rate and soil organic matter content, whether expressed as total soil C or N (*R*<sup>2</sup> = 0.47, slope = 1371, CI % = 275–2467 and *R*<sup>2</sup> = 0.65, slope = 96.0, CI % = 42.4–149.6, Figure 2). Thus, the rate of soil organic matter decomposition was dependent upon the quantity of organic matter present.

**Figure 2.** Relationship between total carbon mineralised (μgCg−<sup>1</sup> soil) and organic matter quality demonstrated as two different forms. (**a**) Total soil nitrogen (N%) *R*<sup>2</sup> = 0.47 and (**b**) total soil carbon (C%) *R*<sup>2</sup> = 0.65. Data were determined from soil collected at nine different sites in a eucalypt forest.

There was a strong positive correlation between field-mineral-N availability and measured nitrification rate in the laboratory (*r* = 0.79). Therefore, the measurement of soil nitrogen availability of field samples corresponded well with measured rates of nitrogen transformation. There was also a strong relationship between rates of N transformation, ammonification, and nitrification (*r* = 0.64), demonstrating that the rate of nitrification was dependent on the rate of net ammonification, and vice versa.

However, there was only a weak correlation between nitrification and total N mineralisation (*r* = 0.33). Indeed, nitrification rates were relatively stable across the range of sites in comparison to the wide range of ammonification rates (Figure 3). There was by contrast, a strong positive relationship between net ammonification and net N mineralisation rates (*r* = 0.94, Figure 3), indicating that the supply of mineral N in this system was related to the conversion of organic N to ammonium.

**Figure 3.** *Cont.*

**Figure 3.** Rates of nitrogen (N) transformation obtained from incubation measurements from soil taken from nine sites. (**a**) Rates of ammonification (mg N g<sup>−</sup><sup>1</sup> soil); as the conversion of C and N (%) into NH4<sup>+</sup>; and the rates of nitrification (mg N g<sup>−</sup><sup>1</sup> soil); as the conversion of NH4<sup>+</sup> into NO3<sup>−</sup>. One standard error (+ and −) is shown for each site measurement. (**b**) Relationship between net N mineralisation rates and net ammonification and net nitrification rates from soil collected from the nine sites.

### *3.4. Relationship between Litter Quality, Quantity, and Litter Decomposition*

Field samples from communities with *A. dealbata* had almost three times more nitrogen in the litter (2.57% ± 0.11%) compared to *E. obliqua* (0.53% ± 0.03%) and *Bedfordia* sp. (0.80% ± 0.05%), and thus a lower C:N ratio. A higher proportion of *A. dealbata* in the litter (%) subsequently led to a higher release of labile carbon during the litter-decomposition experiment (Table 2; Figure 4). Both approaches to fitting this relationship statistically: (i) The median of the resampled points and (ii) the initial linear model to the median points, yielded nearly identical *R*<sup>2</sup> (indicating a negligible effect of pseudoreplication).

**Table 2.** Litter decomposition table showing the intercept, slope, and corresponding 95% confidence intervals for each, the initial model (fitted to the raw data) and the median of model fits to resampled data points; goodness of fit is indicated by the *R*2.


**Figure 4.** The relationship between the rate of labile carbon decomposition and of the nitrogen fixed *A. dealbata* in the litter. The red line represents the initial linear regression model to the median points (*R*2: 81.8%), and the black line shows the median of the resampled points (*R*2: 81.7%). Dotted lines represent the confidence bounds for the resampled data.

The resampled fits of the Michaelis–Menten saturating equation (MM) for the cumulative CO2 evolved from the decomposition of the litter was similar for each of the three species (Table 3; Figure 5). Although both median and resampled fits yielded a similar *R*2, the resampled model did reveal subtle deviations in the data, but had little to no influence on the *R*2, with *A. dealbata* showing the largest variation of *R*<sup>2</sup> between models (1.9% difference). *A. dealbata* had the lowest half-saturation point for the cumulative carbon mineralisation, reaching it faster than *E. obliqua* and *Bedfordia*. However, all species released a comparable total amount of CO2 per gram of litter, as indicated by the similar asymptotes of the MM equation (Table 3).



**Figure 5.** *Cont.*

56

**Figure 5.** Cumulative carbon mineralisation rate over time (40 days) against all data points for each species. The red dotted line represents the Michaelis–Menten equation (MM) fit to the median data points. The black line is the median of the resampled data fits using the MM equation, with corresponding black dotted 95% confidence bounds: (**a**) *E. obliqua* initial MM fit (*R*2: 93.3) and resampled median (*R*2: 93.7%); (**b**) *A. dealbata* initial MM fit (*R*2: 90.9%) and resampled median (*R*2: 92.8%); (**c**) *Bedfordia* sp. initial MM fit (*R*2: 89.1%) and resampled median (*R*2: 90.2%).
