*2.7. Laboratory Litter Experiment*

The impact of litter quality and quantity on decomposition rate was determined experimentally by manipulating the relative contribution of litter from the three co-occurring species in the field sites (*E. obliqua*, *A. dealbata*, and *P. apetala*). Newly senesced leaves of the three tree species were collected from each of the 24 sites. The leaf litter was combined to give a representative sample of each of the three species across the entire field site and subsequently dried in an oven at 110 ◦C for 24 h. Leaves were shredded in an adapted co ffee grinder and then ground to powder in a ball mill (MM200 Mixer Mill, Retsch, Haan Germany).

We developed seven treatments ranging from 100% contribution of a single species to equal proportions (i.e., an even mixture; Table 1). The total weight of litter was controlled at 2 g, with 10% of the total weight as sieved pasteurised sand (20 g), which eliminated the e ffect of existing soil organic matter interfering with the e ffect of litter amendment on the incubations. Three replicates were prepared for each treatment, each with a di fferent dominant species contributing the largest percentage of litter to the mix, yielding a total of 63 replicates, with three 'no litter' controls. Litter was apportioned into appropriate weights based on the percentage contribution of the species (from 0% to 100% contribution: Table 1) and placed in a 60 mL specimen container in a 500 mL glass jar containing the pasteurised sand. A 2 mL soil slurry solution was added to each jar, being prepared by mixing 50 g (fresh weight) of soil taken from the study site (Section 2.1), with 100 mL of deionised water. The soil was added to the incubation containers with a minimum of additional soil organic matter such that there was a consistency across each treatment. Additionally, samples were supplemented with 2.6 mL of deionised water to bring them to 60% field capacity, which is within the range of soil water content (50%–70%) that has been shown to encourage optimal microbial activity [33]. Field capacity was determined by floodinga3g subsample of soil and allowing it to stand for 30 min before draining the free water on a porous ceramic plate and then weighing to obtain water content at field capacity [34]. The subsample was then oven-dried for 24 h at 105 ◦C to obtain soil dry weight. To control for the respiration rate of the organic matter present in the soil slurry, inoculum blanks consisting of sand and the soil slurry (but no added litter) were also prepared. The sealed jars were incubated in the dark at 25 ◦C. Gas samples were obtained after 1, 2, 4, 7, 10, 14, 18, 23, 30, and 35 days of incubation.


**Table 1.** Relative proportion of litter across the seven treatments (ranging from an even mixture of 1, to one dominated by a single species (0). The three species (*Eucalyptus obliqua* L'Hér, *Acacia dealbata* Link, and *Bedfordia salicina* D.C.) are the most common found across the site.

Daily C mineralisation results for the litter experiment were analysed to determine the amount of potentially mineralisable C using a nonlinear differential model (two-pool model with one pool constant), as follows:

$$\frac{d\mathbb{C}}{dt} = \mathbb{C}\_{\mathbb{d}} k\_{\mathbb{d}} e^{(-k\_{\mathbb{d}}t)} + r,\tag{1}$$

where *dCdt* is the cumulative amount of C mineralised at time *t* (per day in this case), *Ca* is the amount of C in the fast or labile C pool, *kat* is the respiration rate constant of CO2 evolution from the active pool, and *r* is the non-labile soil C respiration rate, which is effectively a constant over timescales less than a year (e.g., [33]). We assumed the labile C pool to have come directly from recent plant inputs (e.g., labile C compounds in litter [35]). Decomposition rates were corrected for CO2 release from the inoculum by subtracting the daily mean CO2 release of inoculum blanks from each experimental unit.

We then explored the di fferences in the rate of decomposition of the labile carbon per day and the proportion of *Acacia* in the litter over time. A linear regression was fitted to the resampled nitrogen composition taken at each proportion of *A. dealbata* (see next section). To determine the di fferences in carbon mineralisation of litter over time, each of three species: *E. obliqua*, *A. dealbata*, and *Bedfordia* sp. were measured for cumulative CO2 evolution during litter incubation over a 40 day period and analysed using the Michaelis–Menten equation (a three-term nonlinear saturating model). Although the Michaelis–Menten equation (MME) is commonly used as a simple representation of enzyme kinetics, it can also serve as a useful two-parameter characterisation of any system that exhibits some type of reaction velocity with a non-linear saturation (via estimation of a maximal rate and half-saturation parameter, with a start at the origin). We agree that soils, being complex mixtures, are unlikely to follow any single, simple reaction dynamics. We instead assume that the overall response will be MME-like but fit the model as a phenomenological rather than overtly mechanistic descriptor of the aggregate system response. The fact that it fits extremely well (See results, *R*<sup>2</sup> ranging from 90%–94%) gives confidence in this approach.
