Seed production and abiotic measurements
We collected seeds from surviving focal individuals (n = 609) as they senesced in September and October. We prevented seed dispersal by tying thin mesh organza bags over the flowers shortly prior to seed release. Where a substantial number of seeds were lost from an individual despite these efforts, we used flower head count to estimate viable and inviable seed production based on average production from other individuals in the study (twoG. rosea and eight L. rosea individuals, 2% and 9.5% of each species’ seed production observations). We considered seeds that were filled and lacking disease and predation to be viable.
To quantify the abiotic environment for each plot, we measured litter cover, soil properties, and canopy closure. We used a spherical densiometer in each corner and the centre of each plot to produce average estimates of shade based on the proportion of overhead area that is covered by foliage versus open sky (Jennings et al. 1999). Litter cover was recorded as 0%, <25% (inputted as 12.5% for calculations), 25% or 50% by visual estimate. Soil samples were collected from the centre and two random outer edges of each plot at a 0-15 mm depth (excluding litter) and mixed thoroughly. The soil was air-dried and analysed for a suite of macro- and micro-nutrients by the School of Agriculture and Food Sciences, The University of Queensland.