We used an additive design to test the competitive effect of Avena on the individual-level responses of four native plant species, which included six density treatments (i.e. 0, 2, 4, 8, 16, 32 Avena plants) to determine the per capita effects (Goldberg & Scheiner 2001). The 32 plant-density treatment was equivalent to the maximum density of Avena measured at the field site. The design also included P addition (P)/no P addition (NP) as a fixed factor and three replicates of each treatment (n = 4 × 6 × 2 × 3 replicates = 144 pots). Pots, each containing one native plant, were laid out in three blocks, according to replicate and stratified by treatments. The pots measured 8 × 8 × 18 cm (depth) and were filled with soil collected from a eucalypt woodland remnant unaffected by cultivation that was adjacent to the field site. Plant available bicarbonate extractable P (Colwell 1965) in the 0–5 cm layer was ~7 mg kg−1 and the pH in CaCl2 was ~5·1. Prior to use, the soil was sterilized at 65°C for 3 h and air-dried. Phosphate, 5·16 mg P kg−1 soil, was added to the pots at the time of planting, as Ca(H2PO4)2 (Sigma-Aldrich, Castle Hill NSW) because this form of P is readily available to plants. This level represents plant available P in old-field soil (Standish et al. 2006), assuming that two-thirds of the P added is unavailable to plants (Bolland, Allen & Barrow 2003). The Ca(H2PO4)2 was mixed into the top 5 cm of soil. Otherwise, we did not alter the soil fertility.
Our study species dominate the eucalypt woodland that we are aiming to restore. Two species, Acacia acuminata Benth. (jam wattle; Mimosaceae) and Hakea recurva Meisn. (djarnokmurd; Proteaceae), have naturally recolonized old-fields in the region, whereas Eucalyptus loxophleba Benth. subsp. loxophleba (York gum; Myrtaceae) and Allocasuarina campestris (Diels) L.A.S. Johnson (sheoak; Casuarinaceae) have not (Standish et al. 2007a). However, E. loxophleba subsp. loxophleba is one of three eucalypts that account for a majority of trees planted on farmland in the region since the 1920s (Smith 2008).
Seedlings were grown from seeds collected by Greening Australia (Northam, Western Australia). Acacia acuminata seeds were immersed in boiling water for 40 s to break dormancy and those that floated were discarded. Seeds were germinated on filter paper in Petri dishes, the E. loxophleba subsp. loxophleba and A. acuminata at 25 °C and the A. campestris and H. preissii at 18 °C. At emergence, the seedlings were transferred to trays of sterilized soil and watered. Avena seedlings were grown from seeds collected at the field site in October 2004. Seeds were de-husked and pricked to break dormancy and then germinated on filter paper in Petri dishes at 15 °C. At emergence, Avena seedlings were planted together with the native seedlings into the soil-filled pots. Avena seedlings were less than 2 cm tall and native seedlings were less than 5 cm tall at the time of planting.
Plants were watered by overhead sprinklers, everyday in the first 4 weeks and then three times a week thereafter to promote competition for water. The low-water regime was designed to approximate the long-term average rainfall for the first 3 months of the growing season at the climate station nearest to the field site (Bureau of Meteorology 2005). Dead plants were replaced in the first 4 weeks. To negate a block effect, the position of the blocks was re-randomized at four and 8 weeks after implementing the low-water regime.
Plants were harvested 12 weeks after implementing the low-water regime, and when treatment effects were apparent (3 July 2005). The temperature in the glasshouse ranged from 7–37 °C during the experiment, which is within the range of temperatures experienced at the field site during the first 3 months of the growing season (Bureau of Meteorology 2005). Pots with dead native plants (n = 30) were discarded. The survivorship of seedlings (died or survived) was not independent of species; the number of dead E. loxophleba subsp. loxophleba (19) was greater than that of the other species (χ2 = 36·9, d.f. = 3, P < 0·005). Plant shoots (i.e. above-ground material) were oven-dried at 70 °C to a constant weight. The roots were washed and oven-dried at 70 °C to a constant weight. A sub-sample of Avena shoot samples from P and NP treatments were sent to CSBP Soil and Plant Analysis Laboratories (Bibra Lake, Perth) to determine foliar P (n = 34 samples). Samples were ground and foliar P was determined by digesting plant material in nitric acid using a Milestone microwave and then measured by ICP-AES (McQuaker, Brown & Kluckner 1979).
For each native species (except E. loxophleba subsp. loxophleba), a one-factor ancova was used to test for the effect of Avena biomass (shoot + root) and P treatments on native plant (shoot + root) biomass, shoot biomass, root biomass and root fraction (root biomass as a fraction of total biomass). The treatment effects were consistent for each of these dependent variables, and thus, we present the results for plant biomass only. The factor was P treatment and the covariate was Avena shoot biomass. Before these analyses, log(x + 1) transformations of both dependent and response variables were used to help linearize the relationships. The influence of each data point on each of the fitted regression lines was estimated using the Cook's distance statistic Di; we interpreted values greater than one as being particularly influential (Bollen & Jackman 1990). We estimated the effect sizes of the dependent variables on the response variable using the omega-squared measure (Hays 1994) for P treatment and the Pearson product moment correlation for Avena biomass. ancova was invalid for E. loxophleba subsp. loxophleba because a linear model did not fit the data for P-treated plants (R2 = 0·11, F = 1·42, d.f. = 1, 11; P = 0·26) and there were too few NP-treated plants to fit any model. A one-factor anova was used to test for the effect of P treatment on Avena shoot [P], shoot biomass and root biomass. The error terms were normally distributed and the variances were homogeneous for each of the ancova and anova models.
The field experiment was located on an old-field in the central wheat-growing district of south-western Australia (22·3 ha; 31°20′ S, 117°44′ E). The old-field was cleared of native vegetation in 1930 and then cropped until 1990 when it was abandoned. The residual effects of P-fertilizer, soil compaction, erosion, and reduced organic carbon were evident in the old field 14 years after abandonment (Standish et al. 2006). The experiment received 273 mm of rain, which is less than the mean but within one standard deviation of the annual rainfall (323 ± 85 mm) measured within the same period (i.e. August to July the following year; Bureau of Meteorology 2005).
We used a split-plot design to determine the effect of competition from Avena-dominated grassland on the establishment of four native species planted as seedlings. The splitting factor was the presence/absence of Avena, applied to 6 × 3 m plots and replicated five times (n = 10 plots). A second treatment factor, the presence/absence of microcatchments to improve water availability to seedlings (Whisenant et al. 1995), was applied to sub-plots within each plot (n = 2 sub-plots per plot). Microcatchments, one per seedling, were created by removing soil until the depressions were about 10 cm in depth and 10 cm in width; they accumulated leaf litter to a depth of ~1 cm during the experiment. We planted one block of seedlings of each species into each of the two sub-plots in a 2 × 2 block configuration (n = 4 blocks per sub-plot). Each block contained nine seedlings in a 3 × 3 configuration. Seedlings (< 30 cm tall) were purchased from Westgrow Farm Trees (Quellington Road, Meckering) and planted on 4 August 2004. Each sub-plot received 5 L of water at planting, but only rainfall thereafter. Dead seedlings were replaced with live ones on 3 September 2004 but not thereafter. Plots were fenced to protect them from browsing kangaroos (Macropus spp.) and European rabbits Oryctolagus cuniculus.
The analyses included four replicates rather than five as kangaroos jumped the fence around one plot and ate most of the seedlings. We scored the survival of the seedlings and above-ground biomass of the survivors in the remaining eight plots 1 year after planting (n = 219 survivors of 576 planted). Above-ground biomass was determined after drying to constant weight at 70 °C. We used paired one-tail t-tests to determine the effects of the experimental treatments on seedling survival and biomass, assessed at the plot level to avoid pseudoreplication. Our a priori prediction was that the removal of Avena and the presence of microcatchments would have a positive effect on seedling survival and biomass. To compare survival and biomass among species, we used a one-factor anova followed by Tukey's HSD test to determine the pairwise differences between means. All analyses were done using spss (SPSS Inc. 2007).