Improved native understorey establishment in mine waste rock in Australia’s wet–dry tropics

Keywords: emergence, fertiliser, grass, growth, legume, mining rehabilitation, native understorey, revegetation, surface litter.

Mine site rehabilitation in the wet–dry tropics of northern Australia has generally focused on establishing overstorey species and managing weeds. However, for an ecologically sustainable vegetation community capable of supporting fauna biodiversity, establishment of a range of plant species and vegetation types fulfilling different functional roles is needed. Understorey plants are an integral component of natural and rehabilitated ecosystems. Understorey vegetation is important because of its ability to stabilise soil surfaces and reduce erosion (Greene et al. 1994; Singh et al. 2002), reduce raindrop effects on erosion, increase water infiltration into soil (Greene et al. 1994), improve soil fertility through nutrient cycling and by trapping resources (Bell 2001; Lane 2002; Pérez-Fernández et al. 2016) and providing food, shelter and nesting sites for a range of vertebrate and invertebrate fauna (Waters et al. 1997). Furthermore, understorey establishment is crucial to restoring biodiversity of Eucalyptus savanna, because 70% of the flora diversity can be in the understorey (Brennan 2005; J. Nicholson, unpubl. data). Restoring native flora diversity can increase the resilience of rehabilitated systems (Carrick et al. 2015), and is particularly important when rehabilitating areas of high conservation significance.

Re-establishing native understorey from seeds on previously mined land can be challenging. Direct seeding is commonly used for reintroducing species into rehabilitated areas as it is cost- and time-effective (Ceccon et al. 2016); however, establishment from seeds in the field is generally very low (Merritt and Dixon 2011; Grossnickle and Ivetić 2017). Waste-rock environments are typically very harsh, with high surface temperatures and low water availability (Singh et al. 2002). Managing heat stress and water availability during germination and early establishment are fundamental factors for direct seeding success (Chambers and MacMahon 1994; Grossnickle and Ivetić 2017; Pedrini et al. 2019). Litter or mulch added to the surface of mine waste rock have been found to improve soil moisture and moderate temperature, thus aiding seed germination (Facelli and Pickett 1991a; Macdonald et al. 2015). In northern Australia, surface mulch has been found to alleviate extreme micro-climate conditions and improve initial overstorey species establishment on waste-rock material at Ranger mine (Johnston and Milnes 2007) and Pine Creek mine (Saragih 2017). However, leaf litter can also negatively affect germination and seedling growth by creating a physical obstruction, limiting sunlight, and allelopathy (Facelli and Pickett 1991a, 1991b; Izhaki et al. 2000; Hoque et al. 2003; Navarro-Cano et al. 2010; Lorenzo et al. 2011; Quddus et al. 2014).

Sourcing sufficient quantities of quality seeds is another major challenge because commercial supplies of native understorey seeds are limited. Seed collection from wild populations can be difficult because of variation in stand productivity and extent, weather and bushfire impacts, and timing of seed collection (Waters et al. 1997; Cole et al. 2000; Merritt and Dixon 2011). In addition, for projects where seed collection is restricted to certain provenances, seed availability will be further constrained. For example, Ranger Mine revegetation requires seed to be collected within a designated zone of Kakadu National Park (Energy Resources of Australia 2020). These factors affect the quantities of seeds collected and seed quality. Seed supplies, therefore, need to be used efficiently. Application of topsoil can overcome native seed limitations and increase species diversity on rehabilitation sites, if available and managed properly (Koch and Ward 1994; Koch 2007). However, sufficient supplies of topsoil are usually not available for hard-rock mining rehabilitation, as the final landform is often far greater than the initial disturbed surface area. Under these circumstances, it can be necessary to revegetate with seeds directly onto mine waste-rock substrates.

Physical, chemical and biological characteristics of mine waste rock can make establishing understorey difficult. Coarse waste rock has low water-holding capacity, which can cause severe surface drought and stressful growth conditions for plants (Bradshaw and Chadwick 1980; Tordoff et al. 2000; Sheoran et al. 2010). Media with coarse particles can also have poor nutrient retention, and may not provide adequate root–soil contact needed for seedling establishment and survival (Chambers and MacMahon 1994). Chemical characteristics that limit establishment can include low organic matter content, low concentrations of macronutrients, acidity, salinity, and toxic metal concentrations (Singh et al. 2002; Sheoran et al. 2010; Bolan et al. 2017). Mine wastes typically lack soil microorganisms, such as nitrogen-fixing bacteria and mycorrhizal fungi, which limits long-term revegetation by affecting nutrient cycling processes (Reddell and Milnes 1992).

Soil surface treatments can ameliorate physical, chemical and biological limitations of mine waste rock and improve establishment success. Incorporating finer textured material or organic matter can improve physical issues, such as poor water infiltration, poor hydraulic conductivity, high bulk density, and obstructed root growth (Bradshaw 1997; Castillejo and Castelló 2010; Larney and Angers 2012). Organic matter can also improve nutrient limitations, and increase soil microbial activity and diversity (Tordoff et al. 2000; Castillejo and Castelló 2010; Sheoran et al. 2010; Huang et al. 2012; Larney and Angers 2012; Macdonald et al. 2017). Examples of organic amendments to mine waste include municipal solid and green waste (Castillejo and Castelló 2010; Wijesekara et al. 2016), biosolids and animal manure (Larney and Angers 2012; Wijesekara et al. 2016), papermill sludge (Wijesekara et al. 2016), pine bark and sawdust (Giles and Bellairs 2001), wood chips (Read et al. 2004), and hay (Read et al. 2004; Huang et al. 2011). However, at remote sites away from populated areas that have limited access to organic waste material, a mulch of native vegetation can be more appropriate and accessible for restoration use.

Fertiliser application can reduce nutrient limitations. However, revegetation trials using fertiliser to establish native Australian species have shown mixed responses. Phosphorus fertiliser applied at intermediate rates (10–20 kg ha⁻¹) was found to result in the highest species richness at Alcoa bauxite mine in Western Australia, whereas high application rates (40 kg ha⁻¹) resulted in the lowest species richness (Daws et al. 2013); in contrast, nitrogen fertiliser had no effect on species richness, density or composition. At Ranger mine on waste rock, Chandrasekaran et al. (2000) reported that applying fertiliser at high rates increased Eucalyptus species diversity, decreased Eucalyptus density, increased exotic ground cover, and did not influence Acacia or native understorey species.

The Ranger uranium mine, surrounded by, but separate from, the World Heritage-listed Kakadu National Park, ceased processing ore in 2021 and is forecast to complete major rehabilitation works by 2028. The habitat of the Ranger project area is required to be as similar as possible to the adjacent areas of the park, such that, in the opinion of the Minister with the advice of the Supervising Scientist, the rehabilitated area could be incorporated into the Kakadu National Park (Supervising Scientist 2017). Revegetation will use local native plant species, similar in density and abundance to those of the surrounding park land, be sustainable, and not require a significant maintenance regime other than that appropriate to the adjacent areas of Kakadu National Park (Supervising Scientist 2017).

Although understorey species have frequently been used for mining rehabilitation in Australia, optimum protocols for establishing native grasses and native legumes in waste rock in the wet–dry tropics of northern Australia are still unclear. The aim of this study was to investigate whether understorey emergence and growth on Ranger mine waste rock would be improved by physical- and chemical-amelioration treatments applied to the surface of the waste-rock media. To investigate the effects of the treatments where water was not a factor, a shade-house trial was established where the pots with treated waste rock were watered daily. To investigate the physical- and chemical-amelioration treatments under field conditions, they were also applied to a waste-rock field site at the Ranger mine, where minimal irrigation was applied. This work will contribute to understanding impediments and suggest treatments to enhance understorey diversity for rehabilitation after mining in northern Australia.

Ranger uranium mine is located 8 km east of Jabiru and 260 km south-east of Darwin, Northern Territory, Australia (12°40′49″S, 132°53′46″E). The wet–dry tropical climate at Ranger mine has a mean annual rainfall of 1554 mm, with 95% of the rain falling in the wet season between November to April. Mean maximum temperatures range from 37.6°C in October to 31.8°C in June; mean minimum temperature ranges from 18.7°C in July to 25°C in November–December (Jabiru airport; Bureau of Meteorology 2020).

The in situ understorey trials were established on the Ranger Trial Landform (TLF), which was constructed from waste-rock material in late 2008 and early 2009 to test rehabilitation techniques. The TLF study area was in 100% waste-rock substrate that had been planted with tubestock of mid-storey and overstorey species in 2009 and 2010. Shade-house trials were at Charles Darwin University (CDU), Darwin (12°22′14″S, 130°51′53″E).

Soil analyses were undertaken for six waste-rock samples collected from the surface 10 cm of material at the Ranger TLF study site in May 2018 and for three samples of the Ranger waste-rock material sent to CDU for the shade-house trial. The fraction <2 mm in diameter was sent for soil chemistry and texture analyses (Nutrient Advantage, Werribee, Vic., Australia). Samples were analysed for texture (Gee and Or 2002), pH (1:5 water), pH (1:5 CaCl₂), EC (1:5 water), organic carbon, ammonium, nitrate, organic carbon, potassium, phosphorus (Cowell), chloride, calcium, potassium, magnesium, sodium, aluminium, cation exchange capacity (CEC), sodicity, copper, iron, manganese, boron and sulfur (Supplementary Table S1). Symbiotic microorganisms, Rhizobium and mycorrhizal fungi, were initially poor to absent on the waste-rock material (Reddell and Milnes 1992), but have recently been returning to the site (Energy Resources of Australia 2020).

Six amelioration treatments were investigated. The Ranger waste-rock material has been reported to have high proportions of rock fragments; so, a treatment of 1 cm of fine sand on top of waste rock was trialled. Because there have been mixed findings regarding fertiliser effects on native vegetation establishment in mine-waste material, a nitrogen and phosphorus blend fertiliser treatment was investigated. Two different types of organic matter were investigated because the quality, nutrient content, and size of the matter can have different effects on germination and establishment (Huang et al. 2011, 2012; Quddus et al. 2014; Wijesekara et al. 2016). One treatment was coarse leaf litter placed on the surface of the waste rock, the other was crushed leaf litter incorporated into the top 3 cm of waste rock. A combination of the fertiliser, sand and coarse litter treatments was also used (crushed litter was excluded so as to not double the litter content). The control was waste rock alone. Full details of the applied treatments are provided below for the shade house and field trial.

Four grass (Poaceae) species, namely, Alloteropsis semialata (R.Br.) Hitchc., Aristida holathera Domin, Eriachne armittii F.Muell. ex Benth. and E. obtusa R.Br., and four legume (Fabaceae) species Acacia gonocarpa F.Muell., Galactia tenuiflora (Klein ex Willd.) Wight and Arn., Indigofera saxicola F.Muell. ex Benth. and Tephrosia oblongata R.Br. ex Benth., were used for the study. These species were selected on the basis of three criteria. Species needed to be native to the study area, have low to moderate biomass to reduce fuel loads and minimise fire risk, and have sufficient number of germinable seeds. Perennial species were preferred because they provide ongoing ecosystem services.

The grass seeds did not require treatment; however, the legumes were found to have higher germination when moist-heat treated. Seeds of G. tenuiflora, I. saxicola and T. oblongata were submerged in 80°C water for 1 min, and A. gonocarpa was boiled in 100°C water for 1 min (Table S2).

The shade-house trials used all eight species. Because of limited seed availability, only three of the grasses, namely, A. semialata, E. armittii, and E. obtusa, and two of the legumes, namely, G. tenuiflora and I. saxicola, were used for the field trial.

In April 2018, a trial in shade houses at CDU had six media treatments applied to the Ranger waste rock (Fig. 1). There were five replicate pots per species for each treatment, except for T. oblongata, which had four replicates per treatment because of low seed numbers. For A. semialata, A. holathera, E. armittii, E. obtusa, G. tenuiflora and I. saxicola, 50 seeds per pot were sown. Owing to low seed numbers, A. gonocarpa had 45 seeds per pot and T. oblongata had 42 seeds per pot. The legume seeds were treated before sowing.

Fig. 1.  Waste-rock media amelioration treatments.

The waste rock was sieved using a 16-mm sieve and coarse rock material was removed. Pots of 125-mm diameter had a layer of geotextile fabric (Grunt’s Landscape Fabric, Northland, Melbourne, Vic., Australia) placed at the bottom to prevent loss of material from the pots.

For the control treatments, each pot was filled with 2.129 kg of waste rock up to 1 cm from the top and tapped on the ground three times. The waste-rock surface was lightly disturbed using a fork, seeds were sprinkled evenly, and the surface was lightly smoothed over by hand.

Fertiliser treatments were set up the same as the controls, except 0.129 g of Osmocote landscape formula all-purpose fertiliser (Scotts Australia Pty Ltd, Sydney, NSW, Australia) and 0.110 g of Richgro super phosphate high-phosphate fertiliser (Richgro Garden Products, Perth, WA, Australia) were evenly distributed on the surface of the waste rock before the surface was smoothed over. The fertiliser rate of 0.5 g m⁻² (10% N, 9% P) was based on Lane (2002).

For the sand treatments, each pot was filled with 1.935 kg of waste rock up to 2 cm from the top and then filled with 194 g of builder’s sand to 1 cm from the top; the sand was sourced from Ranger mine. The sand was lightly disturbed before seeds were distributed, then lightly smoothed again.

Litter treatments had the pots filled with 1.935 kg of waste rock up to 2 cm from the top; seeds were distributed and the surface lightly smoothed before 8 g of litter was added at 1 cm thickness. The litter was sourced from the riplines on the TLF and was predominantly a mixture of Eucalyptus, Corymbia and Acacia dry leaves and twigs.

For the organic matter treatment (OM), the TLF litter was crushed using a Ryobi 2400W electric mulching shredder (Techtronic Industries, Kwai Chung, Hong Kong, PR China), then sieved to <5 mm. The mulch was incorporated as organic matter into the waste rock. Pots were filled with waste rock up to 5.5 cm from the top; then a mix of 38.5 g of organic matter and 385 g of waste rock was added to the pot, up to 1 cm from the top. This equated to 10% organic matter by weight of waste rock, as also used by Huang et al. (2011). The surface was lightly disturbed, seeds were evenly distributed, and the surface was smoothed over.

The mixed treatment was a combination of sand, fertiliser and surface litter. The pots were filled with 1.258 kg of waste rock up to 3 cm from the top; 196 g of sand was then added up to 2 cm from the top and was lightly disturbed before seeds and fertiliser were distributed evenly. The sand was lightly smoothed before 8 g of litter was added at 1-cm thickness, up to 1 cm from the top.

Pots were watered to saturation using sprinkler irrigation twice daily at 04:30 and 16:30 hours for 10 min throughout the trial. Pots were free-draining on mesh benches.

Seeds were considered to have emerged when the radicle or plumule was observed emerging from the testa or soil, or in the case of treatments with litter, when the shoot had penetrated through the litter. Emergence data were recorded twice per week for the first 8 weeks, then once per fortnight for the following 14 weeks. All emergence data were adjusted, so that they were based on the proportion of germinable seeds. The germinable seeds ranged from 35 ± 7 to 61 ± 7% of the applied seeds.

Four weeks after sowing, the seedlings were thinned to four plants per pot to prevent over-crowding and resource competition. Thinning was performed at 4 weeks as the seedling emergence rate had reached a plateau. Any seedlings that emerged after thinning were recorded then removed. If a pot had fewer than four plants, any seedlings that emerged were kept; once a pot reached four plants, any further emerging seedlings were recorded and removed.

Plant heights were measured weekly from Weeks 3 to Week 8, then fortnightly from Week 10 to Week 20. Legume heights were measured from the surface of the medium to the tallest green point on the apical meristem; grass heights were measured to the tallest green point on the tallest leaf.

Directly sown trials were set up at the TLF in April 2018. Seeds were sown after the monsoon rains because this has been shown to result in better survival (Fawcett 1995; Lane 2002). Rainfall from January to March 2018 had 44% more rain than average and therefore sowing was delayed until April to avoid heavy late wet-season rains. After 20 April 2018, there were only minor rainfall events, totalling 2.0 mm of total rainfall up to 1 November 2018.

In total, 240 potential plot locations, each 25 × 50 cm, were identified. These potential plots were selected on the basis of each plot occurring in a location with a high proportion of fines, without rocks greater than 5 cm in diameter, and in the flat area between the riplines. Locations that had high proportions of organic litter or were less than 1 m from an established plant were then excluded so as to not bias trials. From the potential plots that met these criteria, 180 plots were randomly selected and the experimental treatments were applied. Plots were randomly allocated to one of the six treatments and five species, providing five independent replicate plots per treatment and species combination.

The six treatments (control, fertiliser, sand, litter, OM, mixed) were randomly allocated to plots, with six replicates per treatment and species. Each plot had 100 seeds sown for the grasses, A. semialata, E. armittii and E. obtusa, and 50 seeds sown for the legumes, G. tenuiflora and I. saxicola.

The TLF treatments were similar to the shade-house treatments, with the quantity of materials adjusted to account for the larger plot area. The control plots were lightly disturbed with a hand-rake; the seeds were evenly sprinkled in the plot before the surface was smoothed over again by hand. The fertiliser plots had 1.875 g of Osmocote and 1.6 g of super phosphate fertiliser applied when the seeds were distributed, after which the plots were smoothed over again. The sand treatment had 2 kg of builder’s sand spread at 1-cm thickness on top of the waste rock in the 25- × 50-cm plot; the sand was then disturbed, seeds were distributed, and the plot was smoothed over. The surface-litter plots used 750 g of litter, which was spread at 1-cm thickness over a 50- × 100-cm area on top of the 25- × 50-cm plot to create an ‘edge’ for each plot. The OM plots used 750 g of <0.5-cm mulched, sieved litter thoroughly mixed with the surface 3 cm of waste rock (∼7.5 kg), then the combined rock and OM was returned to the plot. The mixed-treatment plots first had sand spread at 1-cm thickness, then the fertiliser and seeds were evenly distributed, and finally 750 g of litter was added following the same method as for the litter plots.

During the first 9 weeks of the TLF trial, each plot was watered by hand two or three times per week for 15 s with 1 L of water (∼8 mm per water event). After 9 weeks, an irrigation system was set up and the whole TLF area was watered for 45 min with ∼4.5 mm of water per event, three times per week. Generally, on the TLF, seedlings show mild signs of wilting in the dry season if not irrigated for 4 days (M. L. Parry, pers. obs.). Based on soil moisture data obtained for tree seedlings on the TLF, young seedlings would have experienced intermittent water stress (P. Lu, unpubl. data). The TLF was weeded when necessary throughout the trial period, taking care not to disturb the establishing understorey plants.

At the TLF, emergence data were recorded 3, 5, 7, 9, 13, 17 and 21 weeks after sowing. Plant heights were measured from Week 9 on a monthly basis for 4 months. When there were more than 10 plants in a plot, 10 plants were initially randomly selected and the same plant was repeatedly measured. Emergence proportions were based on the germinable seeds applied.

At the TLF, the distances from the centre of each plot to the base of the nearest tree with a diameter at breast height >5 cm were measured with a Stanley TruLaser TLM 100i (STANLEY, Mississauga, ON, Canada). Canopy cover above each plot was determined using a digital photo and the plant image analysis software Easy Leaf Area (see www.quantitative-plant.org/software/easy-leaf-area, accessed 2018; Easlon and Bloom 2014). Canopy photos were taken mid-morning in July 2018 at 50-cm height by using a digital camera on tripod and were vertical as determined by a spirit level.

To test treatment effects on emergence, a generalised linear mixed model (GLMM) was performed in the IBM SPSS Statistics program (ver. 24, IBM, Armonk, NY, USA). The binomial logistic regression was used with outcome emergence and number of seeds as the denominator, with the control treatment as the reference. Treatment and time were used as fixed effects, time as a repeated-measures variable, and pot per plot as a random effect. Selection of the model covariance structure was performed and the model with the lowest Akaike information criterion (AIC) was chosen. A compound symmetry covariance structure was used for the TLF species. With the shade-house species, the compound symmetry covariance structure had the lowest AIC for E. obtusa, I. saxicola and T. oblongata; the other shade-house species had lowest AICs when a diagonal covariance structure was used. The analyses were originally run to test for an interaction effect between time and treatment. However, no significant interaction effects were found for any of the species; therefore, the GLMM was run without the interaction analysis, which resulted in a lower AIC. If the main test for treatment was significant, post hoc pairwise comparisons were performed to test for significant differences among treatments. The P-values were adjusted to account for multiple comparisons using the Sidak correction method option in SPSS (Abdi 2007).

Differences in height over time among treatments were determined using permutational ANOVA (PERMANOVA) and permutational analysis of homogeneity of dispersions (PERMDISP) in the program PRIMER-E (ver. 7, see https://www.primer-e.com/; Anderson et al. 2008). The model design had four factors. Time and treatment were fixed factors, and pot per plot and plant were random factors; plant was nested in pot per plot, which was in turn nested in treatment. A Euclidean distance measure was used with Type III sums of squares and 999 permutations of the raw data. P-values of <0.05 were considered significant. PERMDISPs were used to test for differences in the dispersion of data among the sample groups. When a significant time and treatment interaction effect was found, post hoc pairwise comparisons were performed to determine which treatments differed significantly from each other and when. A more conservative P-value of <0.01 was used in the pairwise comparisons to account for multiple testing. Differences in inflorescence production in the shade house was similarly analysed using PERMANOVAs and PERMDISPs.

The relationship between seed mass and seedling emergence was analysed using linear regression for each treatment separately.

Two species, namely, A. semialata and G. tenuiflora, had sufficient emergence at the TLF to determine the relationships between emergence and tree distance, and between emergence and canopy cover. The effect of distance to nearest tree and canopy cover leaf area on seedling cumulative emergence at Week 21 were tested by adding these as additional fixed factors to the directly sown A. semialata and G. tenuiflora GLMM models.

Media characteristics were similar for the TLF and the CDU waste-rock materials. Texture was a loamy sand. It was alkaline, with pH (1:5 water) 8.6 ± 0.2 for TLF, and 8.1 ± 0.1 for CDU, with low electrical conductivity of 0.08 ± 0.02 dS m⁻¹ for TLF and 0.08 ± 0.00 dS m⁻¹ for CDU, low sodicity, very low concentrations of organic carbon, available nitrogen and phosphorus, low concentrations of potassium, low CEC and low to very low concentrations of the micronutrients manganese, zinc and boron (Table S1).

In the shade-house trials, treatments significantly affected emergence of most of the understorey species, but not E. armittii or E. obtusa (Fig. 2). None of the treatments was consistently beneficial for seedling emergence compared with the control. Surface litter had the highest mean emergence for four of the eight species; however, only A. gonocarpa had significantly greater emergence than the control (P < 0.05). Acacia gonocarpa emergence was significantly greater in litter than all other treatments, excluding the mixed treatment, which included litter.

Fig. 2.  Emergence in amelioration treatments at the CDU shade house for (a) Alloteropsis semialata, (b) Aristida holathera, (c) Eriachne armittii, (d) Eriachne obtusa, (e) Acacia gonocarpa, (f) Galactia tenuiflora, (g) Indigofera saxicola and (h) Tephrosia oblongata. Emergence is mean percentage of viable seeds and bars show s.e. Different letters indicate significant differences.

Generally, the OM and sand treatments resulted in a lower seedling emergence for the two grasses and four legumes (Fig. 2). OM treatments had a significantly lower seedling emergence than did all other treatments with A. semialata and A. holathera, and most of the other treatments with the four legumes. Sand treatments also had a significantly lower seedling emergence than did most other treatments for the legumes and had a significantly lower emergence than did control treatments for A. semialata.

Emergence levels varied greatly across species and significantly among treatments. Grass emergence ranged between 5 and 97% of viable seeds sown and legume emergence ranged between 15 and 104% of viable seeds sown (Fig. 2). Seedlings continued to emerge until the end of the trial at 20 weeks, but the majority of seedling emergence occurred in the first 4 weeks after sowing.

At the TLF, treatment had a highly significant effect on the number of seedlings that emerged for three of the four species that germinated (Fig. 3). Surface-litter treatments had the highest mean emergence for all species. Galactia tenuiflora had a significantly greater emergence from surface-litter than all other treatments. Alloteropsis semialata had a significantly higher emergence from the litter treatment than from sand and OM. Indigofera saxicola had a significantly higher emergence from litter than from the sand, OM and mixed treatments.

Fig. 3.  Emergence in amelioration treatments at the trial landform for (a) Alloteropsis semialata, (b) Eriachne armittii, (c) Galactia tenuiflora and (d) Indigofera saxicola. Emergence is the mean percentage of viable seeds and bars show s.e. Different letters indicate significant differences.

Sand, OM and mixed treatments generally had the lowest seedling emergence on the TLF (Fig. 3). Sand had no emergence for both legumes. The OM treatment had no emergence of A. semialata and a significantly lower emergence than the control for G. tenuiflora. The mixed treatment had no emergence for E. armittii and I. saxicola.

Understorey seedling emergence at the Ranger TLF was generally low (Fig. 3). Grass emergence ranged between 0 and 11% of viable seeds sown and legume emergence ranged between 0 and 45% of viable seeds sown. E. obtusa had no seedling emergence on the TLF. Seedlings continued to emerge until the end of the trial at 21 weeks; however, the majority of seedling emergence occurred in the first 5 weeks after sowing.

In the shade house, treatment had a significant effect on plant height over 20 weeks. For the majority of the trial, plants in mixed and fertiliser treatments were significantly taller than those in the other treatments (P < 0.01; Fig. 4). Sand and control treatment plants were consistently shortest throughout the study for all species. OM plants were also among the shortest for the first 10–12 weeks of the trial; however, by 14 weeks, the OM plants began growing at faster rates for all of the species, except G. tenuiflora. By Week 20, plants in OM treatments were significantly taller than were control or sand plants for all the species except A. holathera and G. tenuiflora. Two of the four grass species showed a reduction in height during the second half of the trial due to leaf tip death. Although most of the treatments experienced some degree of leaf death, the taller fertiliser and mixed plants were the most affected.

Fig. 4.  Height of surviving plants in amelioration treatments at the CDU shade house for (a) Alloteropsis semialata, (b) Aristida holathera, (c) Eriachne armittii, (d) Eriachne obtusa, (e) Acacia gonocarpa, (f) Galactia tenuiflora, (g) Indigofera saxicola and (h) Tephrosia oblongata. Error bars show s.e.

The following three species produced inflorescences during the CDU trial: A. holathera, E. armittii and E. obtusa (Fig. 5). Treatment had a significant effect on the number of inflorescences and spikelets produced by the plants of these species (P < 0.01 for all, except for E. obtusa spikelets P < 0.05), with plants in the fertiliser and mixed treatments producing significantly more inflorescences and spikelets than those in the other treatments. Aristida holathera had an average of 14.6 ± 1.3 inflorescences in the fertiliser and 17.4 ± 1.6 inflorescences in the mixed treatments, compared with averages ranging from 4.2 ± 0.4 to 5.8 ± 0.4 inflorescences in the other treatments. Eriachne armittii had an average of 5.2 ± 0.8 inflorescences in the fertiliser and 4.8 ± 1.0 inflorescences in the mixed treatments, compared with averages ranging from 0.6 ± 0.2 to 2.8 ± 0.4 inflorescences in the other treatments. For E. obtusa, plants in the fertiliser (2.7 ± 0.7 inflorescences) and mixed treatments (3.2 ± 1.0 inflorescences) were the only ones to develop inflorescences.

Fig. 5.  Mean number of inflorescences and spikelets on plants in amelioration treatments at the CDU shade house for (a) Aristida holathera, (b) Eriachne armittii and (c) Eriachne obtusa. Error bars show s.e. and different letters indicate significant differences.

At the Ranger TLF, the only significant effect for treatment on height was that for G. tenuiflora plants (P < 0.05). Galactia tenuiflora plants in the surface-litter treatment were significantly taller than those in the fertiliser and control treatments (Fig. 6). For two of the other three species, surface-litter treatment had the tallest mean plant heights of the treatments, but the differences were not significant. No plants in the field trial produced inflorescences.

Fig. 6.  Height of surviving directly sown plants in different amelioration treatments at the trial landform for (a) Alloteropsis semialata, (b) Eriachne armittii, (c) Galactia tenuiflora and (d) Indigofera saxicola. With E. armittii, the plant heights stop abruptly for the sand and control treatments because there were no surviving plants left at the next measurement time. Error bars show s.e.

Seed mass affected seedling emergence. There was a highly significant effect of seed mass on seedling emergence in control treatments in the shade house (P < 0.001) and at the TLF (P < 0.0001), with the greatest effect for surface-litter treatments at the TLF (P < 10⁻⁹). At the TLF, species with seeds less than 3 mg had a reduced emergence, and the species with the heaviest seeds, i.e. Galactia tenuiflora, had the highest mean emergence (Fig. 7). The shade-house species with seeds less than 3 mg had a reduced emergence (Alloteropsis semialata (56 ± 3%), Eriachne armittii (46 ± 8%), Eriachne obtusa (10 ± 2%)) compared with seeds greater than 3 mg (Acacia gonocarpa (82 ± 10%), Aristida holathera (94 ± 5%), Galactia tenuiflora (84 ± 9%), Indigofera saxicola (85 ± 11%), Tephrosia oblongata (92 ± 7%)). Control data are shown because other treatments were not significantly greater than the control in the shade house, except for Acacia gonocarpa.

Fig. 7.  Relationship between seed mass and seedling emergence in (a) control and (b) surface-litter treatments at the Ranger Trial Landform. Emergence is the mean percentage of viable seeds.

Distance to nearest tree and canopy cover affected A. semialata emergence (Fig. 8). Seedling emergence increased significantly when seeds were sown closer to trees (P < 0.05), or under greater canopy cover (P < 0.05). Neither factor significantly affected G. tenuiflora emergence.

Fig. 8.  Relationships between canopy cover leaf area and distance to the nearest tree and seedling emergence at the TLF for (a) Alloteropsis semialata and (b) Galactia tenuiflora. Emergence is the mean percentage of viable seeds.

In this study, we found that physical and chemical treatments could improve understorey establishment on waste-rock media, with surface litter improving establishment on waste-rock media in the field. Seedling emergence rates were significantly lower on the waste-rock landform than that in the shade house. On the waste rock, lack of water is an obvious stress, even when the seedlings are being watered three times per week. Seedlings on the TLF show mild signs of wilting in the dry season if not irrigated for 4 days (M. L. Parry, pers. obs.). Under these conditions, the litter treatment had the highest emergence rate. Surface litter presumably improved seedling establishment on the waste rock by creating a suitable microclimate that retained moisture and reduced or moderated temperature (Fowler 1986; Facelli and Pickett 1991a; Commander et al. 2013). In the wet–dry tropics of northern Australia, little to no rainfall is experienced for 6 months of the year and soil temperatures can reach up to 50–60°C on a waste-rock surface (P. Lu, unpubl. data). These conditions create a highly desiccating environment for young seedlings. A meta-analysis by Loydi et al. (2013), which involved 46 studies where plant litter was manipulated in herbaceous communities, suggested that in dry grasslands or under water-limited conditions, litter may have a positive effect on seedling emergence, biomass and survival. Manipulative experiments have also shown that under intermittently dry conditions, grass and herbaceous seedling emergence and growth are improved by litter because it lowers maximum temperatures, increases shade, reduces evaporation, and offers protection from desiccation (Fowler 1986; Eckstein and Donath 2005). In northern Australia, leaf litter on the surface of waste rock improved overstorey seedling establishment at Pine Creek mine, Northern Territory (Saragih 2017), and the addition of surface mulch resulted in a significantly greater ground cover revegetation at Mount Morgan mine in central Queensland (Read et al. 2004). At Ranger mine, Eucalyptus mulch on top of waste rock increased overstorey species seedling survival and establishment rates by up to 40 times (Johnston and Milnes 2007). In other situations, surface litter has had a deleterious effect on seedling emergence as a result of physically obstructing growth, limiting sunlight and allelopathy (Bosy and Reader 1995; Xiong and Nilsson 1999; Izhaki et al. 2000; Hoque et al. 2003; Navarro-Cano et al. 2010; Lorenzo et al. 2011; Quddus et al. 2014). Under intermittently dry conditions with extreme temperatures, as experienced at mine sites in northern Australia, surface litter may create suitable microsites for native understorey seedling emergence and be beneficial for growth.

Surface litter may also improve seedling establishment in the field as a result of increased protection from seed predation. Seed eating by ants, as well as other insects including gryllid crickets, cockroaches and tenebrionid beetles, has been observed at Ranger mine, with significantly more eating incidents on disturbed and waste-rock sites than on natural sites (Andersen and Morrison 1998). It is possible that surface litter helped conceal seeds in this study, which reduced their discovery by predators (Sydes and Grime 1981; Cintra 1997). Direct seeding experiments in seasonal dry tropical forests have found that buried seeds are less predated by ants (Woods and Elliott 2004). However, it has also been suggested that ground cover may increase seed predation by providing habitat for seed predators, therefore reducing seedling emergence (Reader 1991).

Reduced erosion may be another explanation as to why litter treatments had greater seedling establishment in the field. Wet–dry tropical northern Australia experiences annual periods of monsoonal and flooding rainfall; surface litter may benefit native understorey seedling establishment by absorbing rainfall energy and decreasing erosion, thereby reducing seed wash, root exposure, seedling damage and uprooting (Chambers and MacMahon 1994; Read et al. 2004). Although most of the field trial was conducted during the dry season, a short but heavy rainfall event hit the TLF 1 day after the direct seeding trials were set up (34.8 mm at Jabiru Airport over ∼1 h; Bureau of Meteorology 2018). The surface litter may have protected seeds from being washed away during this event, resulting in greater seedling emergence.

The surface-litter treatment in the shade house did not generally improve understorey seedling emergence or growth. This is consistent with other studies that have found that under constantly humid or wet conditions, surface litter has a negative or neutral effect on seedling establishment (Eckstein and Donath 2005; Loydi et al. 2013). This may be because the controlled conditions in shade-house experiments nullify the positive effects of surface litter, which are most pronounced under water-limited conditions (Loydi et al. 2013). Furthermore, because litter maintains high soil moisture and reduces evaporation, it may promote the development of pathogens if moisture levels are excessive over a prolonged period of time, which can increase seed mortality (Facelli et al. 1999; Loydi et al. 2013).

Fine sand and ground organic matter incorporated into waste rock had a deleterious effect on native understorey seedling emergence. This may be due to surface crusting, which was observed with both treatments in the field and shade-house trials. Fine-textured soils and mine spoils have a tendency to form surface crusts (Sheoran et al. 2010). Although the seeds in this study were covered only by millimetres of media, the crust may have prevented seedlings from emerging through the surface (Forcella et al. 2000), or inhibited water infiltration (Daniels and Amos 1985). However, the OM treatment was not made of particularly fine material, so it seems unlikely that this was the reason for a crust to form. Furthermore, Grigg et al. (2006) found that the addition of organic matter to mine spoil decreased surface crusting of tailings waste. It is also possible that elevated organic matter in moist shade-house conditions could promote fungal disease and affect seed viability or seedling mortality. However, this would only explain why emergence was low in the shade house, because conditions on the TLF were desiccating rather than moist. Additionally, if organic matter was promoting seed rot, similar negative effects on seedling emergence would be expected with the other organic matter treatment, i.e. surface-litter treatment. However, surface litter had a neutral to positive effect on seedling emergence. Another possible explanation for the low OM emergence could be that the organic matter had some hydrophobic properties (Mcghie and Posner 1980). It may be that the ground tree litter incorporated into the sand-textured waste rock created a hydrophobic soil surface, similar to the non-wetting sands in western and southern Australia (Mcghie and Posner 1980; Franco et al. 1995, 2000), which inhibit seed germination. It is also possible that the crushed Acacia and Eucalyptus leaves had an allelopathic effect and inhibited seed germination, and root and shoot growth (Hoque et al. 2003; Lorenzo et al. 2011; Quddus et al. 2014).

Fertiliser treatments improved seedling growth in the shade house, indicating that nutrients were a limiting factor for understorey establishment. Nitrogen is an essential plant growth nutrient, and deficient plants have stunted growth and smaller, less succulent leaves that have small cells and thick walls (Russell 1973). This in turn means less available surface for photosynthesis, further affecting plant growth. Because the waste rock has lower nitrogen concentrations than do the undisturbed soils where the understorey species naturally occur (Ashwath 1994), nitrogen deficiency was likely to be a growth-limiting factor for understorey establishment, which was improved by fertiliser application. The waste rock also has low concentrations of potassium and phosphorus, although similarly low concentrations of potassium and phosphorus have been recorded in the surrounding natural soils (Ashwath 1994). Although fertiliser treatments significantly improved understorey seedling growth in the shade house, fertiliser had no effect in the field trials. This suggests that under well watered conditions, understorey plant growth in the waste-rock media is affected by nutrient deficiency, but under desiccating field conditions, water supply is the main growth-limiting factor.

Organic matter treatments to waste rock may improve understorey plant growth. In the shade house, the OM plants had stunted height during the first half of the trial, then after 12 weeks they grew considerably. It is assumed that the ground organic matter was largely intact during the first half of the study and provided little benefit to the seedlings. Then, when the organic matter began breaking down and decomposing, nutrients such as nitrogen were released into the waste rock, increasing seedling growth (Landon 2014). The surface-litter plants in the shade house did not show the same growth pattern as did the OM plants. This is likely to be due to the surface litter being larger and coarser than the ground organic matter, therefore taking longer to break down and release nutrients. Although the OM treatment significantly improved the height of some of the plants in the shade house, this was not evident with the TLF trials. This supports the belief that water, not nutrients, is the first and foremost limiting factor for understorey plant growth in waste rock in the field.

Further research on improving understorey establishment on mine-waste material is needed, particularly in situ field trials because shade house and field results can vary considerably. In this study, surface litter resulted in the greatest seedling emergence in the field; however, growth was still considerably stunted compared with fertiliser-treated plants in the shade house. We recommend that a combination of surface litter and fertiliser should be investigated in field trials, because this may improve surface conditions and ameliorate nutrient limitations, thus increasing understorey emergence and growth. A mixed treatment with both surface litter and fertiliser was already trialled in this study; however, it also included fine sand which is likely to have confounded the positive effects of the other two treatments on the basis of the poor performance of the sand-only pots and plots. It is also recommended that different amounts of surface litter are investigated. Although all of the species had the highest mean emergence in the field with surface litter, emergence of the smaller-seeded species may have been greater with a thinner cover of leaves. A thinner layer of surface litter would still likely improve the seedling microclimate, but may also create less of a physical barrier for seedling emergence. Last, further research is needed on the timing of seeding. Although this study deliberately seeded in the late wet season to avoid flooding rains, future trials may see improved emergence and survival if seeding earlier in the season.

Surface-amelioration treatments can create a favourable micro-environment that aids the successful introduction of native understorey on mine waste and improves seed use efficiency. In situations where there is minimal organic matter and no existing understorey, seeds and young seedlings are vulnerable to the harsh conditions of waste-rock landforms. This may be remediated by application of litter or mulch treatments. Seedling emergence and establishment requires an appropriate micro-niche and without amelioration it may take years, even decades, for a revegetated ecosystem to naturally develop suitable conditions for a diverse understorey to colonise or establish from seeds.

Supplementary material is available online.

The data that support this study will be shared upon reasonable request to the corresponding author.

The authors declare that they have no conflicts of interest.

M. Parry was supported by Charles Darwin University scholarships. This research did not receive any specific funding.