Low canopy cover in a sown Digitaria eriantha Steud. pasture improves establishment of Trifolium subterraneum L.

ABSTRACT

In the summer-dominant rainfall zone of inland northern New South Wales and southern Queensland, tropical grass-temperate legume mixed pastures can provide forage for grazing livestock for most of the year; however, the effect of canopy cover and its interaction with ground cover for regeneration of Trifolium subterraneum (subterranean clover) are not known. An experiment was conducted using established Digitaria eriantha with two levels of canopy cover (defoliated to 0.1 m and undefoliated) and four levels of ground cover (ranging from bare to 75%). Plots were oversown with subterranean clover seed in May–June, and legume establishment and spring herbage production were assessed. Low rainfall throughout the experiment resulted in poor clover establishment and production but seedling densities were significantly higher in the defoliated treatments (14 cf. 6 seedlings/m²). Herbage production was also significantly higher in the defoliated plots (530 cf. 302 kg DM/ha). There was no effect of ground cover. These findings indicate that management strategies that remove excess forage and reduce canopy cover of tropical grass pastures in late summer will maximise regeneration of subterranean clover in autumn.

KEYWORDS:

• Digit grass
• competition
• shade

Introduction

Inland northern New South Wales (NSW) and southern Queensland form a frost-prone zone which has a summer-dominant rainfall pattern with 60%–70% of rainfall occurring during October to March. The zone is suitable for both tropical and temperate pasture and forage species (e.g. Clarkson et al. 1991; Boschma et al. 2009; Murphy et al. 2014), but tropical perennial grasses are common throughout the zone as both native endemic and sown pastures (e.g. Murphy and Sanson 2012; Murphy et al. 2014). Temperate annual legumes have traditionally been sown as companion legumes in these pastures; Trifolium subterraneum L. (subterranean clover) on the light-medium acid-neutral pH soils, and Medicago spp. (medics) on the heavier, more alkaline soils (e.g. Clarkson et al. 1991; Fitzgerald 1994).

Legumes play an important role in mixed pastures by fixing nitrogen and providing protein for grazing livestock. In this zone, studies have been conducted to determine optimal grazing management strategies of mixed grass–legume pastures with a focus on persistence of the sown temperate or native summer perennial grass species (e.g. Garden et al. 2000; Lodge et al. 2003a), or animal production (e.g. Lodge et al. 2003b). Subterranean clover persistence has been a focus of research, although, in perennial grass-subterranean clover pastures, the emphasis has tended to be on management during spring to produce maximum seed set and most studies have been conducted in environments dominated by winter rainfall (e.g. Dear and Cocks 1997; Dear et al. 2001). In contrast, there have been few studies investigating tropical grass management (e.g. Kalmbacher and Martin 1983; Dear et al. 2001); and no studies of how to manage sown tropical grasses in this zone to optimise regeneration of temperate legumes in autumn. To address this knowledge gap, we conducted a field study in inland northern NSW to determine the effect of canopy and ground cover on the establishment of subterranean clover and test the hypothesis that subterranean clover establishment and production would be highest when canopy cover was removed.

Materials and methods

Site

The experimental site was located at Duri 20 km south of Tamworth, NSW (31.16°S 150.52°E; elevation 490 m a.s.l.) on a Red Chromosol soil (Isbell, 1996). Average annual rainfall at Duri is 734 mm with a warm-season dominance; 60% falling during spring–summer. In December 2005, two tropical perennial grasses (Digitaria eriantha Steud. (digit grass) cv. Premier and Chloris gayana Kunth (Rhodes grass) cv. Katambora) and annual forage sorghum (Sorghum bicolor (L.) Moench ssp. bicolor x S. bicolor ssp. drummondii cv. Sweet Jumbo) were established into 2.6 × 3.3 m plots with three replicates (total of 90 experimental plots), with a 1 m wide buffer surrounding the experimental area. An experiment was conducted over the period October 2006–May 2008 to quantify productivity and persistence of the three species with five nitrogen (N; 0, 50, 100, 150 and 300 kg N/ha) and two defoliation frequency (2 and 6 weekly defoliation) regimes. The findings have been reported in Boschma et al. (2014, 2017). At the conclusion of the experiment in May 2008, there were no differences in soil nitrate levels (0–0.9 m) between the 0 and 100 kg N/ha treatment plots (Boschma et al. 2014). From this time until March 2009, all plots in the experiment (i.e. the established perennial grass and sorghum–fallow plots) were maintained weed-free and no fertiliser applied. In November 2008, the sorghum–fallow plots that had been fertilised with nitrogen rates 0–100 kg N/ha were identified in each replicate. Half of the plots were sown with 1 kg/ha (viable seed) of digit grass and the other half were maintained in a weed-free fallow (i.e. bare).

Experimental treatments

In May 2009, an unbalanced two factorial experiment with three replicates was prepared using sorghum–fallow plots, and the newly established and established digit grass plots from the former experiment. All plots selected had been fertilised with 0–100 kg N/ha in the former experiment. The experiment consisted of bare plots (nil ground cover), plus two levels of canopy cover applied to three levels of ground cover (seven treatments with a total of 21 plots (each plot 2.6 × 3.3 m) in the experiment).

On 4 May 2009, in each replicate area, ground cover of all plots containing digit grass were visually assessed (Murphy and Lodge 2002) and single plots were allocated as ‘bare’ (<5% ground cover; i.e. former sorghum–fallow) and duplicate plots were selected to provide each of the three ground cover treatments: ‘low’ (25% ground cover, ranging 20%–30%), ‘medium’ (50% ground cover, ranging 45%–55%) and ‘high’ (75% ground cover, ranging 70%–80%). The duplicate plots of the three ground cover treatments were each allocated to one of the two canopy cover treatments: ‘defoliated’ and ‘undefoliated’. The tall foliage in the undefoliated plots remained untouched while the defoliated plots were mown with a rotary mower to 0.1 m height and the herbage removed from the plots to simulate a heavily grazed or slashed pasture.

Inoculated seed of T. subterraneum L. ssp. subterranean (subterraneum clover) cv. Dalkeith was broadcast sown at 4 kg/ha (germinable seed) on 18 May 2009 and an additional 4 kg/ha broadcast in June 2009. Rain (15 mm) fell over two days following the second sowing and an additional 20 mm of supplementary irrigation applied over four consecutive days (giving a total of 6 days consecutive ‘rainfall’) to assist subterranean clover germination. Single superphosphate (125 kg/ha, 8.8% P and 11% S) was broadcast over the experiment on 18 May 2009.

Data collection

The digit grass pasture in each plot was characterised in May prior to sowing subterranean clover:

1. Foliage and flowering stem height (m) were determined in four random locations in each plot.

2. Stem density (stems/m²) was determined by counting the number of flowering stems in two 0.4 × 0.4 m quadrats.

3. Basal point frequency (%) was determined along a 1.0 m transect randomly placed in the plot three times and the presence of live crown every 50 mm along the transect recorded.

4. Measurements of photosynthetically active radiation (PAR, 0.4–0.7 μm, μmol/m²/s) were recorded in each plot on 15 June 2009 between 1200 and 1300 h (Australian Eastern Standard Time) using a ceptometer (Delta-T Devices Ltd., Cambridge, UK). In each plot, one measurement of incident PAR was made at 1.3 m above the pasture canopy (PAR_O) and transmitted PAR calculated as the average of three random measurements taken below the canopy at the soil surface (PAR_T). The proportion of radiation reaching the soil surface (i.e. not intercepted by the grass pasture canopy) was calculated as 1 − (PAR_O–PAR_T)/PAR_O.

Establishment (plants/m²) of subterranean clover was assessed in August 2009 by counting the number of seedlings in three randomly placed quadrats (0.6 × 0.15 m).

Herbage mass of subterranean clover was assessed in October 2009. Three visual estimates of subterranean clover herbage mass per plot were recorded on a continuous 0–5 scale (0, nil; 5, high). Calibration quadrats (15, 0.40 × 0.40 m) selected to cover the range of herbage mass in the experiment were also scored then cut to 10 mm above ground level, dried at 80°C for 48 h, then weighed. Herbage mass scores were regressed (linear or quadratic R²> 0.80) against actual herbage mass (kg DM/ha) to determine the herbage mass of subterranean clover.

Statistical analyses

The data were analysed as a partial factorial generalised linear model using ASREML v4.1 (Gilmour et al. 2015). For the pasture characteristics foliage height, flowering stem height and stem density, the undefoliated plots of the low, medium and high ground cover treatments were compared (total nine plots). The defoliated treatment plots were not included as all replicate plots had the same value and therefore no variation. For the pasture characteristics basal point frequency and PAR, also subterranean clover seedling density and herbage mass, the fixed model firstly compared ‘bare’ (3 plots) with ‘other’ (i.e. the ground cover treatment plots, total 18) then, within the other plots, canopy cover (i.e. defoliated/undefoliated), ground cover (i.e. low, medium, high) and their interaction. The only term in the random model was ‘rep’. All data had uniform variance, except subterranean clover seedling densities which were transformed by square root. Pairwise comparisons were made using least significant difference (LSD, P = .05) where model effects were significant (P < .05).

Results

Climate

Growth of the tropical grasses ceased in May with the onset of frosts (overnight temperatures <1°C). Total rainfall during the final four months for the grass’ growing season (January–April) was highly variable but total rainfall was about average. From May until the end of the experiment, monthly rainfall was again highly variable with a total of 212 mm rainfall received at the experimental site. Rainfall recorded for this period at the Bureau of Meteorology (BOM) site located 22 km away was only 63% of the long-term average (182 cf. 291 mm) (Figure 1A). Minimum and maximum temperatures at the experimental site during this period were an average 1.2°C and 2.6°C respectively, below those recorded at the BOM site. Minimum temperatures recorded at the BOM site during the experimental period deviated ± 1°C from the long-term average while maximum temperatures were 1.8°C above average (ranged 0.8°C–3.9°C) (Figure 1B).

Figure 1. A, Actual and long-term average (LTA) monthly rainfall (mm) and B, average minimum and maximum daily temperatures (°C), January–October 2009. Actual values are from BOM Station number 55325 located 22 km from the experimental site while long-term average values are from BOM Station number 55054, collocated with 55325.

Pasture characteristics

All undefoliated treatment plots had similar foliage height, however, both flowering stem height and density increased with increasing ground cover treatment (P > .05, Table 1). Basal point frequency and PAR assessed for all plots (i.e. defoliated and undefoliated) varied significantly. Basal point frequency increased with increasing ground cover (P < .05); low ground cover treatments had an average frequency of 10% compared with 39% for the high ground cover treatment plots (Table 1). PAR which reached the soil surface was similar for all the defoliated treatments and the undefoliated low ground cover treatment (>86%, Figure 2) and significantly higher than on the undefoliated medium and high ground cover treatment plots (51%–54%, Figure 2).

Figure 2. Proportion of photosynthetically active radiation (PAR, %) at the soil surface of digit grass plots with four levels of ground cover that were either defoliated or undefoliated. Data points with the same letter are not significantly different (P = .05).

Table 1. Pasture characteristics of the three tropical grass ground cover treatments denoted by foliage height (cm), flowering stem height (cm) and density (stems/m²), and basal frequency (%).

Pasture characteristic	Ground cover
	Low	Mid	High
Foliage height (cm)^a	22	22	23
Stem height (cm)^a	70b	108a	123a
Stem density (stems/m^2)^a	20b	108a	111a
Basal frequency (%)^b	10c	26b	39a

Note: Within each characteristic, means with the same letter are not significantly different (P = .05).

^aOnly undefoliated plots included in the analysis.

^bMain effect (i.e. both defoliation treatments included in the analysis).

Subterranean clover establishment and herbage production

Subterranean clover seedling density was similar between the bare vs. the other ground cover treatments (P > .05). Of the treatments with ground cover (i.e. >5%), seedling density was higher in the defoliated treatments (P < .05); 14 cf. 6 plants/m². There was no effect of ground cover or interaction between the variables (P > .05).

Herbage mass of subterranean clover was lower in the bare than the other ground cover treatments (P < .05) as all seedlings in the bare plots died during the season. Within the other ground cover treatments, herbage mass was higher in the defoliated treatment plots (main effect); subterranean clover averaging 530 cf. 302 kg DM/ha in the undefoliated plots. There was a trend for increasing subterranean clover herbage mass with increasing ground cover; however, neither the main effect nor defoliation-ground cover interaction was significant (P > .05).

Discussion

This study has shown that subterranean clover establishment and herbage production were highest when a digit grass pasture was defoliated to reduce shading by the canopy. While this finding may be broadly consistent with that of other studies, the nature of our species mix was different; most studies have assessed performance of species mixes with similar growth patterns, that is, both temperate or both tropical species (e.g. Kalmbacher and Martin 1983; Dear and Cocks 1997; Dear et al. 1998, 2001), while our study used a temperate annual legume and a tropical perennial grass with contrasting growth patterns.

The effect of shade on subterranean clover is varied in the literature; ranging from fostering establishment and survival to suppressing them. Variation in results is mostly due to the technique used to apply shade treatments; that is, shade cloth and mesh (Lodge 1996; Mauromicale et al. 2010), stubble/mulch (McWilliam and Dowling 1970; Lattimore et al. 1994) or perennial pasture (Kalmbacher and Martin 1983; Fitzgerald 1989; Dear and Cocks 1997; Dear et al. 1998, 2001). Shade provided by shade cloth resulted in no effect (McWilliam and Dowling 1970), an inconsistent effect (Lodge 1996) also delayed emergence and flowering, with a lengthened growing season of subterranean clover (Mauromicale et al. 2010). In contrast, studies conducted with perennial pastures all reported that the perennial suppressed the establishment, development and survival of subterranean clover seedlings (Fitzgerald 1989; Dear and Cocks 1997; Dear et al. 1998, 2001), especially when management (or lack of) resulted in the pasture becoming rank in late summer–autumn (Fitzgerald 1989).

The effect of the perennial grass on the annual legume was more substantial than that of shade provided by an artificial source, while the effect of a stubble/mulch was intermediate (McWilliam and Dowling 1970; Lattimore et al. 1994). The stubble or mulch may have had an impact on soil water and nutrient levels while the crop was growing, but not while the legume was regenerating as the mulch had ceased growth. In a pot study conducted to differentiate stresses imposed by phalaris on subterranean clover, seedling growth was affected (in decreasing order) by soil water, shade and nutrition (Dear et al. 1998).

Subterranean clover cannot avoid competition in a tropical grass pasture, but the level of competition could potentially be manipulated. Competition for soil water could be reduced by decreasing the density of perennial plants (Dear and Cocks 1997; Dear et al. 2001) or possibly by changing their spatial arrangement (Boschma et al. 2010). For example, sowing the perennial at a lower sowing rate and/or in wider row spacing (e.g. 0.5 cf. 0.15 m) may maximise competition between the perennial plants (intra-specific competition) and minimise competition between the annual and perennial (inter-specific competition) (Harper 1977). Similarly, the effect of shading by the perennial pasture could be reduced in late summer by utilising the grass forage with heavy grazing (Fitzgerald 1989; Dear and Cocks 1997), mowing or slashing, or removing the material by conserving the forage as hay or silage.

Digit grass has high growth rates over summer (Murphy et al. 2010b; Boschma et al. 2017) especially when fertilised (Boschma et al. 2017). During autumn when subterranean clover establishment and growth commonly commences, tropical grass growth, including digit grass, has slowed (Murphy et al. 2010b), but stored soil water levels are lowest in tropical grass pastures at this time of the year (Murphy et al. 2010a, 2018) and lower than phalaris (e.g. Dear and Cocks 1997). Rainfall is therefore essential for legume germination and growth.

Rainfall in this zone is highly variable, as indicated by the rainfall received during our study. This presents a major challenge for species regenerating in soils with little-stored soil water, and highlights that not all years will be suitable for legume establishment, growth and seed set. The most realistic strategies to increase the likelihood of annual legume success is to manage the legume to maximise seed set in spring, manage the grass in late summer–autumn to maximise soil water and reduce shade, and choose adapted legume(s) with traits that will assist survival. Advantageous legume traits may include high seedling vigour, deep rootedness, high seed production and high hard seed levels. Legume species could also have different tolerances to shade (e.g. Fitzgerald 1989; Lodge 1996; Devkota et al. 1997), so additional studies are required to ensure these results are applicable to other legumes.

Acknowledgements

The authors acknowledge collaboration of the landholders Clive and Renee Barton and technical support provided by Mark Brennan.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

We gratefully acknowledge financial support of the Future Farm Industries Cooperative Research Centre (CRC) and NSW Department of Primary Industries.