Volume 37, 129-150

Tropical Grasslands (2003) Volume 37, 129–150

Effects of sown grasses and stocking rates on pasture and animal production from legume-based pastures in the seasonally dry tropics

R.J. JONES

CSIRO Sustainable Ecosystems, Davies Laboratory, Townsville, Queensland, Australia

Abstract

Native pasture was oversown with a legume mixture [Macroptilium atropurpureum cv. Siratro (siratro), Stylosanthes hamata cv. Verano (verano) and S. viscosa] together with Urochloa mosambicensis cv. Nixon (sabi) or Chloris gayana cv. Callide (rhodes) or with no sown grass (native). The 3 treatments were stocked continuously at 3 stocking rates (SR) of 0.65, 0.95 and 1.25 steers/ha with 2 replicates. In addition, unreplicated paddocks of the 3 grass treatments stocked at 1.25 steers/ha were fertilised with ammonium nitrate at 100 kg N/ha/yr. All paddocks were fertilised with 100 kg/ha of single superphosphate every 2 years (≈5 kg P/ha/yr). The experiment was sown in 1977 and terminated in 1993.
Native perennial grasses decreased with increasing SR and were virtually eliminated at high SR with sabi. Rhodes established rapidly and was the most dominant grass for the first 5 years. Thereafter (except with N fertiliser), it declined through years of drought. Sabi established slowly but after the 5th year became dominant. After 1984, Bothriochloa pertusa (not recorded in the original native pasture) increased and became the dominant grass at high and medium SRs on the native and rhodes grass treatments. It was less invasive at low SR.
Verano was the most successful legume, becoming the dominant species at high and medium SRs until B. pertusa invaded and reduced legume contribution. Siratro persisted only at low SR, mainly with sabi grass.
With N fertilisation, rhodes grass was the most prominent species and N-fertilised native and sabi treatments contained significant levels of annual grasses and dicot weeds, especially Sida spp. and Alternanthera angustifolius.
Pasture presentation yields declined linearly with SR. Overall, pasture yields at low SR were similar for all grass treatments but at medium and high SRs, yields on the native treatment were lower than for the sown grass treatments. N-fertilised treatments gave higher yields than the legume-based treatments at the same SR, especially with rhodes grass where mean yields were double those on the native grass treatment (7250 vs 3360 kg/ha).
Annual rainfall (July–June) varied from 293–1316 mm over the period but there was no relationship (P>0.05) between annual liveweight gain (LWG) and annual rainfall. Annual LWG/head was, however, linearly related (P<0.01) to ‘green days’/yr. SR had more effect on annual gain/head than did the grass treatments, with a mean linear decline of 40 kg/unit increase in SR. In only 2 years did a sown grass treatment (sabi) give higher (P<0.05) LWG than the native pasture control. Overall, however, gains on sabi were 14% higher than those of the other treatments (P<0.01). Mean gains/head/yr were 143, 125 and 125 kg for the sabi, rhodes and native treatments, respectively.
The N-fertilised treatments gave mean LWGs of 126, 121 and 86 kg/head/yr for the sabi, rhodes and native treatments.
Neither copper supplementation nor a full mineral supplement improved liveweight gain on any treatment despite low Cu concentrations in rhodes grass of <4 mg/kg.