Summary. Two lines of perennial ryegrass (Lolium perenne L.), cv. Aurora and breeding line Ba 11351, from the United Kingdom with elevated concentrations of water-soluble carbohydrates in the shoot were compared with the standard cultivars, Ellett, Vedette and Kangaroo Valley, in pure grass swards under irrigation at Kyabram, Victoria, and Gatton, Queensland, and under natural rainfall at Condah, Victoria, during 1995–97. Near infrared reflectance spectroscopy was used to predict the water-soluble carbohydrate, crude protein, in vitro dry matter digestibility, neutral and acid detergent fibre, and Klason lignin concentrations of the perennial ryegrass herbage. Herbage yield and water-soluble carbohydrate differed between cultivars at each site at most harvests, with the high water-soluble carbohydrate lines usually yielding less and having higher water-soluble carbohydrate concentrations than the 3 standard cultivars. However, the high water-soluble carbohydrate lines also had higher water-soluble carbohydrate concentrations at harvests where their yield was equal to the standard cultivars. The other nutritive value traits differed significantly at more than half of the 32 harvests: the high water-soluble carbohydrate lines had higher crude protein and dry matter digestibility, and lower neutral detergent fibre, the neutral detergent fibre containing less acid detergent fibre and lignin than did the standard cultivars. The high water-soluble carbohydrate lines were more susceptible to crown rust during spring and summer than the standard cultivars at Kyabram and Gatton: heavy infections reduced yield, water-soluble carbohydrate, dry matter digestibility and crude protein. Higher water-soluble carbohydrate may depend on only a few genes, as does rust resistance and it seems likely that high yielding, high water-soluble carbohydrate cultivars can be developed by recombination and selection.

Breeding methodologies for cultivated lucerne (Medicago sativa L.), an autotetraploid, have changed little over the last 50 years, with reliance on polycross methods and recurrent phenotypic selection. There has been, however, an increase in our understanding of lucerne biology, in particular the genetic relationships between members of the M. sativa complex, as deduced by DNA analysis. Also, the differences in breeding behaviour and vigour of diploids versus autotetraploids, and the underlying genetic causes, are discussed in relation to lucerne improvement.Medicago falcata, a member of the M. sativa complex, has contributed substantially to lucerne improvement in North America, and its diverse genetics would appear to have been under-utilised in Australian programs over the last two decades, despite the reduced need for tolerance to freezing injury in Australian environments. Breeding of lucerne in Australia only commenced on a large scale in 1977, driven by an urgent need to introgress aphid resistance into adapted backgrounds. The release in the early 1980s of lucernes with multiple pest and disease resistance (aphids, Phytophthora, Colletotrichum) had a significant effect on increasing lucerne productivity and persistence in eastern Australia, with yield increases under high disease pressure of up to 300% being recorded over the predominant Australian cultivar, up to 1977, Hunter River. Since that period, irrigated lucerne yields have plateaued, highlighting the need to identify breeding objectives, technologies, and the germplasm that will create new opportunities for increasing performance. This review discusses major goals for lucerne improvement programs in Australia, and provides indications of the germplasm sources and technologies that are likely to deliver the desired outcomes.

Summary. Irrigated, pure stands of perennial ryegrass (Lolium perenne cv. Yatsyn), prairie grass (Bromus willdenodii cv. Matua) and tall fescue (Festuca arundinacea cv. AU Triumph) were compared with Italian ryegrass (L. multiflorum cv. Concord) under grazing in the subtropics of south-east Queensland. Pastures were fertilised with 50 kg nitrogen/ha . month as urea and annual dressings of 20 kg phosphorus/ha and 50 kg potassium/ha (as superphosphate and muriate of potash, respectively). There were 4 pasture treatments grazed by multiparous Holstein-Friesian cows at 3 cows/ha in a 1-week-on, 3-weeks-off rotation with 2 replicates and 3 cows/treatment block. Feed on offer was measured weekly and pasture quality, at the mid point of each of the 4 seasons. Detailed measurements on plant and tiller dynamics were recorded on fixed quadrats within the grazing areas from November to May in the second and third years. Yield of pasture on offer was greatest with prairie grass and the difference was most marked in spring and early summer. Yield of fescue was generally higher than that from the other 3 grasses in the autumn. In the second and third summers, the grass weed component in the 2 ryegrass, and to a lesser extent the prairie grass, pastures was greater than the sown grass component. Fescue generally produced forage lower in quality than the other 3 grasses although the differences were small in summer. The forage quality of Italian ryegrass was higher than perennial ryegrass in most seasons and for most attributes measured. Generally prairie grass had similar quality forage to the ryegrasses but at times it was as low as fescue. Fescue was the most persistent grass; it maintained a frequency of occurrence of 88, 56 and 71% in the first, second and third autumn periods, respectively, compared with 36, 37 and 21% for perennial ryegrass. To achieve these persistence figures, perennial ryegrass needed over-sowing in each autumn. Plant density and tiller numbers per plant fell in all grasses from November to May but the fall was significantly less in fescue than in other grasses. It was concluded that all 3 temperate perennial grasses demonstrated traits which were useful for subtropical dairy pastures. Fescue was the most persistent and the only grass which could sustain grazing in autumn. The performance of Italian ryegrass was as good as that of perennial ryegrass in the first year but fell substantially in the second and third years as the level of summer grass invasion suppressed the existing population and made oversowing increasingly less effective. Prairie grass produced the highest dry matter on offer under grazing, its forage quality was generally similar to that of the ryegrasses and it regenerated from self-sown seed.

Testcrosses were made with novel sources of lucerne germplasm. These were evaluated in the field in a subtropical environment to identify the lines which produced the highest yielding hybrids as a guide to future breeding efforts. The novel sources were derivatives of Medicago sativa × M. arborea partial (asymmetric) hybrids (termed sac) and very highly winter-active Omani landraces of M. sativa. As testers, 2 lines were used; a Colletotrichum trifolii race 2 resistant selection from the group 9 Australian-bred and adapted cultivar PacL 901 (selection hereafter termed 901) and the Omani landrace, Oman 2, collected at 17°N latitude, from Salalah, Oman. In the row experiment, substantial and significantly positive tester parent heterosis for overall yield (sum of 13 harvests) was observed in all of the sac × Oman 2 testcrosses, with the mean performance of the 11 testcrosses (1839 g/m row) significantly (P < 0.05) exceeding the mean performance of the sac × 901 testcrosses (1703 g/m row) evaluated. Where 901 was used as the tester, heterosis values relative to the tester for the same sac lines were negative for all testcrosses with 8 of the testcrosses being significantly negative. For the Omani landrace × 901 testcrosses, positive and negative heterosis values for total yield relative to the tester were observed, but none were significantly different from zero. The 901 tester yielded significantly (P < 0.05) more per se than the Oman 2 tester (1956 v. 1470 g/m row), although in an adjacent sward experiment Oman 2 yielded comparably to most of the standard commercial cultivars. The potential of the novel germplasm in the subtropics was verified in sward experiments with synthetics and/or strain crosses with yield increases of up to 42% over the benchmark synthetic Sequel. Further improvements can be expected following selection for disease and pest resistance within the lines and in the case of Oman 2 and sac, converging to maximise complementary gene action.

A glasshouse study was undertaken to determine the effect of defoliation frequency (three times at one leaf stage or once at three leaf stage of the regrowth cycle) and height (20, 50 or 120 mm) on regrowth, plant water soluble carbohydrate (WSC) reserves and root growth of seven Lolium perenne and two Lolium multiflorum cultivars. The sensitivity to defoliation was in decreasing order: biennial, PI perennials (cv. Ellett, AN2327, LP30, LP31), P2 perennials (cv. Kangaroo Valley, Yatsyn, Pacific). The effect of frequent, compared to infrequent, defoliation was to suppress regrowth by l00%, 95% and 80%; stubble WSC (mg/plant) by 97, 89 and 81%; root DM (g/plant) by 76, 60 and 6%, for biennial, P1 and P2, respectively. The effect of defoliation height accentuated this response, with biennials defoliated frequently at 20 mm stubble height all dying. Under defoliation conditions producing optimal yield, the yield was positively related to sensitivity to defoliation, giving regrowths of 2.90, 2.68, 1.53 g DM per plant for biennial, P1 and P2 plants, respectively. In view of the marked defoliation by cultivar interaction, response to defoliation should be considered as a possible selection criterion in any evaluation process.