Native, naturalised, pasture, and new grass introductions are discussed in terms of concepts of moisture, temperature, fertility, and utilisation gradients; characteristics of the New Zealand high country environment; and their utility for introduction into low-input pastoral systems. The perennial, long-lived habit of native grass is mimicked in introduced grasses. Poa colensoi is the best native grass, but there is a need for wider evaluation of serai non-tussock species. There is a greater frequency of spreading habit and annuals in naturalised low-input species. The "try it and see" and genotype/environment analysis are advocated for screening new introductions.

A grazing trial on oversown tussock country at Tara Hills Research Station, started in 1978 and still continuing, has highlighted short comings of clovers and grasses traditionally used for improvement. In 1984 a wide range of legume and grass germplasm was planted into each of the nine stocking rate/management treatments to identify plant material more persistent than that traditionally used, and investigate specific responses under Extreme (high stocking rate), Optimal (moderate stocking rate and management giving optimum liveweight gain/ha) and Las (low stocking rate) grazing management. After 6 years none of the 12 white clover, 7 alsike and 9 lotus lines tested persisted significantly better than Grasslands Huia white clover (Trifolium repens). Plant survival and spread of Caucasian clover (Trifolium ambiguum) was exceptional. In terms of persistence and spread this legume ranked first out of the 56 germplasm lines under examination. The 15 ryegrasses examined did not persist better than Grasslands Nui ryegrass (Lolium perenne). However, for Optimal management, many ryegrasses performed better than Nui. Cocksfoot(Dactylis glomerata) also performed very well, with Grasslands Wana showing greater persistence under Extreme management than other cocksfoots and ryegrasses. Tall oat grass (Arrhenatherum elatius) was very persistent and like Caucasian clover performed well under all managements. Other grasses tested showed poor persistence, as did the herb sheep's burnet (Sanguisorba minor). Keywords: Arrhenatherum elatius, Dactylis glomerata, grazing management, high country persistence, sustainable improvement, Trifolium ambiguum, tussock grasslands

Grasslands Kara cocksfoot, Grasslands Nui ryegrass, Grasslands Roa tall fescue, Grasslands Gala grazing brome and Grasslands Maru phalaris were sown as single grass species with either white or Caucasian clover during December 1993, on a Lismore stony silt loam in Canterbury. Plots were irrigated and rotationally grazed by sheep. Total herbage produced during the first measurement period (September 1994 to July 1995) averaged 13.6 t DM/ha. All white clover-based swards produced similar yields, but with Caucasian clover, phalaris-based swards produced the most and ryegrass and grazing brome the least. Production tended to be lower from pastures sown with Caucasian than white clover (12.9 cf. 14.2,kg DM/ha). During the second measurement period (July 1995 to May 1996) total production was similar with both legumes. When evaluated over both measurement periods, phalaris and tall fescue gave the highest and cocksfoot and ryegrass the lowest production when sown with white clover, while cocksfoot and grazing brome were the lowestproducing swards with Caucasian clover. Ten --months-after-sow.ing,-white-cIover-contributed-in excess of40% of the DM in all swards except with ryegrass and grazing brome where it contributed 20%. The proportion of white clover then declined in all pastures over the following 20 months, ryegrass-based swards having the highest (10%) and cocksfoot the lowest (1%). When sown with Caucasian clover the legume content of swards increased during the 12 months from spring 1994 in swards containing phalaris, tall fescue and grazing brome. The proportion of legume in -cocksfoot swardsmdeclined-continuously from 22 to 4% over the 20 months while it increased continuously from 4 to 28% with ryegrass. The legume content in lo-month-old pastures was higher when sown with white than with Caucasian clover but this ratio was reversed by 28 months. This occurred with all pasture types but was most pronounced with ryegrass and cocksfoot pastures. The present experiment shows Caucasian clover can make a major contribution in irrigated pastures in Canterbury. 'Wattie Bush, RD 22, Peel Forest, Geruldine Keywords: Bromus stamineus, Dactylis glomerata, Festuca arundinacea, grazed pastures, herbage production, Lolium perenne, Phalaris aquatica, Trifolium ambiguum, Trifolium repens

One hundred and twenty one cultivars and accessions of conventional and alternative legume, grass and forb species have been assessed over a widely variable lower sunny face landscape mosaic within the semi-arid tussock grasslands. Results are presented for assessments carried out 7-8 years after planting. Caucasian clover, hairy canary clover, birdsfoot trefoil, crown vetch and luceme were the best adapted legume species. By contrast, conventional clover species performed poorly; red and alsike clovers all died and only a few plants of white and subterranean clovers have survived, all within the more favourable areas of the landscape. Cocksfoot, wheatgrass, smooth brome and tall fescue survived well across the landscape mosaic and showed good vigour and survival. Cocksfoot cultivars and accessions had the best combination of agronomic attributes. In contrast, the survival of all perennial ryegrasses, declined markedly since the 3rd year. These results have important implications for the future development of sustainable pastoral systems in drought-prone landscapes. Keywords dryland pasture species, legumes, grasses

Herbage mass, moisture content and in vitro OMD of autumn saved ryegrasslwhite clover pasture grown on irrigated flats in the Upper Waitaki Basin were monitored over four winters from 1979-82. During the winter of 1981 and 1982 the effect of closing date and nitrogen application was examined. In addition Merino ewes were all-grasswintered from mid June to mid September without supplements on a feed allowance of 1.2 kg DMlhd.day to determine effect on liveweight. Mean herbage mass and OMD loss from mid May to early September were 25 and 13% respectively and the moisture content of the pasture decreased from 73 to 45%. Herbage mass by mid May was less for March and April than for February closing, although for March closing this deficit was offset through use of nitrogen. Approximately 50% of the sward was frosted by mid September with OMD of the f-m_.c-.tAcd rnwn,rti.nIVn.I IC,*"O"/. ,v, .-vrvrmlllnv-~-r~C~.ul-l~ +L-U k0-1.a.y. Ewe liVeWeightS initialiy ieii but recovered during the latter half of the winter despite the constant fall in herbage OMD. Sward desiccation exceeded loss in digestible herbage mass over the winter, suggesting increasing rather than decreasing food value. Keywords: High country, herbage losses, frosting, in-vitro OMD, irrigated pasture, autumn-saved-pasture, nitrogen, feed value, all-grass-wintering, Merino ewes