Abstract. Climate change projections for Australia predict increasing temperatures, changes to rainfall patterns, and elevated atmospheric carbon dioxide (CO 2 ) concentrations. The aims of this study were to predict plant production responses to elevated CO 2 concentrations using the SGS Pasture Model and DairyMod, and then to quantify the effects of climate change scenarios for 2030 and 2070 on predicted pasture growth, species composition, and soil moisture conditions of 5 existing pasture systems in climates ranging from cool temperate to subtropical, relative to a historical baseline. Three future climate scenarios were created for each site by adjusting historical climate data according to temperature and rainfall change projections for 2030, 2070 mid-and 2070 high-emission scenarios, using output from the CSIRO Mark 3 global climate model. In the absence of other climate changes, mean annual pasture production at an elevated CO 2 concentration of 550 ppm was predicted to be 24-29% higher than at 380 ppm CO 2 in temperate (C 3 ) species-dominant pastures in southern Australia, with lower mean responses in a mixed C 3 /C 4 pasture at Barraba in northern New South Wales (17%) and in a C 4 pasture at Mutdapilly in south-eastern Queensland (9%). In the future climate scenarios at the Barraba and Mutdapilly sites in subtropical and subhumid climates, respectively, where climate projections indicated warming of up to 4.48C, with little change in annual rainfall, modelling predicted increased pasture production and a shift towards C 4 species dominance. In Mediterranean, temperate, and cool temperate climates, climate change projections indicated warming of up to 3.38C, with annual rainfall reduced by up to 28%. Under future climate scenarios at Wagga Wagga, NSW, and Ellinbank, Victoria, our study predicted increased winter and early spring pasture growth rates, but this was counteracted by a predicted shorter spring growing season, with annual pasture production higher than the baseline under the 2030 climate scenario, but reduced by up to 19% under the 2070 high scenario. In a cool temperate environment at Elliott, Tasmania, annual production was higher than the baseline in all 3 future climate scenarios, but highest in the 2070 mid scenario. At the Wagga Wagga, Ellinbank, and Elliott sites the effect of rainfall declines on pasture production was moderated by a predicted reduction in drainage below the root zone and, at Ellinbank, the use of deeper rooted plant systems was shown to be an effective adaptation to mitigate some of the effect of lower rainfall.

A biophysical simulation model (the Sustainable Grazing Systems Pasture Model) was developed as an integral part of the Sustainable Grazing Systems National Experiment. It was developed to meet the needs of the researchers both for analysing data and processes at individual sites, and for simulating the outcome of these processes operating in generic pasture systems on a range of soil types, under specific grazing managements. The model was designed to reside on the desktops of individual researchers and for those researchers to be part of its development process.The Sustainable Grazing Systems Pasture Model incorporates the following: a physiological model of pasture species herbage accumulation in response to climatic conditions; the water balance including evapotranspiration, runoff (surface and subsurface), infiltration and drainage; pasture utilisation by grazing animals; a metabolisable energy-based animal growth model; and organic matter and inorganic nutrient dynamics (for nitrogen, phosphorus, potassium and sulfur) including plant uptake, adsorption, leaching, nitrogen fixation by legumes, and atmospheric nitrogen losses. A range of grazing options (set-stocking, rotational grazing and continuous grazing at a variable rate) is available for ewes and lambs, and wethers. Each of the main modules (water, nutrients, pastures and animals) is interconnected. To avoid bias in the influence of any one module, each is described at about the same level of complexity, with the description of any process being restricted to about 5 parameters. The model is hierarchical in structure and most processes are described in terms of a series of fluxes (or, more specifically, flux densities) that have dimensions of amount per area per time.The model can be closely linked to a database specifically developed for the Sustainable Grazing Systems National Experiment to allow easy importing and exporting of climate and experimental data for comparison with model output. This paper gives an overview of the model structure and its output, the process that was used for its development within Sustainable Grazing Systems, and its use by the Sustainable Grazing Systems sites and themes. Comments are provided on the implementation of the development process to assist future programs using a similar approach.

Ninety-one perennial legumes and herbs (entries) from 47 species in 21 genera were evaluated at sites in New South Wales, South Australia and Western Australia over 3 years from 2002 to 2005 to identify plants with superior herbage production, persistence and the potential to reduce ground water recharge. Evaluation was undertaken in three nurseries (general, waterlogged soil and acid soil). Medicago sativa L. subsp. sativa (lucerne) cv. Sceptre was the best performing species across all sites. In the general and acid soil nurseries, Cichorium intybus L. (chicory) cv. Grasslands Puna was the only species comparable with Sceptre lucerne in terms of persistence and herbage production. Trifolium fragiferum L. cv. Palestine and Lotus corniculatus L. SA833 were the best performing species on heavy clay soils prone to waterlogging. Three Dorycnium hirsutum (L.) Ser. accessions persisted well on acid soils, but were slow to establish. Short-lived perennial forage legumes, such as Onobrychis viciifolia Scop. cv. Othello, and three Hedysarum coronarium L. entries, including cv. Grasslands Aokou, had high herbage production in the first 2 years and may be suitable for short-term pastures in phased pasture-crop farming systems. T. uniflorum L. and M. sativa subsp. caerulea SA38052 were highly persistent and could play a role as companion species in mixtures or ground cover species for undulating landscapes. Cullen australasicum (Schltdl.) G.W. Grimes SA4966 and Lotononis bainesii Baker cv. Miles had poor establishment, but were persistent. Chicory, T. fragiferum and L. corniculatus were identified as species, other than lucerne, with the most immediate potential for further selection to increase the diversity of perennial legumes and herbs adapted to southern Australian environments.

The literature relevant to the grazing management of lucerne in temperate Australia is reviewed with emphasis on the factors likely to affect its persistence. Knowledge of lucerne physiology is used to question the validity of the traditional methods of managing grazed stands, which rely mainly on using 10% flowering as a guide to root carbohydrate levels. From these data several alternative management guidelines are proposed that may lead to increased persistence; however, for long-term persistence, there is little doubt that lucerne needs to be grazed leniently and at a late stage of maturity. Several grazing experiments indicate that grazing periods of 16-20 days should have no effect on persistence, provided that the rest period between successive grazings is 35 days or longer. Data from other countries and Australian data from a limited number of experiments also indicate that grazing in either autumn or winter may substantially reduce production and could affect persistence. Three grazing studies in New South Wales were used to highlight critical differences in experimental design which make comparisons among experiments difficult. Standardised sowing rates and grazing management, and statistical procedures which account for the genotype x management x environment interaction, are suggested to improve the extrapolation of results from experiments to other environments. Persistence of different lucerne types under grazing, particularly those recently imported from the U.S.A. or bred in Australia, is considered. While it has been proposed that grazing effects may be related to crown structure, interactions with other factors which affect persistence may also occur. If grazing can be considered to be stressful to a lucerne plant then it could interact with other stresses, caused by moisture deficit, excessive moisture, insect pests and disease, to reduce persistence. Additionally, considerable variation in varietal resistance to some pests and diseases has been recorded in haycut stands, and so there may also be cultivar x grazing effects. All of these factors could combine to affect the persistence of a particular cultivar under grazing. Patterns of lucerne decline were either continuous or step-like. Continuous decline was associated with prolonged grazing, grazing and moisture stress, grazing under waterlogged conditions, or grazing in situations where the incidence of disease was likely to be high. To understand the reasons why plants fail to persist, measurements need to be made frequently and a1 regular intervals, and the moisture and disease status of the site needs to be accurately monitored. The adequacy of different methods of measuring stand persistence is also questioned. The implications for graziers, researchers and lucerne breeders are discussed.