Lucerne performance on duplex soil under Mediterranean climate : field measurement and simulation modelling.

Abstract

The experimental work reported in this thesis quantified the productivity of lucerne over a two-year period (2000-2001) for a Mediterranean climate at Roseworthy in South Australia (34°32′S, 138°45′E), and determined associated dynamics for water and nitrogen in duplex soil. Shoot growth of dryland lucerne was limited primarily by the pattern and amount of incident rainfall, but high temperature (30-35oC) also constrained summer production. These high summer temperatures induced greater production when irrigation was applied, but under the normally dry summer conditions high temperatures combined with soil water deficit (up to 200mm) caused growth to cease. Thus, shoot dry matter yield under rainfed conditions was 4.9 t ha⁻¹ in 2000 (from 7 harvests) and 1.8 t ha⁻¹ in 2001 (from 5 harvests) whereas summer irrigation increased yield to 14.9 t ha⁻¹ in 2000 (7 harvests) and 7.1 t ha⁻¹ in 2001 (5 harvests). Under rainfed conditions the RUE was 0.55 g DM MJ⁻¹ PARi compared with 1.08 g DM MJ⁻¹ PARi in the irrigated treatment in 2000, reducing to 0.4 g DM MJ⁻¹ for the rainfed and 0.7 g DM MJ⁻¹ under limited irrigation in 2001. Lucerne plant population declined from 69 to 20 (plants m⁻²) in the rainfed treatment and the plants partially compensated for this in 2000 by increasing stem density from 300 to 400 m⁻² in 2000 although this declined back to 300 m⁻² in 2001. In all treatments more than 70% of root biomass was in the top 40 cm soil, this was partially due to the vertical distribution of plant available water but also to subsoil constraints to root development below 0.6m. Nevertheless, lucerne was able to extract water and nitrate to 1800 mm soil depth. Large amounts of irrigation >400mm) over summer (Dec 1999-Mar 2000) increased total soil water content, approaching the drained upper limit; causing a 600% increase in shoot dry matter yield, similarly higher growth rate (71 kg DM d⁻¹) and higher RUE (~1.7 g DM MJ⁻¹ ), confirming that water availability was the main constraint to lucerne growth. Delayed benefits of summer irrigation, especially in the subsurface treatment, were also observed later (July to October) when lucerne was able to scavenge excess irrigation water and nitrate stored in the 600-1800 mm soil profile, which resulted in increased shoot growth. Drainage below the effective rooting depth was negligible, even under irrigation, confirming that lucerne can dry soil profiles and reduce deep drainage. Average annual water use efficiency was 9 kg DM ha⁻¹ mm⁻¹ under rainfed conditions compared to ~15 kg DM ha⁻¹ mm⁻¹ under irrigated conditions. Shoot dry mattter production was closely related to evapotranspiration in all treatments, however, under rainfed conditions losses from evaporation were proportionally higher compared to irrigated treatments. Sub-surface drip irrigation proved superior to surface irrigation using 22% less water compared to surface sprinkler irrigation treatment with comparable yields. Biological N₂fixation was strongly related to shoot production with 18 to 27 kg N fixed per tonne of shoot dry matter across all seasons and treatments. Dependence on N₂fixation appeared to be unrelated to soil mineral N concentration and amounts of nitrate in the profile (to 1m) were generally quite low <35 kg N ha⁻¹). Soil water dynamics under both rainfed and surface irrigated treatments were adequately simulated by the Agricultural Production System Simulator (APSIM) with RMSD < 10% of the observed means and R² > 0.80 for the total soil profile (0-2000 mm). Simulation of growth and development was less satisfactory. For example, the RMSD was ~50% of observed mean for shoot biomass (R² = 0.68) in the rainfed treatment, and 36% (R² = 0.77) in the irrigated treatment. Overall, simulation of shoot DM production was close to observed values during the growing season (Apr-Nov), however the model was unable to capture the observed shoot yield in response to summer irrigation, with simulated shoot DM 40% less than the observed value in 2000 and 35% less in 2001. N dynamics were poorly simulated under these soil and climate conditions. Amounts of soil mineral nitrogen (kg NO⁻₃-N ha⁻¹) were adequately simulated in rainfed conditions but consistently over-predicted under irrigated conditions. This evaluation of APSIM highlights both good and poor model performance and the analysis indicates the need for caution when applying the model in situations where observed data is scarce. Areas requiring improvements to the model are identified. Overall this research has improved understanding of the limitations to potential production of lucerne in a Mediterranean environment on duplex soils and shown that APSIM-Lucerne can be used confidently for many applications, particularly soil-water dynamics.