Modelling phenological development, yield and quality of lucerne (Medicago sativa L.) using APSIM next generation : A thesis submitted in partial fulfillment of the requirement for the Degree of Doctor of Philosophy at Lincoln University

Abstract

This research integrated knowledge of lucerne crop physiology into the Agricultural Production Systems sIMulator (APSIM) next generation (APSIM NextGen) model framework to develop and verify a comprehensive lucerne simulation model (APSIM NextGen lucerne model). The model was developed to simulate the growth, development and quality of lucerne cultivars grown under different defoliation management and growth conditions. One of the major challenges for developing a lucerne crop simulation model is to capture the seasonality of perennial reserves and their effect on shoot regrowth in response to different defoliation regimes. In this thesis, model development and testing was based on long-term field datasets with multiple defoliation regimes (28 day: S; 42 day: L; and 84 day: H) and three genotypes of fall dormancy (FD; FD2, FD5 and FD10) under irrigated conditions. The APSIM Plant Modelling Framework (PMF) was used to simulate generic organs (leaf, stem and root) and represent key crop physiological processes, including crop phenological development, canopy expansion, dry matter and N accumulation, remobilization and partitioning. Development was parameterized based on thermal time (Tt) targets and a photoperiod (Pp) response. Seedling crops required a juvenile phase (Ttjuv of 215 to 547 ˚Cd). For both seedling and regrowth crops, the Tt to reach 50% buds visible (Tt0-bv) increased as Pp shortened in autumn, a minimum of 278 ˚Cd for the basic vegetative (TtBVP) period was required at Pp >14h for regrowth crops to reach buds visible stage. After crops reached buds visible stage, another 310 ˚Cd of Tt (Ttbv-fl) was required to reach flowering. Lucerne biomass supply was parameterized as the product of accumulated intercepted total radiation, and radiation use efficiency (RUEtotal, g DM MJ-1 total radiation). The intercepted total radiation was calculated by LAI and an extinction coefficient (k) of 0.81. LAI was parameterized as leaf area expansion rate (LAER) and Pp response. LAER declined as the Pp decrease, being 0.018 m2 m-2 ˚Cd-1 at 16.5 h and 0.008 m2 m-2 ˚Cd-1 at 10 h. However, a Pp response was not observed in seedling crops and regrowth crops in increasing Pp conditions. The RUEtotal was 1.1±0.31 g DM MJ−1 at 18 ˚C for both seedling and regrowth crops. Biomass supply was then allocated based on the relative demand of each organ. Leaf and stem biomass demand were parameterized as positive power functions. Root biomass showed a seasonal pattern. The APSIM NextGen lucerne model provided a mechanistic framework to model root biomass dynamics with structural and storage components. Structural root biomass was defined and estimated as the amount of root biomass (~2500 kg ha-1) that had no root maintenance respiration loss in winter. The ratio of storage to structural root differed among development stages and FD classes. In an increasing Pp, there was no storage root demand. The decrease of root biomass during this period was due to remobilization from root to shoots and root maintenance respiration. A remobilization coefficient value and a regrowth coefficient function were used to calculate root remobilization. A remobilization coefficient value was defined as the percentage of storage root biomass per day (5 for FD5, 1 for FD2 and FD10). The regrowth coefficient function includes two parameters (remobilization duration and remobilization rate). Remobilization duration was defined as Tt since harvest, whereas remobilization rate is an adjusted value for the current remobilization coefficient value (ranging from 0 to 1.5). The regrowth coefficient function represents remobilization started at the maximum remobilization rate (1.5) from the beginning of each regrowth cycle (0 ˚Cd). This remained constant until 300 ˚Cd for FD5 (250 ˚Cd for FD2 and 500 ˚Cd for FD10), and then declined to 0 at 350 ˚Cd for FD5 (300 ˚Cd for FD2 and 550 ˚Cd for FD10). In a decreasing Pp, the increasing root biomass was caused by carbon partitioning. Thus, the model was parameterized to have a maximal root demand with no remobilization. A constant root maintenance respiration coefficient (Rm_root_day) of 0.0005 g g-1.day-1 was applied to model root storage maintenance loss. The model had good prediction on shoot biomass and fair prediction on root biomass for 42 and 84 day defoliation treatments. However, the model did not accurately predict root biomass under a 28 day frequent defoliation (SS) probably due to a limitation of root N reserves. The N module was linked with DM in the PMF. The N supply was estimated as 2.5% of total biomass, whereas N demand was built as N threshold functions for each organ. Root N showed a similar seasonal pattern as root biomass. A root N remobilization coefficient value (% storage root N per day; 2 for FD5, 0.5 for FD2 and FD10) was used for remobilization calculations in an increasing Pp. Applying the N module improved biomass prediction, especially for the 28 day defoliation treatment (SS). Simulation results showed good agreement for predicting phenological development stages (NSE of 0.77 for buds visible and 0.67 for flowering stage), good agreement for canopy expansion (overall NSE = 0.61), good agreement for shoot and root biomass (NSE of 0.68 and 0.53). However, there was fair to poor agreement for leaf N (NSE of 0.16 to -0.14), stem N (NSE of 0.51 to -4.61) and root N (NSE of 0.16 to 0.29) for all three FD classes under different defoliation regimes. This was because leaf biomass was used to parameterize leaf N thresholds which resulted in systemic bias. There was a lack of measured N concentration data for the model testing for most treatments. Thus, additional measurement and a more effective approach for parameterizing N demand are required to improve the model. Overall, these results indicate that the APSIM NextGen lucerne model was successfully created to predict growth and development of crops grown under unlimited environmental conditions. Model validation is required under different climate conditions.