Assessing the sustainability of wheat-based cropping systems using simulation modelling: sustainability

Concepts of agricultural sustainability and possible roles of simulation modelling for characterising sustainability were explored by conducting, and reflecting on, a sustainability assessment of rain-fed wheat-based systems in the Middle East and North Africa region. We designed a goal-oriented, model-based framework using the cropping systems model Agricultural Production Systems sIMulator (APSIM). For the assessment, valid (rather than true or false) sustainability goals and indicators were identified for the target system. System-specific vagueness was depicted in sustainability polygons-a system property derived from highly quantitative data-and denoted using descriptive quantifiers. Diagnostic evaluations of alternative tillage practices demonstrated the utility of the framework to quantify key bio-physical and chemical constraints to sustainability. Here, we argue that sustainability is a vague, emergent system property of often wicked complexity that arises out of more fundamental elements and processes. A 'wicked concept of sustainability' acknowledges the breadth of the human experience of sustainability, which cannot be internalised in a model. To achieve socially desirable sustainability goals, our model-based approach can inform reflective evaluation processes that connect with the needs and values of agricultural decision-makers. Hence, it can help to frame meaningful discussions, from which actions might emerge.

The effects of high stocking rates on milk production from dryland and irrigated Mediterranean pastures

An experiment using herds of similar to 20 cows (farmlets) assessed the effects of high stocking rates on production and profitability of feeding systems based on dryland and irrigated perennial ryegrass-based pastures in a Mediterranean environment in South Australia over 4 years. A target level of milk production of 7000 L/cow.year was set, based on predicted intakes of 2.7 t DM/cow.year as concentrates, pasture intakes from 1.5 to 2.7 t/cow.year and purchased fodder. In years 1 and 2, up to 1.5 t DM/cow.year of purchased fodder was used and in years 3 and 4 the amounts were increased if necessary to enable levels of milk production per cow to be maintained at target levels. Cows in dryland farmlets calved in March to May inclusive and were stocked at 2.5, 2.9, 3.3, 3.6 and 4.1 cows/ha, while those in irrigated farmlets calved in August to October inclusive and were stocked at 4.1, 5.2, 6.3 and 7.4 cows/ha. In the first 2 years, when inputs of purchased fodder were limited, milk production per cow was reduced with higher stocking rates (P < 0.01), but in years 3 and 4 there were no differences. Mean production was 7149 kg/cow.year in years 1 and 2, and 8162 kg/cow.year in years 3 and 4. Production per hectare was very closely related to stocking rate in all years (P < 0.01), increasing from 18 to 34 t milk/ha.year for dryland farmlets (1300 to 2200 kg milk solids/ha) and from 30 to 60 t milk/ha.year for irrigated farmlets (2200 to 4100 kg milk solids/ha). Almost all of these increases were attributed to the increases in grain and purchased fodder inputs associated with the increases in stocking rate. Net pasture accumulation rates and pasture harvest were generally not altered with stocking rate, though as stocking rate increased there was a change to more of the pasture being grazed and less conserved in both dryland and irrigated farmlets. Total pasture harvest averaged similar to 8 and 14 t DM/ha.year for dryland and irrigated pastures, respectively. An exception was at the highest stocking rate under irrigation, where pugging during winter was associated with a 14% reduction in annual pasture growth. There were several indications that these high stocking rates may not be sustainable without substantial changes in management practice. There were large and positive nutrient balances and associated increases in soil mineral content (P < 0.01), especially for phosphorus and nitrate nitrogen, with both stocking rate and succeeding years. Levels under irrigation were considerably higher (up to 90 and 240 mg/kg of soil for nitrate nitrogen and phosphorus, respectively) than under dryland pastures (60 and 140 mg/kg, respectively). Soil organic carbon levels did not change with stocking rate, indicating a high level of utilisation of forage grown. Weed ingress was also high (to 22% DM) in all treatments and especially in heavily stocked irrigated pastures during winter. It was concluded the higher stocking rates used exceeded those that are feasible for Mediterranean pastures in this environment and upper levels of stocking are suggested to be 2.5 cows/ha for dryland pastures and 5.2 cows/ha for irrigated pastures. To sustain these suggested stocking rates will require further development of management practices to avoid large increases in soil minerals and weed invasion of pastures.

Modelling the manager: Representing rule-based management in farming systems simulation models

We trace the evolution of the representation of management in cropping and grazing systems models, from fixed annual schedules of identical actions in single paddocks toward flexible scripts of rules. Attempts to define higher-level organizing concepts in management policies, and to analyse them to identify optimal plans, have focussed on questions relating to grazing management owing to its inherent complexity. “Rule templates” assist the re-use of complex management scripts by bundling commonly-used collections of rules with an interface through which key parameters can be input by a simulation builder. Standard issues relating to parameter estimation and uncertainty apply to management sub-models and need to be addressed. Techniques for embodying farmers' expectations and plans for the future within modelling analyses need to be further developed, especially better linking planning- and rule-based approaches to farm management and analysing the ways that managers can learn.