Background

Few, if any, whole of farm models have been developed that fully integrate the complexity of the different factors involved in smallholder farming. While some whole farm models simulate a range of crops, forages and systems (e.g. IMPACT (Herrero et al. 2007), SEAMLESS (van Ittersum et al. 2008), NUANCES-FARMSIM (Van Wijk et al. 2009) and FarmDESIGN (Groot et al. 2012)), most have lacked intrinsic system components such as labour and economic resources (Castelan-Ortega et al. 2003), temporal scales to capture multiple seasons, or have been region-specific (Parsons et al. 2011a). Labour is often one of the most limiting resources, but is commonly overlooked in farm system models (Norton 2006, Connor et al. 2015).

Precursor models

The development of CLEM is based on a large amount of published scientific research that was included in two precursor models. The functionality of these models (IAT and NABSA) was the initial scope of developing CLEM by using a contemporary programming environment and addressing the shortfalls recognised by users of these models.

The Integrated Analysis Tool for a mixed enterprise (IAT) was developed to integrate key biophysical and economic processes in smallholder farming systems and explore the impacts of changes in management on outcomes for farms (McDonald, 2015, McDonald et al., unpublished). It comprises three core components: long term crop and forage yields simulated with models such as APSIM or expert opinion based on farmer interviews and review of literature; a model for predicting ruminant growth and reproduction; and a model simulating the economic performance of a typical smallholder farm household enterprise (Lisson et al., 2010, McDonald et al. unpublished). The impacts of climate, soil properties and the performance of crops and forages are captured in the external model simulations or the expert opinion providing data. Livestock feeding, growth and reproduction are represented in the livestock component while profitability of farm management is determined within the framework of crop, forage and livestock production, and subject to the limitations of land and labour availability. This structure is highly transferable to many locations and can be modified to suit local conditions.

The IAT was first developed as a decision support tool for Indonesian smallholders (McDonald et al. 2004; Lisson et al., 2010). Since that time, the IAT has been widely used to explore research questions in East Asia (China, Indonesia, Vietnam and Laos), South and West Asia (India and Pakistan), and Africa (Ethiopia, Burkina Faso and South Africa) (e.g. Lisson et al., 2010; McDonald et al. unpublished, Komarek et al., 2012; Parsons et al., 2011b; Shafiullah 2012; Shalander et al. 2017; Rigolot et al. 2015a, 2017; Mayberry et al. 2017, Traill 2017, Monjardino 2018). These research questions have focused on changes to existing cropping practices; adoption of forages and livestock management practices; closing yield gaps; land allocation and intensification; and tactics to improve resilience to climate variability.

The Northern Australian Beef Systems Analyser (NABSA) model (Ash et al., 2015), which is developed from IAT, is applicable to extensive beef systems. This model expands the understanding of herd dynamics and tropical native grass pastures required to simulate beef systems in northern Australia.

The NABSA model has been used to explore a range of scenarios relevant to the northern Australian beef industry, including improving reproduction and growth efficiency through genetic gains in cattle; nutrient supplementation; and alteration of the feed base through introduced pastures and forage crops (Ash et al. 2015; Grice et al. 2013; Monjardino et al. 2014; MacLeod et al. 2015, Crimp et al. 2014).

The IAT and NABSA models have been developed using the Microsoft Excel® platform with model code written in Visual Basic Applications associated with the spreadsheets. Access to IAT and NABSA has historically been obtained through requests to its original developers (current and former CSIRO staff, Mr Cam McDonald, Mr Neil McLeod, Dr Andrew Ash, Dr Shaun Lisson and Ms Di Prestwidge) who have been integral in assisting with training and using the models.

While the models provided very useful modelling functionality, a number of problems had been raised by the growing user base. These include:

  1. the model wasn’t easily accessible for users to download and then use,
  2. the transparency of how the model worked was poor even with detailed user manuals,
  3. there was a lack of flexibility in simulation setup,
  4. there is a significant amount of data required to parameterise a single simulation,
  5. output files are difficult to analyse and summarise,
  6. data visualisation tools were poor and
  7. it was difficult to add extra functions/models

In addition, access to and preservation of the intellectual property developed in the IAT would be improved through development and maintenance of the model on a contemporary platform that can be accessed centrally.

Before developing a new model, the developers approached a number of current and potential model users to determine a wish list of improvements and desired model additions that directed the development of CLEM.

See Fundamentals for more details