SATrends Issue 76 March 2007
  • Supporting Evidence-Based Policy Dialogue
  • Many hands make light work!
  • A tale of master and slave

  • 1. Supporting Evidence-Based Policy Dialogue

    Many goals have been set to reduce poverty and improve the agricultural sector in Southern Africa. Under the Maputo Declaration, member countries of the Southern Africa Development Community (SADC) have promised to increase their agricultural budgets to at least 10% of their national budgets by 2008. SADC countries have also endorsed the Comprehensive Africa Agriculture Development Programme (CAADP), which aims to improve the 'productivity of agriculture to attain an average annual growth rate of 6%'. These initiatives must also contribute to the Millennium Development Goal (MDG1) of cutting hunger in half by 2015.

    In order for these goals to be met, it is imperative that policymakers have a thorough understanding of how the agricultural economy has performed over time as well as a sound basis when planning and shaping agricultural policy. For example, studies have shown that developing countries as a whole allocate less than 10% of their GDP to agriculture, much less than developed countries, which allocate more than 20%. The Maputo Declaration encourages the increase of investment in agriculture, but the question of "how" still remains. What constitutes agricultural investments? How must budgets be allocated among the different sub-sectors within agriculture? How can countries ensure efficient use of the increased agricultural budgets?


    The Regional Strategic Analysis and Knowledge Support System for Southern Africa (ReSAKSS-SA), implemented by ICRISAT and IWMI and other regional and national partners, can help in answering these questions. The promise of ReSAKSS lies in its contribution to promoting evidence-based policy dialogue and investment decision making in the SADC region. ReSAKSS-SA provides three services:

    • monitoring trends on key agricultural growth patterns, which will facilitate countries to track their progress in achieving objectives such as the MDGs
    • strategic analysis to support policy decision making
    • knowledge management and sharing of credible data and information in an efficient and effective manner.

    While ReSAKSS as an initiative is more than a year old, it is quite a new venture in its present form. ReSAKSS-SA is housed at IWMI, Pretoria. A coordinator, backed administratively by IWMI, runs the day-to-day business and receives support of a senior scientist from ICRISAT.

    For more information, contact

    2. Many hands make light work!

    A few years ago we reported the success of the small seed pack idea. The idea was extended to fertilizer packets as well, and this, coupled with extraordinary teamwork from public and private sector partnership, has spelled success for farmers in the Eastern and Southern African region.

    Progress Milling (PM) established the Limpopo Community Development Program (LCDP) in Limpopo province, in southern Africa to coordinate investment aimed at increasing agricultural productivity and economic livelihoods of rural communities, and invested in rural depots for sale of maize and purchase of farmer grain.

    ICRISAT conducted research with LIMPAST, a community based organization (CBO) in Limpopo, on testing fertilizer small-packs for poor farmers. The idea was to demonstrate to resource-poor farmers who do not use any fertilizers that even small quantities of fertilizer, half a beer bottle cap per plant, applied at knee-high maize crop can make a difference in yield. Previously, fertilizer was sold in the PM depots in 50 kg packs. Sales in five previous years were 85 tons per annum.

    fertilizer to resource-poor farmers It is difficult to sell large bags of fertilizer to resource-poor farmers.

    To complement the fertilizer research involving small packs, PM, LIMPAST and ICRISAT approached Sasol-N to become a partner in LCDP for fertilizer sale at PM depots. ICRISAT proposed testing the sale of small packs in addition to the traditional 50 kg bags, for poor farmers. Sasol registered 10/20 kg packs of fertilizer and in 2005/06 supplied PM with three pack sizes of fertilizer for sale at depots. Sasol helped conduct demo-trials with CBOs associated with Progress Milling. Pannar Seeds supported the marketing trial with small packs of seed for sale at the depots. In villages where farmers were familiar with fertilizer, purchase of 50 kg bags dominated sales. However, in villages where use of fertilizer was uncommon, 99% of sales were of small packs (10 kg preferred). One hundred tons of fertilizer was sold in the 2005/06 season alone through Progress Milling. Pannar also reported that 20 extra tons of seed was sold at the depots.

    Success can be expressed in several ways; today 10 LIMPAST extension officers are undertaking further trials on 100 farmers' fields in Limpopo and Mpumalanga Provinces. PM, Sasol, Pannar and LIMPAST are now funding a Development Coordinator position to supervise sales of seed and fertilizer at PM depots, and Sasol has provided a bursary for an agronomist to undertake a Masters Degree on fertilizer use by poor farmers and be responsible for correspondence, local publications and other support services.

    For more information contact

    3. A tale of master and slave

    ICRISAT is one of the four CG institutes that boast a Paracel high performance computer system (HPC). The term "high performance" comes from the four nodes of dual processors in this system, which is the equivalent of having 8 PCs simultaneously executing the same task. This obviously allows faster completion of the job submitted.

    While the ICRISAT HPC has four nodes the ILRI cluster consists of 32 nodes! Now why, you might ask, does one need high performance computers? The development of such systems was motivated by "Grand Challenge Problems" such as the analysis of the human genome, weather and climate data etc - large problems that take time. Thus high-end computing tasks that can be divided into independent subtasks are executed on high performance computing systems to save time and allow simultaneous execution of multiple tasks.

    ICRISAT's high performance computer ICRISAT's high performance computer.

    Taking an example closer home, one might want to use this system to study the population structure of genotyping data from several thousand accessions over several hundred markers. Under the cluster environment, one node designated the "master" coordinates the distribution of either the dataset or functional components of the task amongst the "slave" nodes to execute simultaneously. The master then collates results from the "slaves" at the end of job, compiles and returns the result to the user. Your analysis is complete without locking up your PC for weeks on end, provided of course your favorite analysis software is installed, configured and optimized to work on the HPC!

    Unfortunately not all software can be optimized to work on parallel architectures. The ICRISAT HPC hosts software for two instances of high-end computing jobs, one for population structure analysis in large datasets and another for sequence clustering and assembly for marker identification. With easy to use interfaces, you are unlikely to know where or how your job is running. Check these out at If any of you would like to run specific software on the HPC but are unsure of whether the software can be optimized for the cluster environment, write to us. The process of writing software for parallel architectures is quite different from serial programming and is an interesting study area within the bioinformatics domain. One does not have to have expensive HPCs; it is cheaper to build parallel systems using available computers connected to networks.

    For more information contact