Fertilizer is a major expense to plantations, the largest variable cost to plantation managers. Few doubt its importance to continued high productivity. However, how much do managers really know about the payback from fertilizer on their estates? Knowing the general effect of an input and knowing its specific effect, under normal production conditions are two completely different things. In practice, agronomists can say little about the effect of fertilizer on specific estates on which they haven’t trialed because the effects of soil, climate and interactions with management factors, such as harvest processes, introduce huge uncertainties. Managers continue to tolerate uncertainty about the specific benefits of fertilizer because, until now, they had no way of estimating the specific effects of fertilizer. The idea behind ESE is to enable managers to see how fertilizer performs on their own plantations, under the real-world conditions of the production system. We design ESE so that it imposes minimal additional costs and can be adopted in ever-larger production areas, producing a stream of intelligence about the return on variable costs.

Previously this has not been easy for three basic reasons: (1) varying input was difficult over large production systems, (2) measuring output was not possible over large areas, and (3) there was no conventional method of socializing the results from ESE for management change. The first two limitations do not apply to oil palm, which has a history of detailed recorded keeping of inputs and outputs at the block level. The third limitation applies, but less so, since large plantation operators, at least, are organized to manage the flow of information and to act accordingly. Agronomists are included in this flow and so are available to provide science-based interpretations of plantation performance and to support decisions. We introduce a deliberate variation into the pattern of fertilizer input and analyze its effect in production. The process is one of dialogue between estate managers, senior managers and agronomists, and of observations in the production system.

In practice, getting value out of ESE is not trivial, and in this presentation, we will show our process of data capture, analysis and interpretation that provides the evidence for improved decision-making in fertilizer management of large oil palm operations. We use data from a 6,000ha fertilizer estate trial. The first step in the analyses is to look for treatment effects using conventional analysis of variance. Then, spatial analysis is used for a clearer method of assessing fertilizer response. IPNI and partners are trialing various analytical methods, including geographically weighted regression, Bayesian probability statistics and the use of Homologous Events. Insights will lead to the identification of ‘Sure win areas’ i.e. areas with a high likelihood of positive response to fertilizer, while ‘high risk areas’ identify blocks with a low likelihood of positive response, based on prior analysis. ‘Puzzle areas” identify areas that fail to respond for no clear reason and further detailed (trial) work is needed.