339

Plantation Intelligence Applied Oil PalmOperations: Unlocking Value by Analysing

Commercial DataOBERTHÜR T, DONOUGH C R, COOK S, SUGIANTO H, LIM Y L, COCK J, KAM S P,

AND FISHER M JInternational Plant Nutrition Institute, Southeast Asia Program, 29C-03-08 Maritime Piazza,

Karpal Singh Drive, 11600 Penang, Malaysia

Plantation intelligence [PI] applies the concept of business intelligence, which is analysis of companydata, to oil palm production. Plantations already have collected monthly data of block yield but do notuse them to enhance financial performance. These data were analysed for a whole plantation to rankindividual blocks according to their ability to respond to applied fertiliser. Blocks were classified accordingto their average fertiliser productivity (AFP), which was associated with the block’s soil managementgroup (SMG). AFP varied between years depending on rainfall and SMG. The ranking was used to guidefertiliser management by diverting fertiliser from unresponsive blocks to those that are more responsive.Although the inferences lack statistical validity they appear robust from a practical viewpoint. They areeasy to evaluate in the field, since they require no upscaling from or interpretation of experimental data.They provide managers with a tool to evaluate the variable effect of fertiliser over the whole plantation indifferent years and to improve financial performance.

Keywords: Plantation intelligence, fertiliser response, soil management group, financial performance.

Information management is a vital componentof success for all businesses, but commercialagriculture has lagged in adopting it. Mostagricultural managers lack information aboutperformance, risks and values of fieldoperations on which to make key decisions. Inoil palm plantations, uncertainty has alsoincreased because of rapid expansion andchange. The goal of Plantation Intelligence (PI)is to accelerate the adoption of informationmanagement in the palm oil industry to supportthe profitable adoption of best managementpractices and other changes.

CONCEPTS

Plantation intelligence

Strategy in  the  business  sense  is  planning a

company’s next move. In contrast, ‘tactics’ isthe process of physically carrying out that plan.Business intelligence is any tool, activity orprocess that is used to analyse a company’sdata to support better decision making, toidentify new opportunities and to reduce costs.A company uses both strategic and tacticalintelligence to develop and implement itsbusiness plan. The primary difference betweentactical and strategic intelligence is the timingregarding the business environment.

Strategic intelligence allows decisionmakers to visualise the company’s futuredirection in accordance with its stated missionand goals. Tactical intelligence deals with thepresent. It provides the necessary informationto monitor the company’s current operatingenvironment and identify new opportunities.

The Planter, Kuala Lumpur, 93 (1094): 339-351 (2017)

340

Tactical intelligence offers analysis of currentconditions and identifies the actions needed toachieve the company’s strategic objectives.Tactical intelligence focuses on the resourcesavailable such as people, time and money forachieving the strategic goals. Tactical planninghelps a company make the most efficient useof resources to deliver value to its stakeholders.

PI applies the concepts of both strategicand tactical business intelligence to the palmoil industry. PI is directed towards managementof the components of the production system,which managers and analysts togetherdetermine sources of variation that managersdeem to be important. The process starts withacquiring the data, preliminary analysis anddiscussion with management. The analysis isthen refined, further discussed withmanagement, which is followed by a furthercomprehensive review.

PI link to plantation businessperformance

The American management consultant PeterDrucker told us that we all have a businessmodel in mind when we try to manage anyentity, even if we do not realise it. A businessmodel is the conceptual structure supportingthe viability of a business. It outlines the way afirm captures value, including its purpose, itsgoals and its ongoing plans for achieving them.All business processes and policies are part ofthat model. According to Magretta (2002), abusiness model answers Drucker’s keyquestions: Who is your customer, what doesthe customer value, and how do you delivervalue at an appropriate cost? A firm without aviable business model may not last for long.Oil palm producers, whether large or small, areno different from anyone else and each has abusiness model whether they acknowledge it

or not. Large oil palm plantations undoubtedlyhave business models, which are not usuallypublished in detail, but profit for shareholdersmust be one important objective, amongstothers. Shareholder profit for oil palmbusinesses primarily comes from profitableproduction of palm oil, that is the value of palmoil produced is greater than the cost ofproducing it. The business model is createdaround (a) business case(s), i.e. actualopportunities that the team are looking forwardto harness. A business case clarifies “what ifquestions”. For example: “does the investmentinto a changed fertiliser management practicejustify the expense”. It is usually organisedaround a single action or decision and itsalternatives, and helps to anticipate impactsfrom such action, by being based on a costmodel and a benefit rationale.

A business plan is similar to a businessmodel but it is more detailed, specifying all theelements required to demonstrate the feasibilityof the prospective business. A business plan istherefore a formal statement of business goals,why they are attainable, and the plans to reachthem. The plan may also contain informationabout the organisation or team that is attemptingto reach those goals. Business plans answerquestions about the future of the business, andare regularly updated to account for changesin external conditions (for example climatechange), including changes in perception andbranding by the customer, the client, or thelarger community. Major changes to an existingplan or plans for a new venture usually requirea three to five-year business plan, which is thetimeframe most investors require. The businessplan focuses on the business performance, byquantifying major components of income suchas yield, margins or revenues. Being based onthe business model, it shows where and howthe business makes money.

Plantation intelligence applied oil palm operations: unlocking value by analysing commercial data

341

Plantations collect copious amounts ofoperational data, which cannot be analysedbecause the data are not captured in amanagement information system (MIS). Theonly data that are used are of inputs and costs,which are presented as summaries, whichmasks actual variations in performance. Theirvalue as tools for learning is lost.

The value of the data already beingcollected will come from analyses that provideinsight into companies’ operations. Theanalyses can unlock the value that the datacontain and relate them to real experiences.PI is a mechanism that brings together insightsderived from data analyses of agronomicunderstanding including research to provideguidance to plantation management. It providesvaluable input to the business plans of palm oilplantations by clarifying the agronomicperformance of an operation. This allowsquantification of the opportunities provided bythe management of key resources and keyactivities. It helps to clarify how specificbusiness cases impact on the cost structureand revenue streams.

HOW PI SUPPORTS AN OIL PALMPLANTATION’S BUSINESS PLAN

Palm oil

The global trade in crude palm oil (CPO) hasincreased from 1.5 million tonnes in 1961, worth$1.3 billion, to about 56 million tonnes in 2013/2014, worth over $ 40 billion (in constant 2004-2006 US$). Most of the increase came fromMalaysia and Indonesia, which account forover 80 per cent of global production (USDA,2014). Although the yield per unit area hasincreased somewhat since 1961, most of theincrease came from greatly increased areaplanted to oil palm, which has led to increasing

external criticism of the industry. The criticsclaim that plantations have been established inpristine rainforest and drained peatlands.Burning of these lands releases greenhousegases and impacts the biodiversity of nativeplants and animals, especially of charismaticmegafauna. The truth appears more nuanced,but the industry is conscious of the need to takesteps to avoid widespread censure.

Global demand for CPO is forecast tocontinue to grow for at least the next 10 years,which further increases the pressure to expandthe area of oil palm. One way to reducecriticism while meeting the increased demand,which appears to be feasible, is to reduce thegap between actual plantation yields and thosethat can be achieved with improvedmanagement.

The major ongoing operational costs forpalm oil plantations are for fertiliser and labourfor harvesting. There are data that suggest theamount of labour used for harvesting is causallycorrelated with field yield of fresh fruit bunches(FFB). Plantation managers, however,undoubtedly allocate labour for harvest to eachblock depending on their subjective assessmentof block yield, so causality is moot.Nevertheless, managers are continuallyconfronted with the need to provide adequatelabour for harvesting to ensure that the yieldtaken (FFB harvested) is as close as possibleto the yield made, that is the FFB available forharvest. The problem is confounded by externalsocial and political pressures, further discussionof which are outside the scope of this article.

Yield gaps

Analysts have used the concept of ‘yield gap’to address issues improving farmers’ yields,including forecasting the effects of climatechange on future food security. While avoiding

The Planter, Vol. 93, No. 1094, May 2017

342

the technical aspects, the authors think that theconcept is useful to address some of the issuesconfronting palm oil industry. Scientists talk ofbiological potential yield, that is the yield possiblewhen all conditions for plant growth are optimal.A more practical concept, however, is the bestachievable farm yield, that is the yield that anexperienced, prudent farmer might achieve ongood land in a good season, but hedging againstnormal risks. It indicates what yield a farmingsystem might obtain if all farmers used the bestavailable technology. The yield gap is thedifference between this best achievable yieldand the actual yields that farmers obtain. Thereare many reasons why yield gaps are not zero,chief amongst them that farmers not only takedifferent approaches to risk but they assessrisk differently.

The maximum recorded yields of oil palmare 9-10 tonnes per hectare per year CPO,while maximum yields from whole blocks inplantations are 7-8 tonnes per hectare per year.Best reported plantation yields are about6 tonnes per hectare per year CPO but theaverage over most of Indonesia and Malaysia

is about 4 tonnes per hectare per year, muchhigher than the average elsewhere in the world(Figure 1).

FFB yield

Plantations record the yield of FFB from eachblock as it is harvested. Analysis of these datareveal the intrinsic performance of each blockas a component of the plantation. They alsoshow block yield responds to uncontrollablefactors like rainfall and to controllable factorsof plantation management. These are allcomponents of PI and provide guides to theexecution of the business plan.

FFB yields increase in young palms untilthey reach a plateau from 7-18 years, afterwhich they decline (Figure 2). Plantationsnormally replant after about 25 years. Yieldanalyses remove the effects of palm age andidentify the effects of uncontrollable effects toidentify the intrinsic yield capacity of eachblock.

The intrinsic yield capacity of each blockis little used by plantation management. But it

Figure 1 Examples of reported oil palm yields at various spatial scales

Plantation intelligence applied oil palm operations: unlocking value by analysing commercial data

343

is a valuable tool that could be used to improveprofitability by eliminating low-yielding blocks,either according them low priority in resourceallocation (labour, fertiliser) or removing themfrom production. The logical analysis is grossmargin (Figure 3), which can readily be

calculated from plantation data. If the intrinsicyield is higher than actual block yield becauseof controllable factors that can be rectified,there are two questions for management toconsider. How much additional resources andcost will it take to bring the yield up to the

Figure 2 FFB yield age profile (age on x-axis, FFB on y-axis), showing sharp ascent to 5 years old,followed by plateau. The dotted line at 30 t/ha is provided for easy visualisation

Figure 3 An example of a “Simple Gross Margins” analyses, showing blocks on the x-axis and monetary return on the y-axis. The different colours indicate the estateto which a block belongs

The Planter, Vol. 93, No. 1094, May 2017

344

intrinsic level, and is there enough time left inthe block’s life cycle to recoup the additionalinvestment. PI can answer both questions.

HOW PI SUPPORTS APLANTATION’S CONCRETE

BUSINESS CASE

Application of PI to fertiliser

Each tonne of FFB contains 22-35 kg of themacronutrients N, P, K and Mg (see forexample Ng & Thamboo, 1967; Ng et al.,1968), which must be replaced with fertiliser ifyields are to be maintained. A crop yielding20 tonnes per hectare per year FFB requires450-700 kg per hectare per year of a mixtureof fertilisers containing these macronutrients,which can account for as much as 50 per centof the plantation operational costs. Themanagement of fertiliser is therefore a majordeterminant of profitability of the plantation.

The management unit of oil palm at theplantation level is a block of 20 - 80 ha, plantedat the same time and given uniformmanagement. Agronomists assist inrecommending the allocation of amounts offertiliser applied to each block, usually byextrapolating existing knowledge fromexperiments on yield response to fertilisers. Butthe company management’s overall financialbudget controls fertiliser policy at the plantationlevel. The emphasis at the company budgetinglevel is on cost control and there is little attemptto determine whether the fertiliser applied toeach block is profitable. That is, applications atthe commercial block level are not generallytailored to the actual measured responsivenessof the palms to which they are applied. This isbecause the commercial data are rarelyanalysed systematically to address this issue.

Plantations maintain records of the numbersof FFB harvested from each block. They

estimate the mean weight of the bunches ofeach block from each collected load of bunchesdelivered to the mill and use it to calculate theFFB yield of each block. Bunches within anyone load delivered to the mill may not all comefrom the same block. Nevertheless, numbersof bunches harvested from each block arereliable and due to the very large numbers ofbunches involved small discrepancies in bunchweight are unlikely to be important.

IPNI and collaborators worked withplantation management to help make fertiliserapplication to oil palm more profitable (IPNI,2015). Although fertiliser accounts for abouthalf of a plantation’s annual budgetexpenditures, plantation management does notknow how profitable this expenditure is. Theydo know, however, that if they stop fertilising,yields fall and can take years to recover.

FFB yield varies a lot between blocks, oftenfor reasons that are not identified. Position inthe landscape and soil are important, but otherthings such as genetics, agronomic manage-ment and pests and diseases also play a part.The IPNI researchers sought to use plantationhistorical records to identify blocks with highyield potential from those that yield poorly.

On the face of it, this might seem an easytask, but oil palm is a complicated plant. It takesover three years from the time a fruit bunch isinitiated as a tiny bud weighing a fewmilligrammes to when it is a “fresh fruit bunch”ready for harvest weighing over 20 kg inmature palms. There are several critical stagesduring this development process (Breure,2003). When they initiate about 39 monthsbefore the harvest, the vestigial flowers in anoil palm inflorescence include both male andfemale types. After another 12 to 15 monthsgrowth, the final sex of the developinginflorescence is determined, about 60 per centbecome females that develop into fruit bunches

Plantation intelligence applied oil palm operations: unlocking value by analysing commercial data

345

and the remainder become males that providepollen to fertilise the female inflorescences atanthesis. Stress at this stage can alter the ratioof females to males and hence affectsubsequent yield.

The next vulnerable stage is when theinflorescence starts to expand in size. Stressat this stage can be severe enough that thedeveloping inflorescences abort.

The final vulnerable stage is post-anthesisafter the male flowers release pollen thatinsects convey to the female flowers and thatfertilises them. The pollinated bunch developsand matures over the last six months of theprocess, and obviously stress during this stagecan cause the developing bunches to fail andreduce both fruit and oil yield.

Fertiliser has both direct and indirect effectson crop yield. Adequate nitrogen supply isnecessary to develop a full canopy that isneeded for the palms to produce adequatephotosynthate for fruit production. Potassiuminfluences oil yield and its quality. But becauseinadequate nutrient status during at least thelast two years of fruit development can affectyield of FFB, the researchers need to takeaccount of the fertiliser applied over the wholeof this time.

Examination of the data suggested thatfertiliser applied between sex differentiation andanthesis, that is 6-18 months before harvest,has greatest effect on FFB yield. Fertiliserapplied before sex differentiation or during fruitproduction after anthesis does affect yield butto a lesser extent. Accordingly, the IPNI teamweighted the fertiliser applied in the three yearlyperiods prior to harvest in the ratio 0.5:1.0:0.5.

Data used in the IPNI analysis

Cook and his colleagues obtained data for theyears 2010-2014 for a commercial plantationin central Kalimantan, Indonesia that covers

17 700 ha and consists of 447 individual blocks.Data for each block included total and plantedarea, year of planting, total number of palms,the predominant soil type for each block andits slope class. Soils were classified into soilmanagement groups (SMGs, Table 1) after theplantation was developed, based on soil depth,texture, drainage and estimated fertility.Monthly rainfall data and the number of rainydays applied to the whole plantation. Yield dataconsisted of monthly FFB yield and the numberof bunches for each block. Fertiliser data werethe annual amounts of each fertiliser applied toeach block. The team summed the elementsN, P, K, and Mg in the individual fertilisers toobtain the total amount of elemental NPKMgapplied annually.

Managers know that block yields varywidely on even the best-managed estates overapparently uniform land. But managers wantto know whether blocks or areas perform asexpected. They therefore want to see howyield in a particular block varies from theaverage to understand better what drives orconstrains profitability. Did the FFB yield meetexpectations, given the age of the palms, theiryield history and other issues such as droughtor excess water? Did the block pay for itsfertiliser?

Data for a whole plantation do not providethe design control used in normal field plotexperiments. As expected, the plantation datawere confounded with uncontrolled factors,which made precise interpretation difficult. Theresearchers therefore calculated thecoefficients of variation (CVs) of FFB yieldover all blocks for the five years, which theyused to isolate the effects of uncontrolledexternal variables. This procedure differs fromthe analysis of a designed experiment wherevariation is used to test the statistical confidenceof the effects of applied treatments. The

The Planter, Vol. 93, No. 1094, May 2017

346

researchers therefore presented the results assimple analyses from several perspectives,which together provided insights on whichmanagement can act. An important part of theprocess was to include the skills of the managersand agronomists in all stages of the analysis.Below is a selection of the analysis and dialogueprocess.

Results from analysis, inferences anddecisions supported

Cook and his colleagues used CVs to reducethe confounding effects of uncontrolled externalvariables to focus on the effect of appliedfertilisers on FFB yield. They then identifiedhigh- and low-yielding blocks and their

association with different soil managementgroups to support which blocks merited moreor less fertiliser (Table 2).

Stage one estimated average fertiliserproductivity (AFP) for all blocks without regardto tree age or soil management group (SMG).Younger trees yielded far less than older trees,but also received less fertiliser so that AFP isconfounded with tree age. Trees overfive years old have stable yield and similar rateof fertiliser year-by-year. Each kilogramme ofapplied fertiliser produces about 4 kg of FFB(Figure 4). The data show a wide scatteraround the general relationship. The responseof FFB to applied fertiliser differed betweenSMGs. SMG C gave the highest of 8 kg perkg followed by SMG A with 4 kg per kg. SMGs

TABLE 1SOIL MANAGEMENT GROUPS (SMGS)

SMG: with soil Major soil characteristics Main agronomic limitationstypesa included

A: Typic kandiudult, Moderate to deep (>50 to >100 cm depth); Moderate to low soil fertility.Typic kanhapludult. well drained. High iron content leading to

Texture: sandy clay to sandy clay loam. high P-fixation.Subsoil with platy iron-coated gravel. High porosity hence crops

prone to water stressB: Typic Paleudult Deep; well- to moderately well-drained Moderate to low soil fertility.

(20-50% gleying within 100 cm or 50-80%gleying below 100 cm). Slopes erosion-prone.Texture: sandy clay to sandy clay loam.Developed over sub-recent alluvium.

C: Aquic Paleudult, Deep; imperfectly (50-80% gleying below Low soil fertility, especially P andAeric Haplaquult, 75 cm, or 20-50% gleying below 50 cm) to K.Typic Paleaquult poorly drained (50-80% gleying within 50 cm). Poor drainage hence flood-prone.

Texture: sandy clay to sandy clay loam.D: Typic Shallow (<50 cm) to moderately deep. Very low nutrient and moisture-Haplohumod Structureless sandy soils with cemented layer. holding capacity.

Cemented layer, poor anchorage.Flooding in rainy season,moisture stress in dry season.

a – Soil Taxonomy (USDA)

Plantation intelligence applied oil palm operations: unlocking value by analysing commercial data

347

B and D appeared to be unresponsive(Figure 5).

The SMGs responded differently betweenyears, which suggests that there is aninteraction between SMG’s and climate(Figure 6), principally rainfall. The drought yearof 2011 reduced yields of FFB in 2013 by3 tonnes per hectare, and only SMG Cresponded to fertiliser. Performance reboundedin 2014, with SMG C responding strongly toapplied fertiliser followed by SMG A. There is

no clear trend for SMG A in 2014. There areno data to explain the lack of response in years2010-2012, but low levels of applied fertilisermay be responsible. Blocks in SMG D, a podzolsoil performed poorly in all years. Podzols areknown to be vulnerable to both drought andwaterlogging, so if the intrinsic yield of blockswith SMG D is low, then it could makeeconomic sense to apply less fertiliser to theseblocks.

The data for SMGs across estates

TABLE 2INFERENCES AND DECISIONS SUPPORTED BY RESULTS IN SUCCESSIVE STAGES IN ANALYSIS

Stage in Result(s) from Example(s) of Decision(s)analysis analysis possible inference supported

1 Overall, average fertiliser Secure inference not None.productivity (AFP) is 16 kg possible because ofFFB per kg of additional confounding by tree age.nutrients applied.

2 For palms >5 years old, AFP Overall picture is clearer Further analysis requiredis 4 kg per kg of additional but substantial unexplained before fertiliser responsenutrients applied. variation remains. can be inferred.For palms 5 years old,AFP is 11 kg per kg ofadditional nutrients applied.

3 For palms >5 years old, AFP Substantial unexplained Possible increase fertiliserdiffers for each soil variation remains but SMG applications on SMG A.management group (SMG) C seems the most Review year-to-yearas follows: responsive to fertilisers. performance on SMGsSMG A = 4 kg/kg SMG D is a difficult soil to B, C and D.SMG B = 0 kg/kg manage and yields lowerSMG C = 8 kg/kg than the other SMGs.SMG D = 0 kg/kg.

4 For the period 2010-2014, 2013 a difficult year for SMGs A and C highresponse varies year-by- most SMGs. yielding, with highyear, with strong interactions Strong recovery in 2014 responsive. SMG D lowevident between year and from SMGs B, C and D. yielding but potentiallySMG. SMGs B and C show similar responsive with

pattern of high FFB, low management inputresponse 2010-2012, thenstrong response in 2014.

The Planter, Vol. 93, No. 1094, May 2017

348

Figure 4 Overall yield response to 2-year weighted total fertilisers applied for all blocks of trees>5 years old

Figure 5 Yield response to fertilisers segregated by four soil management groups (SMG). The lines arethe regressions of FFB yield against 2-year weighted sum of NPKMg applied, for all blocks oftrees >5 years old

Plantation intelligence applied oil palm operations: unlocking value by analysing commercial data

349

(Figure 7), show that FFB yields of Estate 3responded more strongly than in Estate 1 forthe same soil types. There is no clearexplanation nor could the team explain thenegative response to fertiliser on SMG B inEstate 2.

CONCLUDING REMARKS

Interpreting analysis of noisy commercialdata

Plantation managers confront many linkedfactors such as variable block productivity, costof fertiliser, labour availability and cost, soil, andweather. Each of these factors varies in spaceand time, which introduces uncertainty. Thegoal of the analysis was to provide a tool toallow managers to evaluate the variable effectof fertiliser over the whole plantation in differentyears, and to understand the factors that control

it. Managers could then make rational decisionson how to modify fertiliser rates according toblock responsiveness. Agronomists can use theinsights from the analysis to improve theunderstanding that they use to recommendapplication rates for plantations.

Although the data did not reach statisticalsignificance, the analysis was practically andcommercially important to managers. Theystarted using the information to guide fertiliserapplication on the plantations for which theywere responsible.

The analysis also helped managers assessand manage risk with regard to the interactionbetween soil, climate and fertiliser. It providedobjective information about field responses toapplied fertiliser on the actual estate and at thesame scale at which management is requiredto make decisions. Although the inferenceslack statistical validity they appear robust froma practical viewpoint. They are easy to evaluate

Figure 6 Yield response variation over 5 years in different soil management groups. Lines represent theregression of FFB yield against 2-year weighted sum of NPKMg applied for each soil managementgroup and year, for all blocks of trees >5 years old. Each row represents a year. Columnsrepresent soil management groups. The dotted line at 30 t/ha is to aid comparison

The Planter, Vol. 93, No. 1094, May 2017

350

in the field, since they require no upscaling fromor interpretation of experimental data.

The analysis made use of data availablefrom the entire production area of a plantation,which is routinely used for accounting andoperational management, but which hasbroader utility. The analysis of fertiliserresponse the IPNI team developed used datafrom the same set of blocks to which theconclusions are intended to apply. In this way,it removed the uncertainty of applyinggeneralised responses that may not be relevantin all circumstances.

Some pitfalls

Analysts need to be aware that external factorssuch as labour availability may causediscrepancies between yield made and yield

taken. The team emphasises that conclusionsderived from analysis of commercial data canonly be applied to the area from which the datacame. Managers may, however, not feel soconstrained and may use their experience toextrapolate insights to other areas they considersimilar.

In contrast to formal experiments,plantation data are unlikely to include zerotreatments so that they are unlikely to providedata structures that give clear results.Moreover, the results are confounded byuncontrolled variables such as climate, whichlie beyond the analysts’ control. Nevertheless,this is the first time plantation data have beenused to identify the response of oil palm tofertiliser at a commercial scale.

The IPNI team offers oil palm plantationsa clear recommendation. All commercially-

Figure 7 Yield response variation over three estates in different soil management groups. Lines representthe regression of FFB yield against 2-year weighted sum of NPKMg applied for each soilmanagement group and year, for all blocks of trees >5 years old. Each row represents an estate.Columns represent soil management groups

Plantation intelligence applied oil palm operations: unlocking value by analysing commercial data

351

managed plantations have similar data that areroutinely collected and available. Analysis ofthese data can indicate those blocks that areresponsive to applied fertiliser and those thatare not. Then managers can divert at least someof the fertiliser from the unresponsive blocksto the responsive blocks. This will lead togreater fertiliser use efficiency as well asincreasing the profitability of the fertiliserapplied. This form of analysis could become acore tool for managers to enhance performanceas the requirement to intensify productionincreases in the future.

ACKNOWLEDGEMENTS

We appreciate and acknowledge the supportprovided for this work by the International PlantNutrition Institute, Canpotex Limited, andplantation partners including WilmarInternational and IJM Plantations Berhad. Wealso appreciate the numerous discussions withindustry players, plantation and estate managersand other practitioners that helped sharpeningand improving our arguments.

REFERENCES

BREURE. 2003. The search for yield in oil palm. In: OilPalm – Management for Large Sustainable Yields(edited by Fairhurst and Härdter). International PlantNutrition Institute and International Potash Institute,Oxford graphic Printers

COOK, S., DONOUGH , C. R., LIM, C. H., LIM, Y.L., COCK, J., KAM, S. P., FISHER, M. J.,VERDOOREN, R. and OBERTHÜR, T. 2016. Aprocess to estimate oil palm yield response tofertiliser from analysis of commercial data fromplantations by International Plant Nutrition Institute(IPNI), Penang, Malaysia (Internal report).

MAGRETTA, J. 2002. Why business model matter.Harvard Business Review 80, 86-92

NG, S. K. and THAMBOO, S. 1967. Nutrient contentsof oil palm in Malaya. I. Nutrients required forreproduction: Fruit bunches and male inflorescences.The Malaysian Agricultural Journal 46, 3-45.

NG, S. K., THAMBOO, S. and DE SOUZA P. 1968.Nutrient contents of oil palm in Malaya. II. Nutrientsin vegetative tissues. The Malaysian AgriculturalJournal 46, 332-391.

USDA. 2014. Oilseeds: World markets and trade. UnitedStates Department of Agriculture, ForeignAgricultural Service. http://apps.fas.usda.gov/psdonline/circulars/oilseeds.pdf.

The Planter, Vol. 93, No. 1094, May 2017