Revisiting foliar diagnosis for oil palm potassium nutrition

International audience

English. Oil palm is a highly potassium (K)-demanding tree requiring fertilisation throughout its life cycle. One of the most common indices to monitor K fertilisation is K content (in % of DW, further noted as %K) in leaflets, typically used in foliar diagnosis (FD). However, uncertainty remains as to whether leaflets are always the most K-responsive organs and furthermore, implementing FD may require separate response curves for different soil and climatic conditions, and newly also i.e. genetic materials (progenies). As such, a universal K indicator ideally responsive to K availability and valid across many progenies could simplify FD. We investigated K responses of oil palm trees in four different progenies grown in North Sumatra, from planting to the maturation stage. Elemental analyses were performed along with biometrical measurements at tree scale, completed by data from agronomic databases. Uni-and multi-variate analyses were performed to identify the best indicators of K response. Tree K balance (difference between K availability and K needs) was computed to determine directly if (i) K supply met tree K demand and (ii) %K in leaflets, rachis or trunk were linked to the K budget. We also interrogate the accuracy of %K expression for K leaf content estimation and answer to it. Standard effects of K availability on height, leaf area, and yield were observed and varied mostly with the progeny. K balance cal-culations showed a systematic K imbalance when bunch production started, suggesting a non-negligible contribution of exchangeable soil K. %K in leaflets, rachis and trunk does not appear to be correlated with K balance. Our work brings out the important result that unlike %K in leaflets, phosphorus (P) and %K in rachis (more precisely %K symbolscript 3 %P) appeared to be tightly correlated with yield (FFB: fresh fruit bunches) two years later, suggesting a possible interdependence between fruits sink force and K-P elemental content in leaves. Since this indicator is under important genetic effect, we propose to use the combination of mineral contents generated by multivariate analysis, which successfully predicts FFB after two years, independently of the progeny.