Modelling seasonal variation in oil palm bunch production using a spreadsheet programme

Annual cycling in oil palm bunch yield is an almost invariable phenomenon even in regions that lack marked seasonal changes in climatic factors, such as radiation or rainfall, likely to have a large influence on yield. Furthermore, such cycles persist even under irrigated conditions. While yield-based endogenous feedback mechanisms have been invoked to partly account for such behaviour, the likely time-lags involved are not generally consistent with the regular annual cycles that are frequently observed. Using data obtained from a long-term trial on a peat soil with a good year-round water supply,the role of various developmental factors in contributing to the resultant yield patterns,was examined.The factors were: a)frond emergence interval (FEI), b)rate of inflorescence and bunch development (FEBRdefined by the number of days from frond emergence to bunch ripening), c)the proportion of nodes with bunches (NWBmainly a function of sex ratio and abortion incidence) and d)single bunch weight (SBW). Frond emergence, male and female inflorescence numbers, abortion and single bunch weight all exhibited regular annual variation in the trial. Yields were simulated using a spreadsheet with the aim of dissecting out the contribution and relative significance of each factor. Even with all factors held constant, there was variation in monthly yield, although it was erratic and failed to result in the single annual peak characteristic of observed yield patterns. Regular annual peaks were,however,obtained by introducing sinusoidal oscillations in the amplitudes of the four factors either individually or in combination.Amplitudes were tested that represented a range of probable behaviour from mild to maximum variation.The best agreement between simulated and observed yields over an 8.5-year period (r2 = 0.6) was obtained by varying NWB (using an amplitude of 50%) while similarly good agreements were achieved by appropriate variations in FEI and in FEBR. SBW had only a small effect. Combining factors did not appreciably improve the correlations over those obtained by the factors individually,although in some cases it resulted in similarly high correlations being achieved using lower amplitudes. These findings show that all the developmental processes examined played some role in accounting for annual yield cycles but the results still leave open the question of what factor(s) are responsible for the cycling of each of the underlying processes