A rapid visual estimation of fruits per lateral to predict coffee yield in Hawaii

Abstract

Estimating coffee yield by measuring components of yield rather than complete harvests can improve the ability to account for heterogeneous conditions common in shade-grown farms and other agroforestry systems. We developed a rapid estimation technique that predicts yield per plant from (1) a visual estimation of the number of fruits per lateral summed across all laterals per plant, and (2) an average fruit dry mass derived from the random harvest of 50 fruits per plant. The technique was then applied to an experiment investigating three levels of managed tree shade (0, 40, and 60%) on Arabica coffee yield. Visual estimates of fruits per lateral proved to be good predictors of counts of fruits summed for each vertical (R2 = 0.97) and yield of the whole plant (R2 = 0.90). Simulations showed this was potentially better than existing methods to account for variable yield per vertical shoot at the whole-plant level and took less time or effort when applied to the plants in our study. We propose that fruits per lateral is an effective alternative component of yield that can account for heterogeneity in shade level, other environmental variables or management practices that can affect yield at small spatial scales in agroforestry systems.