As indicated in other Informed Farmers articles related to mature bearing coffee trees, there is no guarantee that all the coffee cherry on a tree will be ready to pick at the same time, even with perfect flower synchronisation. Often it is desirable to harvest trees with varying amounts and maturities of cherry. Given the adjustments now available, the harvester should be able, theoretically, to harvest virtually all the ripe cherry, leaving the immature cherry on the tree. In practical terms this will never occur unless new coffee cultivars with a large difference between FRF of immature and mature cherry are developed specifically for machine-harvesting.
Figure 1 below, presents data on harvester performance trials conducted at two sites, the McBryde Plantation on Kauai in Hawaii, and Lakeland Downs in North Queensland. The trials compared the performance of different harvesters under conditions where it was desirable for the harvester to be as selective as possible. In the trial, various settings on the harvester were changed and the resulting changes in performance monitored.
At both sites, young, 1.8 m tall, semi-dwarf trees of cultivar Catuai Rojo carrying about 12 t/ha of cherry, approximately half of which was immature, were harvested. The prototype harvester developed for McBryde, the McBryde Quad-Shaker, removed over 70% of the ripe and only 30% of the immature cherry on the tree. At the same site, the Littau berry harvester removed about 35% of the ripe cherry and about 20% of the immature cherry at the settings at which it was operating. The McBryde harvester was clearly more selective, and this selectivity can be seen in the slope of the lines in Figure 1. The steeper the slope, the more selective the harvester.
Figure 1 also demonstrates the difference in performance between different harvesters and how changing the way a harvester is operated changes its performance. It is important to note that because of differences in design we cannot directly compare speed and shaker frequencies between machines.
In the Lakeland Downs trial, a similar Littau berry harvester to that used in Hawaii was trialled against the DPI-SP harvester developed by Queensland Department of Primary Industries (and subsequently used as the basis of the Austoft coffee harvester). The limited difference in FRF between ripe and immature cherry made it more difficult to discriminate between ripe and unripe cherry than in Hawaii, and, as was expected, the berry harvester showed very little selectivity. Increasing the aggressiveness of the shakers removed more fruit and more leaf. The DPI-SP harvester was significantly more selective, and removed less leaf, particularly at the lower speed and shaker settings. The slope of the line indicates the DPI-SP machine at 0.7 km/h or 800 cpm had similar selectivity to that of the McBryde Quad-Shaker machine.
By increasing the shaker speed and forward speed on the DPI-SP harvester, the machine became less selective and removed less ripe cherry. Ripe cherry removal could have been maintained by further increasing the shaker speed, at the expense of increasing leaf removal.
As the proportion of ripe cherry on a lateral increases, or when some cherry has been removed, the characteristics of cherry removal change, because the clusters are no longer intact. Consequently, after one harvester pass, higher levels of cherry removal are achieved at similar shaker speeds subsequent passes. Greater selectivity is also achieved at any shaker setting, and leaf loss relative to cherry removed is reduced.