The development of a fruit-size prediction model for new cultivars Pink Lady, Braeburn, Fuji, Flamingo, Forelle and Rosemery

Bearing orchards of the main apple cultivars in the three apple producing areas (Elgin/Vyeboom, Ceres and the Langkloof) were identified as experimental sites during 1997. Nine trees were labelled in each of these orchards. The trees were commercially thinned, chemically and by hand. It was aimed at thinning being completed by approximately 40 dafb. Three of the nine experimental trees were alternatively stripped every second year after thinning, and all the fruit counted and weighed (per tree) to determine the average fruit massand number of fruit per tree. Orchards of the new apple cultivars, ‘Fuji’, ‘Cripp’s Pink’ and ‘Braeburn’, in the three areas were included in the investigation during 1999 and monitoring of the Golden Delicious, Granny Smith, Bon Chretien, Beurre Bosc and Packham’s Triumph cultivars was discontinued.

A multiple linear regression equation of the form Y = b0  + b1X1 + b2X1² + b3X2 + b4X2² +b5X3 + b6X3²+b7X4 + b8X4²  was fitted to the data where  :

Y   =  average distribution of percentage fruit/weight interval at harvest,

X1 = fruit per cm trunk circumference

X2 = (fruit per cm trunk circumference) ²

X3 = g/fruit

X4 = (g/fruit) ²

X5 = DAFB,

X6 = (DAFB) ²,

X7 = g/fruit interval

X8= (g/fruit interval) ²

The results were verified annually at the end of the season when the crop for three trees was graded and the prediction compared with the actual data. A semi-commercial weight sizer was used to grade the fruit in 10g size intervals.

The present models which were used to estimate the size distribution of the different cultivars in the areas could explain approximately 80 percent of the variation in the size distribution. Crop load and fruit size at 42 DAFB proved to be the most important variables and had a significant effect on the size distribution. The effect of tree spacing and temperatures were not studied due to the insufficient number of samples and these variables should be included in future models especially where tree spacing differs widely from the data which the present models were based on. The present data for crop load and fruit size covered a wide range of values but predictions for orchards where these values are less or more than the present values will not be accurate.

An early prediction of the size distribution will assist growers in identifying those orchards where the percentage of small fruit will be unacceptable and to adapt thinning procedures to reduce these percentages. Growers as well as packers and exporters will benefit if these predictions can be made available on an area basis which will contribute towards a more organised packaging and marketing system.