These included: number of borders potentially 

 influencing flyspeck development in an orchard block, 

 distance between the trees and the borders, severity of 

 flyspeck in alternate host plants in the borders, density 

 of host plants themselves, foliar density of trees, height 

 and diameter of tree canopies, slope and relative 

 elevation of the block with respect to the orchard as a 

 whole, and planting density of the block (no. trees/ 

 acre). We examined all known host plants (from the 

 ground to 6 ft. above ground), not just blackberry. 

 Apples in the adjacent blocks were examined weekly 

 or bi-weekly from mid-July to harvest. 



At the end of each growmg season we looked at 

 the effect of each of the above-mentioned site factors 

 (and all factors combmed) on the amount of flyspeck 

 on the apples, to begin deriving a predictive model for 

 flyspeck mcidence at harvest. Prelimmary stepwise 

 regression analyses done separately for each year 

 suggested the importance of four variables: density 

 of flyspeck on alternate host plants in ttie borders, 

 number of borders, distance from apples to border, 



and slope of the block. The other site factors did not 

 explain substantial amounts of the variation in flyspeck 

 incidence. 



Combining data collected from unsprayed control 

 trees from the years 1995, 1996, 1998, and 1999, we 

 conducted a preliminary assessment usingg flyspeck 

 incidence data from various dates. Data from 1997 

 were not used, because all blocks received summer 

 fungicide sprays. Dates at or near harvest varied from 

 year to year primarily due to cultivar and weather 

 factors. Ultimately, we decided on a range of dates 

 allowing maximum inclusion of orchards in the data 

 set. Data from harvest or near-harvest ranged from 29 

 August to 23 September for the 4 years used in the 

 analysis. 



We concluded the static factor phase of model 

 building by combining these four independent variables 

 with the most inclusive range of harvest dates and a 

 fifth derived variable, inoculum index. Inoculum index 

 was expressed as the product of amount of flyspeck on 

 alternate host plants and the density of those plants in 



t 60 



a. 

 S-40 



e 



20 - 



a 



(A 



c: 



■O 







20 



40 



60 



80 



100 



Actual flyspeck incidence (% apples infected) at harvest 



Figure 1. Predicted versus actual flyspect at harvest in Massachusetts orchards, 2000. 



Fruit Notes, Volume 67, Winter, 2002 



