The ‘change from control’ was calculated for each 10 min time point by subtracting the value in a control plot from the value in each of the three treatment plots within the same block. The mean and 95% confidence interval of the change from control among blocks, defined as the region between the mean plus and minus the standard error of the mean times 1.96, was calculated for each treatment-control pair at each time point. A treatment was determined to be different from the control if the confidence interval of the mean change from control did not include 0. The total time per day that the change from control was different was determined for both positive or negative values of change from control. Because analysis of the mean and confidence interval results indicated distinct patterns of change from control associated with time of day, the total time per day value was also calculated separately for day and night for both positive and negative values. At ranch 1, three of the control plots were compromised early in the season due to accidental removal of leaves by the farm crew. Disease, receptacle, and environmental data from these plots was discarded as missing for analysis. In addition, some environmental data was missing or modified due to mistaken settings on the loggers. First, the four data loggers placed in the plots in one block at ranch 2 stopped recording in October. Second, one of the loggers in 3-row at ranch 1 and one at ranch 2, both in manual removal plots, only recorded instantaneous readings at each time point and not 10 minute averages, round pot therefore no data was available from these plots. Third, one logger in 3-row at ranch 1 and at ranch 2 recorded values approximately 4 and 1 minutes, respectively offset from the rest of the loggers. These timestamps were adjusted to the nearest 10 minutes to align with the rest of the loggers.

At the last rating date, disease incidence was higher in ranch 1 compared to ranch 2 , although the ranches had differences in timing intervals between treatment application from final rating date of 96 and 64 days for ranch 1 and 2 respectively. . On the final rating date, incidence in individual plots ranged from 41.9 to 91.5% in ranch 1 and from 8.6 to 77.8% in ranch 2. Analysis of fixed effects showed evidence for a significant leaf removal × date interaction for ranch 1 2- row and ranch 2, but no evidence for an effect of leaf removal or leaf removal × date for ranch 1 3-row . Examination of the interactions showed that for ranch 1 2-row, evidence for a significant effect of leaf removal was observed on the final rating date only . On this date , the percentage of canes infected by Botrytiscinerea was significantly lower in plots treated with twine removal relative to control plots . At ranch 2, some evidence for an effect of leaf removal was observed on the first two rating dates , but mean separation showed that incidence was similar among all treatments on these dates at P ≤ 0.05. However, examination of individual pairwise comparisons showed some evidence that incidence was higher in manual treated plots compared to the control on Oct 17 . Although leaf removal did not have a significant main or interaction effect on cane Botrytis incidence for 3-row tunnels at ranch 1 , there was a numeric trend of lower incidence in blade removal and control plots on the first two rating dates compared to manual and twine plots . Overall, disease severity levels were similar across ranches during the same time period after treatment. But, for the last rating date at ranch 1 when disease increased, more time had elapsed compared with the last rating date at ranch 2 .

In 2- row tunnels, the shortest lesion length on the final rating date was 0.64 cm at ranch 1 and at ranch 2, while the 3-row tunnels at ranch 1 was 0.95 cm. The longest lesion lengths were 112.39 cm at ranch 1 and 68.58 cm at ranch 2. Evidence of a significant effect of leaf removal treatments on disease severity was observed in all three experiments . In 2 row tunnels at ranch 1, total length of cane Botrytis lesions per cane in manual removal plots were significantly lower than the blade removal treatment . At ranch 1 in 3-row tunnels, lesion length was significantly lower in the blade, manual, and twine removal compared to the control treatment. In contrast, at ranch 2, lesion length in twine and manual removal plots were significantly higher than the control plots, whereas lesion length in blade removal plots did not differ from other treatments or the control. There was a non-significant pattern of lower severity in manual removal plots at ranch 1 when compared to twine and control in the 2-row tunnels and to blade in the 3-row tunnels . There were no statistical or numerical differences in the number of receptacles between treatments in any of the three experiments . The air temperature change from control in individual blocks when averaged by hour generally fell within the range of -2o C to 4o C, although the overall minimum and maximum values over all three experiments were -6.06o C and 7.35o C, respectively . In each of the three experiments, the direction of the change from control did not always agree among blocks, and this disagreement was typically due to one block only. Analysis of the change from control using 95% confidence intervals showed that air temperature in plots treated with leaf removal significantly differed from the air temperature in control plots for appreciable lengths of time per day .

However, the direction was opposite at different times of day. Throughout each of the three experiments, air temperatures in leaf removal treated plots were generally higher than in control plots during daytime hours but lower relative to control plots during nighttime hours. The day-to-day pattern of time with significantly different change from control was similar among treatments, but was generally highest in manual removal plots and lowest in twine removal plots. In 3-row tunnels at ranch 2, for example, temperaturein manual plots had the most hours with a significant difference from the control during the night compared to the other treatments. The relative humidity change from control in individual blocks as hourly averages generally ranged from -10 to 10 percentage points, but the absolute range over all three experiments was -22.1 to 50.3 percentage points . Similar to air temperature, the direction of the change from control typically disagreed for one of the five blocks. Analysis of change from control showed that relative humidity was different between treatment and control plots for noticeable lengths of time per day, but the patterns were dissimilar from air temperature and were not consistent among experiments . In 2-row tunnels, all treatments followed a similar pattern of higher relative humidity in treated plots compared to the control during nighttime hours but lower during daytime hours, but in the final month of the experiment relative humidity in all treatments was higher than the control during both day and night. In 3-row at ranch 1, relative humidity was generally higher in treated plots compared with the control through the first three months of the experiment, round plastic planter regardless of time of day. At ranch 2, relative humidity in blade and manual treated plots was higher than the control at night, but during daytime hours was both higher and lower throughout the experiment. In general, day-to-day patterns among treatments were similar. However, a clear trend was observed in which relative humidity in twine removal plots was higher than in control plots for very little time compared with blade or manual treatments, and only in 3-row tunnel at ranch 1 and ranch 2. In contrast, for 3-row tunnels at both ranches, the time that relative humidity was lower in treated plots compared to the control was similar among all three treatments.Shifts were observed in change from control patterns in the 2-row and 3-row experiments at ranch 1 beginning in mid-April. A shift was most pronounced for the 3- row tunnels where, in treated plots relative to the control, relative humidity shifted from higher to lower, whereas temperature shifted to exclusively higher, regardless of time of day. In 2-row tunnels, a similar shift was observed for relative humidity only during daytime hours, but for temperature no shift was observed. The onset of the shifts observed in April to May in relative humidity and temperature coincided with the removal of suckers by the farm crew that occurred sometime between April 5 and April 11.We found that removal of leaves in the lower canopy of raspberry primocanes may affect cane Botrytis incidence or severity, but the direction of the effect may vary with context. Leaf removal may reduce severity of cane Botrytis in cultivar 1 and may increase severity in cultivar 2 in 3-row production systems. Additionally, leaf removal may reduce incidence in cultivar 1 in 2-row production systems. In our experiments we observed the opposite effect by twine removal between the two ranches for high density canopies. There are numerous aspects that differed among the experiments that might explain the conflict. The main differences are the plant density, the cultivar grown, and the different planting cycles between the two ranches. Variation in plant density among the experiments may have affected the response of cane Botrytis to leaf removal. Within ranch 1, leaf removal treatments provided somereductions in cane Botrytis incidence in the less compact 2-row tunnels but did not influence severity.

In contrast, in the more compact 3-row tunnels, leaf removal treatments influenced severity but not incidence. In a study evaluating use of a contact herbicide to control cane vigor in Scotland, cane Botrytis incidence was 65% in control plots compared to 10% incidence in herbicide-treated plots . Although we did not evaluate chemical removal treatments, our results in a lower density 2-row tunnel system showed lower incidence of cane Botrytis in leaf removal treatments opposed to no leaf removal. Therefore, it is possible that practices that modify the canopy would have a greater impact on the environment within a dense canopy compared with a less dense canopy. This suggestion is supported by our finding that the influence of leaf removal treatments on relative humidity was modest in the 2-row experiment relative to the 3-row experiments. Similar to the tunnel configurations, canopy density varied between ranch 1 and ranch 2 in experiments with the same tunnel configuration. The cultivar at ranch 1 produces a denser canopy than the cultivar at ranch 2 due to a higher number of both canes and suckers. Unlike the 2-row tunnels at ranch 1, however, at ranch 2 disease severity was higher in plots where leaves were removed. It is unlikely that canopy density alone accounted for the opposite response between the two ranches. In addition to agronomic practices that vary between cultivars, the two cultivars we used in our experiments may explain the conflicting direction of influence of leaf removal in two ways. First, the cultivars may differ in their resistance to colonization by B. cinerea and expression of cane Botrytis symptoms, but this information is not publicly available. Second, the cultivars may respond differently to wounding, which in turn may lead to a difference in susceptibility to B. cinerea infection. Thornless cultivars have previously been reported to be less vulnerable to mechanical damage that in turn results in lower incidence of Botrytis-caused disease . However, Knight and Keep and Anonymous evaluated the disease of fruits whereas only Jennings evaluated cane Botrytis, and it is unclear what kind of mechanical damage the plants were subjected to. In our study, we caused mechanical damage to canes and evaluated disease of the canes. As measured from severity in twine removal treatments, the thorny cultivar appeared less susceptible compared with the thornless cultivar that experienced increased severity. It is possible that susceptibility to mechanical damage is not correlated with thorn production in these cultivars. Genetic and physiological characteristics of red raspberry cultivars should be considered when applying leaf removal treatments using twine. The time of year in which the experiments were conducted, specifically when the treatments were applied, may also have contributed to the conflicting results between ranches.