Optimum Sample Size for Determining Disease ... - APS Journals

0 downloads 0 Views 121KB Size Report
Optimum Sample Size for Determining Disease Severity and Defoliation Associated with Septoria ..... present sampling scheme can be adopted to assess other ...
Optimum Sample Size for Determining Disease Severity and Defoliation Associated with Septoria Leaf Spot of Blueberry P. S. Ojiambo and H. Scherm, Department of Plant Pathology, University of Georgia, Athens 30602

ABSTRACT Ojiambo, P. S., and Scherm, H. 2006. Optimum sample size for determining disease severity and defoliation associated with Septoria leaf spot of blueberry. Plant Dis. 90:1209-1213. In a 3-year field study, Premier rabbiteye blueberry plants were sampled at three hierarchical levels (leaf, shoot, and bush) to assess severity of Septoria leaf spot (caused by Septoria albopunctata) and incidence of defoliation. A positive linear relationship (R2 = 0.977, P < 0.0001, n = 2127) was observed between the number of spots per leaf and percent necrotic leaf area, both assessed on individual leaves in mid- to late October. For data summarized at the shoot level, percent defoliation increased nonlinearly (R2 = 0.729, P < 0.0001, n = 224) as disease severity increased, with a rapid rise to an upper limit showing little change in defoliation above 60 spots per leaf. Variance components were calculated for disease severity to partition total variation into variation among leaves per shoot, shoots per bush, and bushes within the field. In all cases, leaves per shoot and shoots per bush accounted for >90% of the total variation. Based on the variance components and linear cost functions (which considered the time required to assess each leaf and select new shoots and bushes for assessment), the optimum sample size for assessing disease severity as number of spots per leaf (with an allowable variation of 20% around the mean) was 75 leaves, one each selected from three shoots per bush on 25 bushes (total time required for assessment: 36.1 min). For disease severity expressed as percent necrotic leaf area, the corresponding values were 144 leaves, two each sampled from three shoots per bush on 24 bushes (total time required: 21.7 min). Thus, given the strong correlation between the two disease variables demonstrated in this study, visual assessment of percent necrotic area was the more efficient method. With an allowable variation of 10% around the mean, a sample of 27 shoots from nine bushes was the optimum sample size for assessing defoliation across the 3 years. Additional keywords: disease assessment, sampling, Vaccinium virgatum

Among the foliar diseases that affect cultivated blueberry (Vaccinium section Cyanococcus), Septoria leaf spot, caused by Septoria albopunctata Cooke, is the most prevalent in Georgia and other southeastern states (7,20). Symptoms include small, circular lesions with white to tan centers and purple margins, each between 1 and 5 mm in diameter and harboring one to five pycnidia (15). In Georgia, these symptoms appear first by early May and then increase rapidly between June and September (18). When left uncontrolled, the disease can result in premature defoliation in late summer or early fall (4,5,17), and this can reduce flower bud set in late fall and yield the following spring (19). With increasing acreage and intensity of production, the disease is becoming an important production problem (21). Efficient disease assessment methods and sampling procedures are critical for

Corresponding author: H. Scherm E-mail: [email protected] Accepted for publication 11 May 2006.

DOI: 10.1094 / PD-90-1209 © 2006 The American Phytopathological Society

epidemiological studies, crop loss assessment, and evaluation of disease management practices (9,22). Although methods for disease assessment and related sampling schemes have been developed for numerous crops (8), no such procedures are available for foliar diseases of blueberry. In the Septoria leaf spot pathosystem, disease severity can be assessed either as number of spots per leaf (lesion density) or percent necrotic leaf area. Although counting of spots on leaves is more timeconsuming, it may provide a more objective measure of disease severity. However, there are no published studies investigating the relationship between these two measures of disease severity, the time needed for making the respective assessments, or how the choice between the two measures of disease severity affects the sample size and time needed for disease assessment. Furthermore, although it has been documented that Septoria leaf spot can lead to premature defoliation (17), no specific information is available on the sample size required to assess this variable. When disease severity is assessed at the leaf level, there is an almost infinite number of ways by which to select leaves for assessment among leaves on a given shoot, among shoots within a bush, and among

bushes within the field. Thus, it would be advantageous to determine the most efficient sampling plan for disease assessment through a sampling optimization procedure. Although optimum sample sizes for assessing foliar disease have been established for various pathosystems (1,2,10, 12,23), no such protocol is available for Septoria leaf spot or any other leaf disease that affects blueberry. Based on these considerations, the main objective of this study was to determine the optimum number of leaves per shoot, shoots per bush, and bushes within the field necessary to assess disease severity and defoliation with a set level of precision. Secondary objectives were to quantify the relationships (i) between Septoria leaf spot severity assessments obtained by counting the number of spots per leaf versus estimating percent necrotic leaf area and (ii) between percent premature defoliation and disease severity in the Septoria leaf spot pathosystem. MATERIALS AND METHODS Field site and data collection. The study was carried out in an experimental blueberry planting at the University of Georgia Horticulture Farm near Athens as part of a larger study on the epidemiology of Septoria leaf spot (17–19). This 0.15-ha planting was established in 1988 and consisted of alternating rows of rabbiteye blueberry (Vaccinium virgatum) cvs. Premier and Climax. Maintenance of the planting, including fertilization, pruning, and weed control, followed generally recommended practices (3). Supplemental overhead irrigation was applied as needed, primarily during the fruit maturation phase in the dry 2002 growing season. No fungicides were applied throughout the 3-year study period. On Premier, which is susceptible to Septoria leaf spot and resistant to other foliar diseases (20), 10, 8, and 10 bushes were selected arbitrarily from different rows within the field in September 2001, October 2002, and September 2003, respectively. On each bush, eight spring shoots (formed during the first flush of growth in the spring) were selected arbitrarily and tagged 20 cm from their tips. Spring shoots were chosen because they develop higher levels of disease and defoliation than shoots formed after harvest of the crop in late summer and fall (18). In mid- to late October, all leaves (typically 10 to 13) on these 20-cm shoot segments were assessed Plant Disease / September 2006

1209

individually for disease severity as both number of spots per leaf and percent necrotic leaf area; the latter variable was estimated following training of the assessor with DiseasePro (16), a computerized disease assessment training program. In this program, peanut leaf spot was used as a model for training because lesion size and number are similar to those of Septoria leaf spot on blueberry. The same assessor evaluated all leaves (844, 541, and 742 in 2001, 2002, and 2003, respectively) in all 3 years. Defoliation was assessed on the tagged shoots in mid-November by counting the number of nodes on which leaves had abscised and expressing it as a percentage of the total number of nodes per shoot. The relative cost associated with assessing disease severity and defoliation was estimated based on the time (T) required to complete these assessments in the field. All other expenditures of time, such as travel to the field and data entry, were not considered (10). For disease severity, the times (in seconds) required to select and move to a new a bush for assessment within the field (TB), to select a new shoot (TS), and to assess disease severity on a single leaf (TL) were recorded with a stopwatch. To determine TB, the time needed to select a bush starting from an arbitrary location within the field was recorded, and this was repeated for 20 bushes across the planting. TS and TL were similarly recorded for shoots and leaves, respectively, to give a total of 20 values for each measure of time. When defoliation was assessed, the times to select the next bush for assessment (TBD) and to select and assess a single shoot for defoliation (TSD) were

determined similarly for each of 20 bushes and shoots. Statistical analysis. The relationship between the number of spots per leaf and percent necrotic leaf area was examined using linear regression analysis for combined data from the 3 years using the PROC REG procedure in SAS (v. 8.2; SAS Institute, Inc., Cary, NC). To determine the relationship between defoliation and disease severity expressed as number of spots per leaf (averaged across all leaves on a given shoot), an exponential equation of the form y = a(1 − e–bx) was fitted to the data using SigmaPlot (v. 8.02; SPSS, Inc., Chicago, IL), where y = defoliation (%), x = number of spots per leaf, and a and b are regression coefficients. To determine the optimum number of leaves, shoots, and bushes for assessment of disease severity, data were analyzed in a three-stage sampling design (leaves within shoots, shoots within bushes, and bushes within the field) using the SAS procedure VARCOMP (13). Based on the analysis, the estimates of variances and variance components associated with sampling leaves, shoots, and bushes were used to derive the respective optimum sample sizes using the equations given by Campbell and Madden (6), adopted from Analytis and Kranz (1): nL(opt) = [(MSL × nL)/(MSS – MSL)]1/2 × (TS/TL)1/2 nS(opt) = {[(MSS – MSL) × nS]/[MSB – MSS]}1/2 × (TB/TS)1/2 nB(opt) = {σ2B + σ2S/[nS(opt)] + σ2L/[nS(opt) nL(opt)]} × [1/V(x)]

In these equations, nL(opt) = optimum number of leaves per shoot; MSL = mean

Fig. 1. Relationship between number of spots per leaf (y) and percent necrotic leaf area (x) on Premier rabbiteye blueberry affected by Septoria leaf spot at the University of Georgia Horticulture Farm. Each year, the two variables were assessed in mid- to late October. The regression equation is y = 2.33 + 6.07x (R2 = 0.977, P < 0.0001, n = 2127). 1210

Plant Disease / Vol. 90 No. 9

square associated with variation among leaves on the same shoot; nL = actual number of leaves per shoot sampled; MSS = mean square associated with variation among shoots on the same bush; nS(opt) = optimum number of shoots per bush; nS = actual number of shoots sampled per bush; MSB = mean square associated with variation among bushes within the field; nB(opt) = optimum number of bushes; σ2L, σ2S, and σ2B = variances associated with leaves, shoots, and bushes, respectively; V(x) = allowable variance around the mean; and TL, TS, and TB were as defined above. The variance around the mean was calculated using the equation V(x) = (0.2 × mean disease severity)2, where the factor 0.2 allows for a 20% variability around the mean. Optimum numbers of leaves, shoots, and bushes were rounded to the next highest integer in all calculations and for presentation in the text. When assessing defoliation, only the numbers of shoots per bush and bushes in the field needed to be optimized because there was only one value of percent defoliation per shoot; thus, data were analyzed in a two-stage sampling design (shoots within bushes and bushes within the field). Based on the estimates of variances and variance components associated with shoots and bushes obtained from the analysis, optimum sample sizes for this variable were determined using the following equations (1,6): nSD(opt) = (σSD/σBD) × (TBD/TSD)1/2, and nBD(opt) = {σ2BD + σ2SD/[nSD (opt)]} × [1/V(x)], where nSD(opt) and nBD(opt) = optimum number of shoots per bush and optimum number of bushes per field, respectively; σ2SD and σ2BD = variances associated with shoots and bushes, respectively; and TSD, TBD, and V(x) were as defined above. V(x) was calculated as described for disease severity except that the allowable variation around the mean was set at 10%. This lower value was adopted after preliminary analyses indicated no substantial increase in the number of bushes to be sampled at this level of precision compared with a 20% variation around the mean. Values of nSD(opt) and nBD(opt) were rounded to the next highest integer in all calculations and for presentation in the text. RESULTS AND DISCUSSION There was a positive, linear relationship (R2 = 0.977, P < 0.0001, n = 2127) between the number of spots per leaf and percent necrotic leaf area (Fig. 1), indicating that the latter variable was a reliable predictor of spot number on individual leaves. Based on the combined data from the 3 years, the equation to estimate number of spots per leaf (y) using percent necrotic leaf area (x) was y = 2.33 + 6.07x. The slope of the regression indicates that, for this particular cultivar, each 1% increment in necrotic leaf area corresponded to approximately six leaf spots. Given the

epidemiological importance of the number of spots per leaf (with each spot containing a discrete number of pycnidia producing secondary inoculum), Septoria leaf spot severity routinely has been evaluated using this disease variable (4,5,17–19). Assessment of disease severity as number of spots per leaf also has been used for Septoria leaf spot of tomato (11). The present study shows that, with adequate training, a visual estimate of percent necrotic leaf area can substitute for the time-consuming task of counting the number of spots per leaf. Based on our measurement of the times required for assessing disease severity, it took about four times longer to count spots on individual leaves than to make visual estimates of percent necrotic area on the same leaves. Mean values of TL, TS,

and TB to assess the number of spots per leaf were 18.3, 9.1, and 4.4 s, respectively, whereas the corresponding mean values for assessing percent necrotic leaf area were 4.2, 8.3, and 4.1 s, respectively. Mean values for TSD and TBD when defoliation was assessed were 4.1 and 4.4 s, respectively. At the shoot level, the relationship between defoliation and the number of spots per leaf was characterized by a rise to an upper limit with little change in defoliation above 60 spots per leaf (Fig. 2). Based on the combined data from the 3 years, the equation to estimate defoliation (y) using number of spots per leaf (x) was y = 91.2 (1 – e–0.047x) (R2 = 0.729, P < 0.0001, n = 224). Previously, we documented a significant relationship between defoliation and

Fig. 2. Relationship between percent defoliation (y) and average number spots per leaf (x) on Premier rabbiteye blueberry affected by Septoria leaf spot at the University of Georgia Horticulture Farm. Each year, disease severity and defoliation were assessed on a per-shoot basis in mid- to late October and in mid-November, respectively. The regression equation is y = 91.2 (1 – e–0.047x) (R2 = 0.729, P < 0.0001, n = 224).

disease severity over time using survival analysis (17). In the present study, even with a single assessment of disease severity at the end of the season, there was a strong association between the two variables. This provides an important link to yield loss assessment because return yield in blueberry is affected, in part, by the timing and magnitude of defoliation (14,24). Disease severity levels were very similar across the 3 years, with the mean value ranging from 30.9 to 35.7 spots per leaf and 4.6 to 5.6% necrotic leaf area each year (Table 1). The greatest sources of variation were variability among leaves on shoots and among shoots within bushes, regardless of whether disease was assessed as number of spots per leaf or percent necrotic leaf area. The variance component associated with leaves accounted for 34.1 to 53.6% of the total variation, whereas variance among shoots accounted for 38.5 to 62.2% (Table 1). The variability associated with bushes was consistently the lowest, ranging from 3.6 to 7.9% of the total. The relative variance component values (expressed as percentages of the total) were similar, on average, for the two measures of disease severity. When defoliation was assessed, the variance component for shoots accounted for >80% of the total variability. Mean defoliation at the respective assessment dates was very similar across the 3 years (Table 1). The optimum number of shoots per bush and bushes per field needed to estimate disease severity as number of spots per leaf or necrotic leaf area were similar (Table 2). However, the total number of leaves per shoot that needed to be assessed to estimate disease severity within 20% of the mean was considerably lower for number of spots per leaf (75 leaves total from three shoots per bush on 25 bushes, when averaged across the 3 years and rounded to the next highest integer) than for necrotic leaf area (144 leaves total from three shoots per bush on 24 bushes). Across the 3 years, a

Table 1. Estimated variance components for assessing disease severity and defoliation associated with Septoria leaf spot, caused by Septoria albopunctata, on Premier rabbiteye blueberry at the University of Georgia Horticulture Farm Sampling level Variance component Variablea Spots per leaf 2001 2002 2003 Necrotic leaf area (%) 2001 2002 2003 Defoliation (%) 2001 2002 2003 a b

Mean value

Percentage of total

Range

Leaf

Shoot

Bush

Leaf

Shoot

Bush

30.9 35.7 31.7

0–289 0–297 0–213

852.8 908.1 568.4

951.3 1,656.6 444.4

110.1 97.3 83.2

44.6 34.1 51.9

49.7 62.2 40.6

5.7 3.7 7.5

4.6 5.6 4.8

0–43.0 0–45.1 0–34.1

24.6 25.7 16.5

24.1 45.9 11.8

2.8 2.7 2.5

47.8 34.5 53.6

46.7 61.8 38.5

5.5 3.6 7.9

60.4 63.8 64.8

13.3–100 0.0–100 7.1–100

…b … …

469.4 569.2 650.9

102.4 77.5 96.1

… … …

82.9 88.1 87.1

17.9 11.9 12.9

Each year, disease severity and defoliation were assessed in mid- to late October and in mid-November, respectively. Defoliation was assessed on a per-shoot basis; thus, there was no variation at the leaf level. Plant Disease / September 2006

1211

total sample of 27 shoots from nine bushes was the optimum sample size for assessing defoliation. The difference in the number of leaves needed for assessing lesion density versus percent necrotic leaf area can be explained based on the time associated with assessing the two variables. The times needed to select and move to new shoots and bushes were very similar for the two variables; therefore, differences in calculated sample sizes were due to the different amounts of time needed to make the leaf assessments (which took about four times as long for counting spots than for estimating percent necrotic leaf area). As a result, the total number of leaves that needed to be assessed was lower, whereas the number of bushes was slightly higher for counting spots. The total time needed to count spots on 75 leaves (including the time needed to select and move to new shoots and bushes) was 36.1 min compared with 21.7 min for assessing 144 leaves for percent necrotic leaf area. Thus, given the lower expenditure of time, visual assessment of percent necrotic area was the more efficient method. Based on these data, we recommend 144 leaves, two each from three shoots per bush on 24 bushes, as the optimum sample size for assessing Septoria leaf spot severity. Further research should be conducted to determine whether the present sampling scheme can be adopted to assess other foliar diseases of blueberry. In addition, methods of reducing the error variance due to variation from leaf to leaf and shoot to shoot need to be investigated. In developing a three-stage sampling plan for apple scab caused by Venturia inaequalis, Analytis and Kranz (1) determined that a total of 360 leaves (sampled

as five leaves per terminal from six terminals per tree on 12 trees) was optimal for estimating percent necrotic leaf area with a 10% variability around the mean; the mean disease severity value in their study was 3.3%, lower than reported here for Septoria leaf spot. When we used the same level of precision (10% allowable variation around the mean) in our calculations with Septoria leaf spot, the total number of leaves that needed to be assessed was 300 for lesion density and 544 for percent necrotic leaf area (data not shown). Although these values are of the same magnitude as those reported by Analytis and Kranz (1), direct comparisons between the two studies are difficult due to differences in cost functions, overall disease levels, and the relative allocation of variation in diseases severity among sampling units. All assessments in the present study were carried out in a planting untreated with fungicide; consequently, disease severity values on individual leaves varied widely (ranging from 0 to 297 spots per leaf). In experimental plots subjected to specific treatments such as fungicide applications, the range of disease severity values may be smaller than observed here, leading to lower variation among leaves and shoots and correspondingly higher sample sizes for leaves and shoots on a smaller number of bushes. This also would apply to disease assessments at the beginning of the season when disease severity and variability among leaves are low. Thus, the optimum sampling sizes reported here represent a situation in which disease is assessed in untreated plots. When large commercial plantings with bushes located farther apart from one another are sampled, the number of bushes sampled would

Table 2. Optimum sample sizes for assessing disease severity and defoliation due to Septoria leaf spot, caused by Septoria albopunctata, on Premier rabbiteye blueberry at the University of Georgia Horticulture Farma Optimum number Variable Spots per leaf 2001 2002 2003 Mean Necrotic leaf area (%) 2001 2002 2003 Mean Defoliation (%) 2001 2002 2003 Mean

Leaves per shoot

Shoots per bush

Bushes

Total no. of leavesb

0.78 0.56 0.86 0.73

2.02 2.89 1.79 2.23

29.34 24.33 18.69 24.12

… … … 75

1.63 1.13 1.84 1.53

2.01 2.78 1.54 2.11

29.87 22.02 17.42 23.10

… … … 144

…c … … …

2.22 2.71 2.74 2.56

8.68 7.23 8.18 8.03

… … … 27

a

Optimum sample sizes were derived using mean squares (from analysis of a two- or three-level nested design structure) and the corresponding times associated with assessing disease severity and defoliation (see text). The allowable variation around the mean used in these calculations was 20 and 10% for disease severity and defoliation, respectively. b Calculated as the product of number of leaves per shoot, number of shoots per bush, and number of bushes per field, with each of the three numbers rounded to the next higher integer. c Defoliation was assessed on a per shoot basis; thus, there was no sampling at the leaf level. 1212

Plant Disease / Vol. 90 No. 9

be reduced relative to those of shoots and leaves due to an increase in TB relative to TS and TL. Conversely, estimation of percent necrotic leaf area is likely to take longer for a novice assessor compared with the trained assessor employed in the present study, which would lead to a reduction in numbers of leaves assessed per shoot relative to those of shoots and bushes. ACKNOWLEDGMENTS Funded in part by the United States Department of Agriculture–CSREES Pest Management Alternatives Program (grant no. 01-34381-11181). We thank A. Savelle for assistance in data collection; K. Stevenson, P. Brannen, A. Culbreath, and S. NeSmith for their useful comments; and W. Turechek for a presubmission review of a draft of the manuscript. LITERATURE CITED 1. Analytis, S., and Kranz, J. 1972. Bestimmung des optimalen Stichprobenumfanges für phytopathologische Untersuchungen. Phytopathol. Z. 74:349-357. 2. Aubertot, J. N., Schott, J. J., Penaud, A., Brun, H., and Dore, T. 2004. Methods for sampling and assessment in relation to the spatial pattern of Phoma stem canker (Leptosphaeria maculans) in oilseed rape. Eur. J. Plant Pathol. 110:183-192. 3. Austin, M. E. 1994. Rabbiteye Blueberries: Development, Production, and Marketing. Agscience, Auburndale, FL. 4. Brannen, P. M., Scherm, H., and Bruorton, M. D. 2002. Fungicidal control of Septoria leaf spot of blueberry, 2001. Fungic. Nematicide Tests 57:SMF46. 5. Brannen, P. M., Scherm, H., and Bruorton, M. D. 2003. Fungicidal control of Septoria leaf spot of blueberry, 2002. Fungic. Nematicide Tests 58:SMF019. 6. Campbell, C. L., and Madden, L. V. 1990. Introduction to Plant Disease Epidemiology. Wiley, New York. 7. Cline, W. O. 2002. Blueberry bud set and yield following the use of fungicides for leaf spot control in North Carolina. Acta Hortic. 574:7174. 8. Cooke, B. M. 1998. Disease assessment and yield loss. Pages 42-72 in: The Epidemiology of Plant Disease. D. G. Jones, ed. Kluwer, Dordrecht, The Netherlands. 9. Danielsen, S., and Munk, L. 2004. Evaluation of disease assessment methods in quinoa for their ability to predict yield loss caused by downy mildew. Crop Prot. 23:219-228. 10. Duthie, J. A., Campbell, C. L., and Nelson, L. A. 1991. Efficiency of multistage sampling for estimation of intensity of leaf spot diseases of alfalfa in field experiments. Phytopathology 81:959-964. 11. Ferrandino, F. J., and Elmer, W. H. 1996. Septoria leaf spot lesion density on trap plants exposed at varying distances from infected tomatoes. Plant Dis. 80:1059-1062. 12. Filajdi , N., and Sutton, T. B.1994. Optimum sampling size for determining different aspects of Alternaria blotch of apples caused by Alternaria mali. Plant Dis. 78:719-724. 13. Littell, R. C., Milliken, G. A., Stroup, W. W., and Wolfinger, R. D. 1996. SAS System for Mixed Models. SAS Institute, Cary, NC. 14. Lyrene, P. M. 1992. Early defoliation reduces flower bud counts on rabbiteye blueberry. HortScience 27:783-785. 15. Milholland, R. D. 1995. Septoria leaf spot and stem canker. Page 16 in: Compendium of Blueberry and Cranberry Diseases. F. L. Caruso and D. C. Ramsdell, eds. American Phytopathological Society, St. Paul, MN. 16. Nutter, F. W., Jr. 1997. Disease severity as-

sessment training. Pages 1-7 in: Exercises in Plant Disease Epidemiology. L. J. Francl and D. A. Neher, eds. American Phytopathological Society, St. Paul, MN. 17. Ojiambo, P. S., and Scherm, H. 2005. Survival analysis of time to abscission of blueberry leaves affected by Septoria leaf spot. Phytopathology 95:108-113. 18. Ojiambo, P. S., and Scherm, H. 2005. Temporal progress of Septoria leaf spot on rabbiteye blueberry (Vaccinium ashei). Plant Dis. 89:1090-1096.

19. Ojiambo, P. S., Scherm, H., and Brannen, P. M. 2006. Septoria leaf spot reduces flower bud set and yield potential of rabbiteye and southern highbush blueberries. Plant Dis. 90:51-57. 20. Scherm, H., Brannen, P. M., Ojiambo, P. S., Savelle, A. T., Krewer, G., and Bruorton, M. D. 2003. Blueberry leaf spots: epidemiology, yield losses, and control. Pages 57-66 in: Proc. Southeast. Blueberry Conf. Savannah, GA. 21. Scherm, H., and Krewer, G. 2003. Blueberry production in Georgia: Historical overview and

recent trends. Small Fruits Rev. 2:83-91. 22. Seem, R. C. 1984. Disease incidence and severity relationships. Annu. Rev. Phytopathol. 22:137-150. 23. Thal, W. M., and Campbell, C. L. 1987. Sampling procedures for determining severity of alfalfa leaf spot diseases. Phytopathology 77:157-162. 24. Williamson, J. G., and Miller, E. P. 2002. Early and mid-fall defoliation reduces flower bud number and yield of southern highbush blueberry. HortTechnology 12:214-216.

Plant Disease / September 2006

1213