Risk Analysis for Latent Infection of Prune by Monilinia fructicola in California
Department of Plant Pathology, University of California Davis, Kearney Agricultural Center, Parlier, CA 93648
--------------------------------------------------------------------------------------------------------
Luo, Y., and Michailides, T. J. 2001. Risk analysis for latent infection of prune by Monilinia fructicola in California. Phytopathology 91:xxx-xxx.
The quantitative relationships between incidence of latent infection (ILI) of prune by Monilinia fructicola and wetness duration (WD) for different bloom and fruit developmental stages and different inoculum concentrations were obtained. Three levels of ILI were considered as criteria for low, moderate, and high risks of latent infection, respectively. Seasonal patterns of WD leading to different risk levels of latent infection were obtained for low (IPL) and high (IPH) inoculum potential conditions in orchards. Longer WD was needed at a resistant than at a susceptible fruit developmental stage to induce similar levels of latent infection. The curves of WD leading to different levels of ILI over the growing season (risky WD curves) were used in risk analysis for latent infection. Multi-year historical WD data from 10 prune-growing locations in California were compared with those from the risky WD curves. The percentage of days (P) with WD leading to a certain risk level of latent infection was calculated for each month from historical weather data. Under the IPL condition, the P distributions for low risk of latent infection were higher in March and April than in May and were the lowest in June for most locations. Under the IPH condition, the number of days that WD leading to low risk of latent infection in May increased compared with those under the IPL condition. The risk analysis approach was evaluated by using separate experimental data as incidence of fruit brown rot obtained from different prune orchards in different years. Consistency between predicted overall risk levels of latent infection and observed incidence of fruit brown rot was obtained. The results demonstrated the usefulness of the risk analysis approach in decision support system for disease management.
Additional keywords: Prunus domestica, stone fruit
--------------------------------------------------------------------------------------------------------
In California, state-wide epidemics of brown rot of stone fruit caused by Monilinia fructicola (G. Wint.) Honey occur infrequently. However, orchard- and region-scale epidemics causing different yield losses frequently occur depending on many factors in orchards such as inoculum potential, weather condition, and cultural practice in orchards (16-19). Two phases of this disease should be considered in decision making for disease management, blossom blight and fruit brown rot. Blossom blight is caused by ascospores or conidia produced from mummified fruit that were infected by M. fructicola, and it usually happens in early spring (4,9,10,24). Severe blossom blight may cause flower rot and reduce the number of fruit. The susceptibility of prune blossoms to infection by M. fructicola over different developmental stages, the relationships between disease incidence and environments, as well as the possible risk of blossom blight associated with different environments were intensively studied (14).
Under favorable conditions, infections could continue on prune fruit after bloom. Infection of fruit is usually caused by conidia produced on infected blossoms, mummies or thinned fruit and could cause serious yield losses (6,9,25). Secondary infections may occur during the growing season when favorable weather conditions are encountered. Many of these infections can be latent for a period until become active brown rot symptoms on fruit with sporulation. Studies on inoculum sources showed that non-abscised, aborted fruit in trees and thinned fruit on orchard floor infected by M. fructicola could serve as secondary inoculum causing fruit brown rot ( 1,10,18). The significance of thinned fruit as a source of secondary inoculum in nectarine orchards in California was also confirmed (8). Inoculum potential in a specific prune orchard is dependent on disease history, orchard environments, cultural practices, and disease control strategies. In prune orchards in California, disease symptoms and sporulation usually do not appear until late in the season when higher humidity and near optimum temperatures are common.
Our previous studies (13) in three prune orchards had determined the seasonal patterns of bloom and fruit susceptibility to latent infection by M. fructicola. The susceptibility to latent infection at bloom stage was at a moderate level, increased to reach to the highest level at about pit hardening stage, subsequently, decreased, reaching the lowest level in early June at embryo growth, and increased again along with fruit development and maturity until harvest. The incidence of latent infection (ILI) increased exponentially with increased wetness duration (WD) and linearly with increased inoculum concentration (IC) for most bloom and fruit developmental stages (13). These findings confirmed similar studies which used detached peach fruit (2,7).
These studies (13) also implied that the same or similar wetness duration and inoculum potential may lead to different latent infection levels at different fruit developmental stages. Therefore, environmental criteria used for disease control might be different throughout the growing season. For example, a certain wetness duration, such as 12 h, might bring about a high risk of latent infection early in the season but low risk mid in the season. Understanding the seasonal patterns of environmental conditions leading to different risk levels of latent infection is therefore important in decision making for disease management.
In California, the commercial recommendation of fungicide spray to control blossom blight is at the green tip stage. Orchards with different levels of inoculum potential may have different possibilities of latent infection, especially in those orchards where the disease infrequently occurred. However, the importance in reducing latent infection after bloom is still underestimated. Our pervious study (13) found that the stage of the highest susceptibility to latent infection was from fruit set to embryo growth, implying that the risk of latent infection causing severe disease is not only at the bloom stage. The microclimatic conditions in orchards after bloom are critical in disease appearance in the season, and an estimation of risk of disease development throughout the growing season becomes critical in disease management.
Wetness duration, IC, and temperature all affected latent infection of prune by M. fructicola (12,13). Similar findings were also reported on other stone fruits (3,5,18,20). How the risk of latent infection changes at different fruit developmental stages and what environments could affect this change are the important questions that need to be addressed for the development of a decision support system for disease control. This study emphasized WD and IC as critical factors for development of risk analysis approach. The objectives of this study were to i) determine the relationships between ILI and WD for different bloom and fruit developmental stages, ii) determine the seasonal patterns of WD leading to different risk levels of latent infection, iii) analyze historical weather patterns to estimate the occurrence of the weather leading to risks of latent infection for different areas in California, and iv) evaluate the risk analysis approach by comparing predicted risk levels of latent infection with observed incidences of fruit brown rot.
Seasonal pattern of bloom and fruit susceptibility to latent infection. In a previous study, experiments were conducted in three prune orchards located in Yolo, Butte, and Tehama counties in California. In each orchard, trees were inoculated at full bloom and four fruit developmental stages: pit hardening, embryo growth, late embryo growth, and before first harvest (23). Five inoculum concentratioins (IC), 1,000, 5,000, 10,000, 20,000, and 30,000 conidia/ml of M. fructicola, were used in inoculation on the branches each bearing 40-60 flowers or fruits by spraying. The inoculated branches were covered with plastic bags to keep high humidity. Six wetness durations (4, 8, 12, 16, 20, and 24 h) for bloom and 4 wetness durations (4, 8, 12, and 16 h) for fruit developmental stages were generated for each inoculation by uncovering plastic bags at different time. Fruit on all inoculated branches were maintained on trees until harvest, and were processed by an overnight freezing incubation technique (16,17) to determine the incidence of fruit with latent infection (ILI). A seasonal pattern of bloom and fruit susceptibility to latent infection by M. fructicola was obtained. A moderate level of ILI occurred at bloom. The fruit susceptibility to infection increased after bloom stage, and the maximum ILI was at about pit hardening stage. Subsequently, susceptibility of fruit to latent infection decreased, reaching the lowest level in early June at the embryo growth stage, and increased again along with fruit development and maturity until harvest. The ILI increased exponentially with increased WD in most treatments at ambient orchard temperatures.
For each IPL and IPH condition, four risk levels of latent infection were assigned as follows: no risk when ILI < 1%, low risk when 1% ≤ ILI < 5%, moderate risk when 5% ≤ ILI < 10%, and high risk when ILI ≥ 10%.
The WDs leading to the corresponding risk levels were calculated by converting the above exponential regressions as: WD = Ln ((ILI – a) / b ) / c, where a, b, and c are all parameters of the regression. Thus, the WD1, WD5 and WD10 were calculated by using the above formula as ILI = 1%, 5%, and 10%, respectively. These three values were referred to the critical WD leading to low, moderate, and high risk of latent infection, respectively, and were determined for each bloom and fruit developmental stage under IPL and IPH conditions calculated by the corresponding regressions.
Approaches of risk analysis for latent infection. Two steps were involved in the risk analysis approaches (Fig. 1). Step 1 used the regressions obtained from the above analysis to determine the seasonal patterns of WD leading to different risk levels of latent infection. Step 2 analyzed the historical weather data for prune-growing locations in California to obtain the frequency distributions of WD leading to different risk levels of latent infection for each location.
The WD1, WD5 and WD10 values for each developmental stage were separately used to draw the curves over the growing season from 15 March to 20 July for IPL and IPH conditions by using the computer software Microsoft Excel (Microsoft Cor., Bothell, WA). Therefore, three curves representing the WDs leading to 1%, 5%, and 10% of ILI, respectively, over the growing season were obtained. The corresponding WD values on the curves are referred to as the risky WD, and the curves are referred to as the risky WD curves below. In addition to this study, these curves can also be used as a reference to estimate risk levels of latent infection over the growing season (Fig. 1).
In order to estimate possible risk of latent infections in California, historical weather data from 1983 to 2000, including hourly air temperature, dew point temperature (ºC), and precipitation (mm) for 10 prune-growing locations, were collected from the California Irrigation Management Information System (CIMIS). These locations were from central San Joaquin Valley to northern Sacramento Valley (Fig. 2). For each hour, wetness was determined by comparing the air temperature and dew point temperature, and by checking the precipitation using a switch variable D as follows:
D = 1, when air temperature < dew point temperature, or precipitation > 0,
D = 0, when air temperature > dew point temperature and precipitation = 0.
Wetness duration (h) was calculated by summarizing D for each day. This calculation was also confirmed by comparing the above calculated wetness duration with the leaf wetness measurement using a data logger (Onset Computer Corporation, Bourne, MA) with a leaf wetness sensor for 3 weeks. The calculated daily wetness durations were equaled to those of the measurement during the 3 weeks.
From the risky WD curves described above, the WD1, WD5, and WD10 values for each day from 15 March through 20 July were determined on a scaled sheet (Fig. 1), and were used to develop a table. This table contained seven columns: date, the corresponding daily WD1, WD5, and WD10 values for the IPL condition and the corresponding WD1, WD5, and WD10 values for the IPH condition (Fig. 1). This table was used to compare with historical wetness duration (WDhis) for each day. The variables RKL, RKM and RKH were used to count the number of historical days associated with low, moderate, and high risk of latent infection, respectively, for both IPL and IPH conditions:
RKL =1 when WD1 ≤ WDhis < WD5, otherwise, RKL = 0;
RKM = 1 when WD5 ≤ WDhis < WD10, otherwise, RKM = 0; and
RKH =1 when WDhis > WD10, otherwise, RKH = 0.
Where RKL, RKM, and RKH are assigned as 0 or 1 to determine if the WDhis leading to low, moderate, and high risk of latent infection, respectively, exists.
The RKL, RKM, and RKH were separately summarized for each month of each year to obtain the number of days with WD leading to low (NDL), moderate (NDM), and high (NDH) risks of latent infection, respectively. The percent days of the month with low risk (PL), moderate risk (PM), and high risk (PH) of latent infection, respectively, were also calculated by dividing NDL, NDM, and NDH by total number of days of the month, respectively. These ND values are referred to as the threshold ND values below. The yearly PL, PM, and PH data were used to obtain their distributions for each month and each location (Fig. 1). The monthly mean PL, PM, and PH and the PL, PM, and PH values with 90% cumulative probability, CPL90, CPM90, and CPH90, respectively, were also calculated for each month. The monthly means of threshold NDL, NDM, and NDH for each location under each IP condition were used as references of historical conditions associated with different risks of latent infection (Fig. 1).
All calculations involved in the risk analysis approaches were performed by programming with SAS (version 8.0, SAS Institute, Cary, NC).
Evaluation of risk analysis approaches. The previously observed fruit brown rot data were used to evaluate the predictability of the above risk analysis approaches. Two evaluations were conducted by using two sets of experimental data. Evaluation 1 used the experimental data obtained from prune orchards in four counties, Tulare, Tehama, Butte, and Sutter in California in 1993 (20). The objective of the experiments was to determine the effects of early and mid summer sprays of fungicides on control of prune brown rot at harvest and postharvest. The average incidence of fruit brown rot from non-inoculated (naturally infected) trees, used as non-treatment control, was used in the evaluation process. The inoculum potential in the orchard in Tulare was estimated as low (IPL), since severe brown rot infrequently occurred in the orchard in comparison with those located in northern California. The inoculum potentials in the remaining three orchards were high (IPH) . The corresponding daily WDs leading to low, moderate, and high risks of latent infection were calculated from each location’s historical weather data using the approach described above. These WD data were used to calculate the NDL, NDM, and NDH, respectively, for each month from 15 March to 20 July. These ND data were then used to compare the monthly mean threshold NDL, NDM, and NDH values for the corresponding locations obtained from the approaches described above. For each month, when ND was less than the corresponding threshold ND value, no corresponding level of risk of latent infection was estimated. Otherwise, such risk level of latent infection was determined. The overall estimations of risk level were then used to compare the observed incidence of fruit brown rot for each experiment.
Evaluation 2 used the experimental data obtained from orchards in Glenn and Butte Counties in California. Similar experiments were conducted from 1994 though 1996 in the same orchards. The objective of the experiments was to study the ecology and epidemiology of prune brown rot and new control strategies (17,19,21). The data used in the evaluation were the means of incidence of fruit brown rot with non-fungicide-treatment, non-inoculation (naturally infected) control (17,19,21). The same method in comparisons between predicted risk and observed disease incidences as the evaluation 1 was used.
Table 1 lists the results of exponential regressions between ILI and WD for bloom and fruit developmental stages combined with three inoculum concentrations. No correlation was determined when IC = 1,000 conidia/ml at the bloom, embryo growth, and before first harvest stages (Table 1). The regressions for other inoculum concentrations and developmental stages were all significant at P < 0.05 (Table 1).
The curves of WD leading to different risk levels of latent infection by M. fructicola over the growing season were produced for the IPL and IPH conditions (Fig. 3). Basically, the seasonal patterns of WD leading to different risk levels of latent infection were related to the seasonal patterns of bloom and fruit susceptibility to latent infection (13). Longer WD is required at the resistant stage than at the susceptible stage to induce similar levels of latent infection (Fig. 3). Along with increased susceptibility to latent infection from the bloom to the pit hardening stage (13), the WD leading to different risk levels of latent infection decreased (Fig. 3). After the pit hardening stage, fruit susceptibility to latent infection dramatically decreased until embryo growth (13), and the WD leading to risk of latent infection increased (Fig. 3A and B). After mid June, the fruit susceptibility to latent infection increased again (13), and WD leading to risk of latent infection decreased (Fig. 3). However, the WDs decrease slower under the IPL (Fig. 3A) than under the IPH (Fig. 3B) condition.
Under the IPL condition, distributions of the percent days with WD leading to low risk of latent infection (PL) in March and April were similar among the 10 locations (Fig. 4). The PL decreased in May at all locations. Only a few years with very low PL occurred in June, but the distribution increased in July. The means of NDL were from 6 to 9.6 days in March and from 10.6 to 14 days in April (Table 2). In May, means of NDL ranged from 3.3 to 8 days (Table 2), and the 90% cumulative probability of PL (CPL90) was from 22 to 37% (6.8 - 11.5 days in May) (data not shown). However, the means of NDL in June were from 0.2 to 1.2 days (Table 2) and CPL90s ranged from 3.6 to 9.3% (1 to 2.8 days in June, data not shown).
Compared with the number of days associated with low risk of latent infection, fewer days with WD leading to moderate risk of latent infection occurred (Table 2). However, the distributions of PM were different among locations (Fig. 5). In general, there were more years with risky WD in northern than in southern California, and some locations, such as Gerber, Orland, and Colusa in northern Sacramento Valley, showed similar PM in May to those in March and April (Fig. 5). Two locations in southern San Joaquin Valley, Parlier and Visalia, showed fewer years with PM compared with the other locations (Fig. 5). Therefore, there were more chances of moderate risk of latent infection in northern than in southern California. These risky WDs mostly occurred in April and the mean NDMs were comparatively less than the mean NDLs (Table 2). The means of NDM ranged from 0.6 to 5 days at the most susceptible stage (April to May), and the maximum CPM90 was about 30% (about 9 days, data not shown). The probability of moderate risk of latent infection late in the season was very low.
Under the IPL consition, a few days with WD leading to high risk of latent infection (NDH) were observed in few years (Table 2). The mean NDHs ranged only from 0 to 1.8 days over the season (Table 2), and the most CPH90 values were below 5% (data not shown). This fact indicates very low probability of high risk of latent infection under the IPL condition.
Comparatively under the IPH condition, there were more chances that WD leading to low risk of latent infection occurred in May than in April (data not shown). For the locations in Sacramento Valley, even in June, there were still quite a few years that showed WD leading to low risk of latent infection (data not shown). The mean NDLs ranged from 3.2 to 9.4 days in May and from 0.5 to 5.6 days in June (Table 2). The maximum CPL90 was about 47% in May and 29% in June. Therefore, the chance of low risk of latent infection was higher under the IPH than under the IPL condition, especially at the resistant stage (Table 2). The means of NDM ranged from 0 to 0.9 days (Table 2) and CPM90s were from 0 to 7% (data not shown). This fact indicates a low probability of moderate risk level. The means of NDH were greater in March and April than in June and July (Table 2). In May, the means of NDH ranged from 0.7 to 2.2 days (Table 2) and the CPH90s from 4 to 17% (data not shown). Therefore, the probability of high risk of latent infection under IPH condition was relatively higher at the susceptible stages than at the resistant stages.
In the evaluation 1, predicted overall risk levels of latent infection were consistent with observed incidences of brown rot of fruit. Two orchards showed the disease incidence within the moderate risk level of latent infection, 8.8% at Tulare (Fig. 6A) and 7.2% at Tehama (Fig. 6B), and the other two orchards showed the disease incidence within the high risk level of latent infection, 17% at Butte (Fig. 6C) and 62% at Sutter (Fig. 6D). For the Tulare orchard in March (Fig. 6A), the NDL was 3 days, less than the mean threshold NDL (7.6 days) (Table 2, for Visalia), but NDM was 3 days, greater than the mean threshold NDM (0.3 days). There was also a day with WD leading to a high risk of latent infection occurred (Fig. 6A). Therefore, a moderate risk of latent infection was predicted for this month. In April, the NDM was 5 days, greater than the mean threshold NDM 2.3 days (Table 2, for Visalia), and either NDL or NDH was not greater than the corresponding threshold NDL or NDH values, respectively. Thus, a moderate risk of latent infection was predicted in April (Fig. 6A). A similar situation was observed in the Gerber orchard (Tehama Co.), and the threshold ND values in Table 2 for Gerber were used in the comparisons. A moderate risk of latent infection was predicted in March and April, and a low risk of latent infection was predicted in May (Fig. 6B). An overall moderate risk of latent infection was predicted (Fig. 6B), and the observed average fruit brown rot (7.2%) was within the range of moderate risk level of latent infection.
Different from the above situations, for the orchard in Butte county, both NDM (4 days) and NDH (4 days) in March (Fig. 6C) were greater than the respective mean threshold NDM (3.8 days) and NDH (2.9 days) (Table 2, for Durham). In April, both NDL (3 days) and NDM (10 days) were greater than the respective mean threshold NDL (1.6 days) and NDM (8.3 days). Therefore, a moderate to high risk of latent infection was predicted in March, and a low to moderate risk of latent infection was predicted in April (Fig. 6C). A low risk of latent infection was also predicted in June, and a moderate risk of latent infection was predicted in July. By summarizing the situations over the growing season, a high risk of latent infection was predicted, and the observed incidence of fruit brown rot was 17% (Fig. 6C). The information for Colusa in Table 2 was used in the evaluation process for the Sutter orchard, since this orchard was close to the Colusa weather station (about 40 km). A high risk of latent infection was predicted in March, and a moderate risk of latent infection was predicted in April. A moderate risk of latent infection was also predicted in June. After considering the situations over the whole season, a high risk of latent infection was predicted, and the observed incidence of fruit brown rot was 62% (Fig. 6D).
In the evaluation 2, the two orchards in Glenn and Butte counties were all considered under the IPH condition, since apothecia were abundant under the trees in spring (T. J. Michailides, unpublished). In the Glenn orchard, the incidence of fruit brown rot was 28.2% in 1994. No risk was predicted in March, but a moderate risk was predicted in April. In May, there were 5 days and 3 days that WD leading to a moderate and a high risk of latent infection occurred (Fig. 7 A), greater than the mean threshold NDM (4.2 days) and NDH (1.8 days), respectively. This weather pattern occurring at the susceptible stages (13) might be a causal reason of the severe brown rot. In 1995, a low risk was predicted in March, June and July, and a high risk was predicted in April and May. The observed incidence of fruit brown rot was 21.3% (Fig. 7B). In 1996, more severe fruit brown rot was observed (incidence = 44.9%). A moderate risk of latent infection was predicted in March, a high risk of latent infection was predicted in April, and a low risk of latent infection was predicted in July (Fig. 7C)
For the orchard in Butte County, the information about risk criteria for Durham (Table 2, for Butte Co.) was used in the evaluation process. The observed incidence of fruit brown rot was 11.8% in 1994 (Fig. 7D). A low risk of latent infection was predicted in March, and a high risk of latent infection was predicted in May. In 1995, although no risky day occurred in March, there were 15 days that WD leading to a moderate risk of latent infection occurred in April (Fig. 7E). A high risk of latent infection was encountered in May, since NDL, NDM, and NDH were all greater than the corresponding threshold values (Fig. 7E). A low risk of latent infection was also predicted in June and July (Fig. 7E). The pattern of continuous risky weather occurred in April and May might lead to a high incidence of fruit brown rot (51.3%). The situations in 1996 were slightly different. A moderate risk was predicted in March (Fig. 7F), and the weather patterns were more favorable for infection in April than in March (Fig. 7F). A low risk of latent infection was also predicted in May and July (Fig. 7F). Occurrence of continuous risky weather patterns in March and April might have led to the high incidence of fruit brown rot at 48.9%.
DISCUSSION
This study determined the seasonal patterns of WD leading to different risk levels of latent infection of prune by M. fructicola. Since the data used in this study were from field experiments at different locations in California, the curves of risky WD over the growing season could be used as reference to guide growers in decision making for disease management. To simplify the use of the results, the IPL could roughly refer to the orchards where brown rot infrequently occurred and apothecia were infrequently found. The IPH could refer to the orchards where abundant apothecia were frequently found in the spring, and brown rot frequently occurred. Thus, the curves in Fig. 3A could be used under the IPL condition to determine if either a current or predicted WD of a certain day or a period of days may relate to what risk level of latent infection. When the WD is below the low-risk curve, fungicide sprays may not be necessary if the grower could sustain a risk of 1% latent infection. However, if the WD is located between the low- and moderate-risk curves on a certain day, a decision for disease control may depend entirely on the grower’s attitude towards risk. When the predicted WD is between the moderate- and high-risk curves, most growers may decide to spray because the predicted ILI levels will be at 5 to 10%. When the predicted WD is above the high-risk curve, a fungicide spray will be highly recommended.
A similar approach could be applied for orchards under the IPH condition (Fig. 3B). Since the data in relation to low risk curve in March and April are not available from the data analysis, the risky WD in these 2 months could not be determined (Fig. 3B). The high-risk curve is higher and sharper at the end of May under the IPH than under the IPL condition. This fact implies that during the period of most resistance to latent infection, even high inoculum potential does not necessarily relate to high risk of latent infection. In other words, much longer WD may be needed to induce a high ILI.
Obviously, not all fruit with latent infection could develop brown rot prior to harvest. Therefore, the ILI may be greater than the observed incidence of fruit brown rot. The conditions under which latent infection could lead to disease appearance are complex. These conditions may include high latent infection levels early and late in the growing season and high humidity during the period of disease appearance. This information is important to determine when a fungicide application is most efficient.
This study not only produced a tool as a reference to decision making for disease control, but also provided information on probabilities of historical occurrence of risk of latent infections. The information is useful especially in decision making when the predicted or observed WDs are between low- and high-risk curves. In making decisions for fungicide spray, growers may need to know a similar historical weather situation and how high the probability of a similar weather and disease situation was. If the current situation may lead to higher risk of latent infection than an average situation, a fungicide spray will be necessary. For example in an orchard at Gerber, located in northern Sacramento Valley, if the predicted WD is 10 h on 10 May, the WD may lead to a moderate risk of latent infection (Fig. 3A) under low inoculum potential (IPL). The average percent days with such WD condition that occurred historically in this month is 10% (3 days), and the P with 90% cumulative probability (CPM90) is 21% (6.3 days). According to the above estimation, the grower may decide to spray because in this period prunes are at a very susceptible stage (13), and only an average of 3 days of this month with the same situation occurred historically. If the disease infrequently occurred in the past in the specific orchard, the current situation may lead to a latent infection level that needs to be reduced. However, the grower may decide not to spray to sustain a certain risk because such risky WD may probably occur only 1 day in this month in the future. In addition to Table 2, other tables, containing the information such as P and CP90, produced by this study, are useful to guide disease control for a specific location, and can be implemented in a decision support system. A grower could make a better decision for disease management by using a precise weather forecast combined with the above risk analysis tools.
Comparing observed incidence of fruit brown rot with predicted risk level of latent infection provided some clues about the importance of seasonal weather patterns in risks of latent infection. This study demonstrated that the risky WD during bloom season was highly related to the incidence of fruit brown rot. For example, higher incidences of fruit brown rot were observed in the Butte and Sutter orchards than in the Tulare and Tehama orchards (Fig. 6), and the predicted risk levels of latent infection in March were consistent with the observations. The weather patterns in May and June might also be important in disease prediction and control since fruit susceptibility to latent infection could change dramatically from susceptible to resistant during this period (13). For example, the higher incidence of fruit brown rot observed in the Sutter orchard than in the other orchards may be due to the fact that only the Sutter orchard showed higher risk levels of latent infection (low risk in May and moderate risk in June) than the other three orchards (Fig. 6).
The evaluation 2 provided information about how possible weather patterns were related to high levels of incidence of fruit brown rot. From the comparisons between weather conditions and observed disease incidence, we conclude that a continuous weather pattern favorable to disease development occurring early in the growing season may bring about highly severe fruit brown rot. More days of favorable weather occurring from March to May can enhance latent infections that may lead to symptoms and sporulation appearance late in the season. The results of the evaluation 2 also provided an outline on possible combinations of risky weather patterns and subsequent levels of disease incidence. For example, if 50% days in April or May showed WD leading to a high risk of latent infection, there would be a high possibility that severe fruit brown rot may occur late in the season. A similar level of disease intensity may also occur when some days (e.g. 30% days of the month) of highly favorable weather occur in both March and April or in both April and May. How different weather patterns at different fruit developmental stages could lead to different disease levels is important information for disease prediction.
Decision making for disease control also needs the information on the effectiveness of fungicides, especially when a fungicide spray is applied at bloom or early in the season. For example, when WD conditions are related to a high risk of latent infection in a period that is within the fungicide residual effect period, an additional fungicide spray may not be necessary. Therefore, a study on fungicide effectiveness together with the risk assessment for disease control is an intensive research topic that may benefit prune production in California.
The seasonal pattern of bloom and fruit susceptibility to latent infection by M. fructicola combined with the seasonal patterns of risky WD conditions outline disease control strategies. In California, fungicide spray(s) at the green tip stage to control blossom blight of prunes is commercially recommended. However, based on our previous study (13), the susceptibility to infection by M. fructicola increased after fruit set especially when fruit are very young. This situation may last from mid April to mid May, and weather conditions during this period are usually flexible in prune-growing areas in California. Infections occurred early in the season might have more chances of disease appearance before fruit become mature than those occurred late in the season. Therefore, the period from April to May is also critical in disease control for which the risk information and decision support are helpful. After mid May, fruit resistance to latent infection increases, and dry weather with much less rain occurs until harvest in these areas. The risk of latent infection during this period decreases, and in most cases, a fungicide spray to control disease may not be recommended. Since the information about the latest period of latent infection that could lead to the disease appearance (sporulation on fruit) before harvest is not yet available, the importance in disease control late in the growing season is still understudied. Even though the fruit susceptibility to latent infection increases in July (13), disease control may not be necessary. But, spraying trees just before the fruit-to-fruit contact stages of prune fruit, that favors brown rot infection (15), may reduce brown rot incidence at harvest. Although a risky WD may be encountered to lead latent infections during this period, disease may not appear before harvest. Moreover, postharvest brown rot may not be important in prunes, especially since prunes are dehydrated immediately after harvest in the majority of the California prune production. However, decision making for disease control in this late period should rely on intensive studies to understand the disease development process late in the season.
Temperature was one factor affecting latent infection of fruit by M. fructicola, that was determined in our previous studies (12,13). Temperatures ranged from 10 to 35ºC throughout the season of the experiments. Since this study involved in a risk analysis of latent infection by considering 10% ILI as the highest risk level of latent infection, a temperature range, relating to less or equal to 10% of ILI calculated from the previous study, was determined. The historical temperatures during the period of 6:00 pm to 10:00 am were used, and the temperature ranges, relating to less or equal to 10% of ILI, encountered in the experiments were found to be within the historical temperature ranges. Therefore, the curves of risky WD could represent the normal temperature situations in fields.
The inoculum potential in different orchards is important information for disease control. Obviously, the inoculum potential in an orchard may change during the growing season as is affected by weather, cultural practices, pathogen survival under different conditions, and other factors. Simply classifying the inoculum potentials into low and high levels as it was done in this study may not represent all possible inoculum potential situations. The sensitivity of inoculum potential in decision making, namely how detailed levels of inoculum potential are needed, is still unknown. Possible methods to determine the inoculum potential in orchards over the growing season are under investigation.
This study was supported by the California Prune Board. We thank D. Morgan, Kearney Agricultural Center, for his technical assistance in previous experiments, data of which were used for the evaluation of risk analysis approaches. We also thank Kris Lynn, Kearney Agricultural Center, for her help in producing the GIS map for this research.
1. Biggs, A. R., and Northover, J. 1985. Inoculum sources for Monilinia fructicola in Ontario peach orchards. Can. J. Plant Pathol. 7:302-307.
4. Byrde, R. J. W., and Willetts, H. J. 1977. The Brown Rot Fungi of Fruit: Their Biology and Control. Pergamon Press. Oxford and New York.
8. Hong, C. X., Holtz, B. A., Morgan, D. P., and Michailides, T. J. 1997. Significance of thinned fruit as a source of the secondary inoculum of Monilinia fructicola in California nectarine orchards. Plant Dis. 81:519-524.
10. Kable, P. F. 1965. Air dispersal of spores of Monilinia fructicola in peach orchards. Aust. J. Exp. Agric. Anim. Husb. 5:166-171.
11. Landgraf, F. A., and Zehr, E. 1982. Inoculum sources for Monilinia fructicola in South Carolina peach orchards. Phytopathology 72:185-190.
16. Michailides, T. J., Morgan, D. P., and Felts, D. 2000. Detection and significance of symptomless latent infection of Monilinia fructicola in California stone fruit. (Abstr.) Phytopathology 90:S48.
17. Michailides, T. J., Morgan, D. P., Felts, D., and Krueger, W. 1996. Ecology and epidemiology of prune brown rot and new control strategies. Pages 109-123 in: Prune Res. Rep. and Index of Prune Res., California Prune Board, Pleasanton.
19. Michailides, T. J., Morgan, D. P., Holtz, P. A., and Hong, C. X. 1995. Biology, ecology, and epidemiology of Monilinia species, and management of prune brown rot with late-spring and early-summer fungicide sprays. Pages 79-100 in: Prune Res. Rep. and Index of Prune Res., California Prune Board, Pleasanton.
20. Michailides, T. J., Morgan, D. P., and Kölliker, R. 1993. Effects of early and mid summer sprays on the postharvest brown rot of French prune. Pages 140-157 in: Prune Res. Rep. and Index of Prune Res., California Prune Board, Pleasanton.
21. Michailides, T. J., Morgan, D. P., Sibbett, S.G., and Teviotdale, B. L. 1994. Management of brown rot of Frech prune by detecting infections in contact surfaces and by early and late summer fungicide applications. Pages 63-86 in: Prune Res. Rep. and Index of Prune Res., California Prune Board, Pleasanton.
22. Phillips, D. J. 1984. Effect of temperature on Monilinia fructicola spores produced on fresh stone fruits. Plant Dis. 68:610-612.
23. Polito, V. S. 1981. Flower and fruit development. Pages 46-52. in: Prune Orchard Management. David E. Ramos, eds. Division of Agricultural Sciences, University of California. Berkeley.
24. Sholberg, P. L., Ogawa, J. M., and Manji, B. T. 1981. Diseases of Prune blossoms, fruits, and leaves. Pages 121-125 in: Prune Orchard Management. Division of Agricultural Sciences, University of California, Berkeley.