THE DEVELOPMENT AND USE OF COUGARBLIGHT 2009C A SITUATION-SPECIFIC FIRE BLIGHT RISK ASSESSMENT MODEL FOR APPLE AND PEAR.
Timothy J. Smith Washington State University 400 Washington Street, Wenatchee, Washington 98801 United States of America
From the 1940's through the 1980's, fire blight caused sporadic, scattered, light to serious damage to pear orchards in the Pacific Northwest United States. Most orchardists did not see fire blight most years. Fire blight damage seemed to the average grower to have no relationship to weather and spray timing, as prophylactic control sprays applied during primary bloom seemed to make little difference to the number of blight strikes. Since the 1990's, apple cultivars and rootstocks in this region have been converted to those that are much more susceptible, and fire blight, while outbreaks remain sporadic, fire blight has become a much more frequent and serious problem. However, most orchards do not have fire blight, even during what we consider a serious fire blight season. In the late 1980's, in an effort to find the most accurate method to predict fire blight infection, a number of fire blight models (Covey 1975, Mills 1955, Steiner 1990, Thompson et al. 1975) were tested, using weather data from the affected regions. These models were mainly built around the concept that blossom blight infection would occur on a day that had a 15.5C (60F) or higher mean temperature and flowers were wetted. The models were difficult to adapt to the pear and apple orchards of the region, as infection rarely occurred during the primary bloom period, as was assumed in the regions where these models had been developed. Also, since blight was uncommon, the contamination of blossoms by Erwinia amylovora was also not common, so blight rarely followed what we would expect to be an infection event. While primary bloom infection is possible on abnormal seasons, in the Pacific Northwest USA, infection usually occurs through the secondary blossoms that occur on pears or apples during the seven to fourteen days after petal fall, or on the shoot tip (tertiary) blossoms that form on certain pear cultivars throughout the growing season. All models tested tended to be highly over-predictive, often indicating that numerous infection events had occurred, when no fire blight resulted in un-sprayed orchards (false positives). The models generally also predicted infection in instances that actual fire blight occurred. More importantly, at times, the models did not predict infection on actual infection events (false negatives). When actual fire blight infection conditions occurred over a large region, it was noted that, usually, a relatively small percentage of orchards in that region were affected, and that disease severity was quite variable. It was determined that no fire blight model present at that time would adequately separate actual fire blight infection events from the more common "near infections."
Temperatures and E. amylovora stigma colony development. The author theorized that a more effective model would need to more accurately evaluate and quantify the temperature conditions leading to the development of a potentially dangerous Erwinia amylovora (Burrill) Winslow et al. colony on the stigma tips of pear and apple flowers. It was noted that days with similar mean temperatures often had very dissimilar hourly temperatures (Figure 1), and it was assumed that this variation could lead to inaccuracies in any model that used mean temperatures as a temperature assessment method. As many accurate entomological models (Coop 1998) use degree days or degree hours derived from sine wave measurements, it was theorized that measuring the number of degree hours that accumulated each hour of the days during the potential E.a. colony growth period would lead to a more accurate assessment of the potential colony growth rate and size. The degree hour values used were based on a asymmetric curve (Figure 2 and Figure 3) adapted from Schoutens' revision of Billings' Potential Doublings table (Schouten 1987), arbitrarily adjusted to a base temperature of 15.5 C, with an optimum peak of 31C, with degree hour values declining to a maximum of 40.5C, above which bacterial numbers were assumed to decline slightly. Degree hour estimated daily totals based on daily high temperatures are outlined in the table provided with the model (C version) (F version). The degree hour heat evaluation method illustrates the difference amongst days with similar mean temperatures (Figure 4). Temperatures over time: Since the stigma tips are open to contamination as soon as the flower opens, and remain in condition to support colony growth only for a low number of days until the flower degrades, it was theorized that temperatures must be evaluated as a total over a few days, rather than a single day as most models (Covey 1975, Steiner 1990) assumed at that time, or continuously throughout the entire blossom period, as others (Thompson et al. 1975, Zoller et al. 1976) did. As reported elsewhere (Smith 1996), a study of numerous actual fire blight infection events and "near infections" showed a more consistent relationship between temperatures accumulated over the four days prior to blossom wetting and actual blossom infection. Figure 5 illustrates four days of hourly temperatures leading up to a flower-wetting potential infection period, which did not result in fire blight infection, despite a 15.5 mean temperature on the day flowers were wetted. Figure 6 shows the four days of temperatures leading up to an actual fire blight infection. Since this assumption was made, and became integral to the Cougarblight model, studies (Gouk 1998, Pusey 2006, 2007) have supported this four day temperature summation assumption by showing that apple stigma tips support the growth of E. amylovora for the first four of the approximate six days that flower remains open and viable. Pusey found that flowers remain succeptible for longer periods, but only under cooler conditions. After the study of numerous infection periods, the minimum four-day Celsius Cougarblight degree hour accumulation that appeared to be necessary for most infections to occur during a blossom wetting was set at approximately 270 (500 F). Further experience has shown that this number should be considered a usually trustworthy guideline only, and the degree of infection risk as determined by degree hour accumulations should be considered as an ever-increasing curve upwards, with "no risk" at the base, rising through "infection possible, but unlikely," to "probable infection," "high risk of infection" and "extreme risk of infection. "Stair-step" thresholds with abrupt changes in risk levels were found to be an improbable model assumption for two reasons: 1. The ultimate size of the colony after a specific amount of time and heat may be greatly influenced by its' CFU numbers when it was first placed upon the stigma surface, and 2. It is not likely that a single degree hour unit would be sufficient to actually move risk from "moderate" 269 four-day degree hour total to a "high" 270 four-day C degree hour total. Due to the "fuzzy" nature of risk thresholds, one should be careful in interpretation of automated model systems. For details about actual fire blight degree hour values, please contact me. I will be pleased to share them. Contact me. Adjusting to the Local Orchard Situation: The potential hazard fire blight presents to any specific orchard varies by variety, rootstock, vigor and age of the tree, and, especially, the recent history of fire blight in the neighborhood. Very often, when blight infection conditions are optimum, most unsprayed orchards escape infection because they are not blooming at that time, or there is very low inoculum in the region. As fire blight infection most often occurs during the four to six weeks following bloom, and the number of secondary bloom varies by year, variety and rootstock, the growers are advised to closely observe their orchards for blossom flushes, and to start running the model whenever open blossoms are present. Since there is no available rapid test for growers to detect the presence and number of E. amylovora in their apple or pear flowers, growers are advised to use the recent history of fire blight in the region around their orchard as a guide to estimate potential pathogen pressure (Model). The most important aspect of this bacterial presence guideline is whether or not fire blight was present in their neighborhood the previous, or has occurred during the current season. Degree hour thresholds are lowered in the model under these potentially high pressure situations, but it is assumed that E. amylovora are present, even though no fire blight has been seen in the region for the previous season. While outbreaks of fire blight are more scattered when infection conditions are experienced in "blight free" areas, this disease generally occurs in a region whenever susceptible hosts are subjected to high or extreme risk infection weather. Growers are advised that infection risk exists everywhere in the USA, it is just more severe some seasons in specific orchards. Blight has shown an ability to remain a potential problem in regions that apparently have had no infections for 20 years or more, flaring in seasons with abnormally conducive weather during blossom periods. The model has long since been automated. Most recently in the (Washington State University Decision Aid System") Most on-line versions compile the hourly degree hour values over a 4 day period, and assign that total to each day, as of 08:00 (8 am). This gives a historical number for each day. (the degree hour total from midnight to 8 am is assigned to the prior day, so the "day" runs from 8 am to 8 am. So, if dew formed from 02:00 to 06:00 in the early morning, the total degree hours for the four days (96 hours) leading up to that wetting event are used in the risk assessment. If the wetting occurs later in the day, the past three days plus the current day are used to determine degree hour totals. These assessments may help the manager decide if an infection event occurred, so they might react accordingly. However, the forecasted degree hours and risk are of greater practical value to the orchard manager. The present day is given a forecasted degree hour value based on actual hourly temperatures for the past three days, plus an estimated value for "today," based on the forecasted daily high temperature. The degree hour forecast for "tomorrow" is based on the actual degree hours for the past two days, plus the estimates for "today and tomorrow." ect. Using forecasted temperatures and estimated degree hour totals, the manager can see infection risk trends in the near future, and can adjust control programs accordingly. Treating during the days leading up to high infection risk periods is especially important when non-antibiotic control products are used. You may be able to suppress the development of E.a. bacterial colonies, which may be much more effective than trying to control infections during or after an infection event. Blossom Wetting: Blossom wetting is often difficult to assess. Moisture measuring and monitoring devices are relatively untrustworthy over time, and, even if they were perfect, it is not possible to place them in every microclimatic condition in an orchard. It is possible to determine that wetting occurred at the monitoring station, but not possible to determine that it did NOT occur in all areas of an orchard region being represented by that station. Rain is the most common form of wetting. Observation of infections occurring in the absence of rain, and a high level of infection occurring only in low areas of orchards with poor air drainage indicates that dew alone will provide the wetness necessary to transport the E. amylovora stigma colony into the flower nectary. It is possible that duration of wetting makes a great difference to a successful infection. Fire blight outbreaks have occurred in some localities without measurable rain occurring during the 10 days to two weeks preceding symptom expression. Data recorded on remotely monitored leaf wetness sensors scattered throughout the Washington State fruit production area indicated that likely infection events coincided with leaf wetness readings of three or more hours, and perhaps as few as two. So, whenever possible, growers are advised to monitor leaf wetness, and consider leaf wetness readings of three or more hours as qualifying as sufficient blossom wetting in a potential infection event. Apple and pear flowers are very often wetted by high volume sprayers during pest control, blossom thinning, or plant growth regulator applications, without any documented instance of resultant fire blight infection in the lower humidity of the western USA. It appears that this short-term wetting does not trigger fire blight infection in Western USA orchards, even when temperatures have been conducive to the development of large E. amylovora stigma colonies. Look-up Chart Version of the Model A simplified version of this model is provided to growers, often with about one hour of training on its use. Growers are advised that they are ultimately responsible for fire blight control decisions, as many of the site-specific aspects related to infection are best evaluated by the model user. Advisers may not be fully aware of the presence of late blossoms, local rain showers, dew formation in poor air drainage areas, or the recent history of fire blight in the specific orchard. As most growers do not have the means to evaluate hourly temperatures, they are directed to assign daily degree hour values from a look-up chart. These estimated daily degree hour values were developed by assigning hourly degree values to a sine wave curve with an amplitude of the daily high temperature. In practice, there is about a plus or minus five percent variation between daily estimated degree hour values and actual daily degree hour values. As growers are taught to consider degree hour totals as an indication of rising and falling risk, rather than as absolute thresholds, the use of estimated, rather than actual, degree hours has resulted in similar control decisions. Agriculture and Agri-Food Canada recently put the Cougarblight Model into Excel spreadsheet format and made it available to all on the internet: Get a file at: Excel Cougarblight Use information: PDF Instructions

Example of 1996, a low blight season
Example of 1998, a high blight season

Covey R .P., 1975. Fire blight control strategies- a look into the future. Proc. 71st An.
Mtg. Washington State Hort. Assoc., 1975: 64-65.

Gouk S.C., 1998. Influence of age of apple flowers on growth of Erwinia amylovora.
Acta Hort., Proc. 8th Int. Workshop on Fire Blight , 489:525-528.

Johnson K. B., and Stockwell V. O., 1998. Secondary spread by antagonistic bacteria
among pear and apple blossoms. Acta Hort., Proc. 8th Int. Workshop on Fire Blight,
489:529.

Ley T. W., 1984. Washington State University Public Agriculture Weather System.
Internet Site: http://frost.prosser.wsu.edu/

Mills W.D., 1955. Fire blight development on apple in Western New York. Pant Dis.
Rptr. 39: 206-207.

Schouten H. J., 1987. A revision of Billing's potential doublings table for fire blight
prediction. Neth. J. Pl.Path., 93:55-60.

Smith T. J., 1996. A risk assessment model for fire blight of apple and pear. Acta Hort.
411:97-100.

Steiner P. W., 1990. Predicting apple blossom infections by Erwinia amylovora using the
MARYBLYT model. Acta Hort. 273:149-158.

Thompson S.V., and Schroth M. N., 1975. Occurrence of fire blight of pears in relation to
weather and epiphytic populations of Erwinia amylovora. Phytopathology 65:353-358.

Witney G.W., and Smith T.J., 1997. Washington State University Chelan/Douglas
Extension. Internet site http://www.ncw.wsu.edu/treefrt.htm

Zoller B. G., and Sisevich J., 1976. Effect of temperature on blossoms and populations of
Erwinia amylovora in Bartlett pear orchards in California during 1972-1976. (Abstract)
Amer. Phytopath. Soc. Proc. 3:322.