You are here

2008 Annual CDM summary

-A A +A

Summary of the 2008 Cucurbit Downy Mildew Epidemic in North America

Thomas Keever
Forecaster / Meteorologist, CDM impPIPE
North Carolina State University


2008 was the first year of Cucurbit Downy Mildew (CDM) epidemic forecasting under the auspices of the new CDM ipmPIPE initiative. The CDM impPIPE is a joint effort of academia, the USDA, and industry, headquartered in the Plant Pathology Department at North Carolina State University. The CDM ipmPIPE provides opportunities not previously available to the disease forecasting system. A network of carefully monitored sentinel plots throughout the central and eastern United States and southern Ontario now exists to aid the tracking of epidemic spread. Collaboration with the NC State Climate Office is yielding improvements in the website (including a new automated Epidemic Status map), increased efficiencies in the reporting and presentation of outbreaks, and additional tools for analysis and evaluation of the forecasts. Please consult our website at


Sentinel Plots

In a first for long-range plant disease forecasting, a network of sentinel plots was incorporated into the CDM forecasting framework in 2008. Eighty-three sentinel plots were planted in 23 states and 1 Canadian province east of the Rocky Mountains. Each plot contained the main hosts for Pseudoperonospora cubensis, including cucumber (2 varieties), squash (2 varieties), cantaloupe, watermelon, and pumpkin. Plots were scouted weekly for signs of disease. Sixty of the 83 sentinel plots reported CDM. Of those that did not, nearly all were in areas a large distance from other outbreaks or in areas where the climate during the growing season was more unfavorable for disease development.

The sentinel plots proved quite beneficial in the tracking of the evolving CDM epidemic despite the comparatively sparse nature of the network. Detection of the northward progression of the epidemic during the spring was possible due to the sentinel plots present in FL and southern GA. This led to better prediction of later outbreaks in the Carolinas. This was in direct contrast to most previous years, in which CDM was discovered in Charleston, SC (for example) in late May or June; yet, there were no reports from growing areas between there and the only other known source region in southern FL during an 8 to 10 week time period.  

Timely information from the field is critical when attempting to track an ongoing plant disease epidemic. Of all initial reports from new regions or new states in 2008, over half came from the sentinel plots. Regular, diligent scouting is very important. This may be the greatest aid that the sentinel plots provided.


Epidemic Overview

The CDM ipmPIPE Forecast Center received 174 reports from 130 counties in the United States and southern Canada in 2008. This included 22 states east of the Rocky Mountains, southern Ontario, southern Quebec, and California. There were 59 reports from the sentinel plots, 88 from commercial fields, 23 from other research plots, and 16 from home gardens.

Disease was most prevalent in two main regions, from southern FL northeast along the Atlantic seaboard into southern New England, and from southern MI through the lower Lakes region of southern Ontario, northern OH, northwest PA, and western NY eastward into the central and eastern sections of PA and NY. Outbreaks were much more isolated to widely scattered in portions of the southern Appalachian Mountains, Ohio Valley, Gulf Coast, and south-central United States. The outbreaks furthest removed from the others were those in central CA.

Michigan, Florida, and North Carolina had the most reports with 17, 16, and 11 counties, respectively. Nine counties had confirmed CDM in Ohio and Pennsylvania. Georgia and New York had 7 reports. Connecticut had 6 reports; South Carolina, Virginia, and Texas had 5. Four reports came from Oklahoma and southern Ontario. All 3 counties in Delaware had confirmed outbreaks. Tennessee and Louisiana also had 3 counties with disease. The remaining states all had 2 or fewer counties with CDM.

Infections on cucumber varieties dominated the 2008 CDM epidemic. Every region that reported disease had confirmed CDM on cucumber, including every state and province. Only 2 of the sentinel plots that reported CDM did not have a report on cucumber. Of the approximately 90 commercial field reports, only 10 did not involve a variety of cucumber (there may not have been cucumbers planted in those fields, as well). A count of “first reports” from new states or new regions revealed approximately 38 outbreaks that may fit that description. Of these, 33 occurred on cucumbers, almost all of which were the Straight 8 variety.

Considering the commercial fields, most of the reports on watermelons came from TX, OK, and FL. Squash infections were scattered, mostly in the Southeast and Northeast. Disease on pumpkin was reported in the Northeast and a few isolated spots in the South. Some infection on cantaloupe was discovered in LA. Outbreaks in some locations were severe. However, all known outbreaks on these other hosts were dwarfed by the impact on cucumber.

Considering the sentinel plots, nearly all of the “first reports” occurred on cucumber. Disease then spread to other hosts in the plot, typically the other cucumber variety or squash initially, during periods of favorable conditions. Many plots reported disease on 5 or more of the 7 hosts present. Much variation among the plots was also observed, however, which was likely due to other factors in addition to the CDM (weather favorability, host susceptibility, condition/growth stage of the plants, presence of other diseases, etc).


Epidemic Chronology

Forecasts in 2008 began on March 3 from sources in southern FL, where CDM is virtually endemic. CDM then crawled northward along the peninsula over the next three months, likely held in check by the availability of host and much drier than normal weather for much of that time.

The first report from central FL came from the Hillsborough County sentinel plot on April 17. Estimated time of first symptoms was April 12. Introduction likely occurred during the April 1-6 time frame as noted in the forecasts. Numerous transport events moved along the peninsula under favorable conditions with many areas under Moderate to High risk. The forecast for April 4 is noted below.


The sentinel plots in Orange County, FL and Marion County, FL went positive for CDM on May 5 and May 16, respectively, with first symptoms estimated at May 5 and May 15. Inoculum was likely deposited in Orange County during the transport episode of April 28, with Strongly Moderate Risk to plants in central FL. The Marion County infection was most likely due to short range transport on May 9. The Orange County event is shown below.


The next outbreak of note was reported on May 30, from Harris County in southeast TX. Estimated time of first symptoms in the commercial field of watermelon was May 23. The most recent rain here fell on May 13-14. Trajectories from the known sources in FL at that time did not move within hundreds of miles of southeast TX. An unknown source, perhaps in Mexico, is suspected in this case though this cannot be supported. Origins of this outbreak remain unknown.

The first report from GA arrived on June 4, from a commercial field of cucumber in Brooks County along the FL state line. Symptoms were fresh. Deposition likely occurred from the nearby sources to the south, either on May 23 or May 29.

The first report from the FL panhandle arrived from the sentinel plot in Gadsden County, FL on June 18. Symptoms on squash were fresh. Many of the forecasts / transport events from June 9 to June 15 indicated Moderate Risk or more to plants in this area. The best choice appears to be June 13, shown below.


Vernon Parish in west-central LA was the site of that state’s first report on June 19, on cucumber in the sentinel plot. Symptoms were estimated at June 12. A discovery was made the next day in an adjacent field of cantaloupe. Rain fell on June 5 and 6 here, a solid timing for appearance of symptoms a week later. Given the lack of other nearby sources, the most plausible case is one in which spores from the fairly weak TX source were washed out here as they brushed by the site on their way northward.

The first report from SC arrived on June 20. Symptoms estimated to have first appeared on June 19 were observed on cucumber in the sentinel plot in Charleston. Consideration of the transport events and actual rainfall at Charleston lead to a high likelihood that introduction of inoculums occurred via the transport events of June 14 from southern GA and/or northern FL. Plants in the Charleston area were at Moderate Risk during this event, illustrated below.


The next major discovery, from Elgin County, Ontario was reported to the ipmPIPE Center on June 20, 2008. Infection was confirmed in a commercial field of cucumber. Estimated appearance of symptoms was June 16-17. As in previous years, thorough and exhaustive examination of all evidence has not yielded a plausible method of epidemic spread to this area via airborne transport of inoculum from known sources. It is strongly suspected that some other method of introduction was involved here, whether by transport from unknown or unreported sources or by some other means.

 The first reports from NY were from 3 counties in the far western sections of the state on July 7-8: Niagara, Erie, and Ontario. Infections on cucumber were discovered in small commercial fields with an estimated date of first symptoms of July 2. The forecasts of June 23 and June 25 describe potential scenarios. The events out of southern Ontario on June 23 (Moderate Risk) and June 26 (High risk) fit the timing of the symptoms. Either or both of these events may have caused these new infections, though actual rainfall data lends greater weight to the June 23 event, shown below.


The first report from MI (Monroe County) arrived on July 3-4. A commercial field in the far southeastern corner of the state showed recent symptoms of CDM on cucumber. Additional information from the MI region indicated that there may have been other outbreaks present in the state. This was borne out a week later when the Forecast Center received a report of CDM in Allegan County, MI which is located on the far southwestern side of Michigan’s Lower Peninsula. It was unknown how long the Allegan County outbreak was present.

No transport events from the southern Ontario outbreak moved toward southern MI, including those from the estimated time of first symptoms. No transport events from the other known sources moved toward southern MI. As in the southern Ontario case, there is suspicion that there are undiscovered or unreported sources, or inoculum has been introduced by other means. There is not enough evidence available to give a reasonable explanation for the MI outbreak or draw a confident conclusion.

With disease established in the Southeast and outbreaks present in southern Ontario and MI, the epidemic situation was ripe for what followed. Increased periods of unsettled weather along the eastern Seaboard and the northern U.S. / southern Canada resulted in the epidemic advancing on two fronts during the heart of summer. The favorable weather conditions encouraged localized to short-range spread near existing sources and enabled long-range transport and deposition from those areas. The weather patterns during these periods also produced elevated-risk transport scenarios that had the potential to impact similar regions on a number of consecutive days. 

From June 27 to June 30, and again from July 4 to July 8, transport events from the southeast U.S. moved northeast through the Atlantic coastal states under mixed to favorable conditions for long-range transport and deposition. Risk to cucurbits in many areas was Moderate, Strongly Moderate, or High during these episodes. Further, previous experience and the historical record show that repeated events over the same areas can greatly increase the threat of epidemic spread.

The first infections in NC in 2008 provided a good example. Reports of CDM arrived on July 10 and 11 from Wilson, Lenoir, Nash, and Johnston counties. Disease had been discovered on cucumber in sentinel plots and commercial fields. Estimated dates of first symptoms ranged from July 5 to July 11.

Any or all of the events originating in the Southeast from June 27 to June 30, or perhaps even those of early July, could have caused these outbreaks. It is not possible to pinpoint a specific transport event based on the information available. The event from southern GA, shown below, is a good illustration.


The first report from DE fit into this same scenario. This outbreak on cucumber in the sentinel plot in Sussex County was reported on July 9. Symptoms had appeared the same day. A specific transport event cannot be identified based on the data; any of several different episodes may have brought live spores to this location. These include the  July 4 event from eastern SC, or even the June 30 event from the southern Ontario region, both shown below.




The first report from OH was on July 9 from Geauga County in the east-central portion of the state. The infection was discovered on cucumber in a commercial field. First symptoms were estimated at July 2. As with the MI sources, this outbreak likely arose from unknown or unreported sources. Transport events from the known sources around the probable time of spore deposition do not lend a plausible explanation.

The first report from MD concerned an infection on Straight 8 cucumber in the sentinel plot in Wicomico County on the DelMarVa peninsula. The report on July 21 indicated fresh symptoms. The first reports from VA (commercial field of slicing cucumber in Suffolk County, sentinel plot infection on Straight 8 in Virginia Beach) arrived two days later with estimated first symptoms of July 16. The origin of these outbreaks lay with existing infections in eastern SC (shown below in the forecasts of July 9-10, Moderate to Strongly Moderate Risk) and/or the newly symptomatic but not yet discovered infections in eastern NC.


The first outbreak in the southern Appalachian Mountain region was from Haywood County, NC. The report was made on July 23. First symptoms in this field of cucumber were estimated at July 18. This infection was due to transport from the southern GA / FL panhandle region. Cucurbits in the western Carolinas were at Moderate Risk for  2 days in a row on July 7 and 8. Other events in the appropriate time frame are far less likely. The July 7 trajectories are shown below.


The first report from NJ came from a field of cucumber in Cumberland County. First symptoms were estimated at July 20 in the July 24 report. This infection likely arose due to transport and effective deposition from infections close by to the south or from the lower Lakes region. Risk to cucurbits here was Moderate July 7 and 8, High July 9, making this period the most likely for initial infection.

On July 30, the first report from Quebec was received. CDM was discovered in a field of cucumber. First symptoms were estimated at July 28. Transport events from the lower Lakes region were responsible for this outbreak. Several opportunities were present, with one of the transport episodes of July 22  (Moderate Risk), July 23 (Strongly Moderate / High Risk), or July 24 (Strongly Moderate / High Risk) being the most likely. The July 23 event from western NY is shown below.


The first report from PA was from cucumbers in the sentinel plot in Erie County on July 31. First symptoms were estimated at July 26. Numerous events with elevated risk to cucurbits in this area occurred from July 18-22, any of which may have led to this outbreak. The High Risk July 21 event is shown below.


The first report from the southern New England region arrived on August 20 from Hartford County, CT. Infection was discovered in a field of butternut squash. Several more reports were received from this area during the following week. Estimated date of first symptoms was August 14. Inoculum was likely introduced during the transport events of August 7-10 from locations to the north and west. There is lesser, but still possible likelihood of contribution from sources to the south. One event from Friday, August 8 is shown below.


On August 22, the first discovery from MA was reported, on cucumber in the sentinel plot in Hampshire County.  Symptoms were recent. This infection was most likely due to deposition of inoculum around August 13-14 from the known sources in DE/NJ and the new (but not yet discovered) infections in CT.

Having somehow avoided any outbreaks previously, a sentinel plot on Long Island finally went positive for CDM on cucumber and was reported on August 29. Symptoms were recent. Airborne spores likely arrived from nearby sources during the third week of August.

Epidemic spread in late summer and fall became much of a hit-or-miss proposition. After a fairly dry summer and few disease discoveries in TX, OK reported its first outbreaks in early September. One was in large commercial fields of watermelon in Custer County, heavily infected especially in the older plantings. Another was at the sentinel plot in Caddo County, on cucumbers and watermelon. Exact origins are unknown. Two factors greatly complicating the analysis are the uncertain length of time the Custer County infection was present before discovery, and the mid-September report of CDM in Atacosa County in western TX, where disease had been present since late May or early June!

Reports from California arrived on September 4, from organic fields of cucumber in Santa Cruz and San Benito counties. Symptoms were estimated to have first appeared in mid-August. Origins are unknown, though it is near certain that airborne spores from sources east of the Rocky Mountains did not cause these outbreaks.

The first reports from KY and TN arrived in mid-September. These were small and mostly isolated reports from cucumbers in the sentinel plots in Breathitt County, KY and Robertson and Cumberland counties, TN. Allen County, KY reported an infection on cucumber in a home garden. Transport events from the known sources lend little support to introduction of inoculum during a seemingly appropriate time frame. It is suspected that events earlier in the summer from sources to the north or east deposited live spores in this region. The ensuing infections then went unnoticed or unreported until the epidemic spread to locations that were regularly scouted.

CDM moved back into the growing areas of the Southeast during their fall planting. Frost / freezes and the end of the natural growing season gradually eliminated host availability from north to south during autumn. The last forecast in 2008 was posted on November 21.