Long-term Trends in Meteorological Conditions Favorable for Dollar Spot in Eastern Portions of the United States

Arthur T. DeGaetano, Associate Professor and Director, Northeast Regional Climate Center, Department of Earth and Atmospheric Science, and Frank S. Rossi, Associate Professor of Turfgrass Science, Department of Horticulture, Cornell University, Ithaca, NY 14953

Corresponding author: Frank S. Rossi. fsr3@cornell.edu

Abstract

Two existing predictive models for dollar spot, caused by Sclerotinia homoeocarpa F.T. Bennet, were modified to a single model to accept archived National Weather Service hourly meteorological observations. The revised model was used to identify trends in the potential for dollar spot epidemics from 1950-2004. The frequency of meteorological conditions conducive to dollar spot has increased at the majority of nearly 100 sites in the northeastern quadrant of the United States. Statistically significant trends in favorable weather conditions for dollar spot existed across the region with the greatest trends in the Southeast and Midwest sub-regions. Compared to 1975, these areas on average experienced 24 more days in 2004 that were more favorable for dollar spot occurrence. It appears that the increase in favorable conditions for dollar spot epidemics was best explained by rainfall frequency. Increased disease pressure could require additional fungicide inputs to maintain high quality golf turf plagued with dollar spot.

Are Conditions for Dollar Spot Increasing?

Recent trends noted in temperature and precipitation measurements suggest warmer and wetter growing season conditions (5,9). Climate models also predict these changes as atmospheric carbon dioxide levels increase (8). Fungal pathogens are one of the most important pest complexes in cool-season golf turf management (17). Fungi that infect turfgrasses require various temperature and moisture modalities for infection; however, extended periods of leaf wetness typically create ideal conditions for dollar spot, caused by Sclerotinia homoeocarpa F.T. Bennet (14).

In isolation, these climate changes intuitively imply that the frequency of fungal diseases in plants will increase. This might be modified by concomitant increases in evaporation (6) or contribute to wet and dry cycles that have been shown to enhance certain foliar pathogens (15). While some work has focused on such changes in disease risk in specific species (1,13), no known studies have investigated the influence of long-term climate variations on turfgrass.

Several empirical models relating weather conditions to turf disease have been proposed in the literature (3,16). Yet a long-term historical climate perspective that assesses disease risk occurrence that has changed over the last half century is not available.

Currently available disease models were developed on instrumented research plots in unique microenvironments, and were not designed for assessing long-term trends. The National Weather Service (NWS) has observed and archived a plethora of meteorological variables throughout the 20th century. These data can provide surrogates for the temporally limited meteorological data that drive existing turf disease models. The objective of this paper is to propose modifications of an existing predictive disease model. This adjustment will ascertain quantifiable changes in disease risk across the Northeastern United States during the last half century.

Adapting Existing Models

Dollar spot was selected for evaluation based on its importance to the golf turf industry, the availability of several empirical models, and the set of meteorological conditions that influence disease occurrence.

Burpee and Goulty (2) describe two dollar spot forecasting equations designed to optimize fungicide applications. Mills and Rothwell (12) recommend fungicide application with elevated disease risk. High disease risk is associated with daily mean temperature greater than 25°C and relative humidity greater than 90% during any three days in any seven day period. Hall (7) implies increased dollar spot risk following: (i) two consecutive days with rainfall and a mean temperature > 22°C; or (ii) three consecutive days with rainfall and a mean temperature > 15°C.

For these models daily mean temperature and relative humidity data are available directly from the National Weather Service (NWS) stations. Rainfall occurrence is implied by the documentation of rain or drizzle in any of the hourly weather occurrence reports or the presence of a non-zero daily precipitation amount. Records of these NWS data begin in 1950 and continue through the present. Thus, they provide a long-term data series that is not available elsewhere. Although more than 300 weather stations currently exist in the region depicted in Figure 1, the number of sites with records that extend to 1970 is considerably less. The locations of these sites are depicted by dots in Figure 1.

Fig. 1. Change in the number of daily dollarspot threshold exceedences from 1975-2004. Solid circles represent increases, while open circles show decreases. Shading denotes stations with statistically significant changes (a = 0.05). Subregional clusters of stations, used in subsequent analyses are also identified.

Disease incidence for model verification was determined over a seven-year period from regional diagnostic laboratory reports and weekly reports from county extension turf specialists.

Although this approach only identified a small sample of the total regional incidences of dollar spot, these two data sources provided a unique measure of relative regional dollar spot activity. Throughout the seven-year period, the reporting practices employed remained constant providing a standardized means of comparing week-to-week and season-to-season variations in disease occurrence. Broad-based disease occurrence was used to modify these existing models to indicate regional rather than site-specific disease occurrence. This upscaling allowed the influence of the broad scale meteorological features to be identified. It was assumed that favorable regional conditions imply an enhanced risk of disease at specific sites. However, it was recognized that site-specific characteristics modulate this risk.

Neither model alone was able to adequately simulate the week-to-week variations in disease reports. Likewise the literature provided no indication of which model was superior. Thus, a single dollar spot index was developed that combined both models. This indirect use of the existing models provided a means of characterizing the weather conditions deemed conducive to dollar spot activity in the existing models. On days when the temperature and relative humidity criteria of the Mills and Rothwell (12) model were met, the index assumed a value of one. If temperature and weather conditions were not met, the index was set to zero. If the rainfall and temperature requirements specified by Hall (7) were met, the daily index was increased by one and for any day the dollar spot index had an integer value of zero, one, or two.

Although the index was able to identify periods of increased dollar spot incidence (based on the regional reports and Weather Service data) during summer, reports of dollar spot outbreaks in spring and autumn were often undetected. To better identify these infection periods, a third component of the index, accounting for leaf wetness duration was introduced. The existing index was increased by one on days in which (i) there were more than 8 h of leaf wetness and (ii) the average daily temperature exceeded 15°C. Using the Weather Service data, leaf wetness was assumed during any hour with reported rainfall or a dew point depression (temperature – dew point temperature) of less than 1.7°C during the subsequent hour. Thus the new index could now attain a maximum value of three on a given station-day. This inflated the index values during the spring and fall, highlighting these potential infection periods.

Based on disease detection data from 2001 and 2002 for counties in New York and New Jersey, average daily dollar spot index values ≥ 1.2 were adopted to identify periods of elevated infection risk. Since the disease reports reflected conditions over groups of counties, individual index values from several stations were averaged to obtain a single index value. In addition, since there was likely to be some time lag between infection and the ultimate receipt and analysis of a sample in the laboratory, a seven-day running mean was computed from the spatially averaged indices. A running mean threshold of 0.6 was adopted to identify periods with infection potential.

As a verification of the modified models, Figure 2 shows the application of the daily and running mean index using an independent set of data from 2004. Reports corresponding to the index peak in mid-May were not available. Index peaks in late June, mid-July, and late September were associated with reported dollar spot occurrences.

Fig. 2. Time series of 7-day running mean dollarspot indices (solid line) and individual daily index values (dotted) at southern New England stations during 2004. The shaded vertical bars show weeks with reported dollarspot occurrences. Hatching denotes weeks with no available reports.

Index peaks in August were not consistent with observations suggesting that intensively managed golf turf was regularly treated at this time of year. Therefore, fungicide treatments to manage the conditions that were reported as uncontrollable in late July to early August carried over to the later infection period.

Given the general relationship between the upscaled indices and regional disease occurrence, it appeared feasible to develop long-term climatological time series for increased dollar spot risk. Records of the relevant hourly meteorological variables exist at almost 100 stations in the northeastern quadrant of the United States (Fig. 2), with most records beginning in 1950. Based on Figure 2, daily indices above 1.2 indicate the potential for dollar spot infection as do 7-day running means > 0.6.

Based on dollar spot thresholds, three measures of disease occurrence were analyzed. Daily tallies when indices were exceeded were used to characterize the overall disease potential during each growing season (April 1 to October 31). Alternatively, a time series of the 90th percentile of each annual 7-day running mean series, gave some indication of the persistence of favorable daily disease conditions in a particular season. Finally, the length of each disease season was quantified by the first spring and last autumn day when disease indices were exceeded.

Each time series was assessed for significant trends using a first difference series test described by Karl and Williams (10). This test alleviates the overemphasis of points at the beginning and end of the time series associated with traditional linear regression. Linear fits are used to quantify the magnitude of the temporal changes in seasonal disease potential.

Trends were analyzed over two time periods: 1950-2004 and 1975-2004. The longer period represented the period of available data. The shorter period illustrates how increased dollar spot incidence results from a combination of anthropogenic climate change and natural climate variability.

During the mid 1970s, variations in the phase of large scale atmospheric circulation patterns (4,11) likely influenced the change. Given this a priori knowledge of increasing maximum and minimum air temperature (4), humidity (5), and precipitation (9), all variables which were included in the disease index calculations, the null hypothesis of no change in disease occurrence was evaluated against the alternative hypothesis of increased disease risk. During the early 1970s, multi-year periods existed at all stations in which only every third hourly observation was available electronically. Missing data compromised the computation of the disease indices, since their calculation required information on daily maxima and/or consecutive occurrences of variables, both of which were influenced by the lower sampling frequency. Therefore, years in which more than 10% of the possible hourly observations from April 1 through October 31 were missing have been omitted.

Dollar Spot Risk Increased

The spatial pattern of slopes associated with the 1975-2004 time series of daily dollar spot indices > 1.2 suggested an increase in disease risk (Fig. 2). Negative trends are scattered except for localized areas along the eastern Great Lakes and in the New York City area. Statistically significant trends towards increased disease risk were observed across the Upper Mississippi Valley to the Ohio Valley and southeast.

Based on the longer 1950-2004 record, only a single station at Bridgeport Connecticut displayed a negative dollar spot index trend. The area of statistically significant positive trends was also more widespread, covering the majority of stations in the Mississippi and Ohio Valleys, West Virginia, Virginia, central Pennsylvania, and northern North Carolina. This indicated that the recent (1975-2004) trends characterize the later half of the 20th century.

Figure 3 illustrates the magnitudes of these trends averaged over stations within the four sub-regional groupings in Figure 2. The trend toward a higher number of times when the dollar spot threshold indices were exceeded was readily apparent in each sub-region. Similar slopes existed in both time periods, with the greatest changes (nearly 1 additional threshold index exceeded per year) in the southeast region, and the smallest increases (approximately 1 additional threshold index exceeded every three years) in the northeast.

Fig. 3. Time series of daily occurrences of dollarspot index values > 1.2 with linear trends lines (dashed 1950-2004 , solid 1975-2004) superimposed. Slopes and their associated F statistics and P-values are given in for the 1950-2004 (upper left) and 1975-2004 (lower right) periods. Time series are averages of stations in the (a) Northeast, (b) Ohio/Michigan, (c) Midwest, and (d) Southeast subregions shown in Figure 2.

The pattern of 1975-2004 trends associated with the 90th percentile 7-day running mean of dollar spot indices was similar. Positive trends dominated (75%), with the trends across most of the Midwest being statistically significant. Negative trends were more numerous than for the daily index values. Such trends also clustered through the northeast (New York, Pennsylvania, New Jersey, and New England). None of these negative trends were statistically significant. The 90th percentile results for the 1950-2004 period were similar.

In spring, the first day in which the daily dollar spot index threshold was exceeded has trended toward earlier dates from 1975 to 2004 at 60% of the stations (Fig 4a). Those sites with positive (toward later dates) trends lie primarily in the Upper Midwest, southeast, and New England. In autumn, the last favorable dollar spot conditions tended to occur later in the season through the region (Fig 4b). Only about 25% of the stations report trends toward earlier dates. These stations are primarily in the Northeast. None of the spring trends and only a few of the autumn trends were significant at the α = 0.10 level.

Fig. 4. As in Figure 2, but for (a) the date of the first spring and (b) last autumn daily dollarspot index threshold exceedence.

Collectively, the period of potential dollar spot epidemics (as given by the daily index values) has increased from 1975-2004 across the majority of the region, with the Northeast being a notable exception. Significant increases in length of weather conditions favorable for dollar spot were centered on Kentucky and extended to southern Indiana, Ohio, and Tennessee. Based on the subregions, the length of favorable conditions for disease increased by 0.85 days per year in the Midwest. This average trend was significant at the α = 0.05 based on a F-test of the least squares regression slope. Conversely in the Northeast, the length of the dollar spot season had a trivial decrease of 0.07 days per year during the 1975-2004 period. The Ohio/Michigan and southeast sub-regions showed an increase in the time favorable for dollar spot of 0.3 and 0.6 days per year. For the 1950-2004 period, the subregional trends in season length were similar (approximately 0.5 days per year) and generally significant at the α = 0.10 level. The Northeast was an exception, experiencing a non-significant 0.1 day per year increase in disease season length over the 1950-2004 period.

Model Implications

Over the last 50 years, meteorological conditions have become more favorable for the occurrence dollar spot across much of the northeastern quadrant of the United States. Three factors may contribute to this trend toward potentially higher disease frequency. Given the form of the disease index, which weights each component equally, it is possible to dissect the index to determine which meteorological condition(s) drive the observed trends. With one exception, the relative humidity term in the southeast, the occurrence of each of the index components increased with time. In all cases, it was the consecutive rain day term that showed the greatest increase through time. Significant increases in this term (on the order of 0.75 to 1.0 more non-zero values per year) were noted in all but the northeast region. The leaf wetness term also showed modest increases (< 0.75 more non-zero values per year) in these sub-regions.

Qualitatively these increases are in line with previously documented trends in average temperature, humidity, rainfall amount, and rainfall frequency. However, as is evident by the form of the disease index equation, dollar spot disease occurrence was influenced by relatively specific combinations of hourly temperature, humidity, and rainfall occurrence. Hence trends in average meteorological conditions only provide a backdrop for changes in the frequency of disease potential.

Collectively, meteorological conditions became more favorable for dollar spot, as statistically significant station-specific and sub-regional trends of increased dollar spot threshold indices were widespread, particularly in the Midwest, Southeast, and Ohio Valley. These increases were partially due to an increase in the season over which meteorologically favorable conditions occurred. However, trends in the occurrence of the first spring and last autumn date in which the threshold indices were exceeded were rarely significant.

Acknowledgments

This work was supported by USDA CSREES Grant 0156430. Partial support was also provided by the National Oceanic and Atmospheric Administration, National Climatic Data Center under contract EA1330-02-RP-0011.

Literature Cited

1. Brasier, C. M., and Scott, J. K. 1994. European oak declines and global warming: A theoretical assessment with special reference to the activity of Phytophthora cinnamomi. Bulletin-OEPP 24:221-232.

2. Burpee, L. L., and Goulty, L. G. 1986. Evaluation of two dollar spot forecasting systems for creeping bentgrass. Can. J. Plant Sci. 66:345-351.

3. Danneberger, T. K., Vargas, J. M. Jr., and Jones, A. L. 1984. A model for weather-based forecasting of anthracnose on annual bluegrass. Phytopathology 74:448-451.

4. Easterling, D. R., Horton, B., Jones, P. D., Peterson, T. C., Karl, T. R., Parker, D. E., Salinger, M. J., Razuvayev, V., Plummer, N., Jamason, P., and Folland, C. K. 1997. Maximum and minimum temperature trends for the globe. Science 277:364-367.

5. Gaffen, D. J., and Ross, R. J. 1999. Climatology and trends of U.S. surface humidity and temperature. J. Climate 12:811-828.

6. Golubev, V. S., Lawrimore, J. H., Groisman, P. Y., Speranskaya, N. A., Zhuravin, S. A., Menne, M. J., Peterson T. C., and Malone, R. W. 2001. Evaporation changes over the contiguous United States and the former USSR: A reassessment. Geophys. Res. Lett. 28:2665-2668.

7. Hall, R. 1984. Relationship between weather factors and dollar spot of creeping bentgrass. Can. J. Plant Sci. 64:167-174.

8. Houghton, J. T. 2004. Global Warming: The Complete Briefing. Cambridge Univ., Cambridge, UK.

9. Karl, T. R., and Knight, R. W. 1998. Secular trends of precipitation amount, frequency, and intensity in the United States. Bull. Amer. Meteor. Soc. 79:1107-1119.

10. Karl, T. R., and Williams, C. N. Jr. 1987. An approach to adjusting climatological time series for discontinuous inhomogeneities. J. Climate Appl. Meteor. 26:1744-1763.

11. Mauget, S. A. 2003. Intra- to multidecadal climate variability over the continental United States: 1932-99. J. Climate. 16:2215-2231.

12. Mills, S. G., and Rothwell, J. D. 1982. Predicting diseases: The hygrothermograph. Greenmaster 18:14-15.

13. Rosenzweig, C., Iglesias, A., Yang, X. B., Epstein, P. R., and Chivian, E. 2001. Climate change and U.S. agriculture: The impacts of warming and extreme weather events on productivity, plant diseases, and pests, center for health and the global environment. Global Change & Human Health 2:90-104.

14. Smiley, R. W., Dernoeden, P. H., and Clarke, B. B. 2005. Compendium of Turfgrass Diseases, 3rd Edn. American Phytopathological Society, St. Paul MN.

15. Uddin, W., Serlemitsos, K., and Viji, G. 2003. A temperature and leaf wetness duration-based model for prediction of gray leaf spot of perennial ryegrass turf. Phytopathology 93:336-343.

16. Vallencourt, L., and Schumann, G. L. 1991. Preliminary model for predicting brown patch (Rhizoctonia spp.) on creeping bentgrass. Phytopathology 81:125.

17. Vargas, J. M. 2005. Management of Turfgrass Diseases, 3rd Edn. John Wiley and Sons, NY.