Key Takeaways

• The Smith-Kerns dollar spot forecast model was developed using bentgrass cultivars with high disease susceptibility. The accuracy of this model on cultivars with low susceptibility to dollar spot was not known.
• This field study assessed the ability of five risk index action thresholds of the model to forecast dollar spot on five creeping bentgrass cultivars and one colonial bentgrass with a wide range of susceptibility to dollar spot.
• The currently recommended 20% risk index action threshold over-predicted dollar spot on low-susceptibility (tolerant) cultivars, meaning disease was predicted when it was not actually observed.
• Thresholds using a risk index greater than 20% or using the risk index slope improved model accuracy for low-susceptibility (tolerant) cultivars.
• This research demonstrates that novel action thresholds may improve dollar spot predication for low-susceptibility (tolerant) bentgrass cultivars, leading to better management decisions.
   

Collecting Data

Dollar spot incidence was assessed by counting the number of infection centers per plot every 2 to 18 days – or more frequently when disease was increasing rapidly – from May to November in 2015, 2016 and 2017. Disease incidence data were summarized over time as the area under disease progress curve (AUDPC) and data were pooled across repetitions to determine the relative susceptibility of cultivars to dollar spot.

Weather data – i.e., air temperature and relative humidity – were collected on-site using a combination data sensor. Air temperature and relative humidity were measured every five seconds and used to calculate hourly averages. The Smith-Kerns model uses five-day rolling averages of hourly relative humidity and air temperature to calculate the probability of dollar spot developing on the site.

Results: Ranking Cultivar Susceptibility to Dollar Spot

Similar to previous reports by the National Turfgrass Evaluation Program (2014), ‘Independence’ ranked in the most susceptible category followed by ‘Penncross’ and ‘Shark’, whereas ‘007’ ranked in a moderately susceptible category and ‘Declaration’ and ‘Capri’ in the least susceptible category.

The initial appearance of dollar spot symptoms was often delayed and peak disease incidence was lower for less susceptible (more tolerant) cultivars, but overall trends of disease progress were similar among all cultivars in the study. Dollar spot activity also mirrored the Smith-Kerns Model risk index output over time (slope). As demonstrated in Figure 1, all cultivars had periods of decreasing dollar spot activity, and these periods of turf recovery frequently occurred when the risk index was decreasing.

Results – Action Threshold Accuracy

Assessing the accuracy of action thresholds
Five action thresholds of the Smith-Kerns Model risk index (RI) were evaluated for accuracy in forecasting dollar spot for each bentgrass cultivar listed in Table 1. The first action threshold used the currently recommended risk index threshold of 20% (Smith et al., 2018). The second action threshold considered the relationship between dollar spot and the slope of the risk index (RI slope). This action threshold was based on preliminary observations at Rutgers University that negative slopes of the risk index were strongly associated with periods of inactive dollar spot and positive risk index slopes were strongly associated with increasing dollar spot incidence. The third action threshold combined the RI 20% and RI slope action thresholds (RI 20% + RI slope). The fourth action threshold evaluated adjusting the risk index from 1% to 60% in 1% increments to select a RI threshold associated with the maximum accuracy in predicting dollar spot for each cultivar (RI_max). The fifth action threshold selected the RI_max while also considering RI slope (RI_max + RI slope).

Dollar spot disease data were compared with the forecasts of the five action thresholds to assess prediction accuracy.

RI 20%
The accuracy of the RI 20% action threshold was greatest on highly susceptible cultivars and lowest on the least susceptible cultivar, ‘Declaration’. This is not surprising because the Smith-Kerns Model, which uses a RI 20% action threshold, was developed with disease data from highly susceptible cultivars (Smith et al., 2018). The association of decreasing accuracy of the RI 20% threshold with decreasing host susceptibility (increasing tolerance) suggests that additional action thresholds may increase prediction accuracy on less susceptible cultivars like ‘Declaration’, ‘Capri’ and ‘007’. Over-predictions (disease increase predicted but not observed) were the primary cause for inaccurate forecasts of this action threshold. Under-predictions – a worst-case scenario for a golf course superintendent where no disease is predicted but dollar spot is observed – occurred on no more than 6% of the observations and were more frequent on highly susceptible cultivars.

RI slope
The RI slope action threshold increased prediction accuracy for all cultivars, except for the highly susceptible cultivar ‘Independence’, compared to the RI 20% action threshold. Low-susceptibility cultivars saw the greatest improvement (11% to 14%) in accuracy. The RI slope action threshold over-predicted dollar spot activity more than it under-predicted for all cultivars except the highly susceptible cultivars ‘Independence’ and ‘Penncross’.

RI 20% + RI slope
Compared with the RI 20% action threshold, the RI 20% + RI slope threshold improved prediction accuracy for all cultivars, with the greatest improvements occurring for lower susceptibility cultivars. Increased accuracy with the RI 20% + RI slope action threshold was due to fewer (17% to 26% less) over-predictions; albeit with a slight increase (between 5% and 13%) in under-predictions compared to the RI 20% action threshold.

RI_max
Prediction accuracy improved as the risk index increased above 20% for low-susceptibility cultivars. An RI_max of 58% increased accuracy by 25% for ‘Declaration’ and 22% for ‘Capri’ compared to the RI 20% action threshold. Once again, improved accuracy was due to reduced over-predictions. Similarly, risk indexes of 47% for ‘007’ and 50% for ‘Shark’ increased the number of accurate predictions, but also increased the number of under-predictions by 13% to 22% compared to the RI 20% action threshold.

Adjusting the risk index away from the 20% model standard did not greatly improve prediction accuracy for the highly susceptible cultivars ‘Independence’ and ‘Penncross’, which corroborates the work of Smith et al. (2018). Action thresholds using a risk index greater than 20% may help improve accuracy of fungicide scheduling on moderate- to low-susceptibility (tolerant) cultivars, but not on highly susceptible cultivars. However, improved accuracy may come at the cost of more under-predictions, which could result in increasing the number and severity of dollar spot outbreaks. This is an important consideration since an increased likelihood of under-predicting the occurrence of dollar spot may affect a superintendent’s willingness to rely on a disease forecast model to schedule fungicide applications.

RI_max + RI slope
Accuracy was further improved when using an RI_max + RI slope action threshold compared to the RI_max action threshold. While the improvement in accuracy of the RI_max + RI slope action threshold on less susceptible cultivars was relatively small, under-predictions on these cultivars were notably reduced.

Under-Predictions: Research Versus Real World

Large increases in dollar spot incidence across all cultivars occasionally occurred during some periods of under-prediction in the current trial (data not shown). Since no fungicides were applied in response to these action thresholds, under-predictions were occurring as inoculum levels steadily increased throughout the duration of the trial. Under real-world conditions where action thresholds would trigger a golf course superintendent to make fungicide applications, the inoculum loads would presumably be much lower and under-predictions may not be problematic. Additional research is needed over several locations, years and cultivar susceptibility levels to determine the risk that under-predictions may pose for practitioners in the field when novel risk index action thresholds are used for scheduling fungicide applications.

During the autumn months, there were occasional periods when dollar spot increased while the risk index was below 10% (data not shown). Soil temperatures are often higher than air temperatures in autumn, a phenomenon which may affect the incidence and severity of dollar spot outbreaks. The model may need to be further refined for predicting disease outbreaks during this period. A recently developed real-time, polymerase chain reaction method is currently being used to detect and quantify the population of C. jacksonii in turf. This method should help determine the extent to which the C. jacksonii populations are affected by seasonal weather patterns, fungicide applications or cultivar susceptibility (Groben et al., 2020).

Conclusions

The current RI 20% action threshold is a useful tool for scheduling fungicides on highly susceptible cultivars like ‘Penncross’ and ‘Independence’ but can lead to over-predictions and unnecessary fungicide use, particularly on low-susceptibility cultivars. Turf managers may be able to use novel action thresholds on low-susceptibility cultivars such as ‘007’, ‘Capri’ and ‘Declaration’ that incorporate a risk index greater than 20% and/or RI slope to reduce fungicide inputs and maintain acceptable dollar spot control. Additional field studies are underway to test this hypothesis.

Acknowledgments

The authors would like to thank T.J. Lawson for assistance with field work. This research was supported by the United States Golf Association, New Jersey Agricultural Experiment Station, Rutgers Center for Turfgrass Science, Golf Course Superintendents Association of America, Golf Course Superintendents Association of New Jersey, Tri-State Turf Research Foundation and the New Jersey Turfgrass Foundation.

This article is based on research published in the journal Crop Science. You can find additional information, including all tables and figures, here.

References

Burpee, L. L., & Goulty, L.G. (1986). Evaluation of two dollar spot forecasting systems for creeping bentgrass. Canadian Journal of Plant Science, 66(2), 345-351.

Groben, G., Clarke, B.B., Murphy, J.A., Koch, P.L., Crouch, J., Lee, S., & Zhang, N. (2020). Real-time PCR detection of Clarireedia spp., the causal agents of dollar spot in turfgrasses. Plant Disease, 104(12), 3118-3123. 

Latin, R. (2021). A practical guide to turfgrass fungicides, 2nd edition. APS Press.

Lyman, G.T., Throssell, C.S., Johnson, M.E., Stacey, G.A. and Brown, C.D. (2007). Golf course profile describes turfgrass, landscape, and environmental stewardship features. Applied Turfgrass Science, 4(1), 1-25. 

NTEP Staff. (2014). Dollar spot ratings of bentgrass cultivars grown on a fairway or tee. 2008 National Bentgrass Test – Fairway/Tee. 2009-13 Data – Final Report, Table 39A.

Smith, D. L., Kerns, J.P., Walker, N.R., Payne, A.R., Horvath, B., Inguagiato, J.C., Kaminski, J.E., Tomaso-Peterson, M., & Koch, P.L. (2018). Development and validation of a weather-based warning system to advise fungicide application to control dollar spot on turfgrass. PloS ONE, 13(3), e0194216