Tar Spot Prediction in Corn: The Weather Matters

CPN 5012. Published December 18, 2023. DOI: doi.org/10.31274/cpn-20231220-1.

Richard W. Webster, North Dakota State University; Camila Nicolli, University of Arkansas; Tom W. Allen, Mississippi State University; Mandy D. Bish, University of Missouri; Kaitlyn Bissonnette, University of Missouri; Jill C. Check, Michigan State University; Martin I. Chilvers, Michigan State University; Maíra R. Duffeck, The Ohio State University; Jane Marian Luis, The Ohio State University; Brian D. Mueller, University of Wisconsin-Madison; Pierce A. Paul, The Ohio State University; Paul P. Price, Louisiana State University; Alison E. Robertson, Iowa State University; Tiffanna J. Ross, Purdue University; Clarice Schmidt, Iowa State University; Roger Schmidt, University of Wisconsin-Madison; Teryl Schmidt, University of Wisconsin-Madison, Sujoung Shim, Purdue University; Darcy E. P. Telenko, Purdue University; Kiersten Wise, University of Kentucky; Damon L. Smith, University of Wisconsin-Madison.

Summary

• Temperature is Crucial. Temperature is the most important environmental variable influencing tar spot development. Optimum temperatures for tar spot development occur when there are extended periods (30 days) of mild temperatures (64-73°F; 18-23℃). Notably, monthly temperatures exceeding 73°F (23°C) reduce the chances of tar spot progression. This confirms prior studies that showed tar spot develops under moderate temperatures.

• Moisture Plays a Role. Tar spot developed when relative humidity was under 90% over a 2-3-week span. Interestingly, while moisture early in the infection process might aid spore germination, extended periods of excessive moisture (RH > 90%), especially at higher temperatures, can hinder disease progression.

• Prediction Tools. These developments in understanding how weather drives tar spot development were used to develop several predictive models. The most accurate model was validated to be accurate 90.1% of the time that it was used. These predictive models were integrated into a smartphone tool publicly available called Tarspotter.

Introduction

Tar spot, caused by the fungus Phyllachora maydis, has emerged as a significant threat to corn (Zea mays L.) production in the United States (Figure 1) as well as Ontario, Canada. Tar spot was initially documented in Mexico in the early 20th century and has since become established throughout much of Latin America (Maublanc 1904). In 2015 P. maydis made its initial appearance in the midwestern United States, specifically in northern Indiana and Illinois (Ruhl et al. 2016). Since then, it has swiftly spread across the U.S. corn belt and has reached states including Georgia as well as Ontario, Canada. The capacity of this pathogen to induce severe epidemics became evident in 2018 when it resulted in a yield loss of approximately 5 million metric tons, translating into an approximate economic loss of over 680 million USD (Mueller et al. 2020).

Despite having been known as a corn pathogen for over a century, the biology and epidemiology of the P. maydis populations in the U.S. remain unknown. Environmental factors undeniably play a crucial role in influencing the disease cycle of P. maydis, from a practical standpoint of spore production, infection, and disease development. For instance, certain temperature ranges and humidity concentrations have been identified as ideal for tar spot development (Hock et al. 1995; Breunig et al. 2023). As this pathogen continues to advance across the U.S., there is a pressing need for a targeted management strategy to properly manage this disease. Currently, while some genetic resistance has been identified in corn germplasm, many commercial hybrids appear to be susceptible. This highlights the importance of fungicides, which have proven to be the most efficacious tool for protecting corn crops against tar spot (Telenko et al. 2022). Presently, the potential of predictive modeling as a tool to optimize fungicide application timing has been demonstrated in numerous pathosystems (Shah et al. 2013, 2014, 2019; Willbur et al. 2018a, 2018b; Kaundal et al. 2006). Recently, this approach has combined traditional statistical methods with modern machine learning techniques to enhance prediction accuracy. Amid the continuing challenges of new and emerging plant diseases, understanding and predicting tar spot development has become essential. The present study aims to bridge knowledge gaps in tar spot development and create predictive models to inform timely and precise management decisions.

Research Goals

• To uncover the environmental variables that are most significant for the development of tar spot in corn.

• To develop models that can predict future tar spot epidemics in the United States and Canada, with the aim of enhancing in-season management decisions.

The Research

Between 2018 and 2022, small plot field trials were conducted across various states including Illinois, Iowa, Indiana, Kentucky, Michigan, Missouri, Ohio, and Wisconsin. Locally adapted hybrids were planted at each site, and the management practices such as seeding rates, nitrogen fertilization, and herbicides followed local recommendations. While the 2018 and 2019 trials included fungicide applications as treatments, only the non-treated plots were considered for the current study. Commercial field sites, operating under regional farmer management, were additionally assessed across this region between 2021 and 2022. These commercial fields primarily served for model validation purposes by considering disease development over time.

Data collection on the severity of tar spot was performed between the R1 growth stage (silking) and continued up to the R5-R6 growth stage (dent to full maturity). Disease severity was evaluated in each trial between one and seven times, depending on the site-year. In small plot trials, the severity of tar spot was assessed visually based on the percentage of P. maydis fungal structures (stroma) on the ear leaf of 5-10 plants per plot, which were then averaged across the entire plot. In commercial fields, a similar protocol was used but for five random corn plants. Then a binary delta variable (yes or no) was defined as any increase in tar spot severity between two consecutive rating dates. Such that, the subsequent models being developed aimed to predict whether tar spot would increase in severity between two time points.

Weather data specific to each site was sourced from historical weather services, gathering hourly average details including ambient air temperature, relative humidity, wind speed, dew point, and precipitation. From these raw values, secondary metrics such as dew point depression (DPD) were calculated. Furthermore, binary variables were established for specific thresholds such as wetness hour (WH) and relative humidity (RH) levels. Using the collected hourly weather data, daily statistics (mean, minimum, and maximum) were derived for all mentioned variables, which were then used to calculate 30-day, 21-day, and 14-day moving averages. These averages, in conjunction with previously established differences (binary delta values), were paired to give insights into tar spot progression based on weather data. Additionally, correlation analyses and logistic regression model development, and evaluations against machine learning algorithms were performed to improve prediction accuracy and model robustness.

This study resulted in an impressive dataset of 588 observations from the Midwest U.S., evaluating the progression of P. maydis stroma. This dataset was randomly divided into training and validation sets to help test the accuracy of newly derived models. Analyses of weather parameters from the historical data service highlighted the significance of the 30-day moving averages of daily minimum and mean ambient temperatures, with correlations of -0.39 and -0.38, respectively, for the progression of tar spot epidemics, meaning that as daily minimum and mean ambient temperatures increase, conditions become less favorable for tar spot development. Examination of relative humidity (RH) emphasized the importance of extended periods of relative humidity below 90% for the development of P. maydis. This effect was especially pronounced during nighttime hours. A daily total wetness hour parameter was introduced, which also exhibited negative correlations with P. maydis in the 21-day and 14-day averages, meaning that increased wetness especially at high temperatures decreases favorable conditions for tar spot development. This work led to the improved understanding of moderate daily mean temperatures (64-73 °F; 18-23 ℃) being the leading environmental factor for tar spot development. Additionally, tar spot development was observed to be highly correlated with periods of dry conditions, especially following heavy wetting events such as thunderstorms or irrigation, meaning that wet/dry environmental cycles can contribute to tar spot development rather than prolonged wet conditions.

From this work, eight logistic regression (LR) models were built (Table 1 in paper). To help predict tar spot development two models emerged as the most likely to predict tar spot, with 86.8% and 86.3% accuracy, respectively. Fusing these models together, led to an ensemble model with improved tar spot prediction accuracy at 87.4%. Additionally, machine learning algorithms were developed, with predictive accuracy of 90.1%, and 85.7% (Table 1 in paper). These models were incorporated into the existing prediction smartphone application Tarspotter.

Tarspotter: A Smartphone Application for Predicting Tar Spot

Tarspotter (https://ipcm.wisc.edu/apps/tarspotter/) was developed as a mobile device application by the Nutrient and Pest Management Program at UW-Madison to predict the likelihood of an increase in tar spot in corn based on the models detailed here. Using the field location, recent weather data preferred action threshold, this app predicts the level of risk of finding tar spot. Using the app can alert farmers to the potential for tar spot development and allow them to respond with field scouting and perhaps fungicide applications, if warranted. Tarspotter can be found in the app store and can be downloaded on smartphones and tablets for free.

Conclusion

This research has significantly advanced our understanding of tar spot development in corn not only in the U.S. but elsewhere. The combined knowledge of the temperature and moisture effect on tar spot, along with predictive models and subsequent smartphone apps, will equip farmers with valuable tools to manage this significant threat. Using these tools, fungicide use can be more precisely suggested to ensure optimal timing of applications, reduce unnecessary application, and improve economic and environmental sustainability.

This research update is based on the work described in the following peer-reviewed research article

Webster, R. W., Nicolli, C., Allen, T.W., et al. 2023. Uncovering the environmental conditions required for Phyllachora maydis infection and tar spot development on corn in the United States for use as predictive models for future epidemics. Sci. Rep. 13:17064. Article / Google Scholar

References

Breunig, M., Bittner, R., Dolezal, A., Ramcharan, A. and Bunkers, G. 2023. An assay to reliably achieve Tar Spot symptoms on corn in a controlled environment. bioRxiv. Article / Google Scholar

Hock, J., Kranz, J., and Renfro, B.L. 1995. Studies on the epidemiology of the tar spot disease complex of maize in Mexico. Plant Pathol. 44:490–502. Article / Google Scholar

Kaundal, R., Kapoor, A.S., and Raghava, G.P.S. 2006. Machine learning techniques in disease forecasting: a case study on rice blast prediction. BMC Bioinform. 7:485. Article / Google Scholar

Maublanc, A. 1904. Especes Nouvelles de champignons inferieurs. Bulletin de la Societe Phytopathologique Francaise. 20:72.

Mueller, D.S. et al. Corn yield loss estimates due to diseases in the United States and Ontario, Canada, from 2016 to 2019. 2020. Plant Health Prog. 21:238–247. Article / Google Scholar

Ruhl, G. et al. 2016. First report of tar spot on corn caused by Phyllachora maydis in the United States. Plant Dis. 100:1496. Article / Google Scholar

Shah, D.A., De Wolf, E.D., Paul, P.A. and Madden, L.V. 2019. Functional data analysis of weather variables linked to Fusarium head blight epidemics in the United States. Phytopathology 109:96–110. Article / Google Scholar

Shah, D.A., De Wolf, E.D., Paul, P.A. and Madden, L.V. 2014. Predicting Fusarium head blight epidemics with boosted regression trees. Phytopathology 104: 702–714. Article / Google Scholar

Shah, D.A. et al. 2013. Predicting Fusarium head blight epidemics with weather-driven pre- and post-anthesis logistic regression model. Phytopathology 103:906–919. Article / Google Scholar

Telenko, D.E. et al. 2022. Fungicide efficacy on tar spot and yield of corn in the Midwestern United States. Plant Health Prog. 23:281–287. Article / Google Scholar

Willbur, J. F. et al. 2018a. Weather-based models for assessing the risk of Sclerotinia sclerotiorum apothecial presence in soybean (Glycine max) fields. Plant Dis. 102:73–84. Article / Google Scholar

Willbur, J. F. et al. 2018b. Validating Sclerotinia sclerotiorum apothecial models to predict Sclerotinia stem rot in soybean (Glycine max) fields. Plant Dis. 102:2592–2601. Article / Google Scholar

Acknowledgements

Authors

Richard W. Webster, North Dakota State University; Camila Nicolli, University of Arkansas; Tom W. Allen, Mississippi State University; Mandy D. Bish, University of Missouri; Kaitlyn Bissonnette, University of Missouri; Jill C. Check, Michigan State University; Martin I. Chilvers, Michigan State University; Maíra R. Duffeck, The Ohio State University; Jane Marian Luis, The Ohio State University; Brian D. Mueller, University of Wisconsin-Madison; Pierce A. Paul, The Ohio State University; Paul P. Price, Louisiana State University; Alison E. Robertson, Iowa State University; Tiffanna J. Ross, Purdue University; Clarice Schmidt, Iowa State University; Roger Schmidt, University of Wisconsin-Madison; Teryl Schmidt, University of Wisconsin-Madison, Sujoung Shim, Purdue University; Darcy E. P. Telenko, Purdue University; Kiersten Wise, University of Kentucky; Damon L. Smith, University of Wisconsin-Madison.

Earn Certified Crop Advisor CEUs

Successfully complete a quiz for this publication to earn 0.5 CCA CEUs.