Relating Peanut Rx Risk Factors to Epidemics of Early and Late Leaf Spot of Peanut

Previous research has demonstrated the efficacy of prescription fungicide programs, based upon Peanut Rx, to reduce combined effects of early leaf spot (ELS), caused by Passalora arachidicola (Cercospora arachidicola), and late leaf spot (LLS), caused by Nothopassalora personata (syn. Cercosporidium personatum), but the potential of Peanut Rx to predict each disease has never been formally evaluated. From 2010 to 2016, non-fungicide-treated peanut plots in Georgia and Florida were sampled to monitor the development of ELS and LLS. This resulted in 168 cases (unique combinations of Peanut Rx risk factors) with associated total leaf spot risk points ranging from 40 to 100. Defoliation ranged from 13.9 to 100%, and increased significantly with increasing total risk points (conditional R² = 0.56; P < 0.001). Leaf spot onset (time in days after planting [DAP] when either leaf spot reached 1% lesion incidence), ELS onset, and LLS onset ranged from 29 to 140, 29 to 142, and 50 to 143 DAP, respectively, and decreased significantly with increasing risk points. Standardized AUDPC of ELS was significantly affected by risk points (conditional R² = 0.53, P < 0.001), but not for LLS. After removing redundant Peanut Rx factors, planting date, rotation, historical leaf spot prevalence, cultivar, and field history were used as fixed effects in mixed effect regression models to evaluate their contribution to leaf spot, ELS or LLS prediction. Results from mixed effects regression confirmed that the selected Peanut Rx risk factors contributed to the variability of at least one measurement of development of combined or separate epidemics of ELS and LLS, but not all factors affected ELS and LLS equally. Historical leaf spot prevalence, a new potential preplant risk factor, was a consistent predictor of the dominant disease(s) observed in the field. Results presented here demonstrate that Peanut Rx is a very effective tool for predicting leaf spot onset regardless of which leaf spot is predominant, but also suggest that associated risk does not reflect the same development for each disease. These data will be useful for refining thresholds for differentiating high, moderate, and low risk fields, and reevaluating the timing of fungicide applications in reduced input programs with respect to disease onset.

Focus expansion and stability of the spread parameter estimate of the power law model for dispersal gradients

Empirical and mechanistic modeling indicate that pathogens transmitted via aerially dispersed inoculum follow a power law, resulting in dispersive epidemic waves. The spread parameter (b) of the power law model, which is an indicator of the distance of the epidemic wave front from an initial focus per unit time, has been found to be approximately 2 for several animal and plant diseases over a wide range of spatial scales under conditions favorable for disease spread. Although disease spread and epidemic expansion can be influenced by several factors, the stability of the parameter b over multiple epidemic years has not been determined. Additionally, the size of the initial epidemic area is expected to be strongly related to the final epidemic extent for epidemics, but the stability of this relationship is also not well established. Here, empirical data of cucurbit downy mildew epidemics collected from 2008 to 2014 were analyzed using a spatio-temporal model of disease spread that incorporates logistic growth in time with a power law function for dispersal. Final epidemic extent ranged from 4.16 ×10⁸ km² in 2012 to 6.44 ×10⁸ km² in 2009. Current epidemic extent became significantly associated (P < 0.0332; 0.56 < R² < 0.99) with final epidemic area beginning near the end of April, with the association increasing monotonically to 1.0 by the end of the epidemic season in July. The position of the epidemic wave-front became exponentially more distant with time, and epidemic velocity increased linearly with distance. Slopes from the temporal and spatial regression models varied with about a 2.5-fold range across epidemic years. Estimates of b varied substantially ranging from 1.51 to 4.16 across epidemic years. We observed a significant b ×time (or distance) interaction (P < 0.05) for epidemic years where data were well described by the power law model. These results suggest that the spread parameter b may not be stable over multiple epidemic years. However, b ≈ 2 may be considered the lower limit of the distance traveled by epidemic wave-fronts for aerially transmitted pathogens that follow a power law dispersal function.

A Model for Predicting Onset of Stagonospora nodorum Blotch in Winter Wheat Based on Preplanting and Weather Factors

Stagonospora nodorum blotch (SNB) caused by Parastagonospora nodorum is a serious disease of wheat worldwide. In the United States, the disease is prevalent on winter wheat in many eastern states, and its management relies mainly on fungicide application after flag leaf emergence. Although SNB can occur prior to flag leaf emergence, the relationship between the time of disease onset and yield has not been determined. Such a relationship is useful in identifying a threshold to facilitate prediction of disease onset in the field. Disease occurred in 390 of 435 disease cases that were recorded across 11 counties in North Carolina from 2012 to 2014. Using cases with disease occurrence, the effect of disease onset on yield was analyzed to identify a disease onset threshold that related time of disease onset to yield. Regression analysis showed that disease onset explained 32% of the variation in yield (P < 0.0001) and from this relationship, day of year (DOY) 102 was identified as the disease onset threshold. Below-average yield occurred in 87% of the disease cases when disease onset occurred before DOY 102 but in only 28% of those cases when onset occurred on or after DOY 102. Subsequently, binary logistic regression models were developed to predict the occurrence and onset of SNB using preplanting factors and cumulative daily infection values (cDIV) starting 1 to 3 weeks prior to DOY 102. Logistic regression showed that previous crop, latitude, and cDIV accumulated 2 weeks prior to DOY 102 (cDIV.2) were significant (P < 0.0001) predictors of disease occurrence, and wheat residue, latitude, longitude, and cDIV.2 were significant (P < 0.0001) predictors of disease onset. The disease onset model had a correct classification rate of 0.94 and specificity and sensitivity rates >0.90. Performance of the disease onset model based on the area under the receiver operating characteristic curve (AUC), κ, and the true skill statistic (TSS) was excellent, with prediction accuracy values >0.88. Similarly, internal validation of the disease onset model based on AUC, κ, and TSS indicated good performance, with accuracy values >0.88. This disease onset prediction model could serve as a useful decision support tool to guide fungicide applications to manage SNB in wheat.

Predicting Pre-planting Risk of Stagonospora nodorum blotch in Winter Wheat Using Machine Learning Models

Pre-planting factors have been associated with the late-season severity of Stagonospora nodorum blotch (SNB), caused by the fungal pathogen Parastagonospora nodorum, in winter wheat (Triticum aestivum). The relative importance of these factors in the risk of SNB has not been determined and this knowledge can facilitate disease management decisions prior to planting of the wheat crop. In this study, we examined the performance of multiple regression (MR) and three machine learning algorithms namely artificial neural networks, categorical and regression trees, and random forests (RF), in predicting the pre-planting risk of SNB in wheat. Pre-planting factors tested as potential predictor variables were cultivar resistance, latitude, longitude, previous crop, seeding rate, seed treatment, tillage type, and wheat residue. Disease severity assessed at the end of the growing season was used as the response variable. The models were developed using 431 disease cases (unique combinations of predictors) collected from 2012 to 2014 and these cases were randomly divided into training, validation, and test datasets. Models were evaluated based on the regression of observed against predicted severity values of SNB, sensitivity-specificity ROC analysis, and the Kappa statistic. A strong relationship was observed between late-season severity of SNB and specific pre-planting factors in which latitude, longitude, wheat residue, and cultivar resistance were the most important predictors. The MR model explained 33% of variability in the data, while machine learning models explained 47 to 79% of the total variability. Similarly, the MR model correctly classified 74% of the disease cases, while machine learning models correctly classified 81 to 83% of these cases. Results show that the RF algorithm, which explained 79% of the variability within the data, was the most accurate in predicting the risk of SNB, with an accuracy rate of 93%. The RF algorithm could allow early assessment of the risk of SNB, facilitating sound disease management decisions prior to planting of wheat.

Quantifying the Effects of Wheat Residue on Severity of Stagonospora nodorum Blotch and Yield in Winter Wheat

Stagonospora nodorum blotch (SNB), caused by the fungus Parastagonospora nodorum, is a major disease of wheat (Triticum aestivum). Residue from a previously infected wheat crop can be an important source of initial inoculum, but the effects of infected residue on disease severity and yield have not previously been quantified. Experiments were conducted in Raleigh and Salisbury, North Carolina, in 2012, 2013, and 2014 using the moderately susceptible winter wheat cultivar DG Shirley. In 2014, the highly susceptible cultivar DG 9012 was added to the experiment and the study was conducted at an additional site in Tyner, North Carolina. Four (2012) or six (2013 and 2014) wheat residue treatments were applied in the field in a randomized complete block design with five replicates. Treatments in 2012 were 0, 30, 60, and 90% residue coverage of the soil surface, while 10 and 20% residue treatments were added in 2013 and 2014. Across site-years, disease severity ranged from 0 to 50% and increased nonlinearly (P < 0.05) as residue level increased, with a rapid rise to an upper limit and showing little change in severity above 20 to 30% soil surface coverage. Residue coverage had a significant (P < 0.05) effect on disease severity in all site-years. The effect of residue coverage on yield was only significant (P < 0.05) for DG Shirley at Raleigh and Salisbury in 2012 and for DG 9012 at Salisbury in 2014. Similarly, residue coverage significantly (P < 0.05) affected thousand-kernel weight only of DG 9012 in 2014 at Raleigh and Salisbury. Our results showed that when wheat residue was sparse, small additions to residue density produced greater increases in SNB than when residue was abundant. SNB only led to effects on yield and test weight in the most disease-conducive environments, suggesting that the economic threshold for the disease may be higher than previously assumed and warrants review.

Postharvest Disease Development on Southern Highbush Blueberry Fruit in Relation to Berry Flesh Type and Harvest Method

Postharvest decay, incited by various fungal pathogens, is a major concern in most blueberry production areas of the United States. Because the risk of infection is increased by fruit bruising, which in turn is increased by machine-harvesting, it has been difficult to harvest fruit from the early-maturing but soft-textured southern highbush blueberries (SHB) mechanically for the fresh market. This could change fundamentally with the recent development of SHB genotypes with crisp-textured ("crispy") berries, i.e., fruit with qualitatively firmer flesh and/or more resistant skin. Four replicate row sections of two or three SHB genotypes having crispy fruit and three with conventional fruit were either hand- or machine-harvested at a commercial blueberry farm in northern Florida in April 2009 and May 2010. Harvested fruit were sorted, packed, and placed in cold storage (2°C) for up to 3 weeks. Average counts of aerobic bacteria, total yeasts and molds, coliforms, and Escherichia coli on fruit samples before the cold storage period were below commercial tolerance levels in most cases. In both years, natural disease incidence after cold storage was lowest for hand-harvested crispy fruit and highest for machine-harvested conventional fruit. Interestingly, machine-harvested crispy fruit had the same or lower disease incidence as hand-harvested conventional fruit. Across all treatments, natural postharvest disease incidence was inversely related to fruit firmness, with firmness values >220 g/mm associated with low disease. In separate experiments, samples from the 0-day cold storage period were inoculated at the stem end with Alternaria alternata, Botrytis cinerea, or Colletotrichum acutatum, and disease incidence was assessed after 7 days in a cold room followed by 60 to 72 h at room temperature. In response to artificial inoculation, less disease developed on crispy berries. No significant effect of harvest method was observed, except for A. alternata inoculation in 2009, when hand-harvested fruit developed a lower level of disease than machine-harvested fruit. Taken together, this study suggests that mechanical harvesting of SHB cultivars with crisp-textured berries is feasible from a postharvest pathology perspective.