Weather based forewarning model for Alternaria leaf spot of safflower (C arthamus tinctorius L.) in scarcity zone of Maharashtra

An experiment was carried out in post rainy ( rabi ) seasons of 2006-07 to 2010-11 for studying the effect of environmental factors and crop phenology on Alternaria leaf spot disease development in safflower under three different sowing conditions viz ., early, normal and late at Zonal Agricultural Research Station, Solapur in Maharashtra, India. The studies revealed that subnormal temperature coupled with above normal humidity and rainfall contributed significantly for the disease incidence and its spread under different sowing situations. Safflower plants were susceptible to Alternaria carthami at all growing stages, but susceptibility increased as the plants matured. Further, the percent disease index (PDI) has progressed at linear rate throughout the plant growth and it was negatively correlated with maximum temperature under late sown condition, while it was positively correlated with rainfall, minimum temperature, relative humidity (morning and evening) in addition to age of crop. By employing step down linear regression models, the incidence of Alternaria leaf spot on safflower can be predicted to an extent of 97.6%, 95.3% and 92.2% accuracy under early, normal and late sowing conditions, respectively while with non-linear models the prediction rate for the leaf spot under above sowing situations was improved to 99.8%, 99.6% and 99.4%, respectively.


MATERIALS AND METHODS
The effect of weather factors like temperature (maximum and minimum), relative humidity (morning and evening in per cent), rainfall (mm) along with one crop factor i.e. age of the crop, on the incidence and development of leaf spot disease were studied at the Zonal Agricultural Research Station, Solapur of Maharashtra, India. The weekly meteorological observations recorded by automatic weather station (AWS-KEPL) at Zonal Agricultural Research Station, Solapur were used for this experiment. The field experiment was conducted for five consecutive years from 2006-07 to 2010-11 in post rainy (rabi) seasons on medium black soil with three sowing dates. The cultivar Phule Kusuma was sown in three plots measuring 500 m 2 each at 15 to 20 days interval i.e. second fortnight of August (early), second fortnight of September (normal) and first/second fortnight of October (late). The crop was fertilized at the rate of 50 kg N and The observations were made on disease severity starting from 20 DAS and till the end of the crop. The rate of increase of disease was calculated based on the average disease gradient (ADG) and percent disease index (PDI) was calculated (Mayee and Datar, 1986). Observations on maximum and minimum temperatures (T max and T min , respectively), relative humidity at morning and evening (RH-I and RH-II, respectively) and rainfall were recorded weekly from sowing to maturity. The averages of the meteorological week (MW) wise weather parameters over the period of experimentation were used in the correlation of the average disease intensity with different weather parameters along with the age of the crop. The data was again subjected to step down regression equations by eliminating the non-significant factors and including only significant factors and linear and non-linear regression equations were developed for disease prediction.

RESULTS AND DISCUSSION
The experimental data (average of five years i.e. from 2006-07 to 2010-11) is presented in Table 1.

Correlation between rate of increase of disease with weather parameters and age of the crop
The rate of disease build up at different stages of crop growth was correlated with weather parameters prevailed during the respective stage and the age of the crop under the three different sowing conditions. The correlation coefficients are presented in Table 2. The results revealed that maximum temperature was positively correlated with the disease development under early and normal sowings with correlation coefficient (r) values of 0.717 and 0.375, respectively. Minimum temperature was significantly positively correlated with the rate of increase of disease under different sowing conditions (r = 0.726, 0.557 and 0.641). The Alternaria leaf spot of safflower mainly depends on the weather. Excessive sporulation of the conidia and favourable incubation temperature has resulted in progressive increase of the disease during the study period. Murumkar et al., (2008) reported that Alternaria leaf spot of safflower had positive correlation with rainfall, temperature and relative humidity under Solapur conditions. Arun Kumar et al., (2011) reported similar results with Alternaria blight of chrysanthemum. Bhaskaran and Kandaswamy, (1980) have reported that prolonged minimum temperature of 20°C is highly congenial for the development of Alternaria leaf blight in sunflower.
The age of the crop played a crucial role in the disease development and spread during the crop growth period. As the age of the plant advanced, the percent of

Disease prediction models
The data was subjected to step down regression by eliminating the non-significant factors and including only significant factors under all sowing conditions for predicting the incidence of Alternaria leaf spot of safflower for three different sowing situations using linear and nonlinear models ( Table 2). Coefficient of determination (R 2 ) improved significantly when non-linear regression models were fitted for predicting the Alternaria leaf spot incidence. By employing step down linear regression models, the incidence of Alternaria leaf spot on safflower can be predicted to an extent of 97.6%, 95.3% and 92.2% accuracy under early, normal and late sowing conditions, respectively. While with non-linear models the prediction rate for the leaf spot under above sowing situations improved to 99.8%, 99.6% and 99.4%, respectively ( Fig.  1A to 3B). Vijayalakshmi et al., (2009) reported that nonlinear regression models are more precise over linear models for prediction of pests and diseases in groundnut. Moreover, Suresh et al., (2012) also reported that among the regression models, non-linear model found to be the more precise for predicting the incidence of Alternaria leaf spot of safflower under southern Telangana zone of Andhra Pradesh. In present study, observed and predicted values from linear and non-linear model of PDI of Alternaria leaf spot of safflower are tested with Chisquare test for goodness of fit. As x 2 Cal is less than x 2 Tab for early, normal and late sowing situations for linear and non-linear model, hence the equations are best fit. The high R 2 values under regression models, particularly in non-linear model confirm the validity of the model in estimating the percent disease index.

CONCLUSION
Forecasting the incidence and spread of leaf spots helps in adopting effective fungicidal spray schedule. The correlation and regression analysis clearly indicated the importance of weather factors and crop age in the prediction of Alternaria leaf spot of safflower. Among the regression models, non-linear model found to be the more precise, compared to linear model, for predicting leaf spot incidence and further spread. Hence, these models can be utilized in Agro-Advisories for Alternaria leaf spot prediction of safflower after validation.