J. ISSAAS Vol. 26, No. 2: 158-169 (2020) 158 DEVELOPMENT OF WEATHER-BASED EMPIRICAL FORECASTING MODELS OF TOMATO LEAF CURL DISEASE IN NORTHERN MINDANAO, PHILIPPINES Bernadith T. Borja 1,2 , Ireneo B. Pangga 2 *, Filomena C. Sta. Cruz 2 and Pompe C. Sta. Cruz 3 1 College of Agriculture, University of Southern Mindanao, Kabacan, North Cotabato, Philippines 2 Institute of Weed Science, Entomology and Plant Pathology, College of Agriculture and Food Science, University of the Philippines Los Baños, Laguna, Philippines 4031 3 Institute of Crop Science, College of Agriculture and Food Science, University of the Philippines Los Baños, Laguna, Philippines 4031 *Corresponding author: ibpangga@up.edu.ph (Received: October 25, 2020; Accepted: November 26, 2020) ABSTRACT Tomato leaf curl is a major virus disease in the Philippines. This study was conducted to develop and validate tomato leaf curl empirical forecasting models in Northern Mindanao. Tomato leaf curl disease incidence (%) and index (%) were assessed and number of whiteflies was counted per plant every two weeks after transplanting in the 2018 dry and wet seasons in field experiments in two sites: Malaybalay, Bukidnon and Claveria, Misamis Oriental. The relationships of disease incidence and index with whitefly number and biweekly-averaged weather variables were analyzed using linear correlation analysis. Disease incidence and index were positively correlated with minimum, maximum and average temperatures, but were positively and negatively correlated with relative humidity (RH), wind speed and rainfall in different sites and seasons. Whitefly number was negatively correlated with disease incidence and index in specific sites and seasons. Using stepwise multiple linear regression analyses, empirical models were developed that included temperature, RH, rainfall, wind speed and whitefly number that explained 90-94% of the variation in disease incidence. Multiple linear regression models specific to site and season were evaluated as good predictive models based on low average prediction error (MDIFF), and narrow length of prediction error (LPE). Key words: correlation analysis, multiple linear regression, whitefly INTRODUCTION Tomato leaf curl is one of the most destructive diseases of tomato affecting many tropical and subtropical regions in the world (Navas-Castillo et al. 2011, Zerbini et al. 2017). It is caused by several viral species of the genus Begomovirus, family Geminiviridae that has a circular, single- stranded DNA genome with two incomplete icosahedral geminate particles (Zerbini et al. 2017). In the Philippines the disease is caused by Tomato leaf curl Philippines virus (ToLCPV), Tomato leaf curl Cebu virus (ToLCCeV) and Tomato leaf curl Mindanao Virus (ToLCMinV) (Kon et al. 2002; Tsai et al. 2011; Brown et al. 2015; Zerbini et al. 2017). The virus is transmitted by an insect vector, whitefly (Bemisia tabaci Genn.), classified in the family Aleyrodidae, in a circulative persistent manner but not transmitted via seed or mechanically (Uchibori et al. 2013). The tomato leaf curl virus affects yield by greatly reducing the number of fruit produced. Severe yield losses which, depending on the age of the plant at the time of infection can reach up to 100% (Levy and Lapidot, 2008). In the Philippines, particularly in Northern Mindanao, most of the farmers cease to plant tomatoes due to leaf curl occurrence in every tomato season that affects the productivity of tomatoes resulting in 40- 60% reduction of crop yield and even higher with very severe infections (Lapoot, personal communication, 2018).