# Source paper archive - Source title: Expansion and Comprehensive Evaluation of the Peacock Eye Disease Management (PedMan) System across Olive Cultivars Differing in Susceptibility to Venturia oleaginea - Author(s): Gilat, Y., Ygzao, D., Shtienberg, D., Ezra, D. - Source landing URL: https://www.biorxiv.org/content/10.64898/2026.11.08.765137v1 - PDF URL: https://www.biorxiv.org/content/10.64898/2026.11.08.765137v1.full.pdf - PDF SHA-256: a5cf9966ad0241a23ae75d6e7f7eff4c4608802b087ce10aa51684cad2e7db91 - Converter: pymupdf4llm 0.0.27 - Limits: Automated PDF text extraction; reading order, equations, tables and formatting may be incomplete or incorrect. Images are omitted, and no OCR was performed. Markup and links are neutralized for safe downloading. Consult the original PDF for the authoritative version. --- [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 1 1 2 **Expansion and Comprehensive Evaluation of the Peacock Eye Disease Management** 3 **(PedMan) System across Olive Cultivars Differing in Susceptibility to** _**Venturia**_ 4 _**oleaginea**_ 5 6 **Yotam Gilat,** **[1,2#]** **David Ygzao,** **[1,2#]** **Dani Shtienberg,** **[1]** **and David Ezra** **[1]** [*] 7 8 1 Department of Plant Pathology and Weed Research, Institute of Plant Protection, 9 Agricultural Research Organization - Volcani Institute, Rishon LeZion 7528809, Israel 10 2 The Robert H. Smith Faculty of Agriculture, Food and Environment, The Hebrew 11 University of Jerusalem, Rehovot 7610001, Israel 12 13 - Corresponding author: DE, dezra@volcani.agri.gov.il 14 15 16 # Authors with equal contribution 17 18 **Keywords** : control, decision support system, host resistance, _Venturia oleaginea_ 19 20 21 22 1 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 2 23 **Abstract** 24 Peacock eye disease, caused by _Venturia oleaginea_, is an important foliar disease of olive that 25 is primarily managed through fungicide applications. Integrating host resistance into disease 26 warning systems could improve spray timing and reduce unnecessary fungicide use. PedMan 27 (Peacock eye disease Manager) was previously developed and validated for highly susceptible 28 olive cultivars. The objectives of the present study were to expand PedMan to cultivars 29 differing in susceptibility to _V. oleaginea_ and to evaluate the expanded system under 30 commercial grove conditions. A sensitivity analysis of the rainfall and temperature criteria used 31 for highly susceptible cultivars was conducted to establish criteria for moderately susceptible 32 and moderately resistant cultivars. The rainfall threshold of ≥15 mm was retained, whereas the 33 temperature range triggering an infection alert was narrowed from 12–18 °C for highly 34 susceptible cultivars to 13–17 °C for moderately susceptible and 14–16 °C for moderately 35 resistant cultivars. The expanded system was subsequently evaluated in 33 field trials and 36 observations under natural infection conditions. PedMan recommendations were classified as 37 correct or incorrect based on statistical comparisons of leaf abscission between fungicide 38 treated and untreated trees. Overall, 30 of 33 recommendations were correct, resulting in an 39 accuracy of 90.9%. Sensitivity was 87.0%, while specificity and positive predictive value were 40 both 100%. These results demonstrate that incorporating cultivar susceptibility into PedMan 41 broadens its applicability and provides reliable cultivar-specific fungicide recommendations. 42 The system has potential to reduce unnecessary fungicide applications while maintaining 43 effective management of peacock eye disease across olive cultivars differing in susceptibility. 44 45 2 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 3 46 47 **1.** **Introduction** 48 Peacock eye disease, also known as olive leaf spot, is one of the most important foliar diseases 49 of olive ( _Olea europaea_ ) and is caused by the fungus _Venturia oleaginea_ (syn. _Spilocaea_ 50 _oleagina_ ). The disease is widely distributed throughout the Mediterranean Basin and other 51 olive-growing regions and can cause substantial reductions in tree productivity (Graniti, 1993; 52 Buonaurio et al., 2023). The disease primarily affects leaves, where characteristic circular, dark 53 lesions surrounded by yellow halos develop on the upper leaf surface. Severe and repeated 54 infections result in premature leaf abscission and extensive defoliation, reducing the 55 photosynthetic capacity and vigor of affected trees and potentially leading to reduced flowering 56 and yield (Graniti, 1993; Buonaurio et al., 2023). Disease development is strongly influenced 57 by environmental conditions, particularly temperature, rainfall, and leaf wetness, which 58 determine conidial release, dispersal, germination, and penetration into the leaf tissue (Graniti, 59 1993; Viruega and Trapero, 1999; Viruega et al., 2011; Obanor et al., 2011). Current disease 60 management practices rely primarily on timely fungicide applications and the use of cultivars 61 with reduced susceptibility to the pathogen. Copper-based fungicides remain the principal 62 means of chemical control and are widely used as protectant treatments, while other fungicide 63 groups, including dodine, strobilurins, and triazoles, have also demonstrated efficacy against 64 the disease (Obanor et al., 2008; Buonaurio et al., 2023). However, effective chemical control 65 depends strongly on application timing relative to infection events, and conventional calendar 66 based spray programs may result in unnecessary applications when environmental conditions 67 are unfavorable for infection. The use of genetically resistant or less susceptible cultivars 68 therefore represents an important component of integrated disease management and can reduce 69 dependence on fungicides (López-Doncel et al., 1999; Buonaurio et al., 2023). Thus, 3 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 4 70 combining cultivar resistance with environmentally based, risk-oriented fungicide applications 71 provides an opportunity to improve disease control while reducing unnecessary fungicide use. 72 Polygenic or quantitative resistance to plant diseases is generally expressed as a continuous 73 range of phenotypic responses rather than as a simple resistant–susceptible dichotomy. Such 74 resistance is often referred to as partial, quantitative, horizontal, or non-race-specific resistance 75 and is typically controlled by several genes, each contributing a portion of the overall resistance 76 (Parlevliet, 1976; Agrios, 2005; Niks et al., 2015). For practical implementations, genotypes 77 exhibiting quantitative variation in disease severity are often classified into discrete response 78 classes, commonly including highly resistant (HR), moderately resistant (MR), moderately 79 susceptible (MS), and highly susceptible (HS). This classification should be regarded as an 80 operational categorization of a continuous resistance phenotype rather than as evidence for four 81 genetically distinct resistance groups. 82 The response of _O. europaea_ cultivars to _V. oleaginea_ is quantitative and ranges from highly 83 susceptible to highly resistant (Graniti, 1993; Buonaurio et al., 2023). To the best of our 84 knowledge, no olive cultivar has yet been reported to be completely immune to the pathogen. 85 Numerous studies conducted in different countries have evaluated the responses of local and 86 international olive cultivars to the pathogen, including MacDonald et al. (2000) in New 87 Zealand; López-Doncel et al. (1999), Moral et al. (2015) and Estudillo et al. (2025) in Spain; 88 Ciccarese et al. (2002) in Italy; Mahfoud et al. (2018) in Syria; Abuamsha et al. (2013) in 89 Palestine; Afiad-Lachger et al. (2020) and Najafi et al. (2023) in Iran; and Bernaschina et al. 90 (2019) in Uruguay. In most cases, the classification of a given cultivar is consistent across 91 studies. For example, cv. Souri was classified as highly susceptible in all studies in which it 92 was evaluated (MacDonald et al., 2000; Mahfoud et al., 2018; Najafi et al., 2023; Ygzao et al., 93 2026a; 2026b)). However, it should be noted that, in some cases, the response of a given 4 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 5 94 cultivar to the pathogen varies among studies and geographical regions. For example, in New 95 Zealand and Uruguay, cv. Barnea was classified as highly or moderately susceptible 96 (MacDonald et al., 2000; Obanor et al., 2011; Bernaschina et al., 2019), whereas in Spain it 97 was categorized as “doubtful” (López-Doncel et al., 1999). In Israel, cv. Barnea has been 98 classified as moderately resistant. It should also be emphasized that, for a given cultivar, 99 disease severity may vary considerably depending on local environmental conditions and 100 inoculum pressure. Thus, a highly susceptible cultivar may exhibit only slight infection under 101 conditions unfavorable for pathogen development or when the initial inoculum level is low. 102 Under such conditions, highly resistant cultivars may remain completely uninfected. 103 Conversely, under environmental conditions highly conducive to pathogen development and 104 in the presence of abundant inoculum, even highly resistant cultivars may develop moderate 105 or, in some cases, severe disease. 106 Our recent studies have provided new insights into the epidemiology and management of 107 peacock eye disease under Israeli conditions. We identified two distinct infection periods 108 during the growing season, occurring in autumn, early winter and in late spring, early summer, 109 and demonstrated that _V. oleaginea_ is effectively monocyclic within a growing season, 110 completing one major disease cycle per season (Ygzao et al., 2026c). Field experiments 111 conducted in commercial olive groves further showed that one or a few fungicide applications 112 in the autumn were sufficient to effectively suppress disease development during the same 113 season (Ygzao et al., 2026b), supporting the epidemiological findings and indicating that 114 fungicide applications can be timed according to infection risk rather than fixed calendar 115 intervals. Based on these findings and our studies of the environmental conditions governing 116 infection (Ygzao et al., 2026a), we developed PedMan (Peacock eye disease Manager), a 117 decision support system (DSS) designed to identify infection events and provide 118 recommendations for optimal fungicide application timing. PedMan integrates rainfall and 5 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 6 119 temperature data to predict infection events, defined as periods favorable for conidial 120 liberation, dispersal, germination, and penetration. For highly susceptible cultivars, PedMan 121 generates an infection alert when rainfall reaches ≥15 mm in a single day or accumulates to 122 ≥15 mm over consecutive days, provided that the average minimum temperature during the 123 rainy days is 12–18 °C (Ygzao et al., 2026a; Ezra et al., 2025). The system can operate using 124 either observed or forecast weather data and recommends fungicide application within 14 days 125 before or after an infection alert, while maintaining a minimum interval of 14 days between 126 consecutive applications and accounting for pre-harvest intervals as harvest approaches. 127 PedMan was evaluated under commercial grove conditions in the highly susceptible cv. Souri 128 and demonstrated reliable performance in predicting disease risk and supporting effective 129 disease management (Gilat et al., 2026). 130 The integration of quantitative cultivar resistance into disease forecasting systems provides a 131 promising approach for adapting fungicide recommendations to the inherent susceptibility of 132 the host. Quantitative resistance can be incorporated into disease forecasting and DSS 133 frameworks in several ways, including adjustment of epidemiological or fungicide-risk 134 thresholds, modification of the predicted rate of disease development, or implementation of 135 cultivar-specific rules for initiating and repeating fungicide applications. These concepts have 136 been demonstrated experimentally in several pathosystems. For example, Abuley and Nielsen 137 (2019) incorporated different levels of cultivar resistance to potato early blight ( _Alternaria_ 138 _solani_ ) into the TOMCAST forecasting model. Their results demonstrated that higher 139 TOMCAST thresholds could be successfully applied to more resistant cultivars, whereas lower 140 thresholds were required for susceptible cultivars. A similar principle has been incorporated 141 into late blight DSSs, in which cultivar resistance is considered together with observed and 142 forecast weather conditions, disease development, and fungicide persistence to determine 143 cultivar-specific spray schedules (Small et al., 2015). In the BlightPro DSS, for example, 6 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 7 144 cultivar resistance was explicitly incorporated into disease simulations and field-based 145 evaluations of fungicide scheduling, resulting in fewer applications for moderately resistant 146 cultivars while maintaining effective disease suppression (Small et al., 2015; Wu et al., 2017). 147 A similar approach is particularly relevant to the management of olive peacock eye disease. 148 Building on these concepts, the PedMan DSS, originally developed and validated for highly 149 susceptible olive cultivars (Ezra et al., 2025; Gilat et al., 2026), can be extended by 150 incorporating cultivar-specific quantitative resistance as an additional component of its 151 decision algorithm. Consequently, highly susceptible cultivars would retain relatively 152 conservative thresholds, whereas moderately susceptible and moderately resistant cultivars 153 would require a greater accumulation of infection risk before a fungicide application is 154 recommended. Such an approach would enable PedMan to integrate pathogen-favorable 155 weather conditions with host susceptibility and thereby provide cultivar-specific spray 156 recommendations, potentially reducing unnecessary fungicide applications while maintaining 157 adequate protection against peacock spot. 158 The long-term goal of our research was to develop effective and environmentally friendly 159 means for managing peacock eye disease of olive. In the present study, we expanded the 160 PedMan system to enable its reliable application across olive cultivars differing in their 161 susceptibility to _V. oleaginea_ . We then evaluated the performance of the expanded system 162 under natural infection conditions in commercial olive groves representing different levels of 163 cultivar susceptibility. Finally, the results from the present study and the parallel report (Gilat 164 et al., 2026) were analyzed together to provide a comprehensive validation of the PedMan 165 system across olive cultivars differing in susceptibility to _V. oleaginea_ . 166 7 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 8 167 **2.** **Materials and Methods** 168 _2.1. Expanding the criteria used by PedMan to cultivars exhibiting different levels of_ 169 _response to V. oleaginea_ 170 The criteria currently used by PedMan to issue an infection alert were developed and evaluated 171 for olive cultivars that are highly susceptible to _V. oleaginea_ (Gilat et al., 2026). The objective 172 of the present study was to adapt these criteria for cultivars that are moderately susceptible or 173 moderately resistant to _V. oleaginea_ by reducing the probability of issuing infection alerts by 174 approximately 25% and 50%, respectively. To achieve this objective, we used weather data 175 recorded at 11 meteorological stations located in the main olive-growing region of Israel during 176 the 2020/1–2023/4 growing seasons (Gilat et al., 2026). These data were used in simulation 177 runs in which PedMan identified the occurrence and timing of infection alerts during the 178 autumn-early winter (September–December) and spring (March–May) infection periods. In the 179 simulations, the rainfall criterion currently used for highly susceptible cultivars (a rainfall event 180 of ≥15 mm within a single day or accumulated over several consecutive days) was increased 181 to either ≥25 or ≥45 mm. Similarly, the temperature criterion (mean daily minimum 182 temperature during rainy days between 12 and 18 °C) was narrowed to either 13–17 °C or to 183 14–16 °C. In a previous study (Ygzao et al., 2026a), we found that 83% of _V. oleaginea_ conidia 184 germinated within the temperature range of 12–18 °C. Restricting the temperature range to 13– 185 17 °C reduced conidial germination to 65.8% (a 20.7% reduction), whereas restricting it to 14– 186 16 °C reduced germination further, to 40.2% (a 51.5% reduction). 187 The number of infection alerts generated for each combination of rainfall threshold and 188 temperature range was compared with the number of alerts generated using the criteria for the 189 highly susceptible cultivars, at each weather station. This analysis enabled us to identify the 190 combinations of rainfall and temperature thresholds expected to reduce the frequency of 8 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 9 191 infection alerts by approximately 25% and 50%, corresponding to cultivars with moderate 192 susceptibility and moderate resistance, respectively. 193 Characterization of the response of the main olive cultivars grown in Israel to _V. oleaginea_ was 194 based on the multi-year experience of leading extension officers (Reuven Birger, Mugira 195 Yunis, and Nizar Abed-Al-Ahadi), results from inoculation experiments using detached leaves, 196 and evaluations conducted during the 2022/3 to 2024/5 growing seasons in the Israeli olive 197 germplasm collection at the Volcani Institute in Rishon LeZion, Israel (Gilat et al., unpublished 198 data). Based on these data, cvs. Souri and Nabali were classified as highly susceptible; cvs. 199 Manzanillo, Coratina, Picual and Arbequina as moderately susceptible; cvs. Picholine, 200 Koroneiki and Barnea as moderately resistant; and cvs. Askal, Mission, Kadesh, Ma'alot and 201 Miilya as highly resistant. As for the highly susceptible cultivars (Ezra et al., 2025), the criteria 202 considered when issuing infection alerts for moderately susceptible cultivars were limited to 203 rainfall quantity and minimum temperature range, as specified below. For moderately resistant 204 cultivars, in addition to these criteria, the time of year (autumn, winter, or spring) and the 205 presence of symptomatic leaves during the previous or current growing season were also 206 considered. According to PedMan, spraying is not recommended for cultivars that are highly 207 resistant to _V. oleaginea_ . 208 209 _2.2. Evaluating the performance of the revised PedMan system under natural conditions in_ 210 _commercial olive groves_ 211 _2.2.1. Experimental locations_ 212 PedMan recommendations were evaluated in two grove experiments during the 2024/5 213 growing season and in 22 commercial grove observations during the 2025/6 growing season. 9 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 10 214 The two experiments were conducted in Kfar Ben Nun (31.859262° N, 34.951702° E), located 215 in the Inner Plains region of Israel. Experiment 1 was established in olive trees of cv. Picual, 216 whereas Experiment 2 was conducted in cv. Manzanillo. The commercial observations were 217 carried out at two locations. The first consisted of a 22.5-ha olive grove near Kibbutz Revadim 218 (31.788228° N, 34.827163° E) in the Inner Plains region. The second comprised three olive 219 plots near Kibbutz Yodfat in the Lower Galilee Mountains: Plot 1 (4.5 ha; 32.781290° N, 220 35.258152° E), Plot 2 (7.5 ha; 32.793788° N, 35.250454° E), and Plot 3 (31.5 ha; 32.788709° 221 N, 35.246459° E). The Revadim grove consisted of 12 subplots (R1–R12), of which six were 222 planted with cv. Coratina, two with cv. Picual, two with cv. Barnea, and two with cv. Askal. 223 Plot 1 at Yodfat consisted of two subplots planted with cv. Coratina (Y1a and Y1b). Plot 2 224 consisted of three Coratina subplots (Y2a–Y2c). Plot 3 comprised six subplots: two planted 225 with cv. Picual (Y16 and Y17), two with cv. Picholine (Y4 and Y5), and two with cv. Barnea 226 (Y10 and Y11). 227 At both locations, the olive trees were more than 30 years old and were planted at spacings of 228 6–7 m within rows and 7 m between rows. Before the initiation of the study, none of the groves 229 had received fungicide treatments against _V. oleaginea_ . During the experiments and 230 observations, irrigation, fertilization, pruning, weed and insect control, and harvesting were 231 performed by the growers according to the recommendations of the Israeli Extension Service. 232 Meteorological data required by PedMan, including daily rainfall and minimum air 233 temperature, were obtained from the Israel Meteorological Service and the Ministry of 234 [Agriculture and Food Security (www.agrometeorology.co.il). Data were collected from the](http://www.agrometeorology.co.il/) 235 Nakhshon, Revadim, and Hinaton meteorological stations, which are located within 2 km of 236 the study sites at Kfar Ben Nun, Revadim, and Yodfat, respectively. 237 10 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 11 238 _2.2.2. Experimental procedures and treatments_ 239 The original objective of the 2024/5 field experiments was to evaluate the positive 240 recommendations of PedMan by applying fungicides according to the system's 241 recommendations. However, during the autumn of the 2024/5 growing season, the minimum 242 temperature during all rainfall events at Kfar Ben Nun remained below 12°C. Consequently, 243 none of the rainfall events met the PedMan criteria for an infection event, and the system did 244 not recommend fungicide application. To evaluate the reliability of PedMan's negative 245 recommendation (i.e., withholding fungicide applications), fungicides were applied following 246 a rainfall event exceeding 15 mm that occurred on 26 December 2024 (Figure 1A), despite the 247 absence of a PedMan recommendation. Accordingly, the experiments included two treatments: 248 (i) **Control**, in which trees were sprayed with fungicides against _V. oleaginea_ despite PedMan 249 recommending no treatment, and (ii) **PedMan**, in which trees were left unsprayed in 250 accordance with the system's recommendation. 11 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 12 251 252 **Figure 1.** Weather conditions recorded at meteorological stations located near the experimental 253 sites during the 2024/5 growing season at Kfar Ben Nun ( **A** ) and near the observation sites at 254 Revadim ( **B** ) and Yodfat ( **C** ) during the 2025/6 growing season. Bars represent daily rainfall. 255 Circles indicate the average minimum temperature during rainfall events exceeding 15 mm. 256 Black filled circles denote rainfall events that met the PedMan criteria for infection in highly 257 susceptible, moderately susceptible, and moderately resistant cultivars. Blue filled circles 258 denote rainfall events that met the PedMan criteria for infection in highly susceptible and 259 moderately susceptible cultivars, whereas red filled circles denote rainfall events that met the 12 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 13 260 PedMan criteria for infection only in highly susceptible cultivars. Open circles indicate rainfall 261 events that did not meet the PedMan criteria for infection for all cultivar response classes. Red 262 dashed lines show the seasonal trend in minimum temperature. Black arrows indicate fungicide 263 applications made according to PedMan recommendations, whereas gray arrows indicate 264 applications made contrary to PedMan recommendations. 265 266 The experiments were arranged in a randomized complete block design with two treatments, 267 four replicates per treatment, and one tree per replicate. The fungicides used were Skipper 268 (difenoconazole, 250 g L⁻¹ EC; Tapazol Chemical Works Ltd., Beit Shemesh, Israel) and 269 Nechostan (190 g L⁻¹ copper as 340 g L⁻¹ tribasic copper sulfate; Nufarm, distributed in Israel 270 by Adama-Agan, Ashdod, Israel). Both fungicides are registered in Israel for the control of 271 peacock eye and are widely used in commercial olive production. They were applied as a tank 272 mixture at concentrations of 0.06% (Skipper) and 0.3% (Nechostan). Fungicides were applied 273 with a handgun sprayer using 15 L of spray solution per tree. 274 275 _2.2.3. Observation procedures and treatments_ 276 As in the field experiments, each of the 22 commercial grove observations consisted of two 277 treatments: **Control** and **PedMan** . The assignment of treatments depended on the 278 recommendation generated by PedMan (Table 1). When PedMan recommended fungicide 279 application (positive recommendation), trees assigned to the PedMan treatment were sprayed, 280 whereas those assigned to the control treatment remained unsprayed to evaluate the benefit of 281 the recommended application. Conversely, when PedMan recommended that no fungicide 282 should be applied (negative recommendation), trees assigned to the PedMan treatment were 13 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 14 283 left unsprayed, whereas those assigned to the control treatment were sprayed to evaluate the 284 validity of withholding the application. 285 At the Revadim site, the grower sprayed one row (30–40 trees) in each of the 12 subplots on 286 19 November 2025 (Figure 1B). Depending on the PedMan recommendation for each subplot, 287 the sprayed row served either as the PedMan treatment (when spraying was recommended) or 288 as the control treatment (when no spray was recommended). The remaining unsprayed trees 289 constituted the corresponding comparison treatment. At the Yodfat site, PedMan recommended 290 fungicide application in the Coratina and Picual subplots. Accordingly, three rows of trees in 291 each subplot were left unsprayed as the control treatment, whereas the remaining trees were 292 sprayed according to PedMan recommendations. In contrast, PedMan did not recommend 293 fungicide application in the Picholine and Barnea subplots. Therefore, three rows of trees in 294 each subplot were sprayed contrary to the PedMan recommendation (control treatment), 295 whereas the remaining trees were left unsprayed (PedMan treatment). Fungicide applications 296 at both sites were made on 19 November 2025 (Figure 1B and C). In all observations, growers 297 applied a tank mixture of Skipper (0.06%) and Nechostan (0.3%) using a Degania axial fan 298 sprayer at a spray volume of 2,000 L ha⁻¹. 299 300 301 302 303 304 14 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 15 305 **Table 1** . Details about the observations carried out in the 2025/6 growing season in 306 commercial olive groves to evaluating the performance of PedMan recommendations in 307 cultivars with dissimilar responses to _V. oleaginea_ . Cultivar Subplot Previous PedMan Treatment [z] Location Name response [y] code infection Recomm. Control PedMan code infection Recomm. Revadim Coratina MS R1 R3 R6 R8 R10 R11 Picual MS R4 R7 Barnea MR R2 R5 Askal HR R9 R12 Yodfat Coratina MS Y1a Y1b Y2a Y2b Y2c Yes Yes Yes Yes Yes Yes Yes Yes No No Yes Yes No No No No No Sprayed Sprayed Sprayed Sprayed Sprayed Sprayed Sprayed Sprayed Not sprayed Not sprayed Not sprayed Not sprayed Sprayed Sprayed Sprayed Sprayed Sprayed To spray To spray To spray To spray To spray To spray To spray To spray Not to spray Not to spray Not to spray Not to spray To spray To spray To spray To spray To spray Not sprayed Not sprayed Not sprayed Not sprayed Not sprayed Not sprayed Not sprayed Not sprayed Sprayed Sprayed Sprayed Sprayed Not sprayed Not sprayed Not sprayed Not sprayed Not sprayed Picual MS Y16-17 Yes To spray Not sprayed Sprayed Picholine MR Y4 Y5 Barnea MR Y10 Y11 No No No No Not sprayed Not sprayed Not sprayed Not sprayed Not to spray Not to spray Not to spray Not to spray Sprayed Sprayed Sprayed Sprayed 15 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 16 308 309 y Cultivar resistance class: MS = moderately susceptible; MR = moderately resistant; HR = 310 highly resistant. 311 z in cases where PedMan recommendation was 'not to spray' the sprays were applied in the 312 control treatment to evaluate the negative accuracy of PedMan (applying sprays in contrast to 313 system's recommendation). 314 315 _2.2.4. Evaluation of PedMan recommendations in the observations_ 316 The impact of peacock eye disease on tree health was assessed by estimating leaf abscission. 317 Leaf abscission was selected as the response variable because it integrates the cumulative effect 318 of repeated infections on tree health and is the principal cause of yield reduction associated 319 with peacock eye disease (Graniti, 1993; Buonaurio et al., 2023). Leaf abscission (expressed 320 as %) was estimated as previously described (Ygzao et al. 2026c; Gilat et al., 2026). Briefly, 321 each tree canopy was visually assessed from four directions, and the percentage of leaf 322 abscission was estimated for each orientation. A score of 0% indicated a full, healthy canopy 323 with no visible leaf loss; A score of 50% indicated that approximately half of the leaves had 324 abscised and the canopy appeared noticeably transparent, and A score of 100% indicated 325 complete defoliation. The four directional estimates were averaged to obtain a single value for 326 each tree. In the field experiments, leaf abscission was assessed for each experimental tree (one 327 tree per replicate) every 2-3 weeks starting at early January and ending at late May, 2026. In 328 the commercial grove observations, 20 trees were evaluated for each treatment within each 329 subplot. Assessments were conducted in April or May, corresponding to the peak of the second 330 annual peacock eye disease episode (Ygzao et al., 2026c). 16 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 17 331 Leaf abscission in olive trees may also result from other biotic and abiotic factors, including 332 other foliar pathogens, insect pests, drought, waterlogging, salinity, nutrient imbalance, and 333 temperature extremes (Martin and Sibbett, 2005). Nevertheless, the differences in leaf 334 abscission observed between fungicide-treated and untreated trees in the present study were 335 attributed primarily to peacock eye disease, as all other cultural practices were identical 336 between treatments within each experiment or observation. For each experiment and 337 observation, control efficacy (CE, %) was calculated by comparing the incidence of leaf 338 abscission in the PedMan treatment ( _LA_ _t_ ) with the incidence of leaf abscission in the control 339 treatment ( _LA_ _c_ ) according to the following formula: 340 𝐶𝐸 (%) = 100 −[𝐿𝐴𝑐 [𝐿𝐴𝑡][] × 100] 341 342 _2.3. Data analyses_ 343 In accordance with the two types of operational decisions generated by PedMan, the data were 344 analyzed separately for cases in which fungicide application was recommended (spray 345 recommendations; cvs. Coratina and Picual) and cases in which fungicide application was not 346 recommended (no-spray recommendations; cvs. Manzanillo, Barnea, Picholine, and Askel). 347 The field observations were analyzed using two complementary approaches. First, differences 348 in the incidence of leaf abscission between treated and untreated trees were assessed using a 349 two-sided Student’s _t_ -test assuming equal variances. Second, to account for potential spatial 350 pseudo replication among trees within rows and the unbalanced structure of the observations, 351 a linear mixed-effects model (LMM) was fitted using restricted maximum likelihood (REML) 352 estimation. The LMM incorporated treatment as a fixed effect and row as a random effect, 353 thereby accounting for the non-independence of observations within rows. Subplot was 354 incorporated as a random intercept effect to serve as a spatial block factor. Fixed effects in both 17 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 18 355 models comprised 'location', 'cultivar' and 'treatment', and the 'treatment' × 'cultivar' interaction 356 term. Statistical analyses were executed using JMP software (Version 16; SAS Institute, Cary, 357 NC). The linear mixed-effects model for each independent validation phase was expressed 358 using the following algebraic equation: 359 𝑌 𝑖𝑗𝑘 = 𝜇+ 𝛼 𝑖 + 𝛽 𝑗 + (𝛼𝛽) 𝑖𝑗 + 𝛾 𝑘 + 𝜂 𝑙 + 𝜀 𝑖𝑗𝑘 360 361 Where: 362 - 𝑌 𝑖𝑗𝑘 represents the observed incidence of leaf abscission (%) quantified at the row 363 level. 364 - 𝜇 is the overall experimental mean baseline intercept. 365 - 𝛼 𝑖 is the fixed effect of the 𝑖 -th Treatment (where 𝑖 = PedMan recommendation vs. 366 Control). 367 - 𝛽 𝑗 is the fixed effect of the 𝑗 -th Cultivar. 368 `o` For the positive accuracy phase: 𝑗 = Coratina, Picual. 369 `o` For the negative accuracy phase: 𝑗 = Barnea, Fishulin, Askal. 370 - (𝛼𝛽) 𝑖𝑗 is the fixed interaction effect between the 𝑖 -th Treatment and the 𝑗 -th 371 Cultivar, evaluating the variance in treatment efficacy across different cultivars. 372 - 𝛾 𝑘 is the fixed environmental covariate effect of the 𝑘 -th location 𝑘 = (Revadim, 373 Yodfat) to account for regional macroclimatic baseline variation. 374 - 𝜂 𝑙 is the random intercept effect of the 𝑙 -th treatment, acting as a spatial blocking 375 factor to pair the treatment and control rows within each subplot. Normality 18 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 19 376 assumptions were tested and verified to be independently distributed: 𝜂 𝑙 377 ~ 𝑁 (0, 𝜎 𝑝𝑙𝑜𝑡2 ). 378 - 𝜀 𝑖𝑗𝑘 is the leftover residual random error term representing within-plot row variance, 379 Normality assumption was tested and verified to be independently distributed: 2 380 𝜀 𝑖𝑗𝑘 ~ 𝑁 (0, 𝜎 𝑟𝑒𝑠𝑖𝑑𝑢𝑎𝑙 ) 381 382 383 _2.4. Evaluation of PedMan recommendations across olive cultivars differing in susceptibility_ 384 _to_ V. oleaginea 385 PedMan recommendations were implemented in 33 field experiments and observations in total. 386 Of which, 9 were presented in the parallel report (Gilat at al., 2026) and 24 in the current report. 387 The performance of the system was evaluated using a binary classification framework based 388 on the statistical evidence obtained from each field trial or observation. The 33 evaluations 389 were classified according to whether the experimental results supported the corresponding 390 PedMan recommendation. When PedMan recommended fungicide application, the 391 recommendation was considered as correct when a statistically significant difference in leaf 392 abscission was detected between trees receiving the fungicide treatment recommended by 393 PedMan and untreated control trees. Conversely, when PedMan recommended no fungicide 394 application, the recommendation was considered as correct when no statistically significant 395 difference in leaf abscission was detected between trees left untreated according to the PedMan 396 recommendation and trees that received fungicide despite the no-spray recommendation. Thus, 397 a statistically significant treatment effect supported a spray recommendation, whereas the 398 absence of a statistically significant treatment effect supported a no-spray recommendation. 19 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 20 399 The resulting decisions were summarized in a contingency table, where the recommendation 400 to spray defined as the positive class and withholding spray as the negative class. Outcomes 401 were partitioned into true positive (TP), true negative (TN), false positive (FP), and false 402 negative (FN). Because classification accuracy alone is insufficient for evaluating performance 403 on potentially imbalanced binary tasks (Altman et al., 1994; Foody, 2023; Chicco and Jurman, 404 2023), we jointly evaluated overall accuracy, sensitivity, specificity, positive predictive value 405 (PPV) and negative predictive value (NPV). Overall accuracy represents the proportion of all 406 predictions that were correct, including both correct positive and correct negative predictions; 407 sensitivity represents the proportion of evaluations in which the statistical analysis supported 408 fungicide application and PedMan correctly recommended spraying; specificity represents the 409 proportion of evaluations in which the statistical analysis did not support fungicide application 410 and PedMan correctly recommended no spraying. PPV and NPV were considered jointly 411 because they characterize different aspects of binary classification performance and accuracy 412 alone does not fully describe the performance of a classification system (Altman et al., 1994; 413 Foody, 2023; Chicco and Jurman, 2023). These performance metrics were calculated as 414 follows: TP + TN TP 415 Overall Accuracy (%) = TP + TN + FP + FN [× 100; ][ Sensitivity (%)][ =] TP + FN [× 100] 416 TN TP TN 418 Specificity (%) = TN + FP [× 100; ][PPV (%)][ =] TP + FP [× 100;] [ NPV (%)][ =] TN + FN [× 100 ] 417 419 As complementary measures of overall classification performance, balanced accuracy, the F1 420 score, and the Matthews correlation coefficient (MCC) were calculated from the contingency 421 table. Balanced accuracy quantifies the average ability of PedMan to correctly classify both 20 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 21 422 disease-risk and non-disease-risk cases, giving equal weight to the sensitivity and specificity 423 of the predictions. F1 score quantifies the balance between PedMan’s precision and sensitivity, 424 reflecting its ability to correctly identify disease-risk cases while minimizing both missed cases 425 and false-positive predictions. MCC quantifies the overall agreement between PedMan 426 predictions and the observed disease-management outcomes, accounting for both correct and 427 incorrect positive and negative classifications. It was included because it incorporates all four 428 cells of the contingency table and provides a single summary measure of agreement between 429 the predicted and reference classifications (Chicco and Jurman, 2023). Because of the relatively 430 small validation dataset and the occurrence of observed proportions equal to 100%, two-sided 431 95% confidence intervals for sensitivity, specificity, PPV, and NPV were calculated using 432 exact binomial confidence limits. The 95% confidence interval for overall accuracy was 433 calculated using the Wilson score method. All confidence intervals were two-sided and 434 calculated at the 95% confidence level. Balanced Accuracy (%) = [Sensitivity][ +][ Specificity] [ Specificity] 2 × 100; 𝐹 1 = 2 × Precision [Precision] + [ ×][ Recall] Recall 435 Balanced Accuracy (%) = × 100; 𝐹 1 = 2 × 2 Precision + Recall [, ] 437 438 436 439 **3.** **Results** (TP × TN) −(FP × FN) MCC = √(TP + FP)(TP + FN)(TN + FP)(TN + FN) 440 _3.1. Expanding the criteria used by PedMan to cultivars exhibiting different levels of_ 441 _response to V. oleaginea_ 442 Data collected over four growing seasons from 11 meteorological stations representing all 443 olive-growing regions in Israel were used to perform simulation runs aimed at defining rainfall 444 and temperature criteria for cultivars with different levels of susceptibility to _V. oleaginea_ . 21 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 22 445 First, only the temperature criterion was calibrated; Narrowing the thermal window used by 446 the system for the highly susceptible cultivars (12–18 °C) to 13–17 °C reduced the mean 447 number of infection alerts issued by the system by 25.1%, whereas further narrowing it to 14– 448 16 °C, resulted in a 51.3% reduction (Figure 2A and B). Next, only the rainfall threshold was 449 modified. Increasing the rainfall threshold used for highly susceptible cultivars (>15 mm) to 450 >25 or >45 mm, reduced the mean number of infection alerts by 34.9% and 55.1%, respectively 451 (Figure 2C and D). Finally, both the temperature and rainfall criteria were modified 452 simultaneously. Combining a rainfall threshold of >25 mm with a temperature range of 13–17 453 °C reduced the mean number of infection alerts by 54.0%, whereas combining a rainfall 454 threshold of >45 mm with a temperature range of 14–16°C reduced the number of alerts by 455 84.1% (Figure 2E and F). 456 Because our objective was to reduce the probability of issuing infection alerts by approximately 457 25% and 50% for moderately susceptible and moderately resistant cultivars, respectively, only 458 the temperature criterion was modified. Accordingly, a temperature range of 13–17°C was 459 adopted as the threshold for issuing infection alerts for moderately susceptible cultivars, 460 whereas a narrower range of 14–16°C was adopted for moderately resistant cultivars. 22 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 23 461 462 **Figure 2.** Sensitivity analysis of the thresholds used by PedMan to generate infection alerts 463 (IA). The mean number of infection alerts issued during autumn (black squares) and spring 464 (gray circles) over four growing seasons (2020/1–2023/4) at 11 weather stations was calculated 465 using the standard PedMan thresholds (R >15 /T 12–18 ) and compared with the mean number of 466 alerts generated using different rain and temperature definitions. 467 468 _3.2. Evaluating the performance of the revised PedMan system under natural conditions in_ 469 _commercial olive groves_ 470 471 _3.2.1. Evaluating the positive predictive performance of PedMan recommendations_ 472 The positive predictive performance of PedMan recommendations was evaluated during the 473 2025/6 growing season in 14 observations carried out in commercial olive groves planted with 474 cultivars Coratina and Picual. Comparing the differences between the treatments in each 23 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 24 475 observation revealed that following PedMan recommendation to spray reduced leaf abscission 476 significantly ( _P_ < 0.05) in all cases (Table 2). The entire data set was then analyzed using the 477 Linear Mixed Model. The fixed effects ('location', 'cultivar' and 'treatment') explained 80.7% 478 of the total defoliation variance (Marginal _R_ _[2]_ = 0.807), while the total model (accounting for 479 the random subblock effect) profiles explained 96.7% of the variance (Conditional _R_ _[2]_ = 0.967). 480 The random spatial subplot effect accounted for 72.7% of the remaining unresolved variance 481 components (Table 3), confirming strong environmental spatial dependency among trees 482 within the same grove subplot. 483 'Treatment' exerted strongest influence on leaf abscission ( _F_ = 242.49; _P_ < 0.001), while 484 'cultivar' also significantly affected leaf abscission ( _F_ = 6.10; _P_ = 0.031). In addition, a 485 significant 'cultivar' × 'treatment' interaction ( _F_ = 15.20; _P_ = 0.002) indicated that the 486 magnitude of the treatment effect differed between cultivars. Because the 'cultivar' × 'treatment' 487 interaction was significant, treatment effects were subsequently sliced by cultivar _via_ post-hoc 488 Least-Squares means contrasts for each cultivar. For cv. Coratina, PedMan-guided spraying 489 significantly suppressed premature leaf abscission ( _P_ < 0.0001), reducing leaf abscission from 490 18.0 to 9.4%, a 47.8% reduction. For cv. Picual, PedMan-guided spraying reduced mean leaf 491 abscission from 25.6 to 11.5%, corresponding to a 56.6% improvement ( _P_ < 0.001). These field 492 observations therefore confirmed the positive predictive performance of PedMan spray 493 recommendations under commercial grove conditions. 494 495 496 497 24 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 25 498 **Table 2** . Evaluating the positive predictive performance of PedMan recommendations 499 (applying sprays in accordance to system's recommendation) in cultivars that were classified 500 as moderately susceptible to _V. oleaginea_ 501 Cultivar Subplot Incidence of leaf abscission (%) [x] Relative Control PedMan Name Resistance class [w] code improvement (%) Coratina MS R1 21.2 ± 1.40 13.8 ± 1.58 34.9*** [y] R3 15.4 ± 1.18 9.2 ± 0.87 40.3*** R6 17.6 ±1.87 10.3 ± 1.39 41.5** R8 17.0 ± 1.69 11.3 ± 1.35 33.5** R10 18.4 ± 1.32 11.0 ± 1.04 45.7** R11 20.7 ± 1.79 10.2 ± 1.18 50.7*** Y1a 13.6 ± 1.38 7.2 ± 1.11 47.1*** Y1b 16.7 ± 2.10 4.4 ± 0.60 73.7*** Y2a 20.0 ± 1.58 7.6 ± 1.04 62.0*** Y2b 20.0 ± 1.74 12.5 ± 1.23 37.5** Y2c 17.1 ± 2.13 7.0 ± 2.00 59.1*** Mean 18.0 ± 0.71 a [z] 9.4 ± 0.82 b 47.8 ± 3.79 Picual MS R4 22.0 ± 1.56 8.5 ± 0.68 61.4*** R7 31.7 ± 2.29 17.7 ± 1.92 44.2*** Y16-17 23.2 ± 2.47 8.3 ± 1.21 64.2*** Mean 25.6 ± 3.05 a 11.5 ± 3.10 b 56.6 ± 6.26 25 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 26 502 503 w Cultivar resistance class: MS = moderately susceptible. 504 x Numbers represent the incidence of leaf abscission (± SE) as determined at the peak of the 505 second disease episode. 506 Y Statistical significance of differences between treatments was determined using Student’s _t_ 507 test: ** _P_ < 0.01; *** _P_ < 0.001. 508 z Means within each cultivar followed by different letters differ significantly according to 509 Tukey's HSD test ( _P_ < 0.0001) as determined by the multiple comparisons for the effects of 510 the interaction 'cultivar' × 'treatment', Table 3. 511 512 **Table 3.** Linear mixed-model analysis of factors affecting leaf abscission in the observation 513 carried out for evaluating the positive predictive performance of PedMan recommendations 514 (applying sprays in accordance to system's recommendation) in cultivars that were classified 515 as moderately susceptible to _V. oleaginea_ **A.** Random effects covariance parameter estimates Variance component Estimate Std error CI 95% Pct of total Subplot 6.602 3.382 -0.028 to 13.23 72.66 Residual 2.483 1.014 1.27 to 6.76 27.33 Total 9.085 3.382 4.91 to 22.22 100.00 **B.** Fixed effects tests 26 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 27 Source Numerator df F ratio Prob > F Location 1 2.681 0.129 Cultivar 1 6.100 0.031 Treatment 1 242.49 <0.001 Cultivar × treatment 1 15.20 0.002 516 517 _3.2.2. Evaluating the negative predictive performance of PedMan recommendations_ 518 The negative predictive performance of PedMan recommendations was evaluated in two grove 519 experiments conducted during the 2024/5 growing season and in eight field observations 520 carried out during the 2025/6 growing season in commercial olive groves planted with the 521 cultivars Picual, Manzanillo, Barnea, Picholine and Askal. In the grove experiment conducted 522 with cv. Picual, withholding fungicide applications according to the PedMan recommendation 523 had no adverse effect on leaf abscission throughout the growing season. Differences in leaf 524 abscission between the two treatments remained insignificant throughout the season. 525 Furthermore, the incidence of leaf abscission in the last assessment date in both treatments did 526 not differ significantly ( _P_ - 0.05) from the incidence in the first assessment (Figure 3A). These 527 results indicate that fungicide application was unnecessary and confirming the negative 528 predictive performance of the PedMan recommendation. In contrast, in the grove experiment 529 conducted with cv. Manzanillo, trees sprayed despite the PedMan recommendation not to spray 530 exhibited significantly lower leaf abscission during April and May, corresponding to the peak 531 of the second disease episode. Similarly, the incidence of leaf abscission in the last two 532 assessment dates in the PedMan treatment differ significantly ( _P_ < 0.05) from the incidence in 533 the first assessment. Differences in the incidence of leaf abscission between these date for the 534 control treatment were insignificant (Figure 3B). These results indicate that, under these 535 experimental conditions, the PedMan recommendation not to spray was incorrect. 27 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 28 536 **Figure 3.** Seasonal progress of leaf abscission in two grove experiments conducted during the 537 2024/25 growing season to evaluate the negative predictive performance of PedMan 538 recommendations. In the control treatment, fungicides were applied despite the PedMan 539 recommendation not to spray, whereas no fungicides were applied in the PedMan treatment. 540 ( **A** ) cv. Picual; ( **B** ) cv. Manzanillo. Vertical bars represent the standard error of the mean. 541 Asterisks indicate significant differences between treatments at the corresponding sampling 542 date ( _P_ < 0.05). 543 544 The negative predictive performance of PedMan recommendations was further evaluated under 545 commercial grove conditions using eight independent field observations. Comparing the 546 differences between the treatments in each observation revealed that following PedMan 547 recommendation not to spray did not affect leaf abscission significantly ( _P_ < 0.05) for cultivars 548 Barnea and Picholine, but the differences between the treatments were significant in the two 549 observations of cv. Askal (Table 4). When the entire data set was analyzed using the Linear 550 Mixed Model, comparing the differences between the treatments in each observation revealed 551 that following PedMan recommendation to spray reduced leaf abscission significantly ( _P_ < 552 0.01) in all cases. The fixed effects of the Linear Mixed Model ('location', 'cultivar' and 553 'treatment') explained 87.6% while the total model (accounting for the random subplot) profiles 28 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 29 554 captured 98.3% of the variance. The random spatial subplot effect accounted for 79.5% of the 555 remaining unresolved variance components, solidifying the equivalent result in the positive 556 recommendations (Table 5). 557 'Cultivar' significantly affected premature leaf abscission ( _F_ = 15.09; _P_ = 0.013). In contrary, 558 neither 'location' ( _P_ = 0.0645) nor 'treatment' ( _P_ = 0.2315) had a significant overall effect. 559 However, the significant 'cultivar' × 'treatment' interaction ( _F_ = 7.97; _P_ = 0.027) indicated that 560 host responsiveness to withheld versus applied fungicide sprays varied across different 561 cultivars. Therefore, treatment effects were evaluated separately for each cultivar. For all three 562 cultivars evaluated (Barnea, Picholine, and Askal), applying fungicides despite the PedMan 563 recommendation not to spray did not significantly reduce leaf abscission ( _P_ - 0.05). Mean leaf 564 abscission remained statistically similar between treated and untreated plots in all cultivars 565 (Table 5). These results demonstrate that fungicide application contrary to the PedMan 566 recommendation provided no measurable benefit, thereby confirming the negative predictive 567 performance of the PedMan decision-support system under commercial grove conditions. 568 569 **Table 4** . Evaluating the negative predictive performance of PedMan (applying sprays in 570 contrast to system's recommendation) in cultivars that were classified as moderately of highly 571 resistant to _V. oleaginea_ 572 Cultivar Subplot Incidence of leaf abscission (%) [x] Relative Name Resistance code Control PedMan improvement (%) code Control PedMan class [w] 29 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 30 Barnea MR R9 11.8 ± 1.19 12.9 ± 1.45 -9.3 [NS] [y] R12 13.0 ± 1.60 11.2 ± 1.08 13.8 [ NS] Y10 18.0 ±1.38 21.0 ± 1.43 -16.7 [ NS] Y11 16.7 ± 1.42 17.3 ± 1.97 -3.6 [ NS] Mean 15.6 ± 2.21 a [z] 14.8 ± 1.47 a -3.92 ± 6.50 Picholine MR Y4 5.2 ± 1.32 4.0 ± 0.69 23.1 [ NS] Y5 6.6 ± 1.15 5.6 ± 1.31 15.2 [ NS] Mean 5.9 ± 0.70 a [z] 4.8 ± 1.13 a 19.11 ± 3.96 Askal HR R2 9.0 ± 0.97 12.0 ± 0.99 -33.3* R5 13.7 ± 1.04 19.3 ± 1.37 -40.9** Mean 11.3 ± 2.35 a 15.6 ± 3.65 a -37.10 ± 3.77 573 574 w Cultivar resistance class: MR = moderately resistant, HR = highly resistant. 575 x Numbers represent the incidence of leaf abscission (± SE) as determined at the peak of the 576 second disease episode. 577 Y Statistical significance of differences between treatments was determined using Student’s _t_ 578 test: NS = Not Significant; * _P_ < 0.05; ** _P_ < 0.01. 579 z Means within each cultivar followed by different letters differ significantly according to 580 Tukey's HSD test ( _P_ < 0.0001) as determined by the multiple comparisons for the effects of 581 the interaction 'cultivar' × 'treatment', Table 5. 582 30 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 31 583 584 **Table 5.** Random and fixed effects of the mixed model used for estimating the parameters 585 affecting leaf abscission in the observation carried out for evaluating the negative predictive 586 performance of PedMan recommendations (applying sprays in contrast to system's 587 recommendation) in cultivars that were classified as moderately susceptible to _V. oleaginea_ A. Random effects covariance parameter estimates Variance component Estimate Std error CI 95% Pct of total Subplot 4.930 3.955 -2.82 to 12.68 79.53 Residual 1.286 0.802 0.49 to 7.63 20.46 Total 6.199 3.955 2.40 to 38.12 100.00 B. Fixed effects tests Source Numerator df F ratio Prob > F Location 1 6.412 0.064 Cultivar 2 15.085 0.014 Treatment 1 1.853 0.231 Cultivar × treatment 2 7.968 0.028 588 589 _3.3. Validation of PedMan recommendations across olive cultivars differing in susceptibility_ 590 _to_ V. oleaginea 591 Across the 33 field trials and observations, PedMan provided 30 correct recommendations and 592 three incorrect ones (Table 6), corresponding to an overall accuracy of 90.9% (95% CI: 76.8– 31 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 32 593 96.9%). Of the 23 evaluations in which the statistical analysis supported fungicide application, 594 PedMan correctly recommended spraying in 20 cases, resulting in a sensitivity of 87.0% (95% 595 exact CI: 66.7–97.2%). In all 10 evaluations in which the statistical analysis did not support 596 fungicide application, PedMan correctly recommended no spraying, resulting in a specificity 597 of 100% (95% exact CI: 69.2–100.0%). Accordingly, all 20 spray recommendations were 598 supported by the corresponding statistical analyses, resulting in a PPV of 100% (95% exact CI: 599 83.2–100.0%). Among the 13 no-spray recommendations, 10 were supported by the statistical 600 analysis, resulting in a NPV of 76.9% (95% exact CI: 46.2–93.8%). The balanced accuracy 601 was 93.5%, the F1 score was 93.0%, and the MCC was 0.818. All three incorrect 602 recommendations were false-negative decisions, in which PedMan recommended withholding 603 fungicide application although a statistically significant difference in leaf abscission was 604 detected between untreated and fungicide-treated trees. 605 606 **Table 6.** Contingency table summarizing the validation of PedMan spray recommendations 607 based on field trial and observation outcomes [y] Field evaluation [z] PedMan recommendation Spraying Spraying not Total warranted warranted Spray 20 (TP) 0 (FP) 20 No spray 3 (FN) 10 (TN) 13 Total 23 10 33 608 609 Y Results are from the current (Tables 2 and 4; Figure 3) and the parallel (Gilat et al., 2026) 610 studies 32 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 33 611 z TP = true positive; FP = false positive; FN = false negative; TN = true negative 612 613 **4.** **Discussion** 614 PedMan was initially developed for olive cultivars highly susceptible to _V. oleaginea_ (Ezra et 615 al., 2025; Gilat et al., 2026). In the present study, we extended the system to cultivars differing 616 in susceptibility by adapting the infection-alert criteria according to host resistance. This 617 approach was based on the strategy of Shtienberg et al. (2000), who adjusted rainfall thresholds 618 for chickpea cultivars differing in resistance to _Didymella rabiei_ . Because infection by both 619 pathogens is closely associated with rainfall, we evaluated whether the rainfall and temperature 620 criteria established for highly susceptible olive cultivars could be modified according to 621 cultivar susceptibility. Sensitivity analysis indicated that the rainfall threshold of ≥15 mm could 622 be retained across susceptibility classes, whereas the temperature range could be progressively 623 narrowed from 12–18°C for highly susceptible cultivars to 13–17°C for moderately susceptible 624 and 14–16°C for moderately resistant cultivars (Figure 2). Thus, the expansion of PedMan was 625 achieved by maintaining a common rainfall threshold while progressively restricting the 626 temperature conditions under which an infection alert is generated. 627 The evaluation results support the effectiveness of this approach. PedMan correctly identified 628 all 20 cases in which fungicide application was warranted and correctly recommended no spray 629 in 10 of the 13 cases in which fungicide application was not warranted (Table 6). In highly 630 susceptible cultivars, fungicide application according to PedMan recommendations reduced 631 leaf abscission by 60.8 ± 4.1% (mean ± SE; Gilat et al., 2026), whereas the reduction was 47.9 632 ± 3.3% in moderately susceptible cultivars (Table 2). In cv. Askal, which was classified as 633 highly resistant, the erroneous no-spray recommendation resulted in a 37.1 ± 3.8% reduction 634 in leaf abscission following fungicide application (Table 4). Had cv. Askal been classified as 33 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 34 635 moderately rather than highly resistant, PedMan would have recommended fungicide 636 application, resulting in a correct recommendation and an effect consistent with those observed 637 for the other two resistance classes. This finding highlights the importance of accurately 638 characterizing cultivar responses to _V. oleaginea_ and suggests that refinement of resistance 639 classifications could further improve PedMan performance. 640 Across all 33 field trials and observations, PedMan correctly classified 30 cases with 90.9% 641 overall accuracy, 87.0% sensitivity, 100% specificity, 100% PPV, 76.9% negative predictive 642 value, 93.5% balanced accuracy, an F1 score of 93.0%, and an MCC of 0.818. The absence of 643 false-positive spray recommendations is particularly important for sustainable disease 644 management, as all 20 spray recommendations were supported by a significant reduction in 645 leaf abscission. Conversely, the three false-negative decisions represent the principal limitation 646 of the system because they indicate missed opportunities for disease control. Given the 647 relatively small validation dataset, the confidence intervals around sensitivity and NPV should 648 be considered when extrapolating these estimates to other cultivars, environments, seasons, 649 and disease-pressure levels. 650 Despite their potential, the practical application of forecasting models and DSSs for peacock 651 eye disease remains limited, particularly because of the scarcity of independent field evaluation 652 (Thomidis et al., 2021). Consequently, the performance of PedMan could only be compared 653 with prediction models and DSSs developed in other pathosystems. Weather-based models for 654 hop downy mildew achieved 83.3–87.5% sensitivity and 84.4–88.8% specificity (Gent and 655 Ocamb, 2009), while the Nærstad potato late blight model achieved 83% accuracy, 89% 656 sensitivity, and 62% specificity (Hjelkrem et al., 2021). Fusarium head blight models achieved 657 up to 85% accuracy and 100% sensitivity, with 78.4% specificity (Giroux et al., 2016), whereas 658 wheat leaf-blotch DSSs showed substantial interannual variation, with 54–95% correct 34 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 35 659 recommendations depending on year and model (Jørgensen et al., 2020). Compared with these 660 systems, PedMan showed comparable sensitivity but particularly high specificity, indicating a 661 strong ability to avoid unnecessary fungicide applications. This feature is especially relevant 662 because the practical value of a DSS depends not only on predictive accuracy but also on its 663 ability to reduce fungicide use without compromising disease control. Field evaluations of 664 PLANT-Plus and BlightPro demonstrated that DSS-guided programs can reduce or optimize 665 fungicide applications while maintaining disease control and yield (van der Waals et al., 2003; 666 Small et al., 2015), and a meta-analysis of 80 field experiments reported a median 43% 667 reduction in fungicide use without increased disease risk under DSS-based strategies (Lázaro 668 et al., 2021). 669 The present results therefore place PedMan within the upper range of published plant disease 670 forecasting and DSs, particularly with respect to specificity and the avoidance of unnecessary 671 fungicide applications. Nevertheless, its false-negative decisions indicate that further 672 refinement and evaluation are needed to ensure that the high specificity observed here is 673 maintained without compromising sensitivity. Such evaluation should include a broader range 674 of cultivars, environments, seasons, and disease-pressure levels, with particular attention to 675 cultivar resistance classification and infection-alert criteria. 676 PedMan has been programmed and made available online, enabling its independent use by 677 [Israeli olive growers (https://www.tavazit.com](https://www.tavazit.com/) – Hebrew version, on-line translation is 678 evaluable) during the coming growing seasons. Its application may also be relevant to other 679 olive-growing regions characterized by a Csa – hot summer Mediterranean climate (Torres et 680 al., 2017; Fraga et al., 2021). However, before implementation outside Israel, PedMan 681 predictions and spray recommendations should be independently evaluated under local climatic 682 and epidemiological conditions. Such evaluation is particularly important because infection 35 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 36 683 risk depends on the interaction between weather conditions, cultivar susceptibility, and 684 pathogen epidemiology, and because the performance of forecasting systems may vary 685 substantially among environments. 686 687 **Acknowledgments** 688 We wish to thank the extension officers Reuven Birger, Mugira Yunis and Nizar Abed-Al 689 Ahadi who contributed to this research; Nir Kilman and Navot Razon, the growers who 690 implemented PedMan in their olive groves. This research was supported in part by the Israeli 691 Council of Fruit Trees, and the Chief Scientist of Israel Ministry of Agriculture grant #20-02 692 0039. 693 694 **References** 695 Abuamsha, R., Abueid, M., Hajjeh, H., Salman, M., 2013. Evaluation of the incidence and 696 severity of olive leaf spot caused by _Spilocaea oleagina_ in different olive cultivars in 697 Palestine. _J. Agric. Environ. Int. Dev._ 107, 201–212. 698 Abuley, I.K., Nielsen, B.J., 2019. Integrating cultivar resistance into the TOMCAST model to 699 control early blight of potato, caused by _Alternaria solani_ . _Crop Prot._ 117, 69–76. 700 Afiad-Lachger, Z., El Guilli, M., Lahlali, R., Amrani, S., 2020. Evaluation of the 701 susceptibility of some Moroccan and international olive cultivars to _Spilocaea oleagina_ under 702 field conditions. _Plant Pathol. J._ 36, 415–424. 703 Agrios, G.N., 2005. _Plant Pathology_, fifth ed. Elsevier Academic Press, Burlington, MA, 704 USA. 36 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 37 705 Altman, D.G., Bland, J.M., 1994. Statistics Notes: Diagnostic tests 1: Sensitivity and 706 specificity _. BMJ_ 308, 1552. https://doi.org/10.1136/bmj.308.6943.1552. 707 Bernaschina, Y., Alaniz, S., Conde-Innamorato, P., Leoni, C., 2019. Response of olive ( _Olea_ 708 _europaea_ ) cultivars against _Venturia oleaginea_ causing olive scab in Uruguay. _Agrociencia_ 709 _Uruguay_ 23, e59. https://doi.org/10.31285/AGRO.23.1.13. 710 Buonaurio, R., Almadi, L., Famiani, F., Moretti, C., Agosteo, G.E., Schena, L., 2023. Olive 711 leaf spot caused by _Venturia oleaginea_ : An updated review. _Front. Plant Sci._ 13, 1061136. 712 **https://doi.org/10.3389/fpls.2022.1061136** 713 Chicco, D., Jurman, G., 2023. The Matthews correlation coefficient (MCC) should replace 714 the ROC AUC as the standard metric for assessing binary classification. _BMC Bioinform._ 16, 715 4. https://doi.org/10.1186/s13040-023-00322-4. 716 Ciccarese, F., Ambrico, A., Longo, O., Schiavone, D., 2002. Search for resistance to 717 _Verticillium_ -wilt and leaf spot in olive. _Acta Hortic._ 586, 717–720. 718 Estudillo, C., Pérez-Rial, A., Guerrero-Páez, F.A., Díez, C.M., Moral, J., Die, J.V., 2025. 719 Characterization of olive-resistant genes against _Spilocaea oleagina_, the causal agent of scab. 720 _Agronomy_ 15, 452. **https://doi.org/10.3390/agronomy15020452** 721 Ezra, D., Ygzao, D., Gilat, Y., Shtienberg, D., 2025. Peacock's eye disease of olive, do we 722 finally understand how to manage it effectively? _Phytopathol. Mediterr._ 64, 354 (Abstract). 723 Foody, G.M., 2023. Challenges in the real-world use of classification accuracy metrics: From 724 recall and precision to the Matthews correlation coefficient. _PLoS One_ 18, e0291908. 725 https://doi.org/10.1371/journal.pone.0291908. 37 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 38 726 Fraga, H., Moriondo, M., Leolini, L., Santos, J.A., 2021. Mediterranean olive orchards under 727 climate change: A review of future impacts and adaptation strategies. _Agronomy_ 11, 56. 728 https://doi.org/10.3390/agronomy11010056. 729 Gent, D.H., Ocamb, C.M., 2009. Predicting infection risk of hop by _Pseudoperonospora_ 730 _humuli_ . _Phytopathology_ 99, 1190–1198. https://doi.org/10.1094/PHYTO-99-10-1190. 731 Gilat, Y., Ygzao, D., Shtienberg, D., Ezra, D., 2026. Evaluation of the peacock eye disease 732 management (PedMan) system in highly susceptible olive cultivar and using the system for 733 addressing key biological questions. _Front. Agron._, submitted. 734 https://doi.org/10.64898/2026.06.02.729514 735 Giroux, M.E., Bourgeois, G., Dion, Y., Rioux, S., Pageau, D., Zoghlami, S., Parent, C., 736 Vachon, E., Vanasse, A. 2016. Evaluation of forecasting models for Fusarium head blight of 737 wheat under growing conditions of Quebec, Canada _. Plant Dis._ 100, 1192–1201. 738 https://doi.org/10.1094/PDIS-04-15-0404-RE. 739 Graniti, A., 1993. Olive scab: A review. _EPPO Bull_ . 23, 377–384. 740 Hjelkrem, A.G.R., Eikemo, H., Le, V.H., Hermansen, A., Nærstad, R. 2021. A process-based 741 model to forecast risk of potato late blight in Norway (the Nærstad model): model 742 development, sensitivity analysis and Bayesian calibration. _Ecol. Model_ . 450 _,_ 109565. 743 https://doi.org/10.1016/j.ecolmodel.2021.109565. 744 Jørgensen, L. N., Matzen, N., Ficke, A., Nielsen, G.C., Jalli, M., Ronis, A., Andersson, B., 745 Djurle, A. 2020. Validation of risk models for control of leaf blotch diseases in wheat in the 746 Nordic and Baltic countries. _Eur. J. Plant Pathol_ . 157 _,_ 599–613. 747 https://doi.org/10.1007/s10658-020-02025-6. 38 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 39 748 Lázaro, E., Makowski, D., Vicent, A. 2021. Decision support systems halve fungicide use 749 compared to calendar-based strategies without increasing disease risk. _Comm. Earth Environ._ 750 2, 224. https://doi.org/10.1038/s43247-021-00291-8. López-Doncel, L.M., García-Berenguer, 751 A., Trapero, A., 1999. Resistance of olive tree cultivars to leaf spot caused by _Spilocaea_ 752 _oleagina_ . _Acta Hortic_ . 474, 549–553. 753 MacDonald, A.J., Walter, M., Trought, M., Frampton, C.M., Burnip, G., 2000. Survey of 754 olive leaf spot in New Zealand. _N. Z. Plant Prot_ . 53, 126–132. 755 Mahfoud, Y., Choumane, W., Doai, F., 2018. Screening of some olive cultivars ( _Olea_ 756 _europaea_ L.) for tolerance to _Spilocaea oleagina_ . _Int. J. Agric. Environ. Res_ . 4, 23–30. 757 https://doi.org/10.53555/eijaer.v3i2.25. 758 Martin, G.C., Sibbett, G.S., 2005. Botany of the olive, in: Sibbett, G.S., Ferguson, L. (Eds.), 759 _Olive Production Manual_, second ed. University of California, Agriculture and Natural 760 Resources, pp. 15–19. 761 Moral, J., Alsalimiya, M., Roca, L.F., Díez, C.M., León, L., de la Rosa, R., Barranco, D., 762 Rallo, L., Trapero, A., 2015. Relative susceptibility of new olive cultivars to _Spilocaea_ 763 _oleagina_, _Colletotrichum acutatum_, and _Pseudocercospora cladosporioides_ . _Plant Dis_ . 99, 764 58–64. 765 Najafi, S., Saremi, H., Jafary, H., Dadras, A., 2023. Evaluation of the relative resistance of 766 different olive cultivars to olive peacock spot disease caused by _Venturia oleaginea_ . _J. Plant_ 767 _Dis. Prot._ 130, 361–369. https://doi.org/10.1007/s41348-022-00697-2. 39 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 40 768 Niks, R.E., Qi, X., Marcel, T.C., 2015. Quantitative resistance to biotrophic filamentous plant 769 pathogens: Concepts, misconceptions, and mechanisms. _Annu. Rev. Phytopathol_ . 53, 445– 770 470. 771 Obanor, F.O., Jaspers, M.V., Jones, E.E., Walter, M., 2008. Greenhouse and field evaluation 772 of fungicides for control of olive leaf spot in New Zealand. _Crop Prot_ . 27, 1335–1342. 773 Obanor, F.O., Walter, M., Jones, E.E., Jaspers, M.V., 2011. Effects of temperature, inoculum 774 concentration, leaf age, and continuous and interrupted wetness on infection of olive plants 775 by _Spilocaea oleagina_ . _Plant Pathol_ . 60, 190–199. 776 Parlevliet, J.E., 1976. Partial resistance of barley to leaf rust _Puccinia hordei_ III; The 777 inheritance of the host plant effect on latent period in four cultivars. _Euphytica_ 25, 241–248. 778 Shtienberg, D., Vintal, H., Brener, S., Retig, B., 2000. Rational management of _Didymella_ 779 _rabiei_ in chickpea by integration of genotype resistance and curative application of 780 fungicides. _Phytopathology_ 90, 834–842. 781 Small, I.M., Joseph, L., Fry, W.E., 2015. Evaluation of the BlightPro decision support system 782 for management of potato late blight using computer simulation and field validation. 783 _Phytopathology_ 105, 1545–1554. https://doi.org/10.1094/PHYTO-05-15-0117-R. 784 Thomidis, T., Michos, K., Chatzipapadopoulos, F., and Tampaki, A., 2021. Evaluation of 785 two predictive models for forecasting olive leaf spot in northern Greece. _Plants_ 10, 1200. 786 Torres, M., Pierantozzi, P., Searles, P., Rousseaux, M.C., García-Inza, G., Miserere, A., 787 Bodoira, R., Contreras, C., Maestri, D., 2017. Olive cultivation in the Southern Hemisphere: 788 Flowering, water requirements and oil quality responses to new crop environments. _Front._ 789 _Plant Sci._ 8, 1830. 40 [bioRxiv preprint doi: https://doi.org/10.64898/2026.11.08.765137; this version posted November 10, 2026. The copyright holder for this](https://doi.org/10.64898/2026.11.08.765137) preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in [perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license.](http://creativecommons.org/licenses/by-nc-nd/4.0/) Gilat et al. b, 41 790 van der Waals, J.E., Denner, F.D.N., van Rij, N., Korsten, L. 2003. Evaluation of PLANT 791 Plus, a decision support system for control of early blight on potatoes in South Africa. _Crop_ 792 _Prot._ 22, 821–828. https://doi.org/10.1016/S0261-2194(03)00049-8. 793 Viruega, J.R., Trapero, A., 1999. Epidemiology of leaf spot of olive tree caused by _Spilocaea_ 794 _oleagina_ in southern Spain. _Acta Hortic_ . 474, 531–534. 795 Viruega, J.R., Roca, L.F., Moral, J., Trapero, A., 2011. Factors affecting infection and 796 development of olive leaf spot caused by _Spilocaea oleagina_ . _Plant Dis_ . 95, 1139–1146. 797 Wu, Y., Small, I.M., Fry, W.E., 2017. Evaluation of the BlightPro decision support system 798 for management of potato late blight using computer simulation and field validation. 799 _Phytopathology_ 107, 1247–1255. 800 Ygzao, D., Shulhani, R., Shtienberg, D., Ezra, D., 2026a. _Venturia oleaginea_, the causal 801 agent of peacock eye disease in olive, maintains monocyclic and polyetic characteristics in a 802 Mediterranean climate. _Plant Dis_ . 110, 338–346. 803 Ygzao, D., Shtienberg, D., Ezra, D., 2026b. Optimizing fungicide application timing to 804 control olive Peacock Eye disease: Impact on current and succeeding growing seasons. _Plant_ 805 _Dis._ 110, 327–337. 806 Ygzao, D., Gilat, Y., Shtienberg, D., Ezra, D., 2026c. Why peacock eye severity varies with 807 olive tree orientation: Defining criteria to predict infection events. _Plant Dis_ ., in press. 808 41