Epidemiology of boxwood blight in hotspots of western North Carolina and Virginia and validation of the boxwood blight infection risk model

Epidemiology of boxwood blight in hotspots of western North Carolina and Virginia and validation of the boxwood blight infection risk model | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Epidemiology of boxwood blight in hotspots of western North Carolina and Virginia and validation of the boxwood blight infection risk model Ihsanul Khaliq, Herve F. Avenot, Anton Baudoin, Leonard Coop, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4651076/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 05 Nov, 2024 Read the published version in Scientific Reports → Version 1 posted 10 You are reading this latest preprint version Abstract Boxwood blight is a highly invasive emerging disease. Since the first US report in North Carolina and Connecticut in 2011, boxwood blight has spread to over 30 US states, risking more than 90% of boxwood production. A boxwood blight infection risk model was developed from limited studies in controlled environments. Our study investigated the disease field epidemiology and validated the model’s prediction, using leaf wetness estimated by leaf wetness sensor or algorithms, by analysing weekly blight monitoring data collected on detector plants exposed to the prevailing environmental conditions from spring through fall of 2014 to 2017. Boxwood blight was recorded in 61 of 86 weeks, with the highest infected leaf count recorded in late summer or early fall. Rainfall, high relative humidity outside rainy periods and optimal temperatures during prolonged leaf wetness had a significant positive effect on boxwood blight development. Classification analyses showed that disease predictions from the model using leaf wetness estimated by leaf wetness sensor were more closely aligned with observations from the field than predictions based on algorithms. This study improved our understanding of disease field epidemiology, provided leads to improve the existing model, and generated essential knowledge for formulating effective strategies for blight mitigation. Biological sciences/Microbiology Biological sciences/Microbiology/Pathogens Biological sciences/Ecology/Invasive species Weather effect forecasting seasonal patterns temporal patterns trap plants weekly exposure Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Invasive pathogens are a major threat to plant health, diversity, and distribution in both natural and managed ecosystems worldwide 1 , 2 . Boxwood blight is a highly invasive emerging disease that causes leaf spots, stem lesions, and defoliation on all species of boxwood, one of the most iconic evergreen shrub and keystone forest species 3 . Boxwood blight was first discovered in the United Kingdom 4 and New Zealand in the mid-1990s 5 , and has now spread widely across Europe, Asia and North America. In the United States, boxwood blight was initially identified in North Carolina and Connecticut in 2011 6 , and has now spread to 30 states 7 , risking more than 90% of boxwood production in the United States 8 , 9 . Total losses from boxwood blight have been significant, with Connecticut state alone incurring losses exceeding $ 3 million within the first year of detection 10 . Boxwood blight is caused by two invasive ascomycete fungi: Calonectria pseudonaviculata 11 and C. henricotiae 12 , but only C. pseudonaviculata has been detected in the United States. Infested soil serves as an important source of inoculum 13 . The sexual stage/spores of the pathogen has not been observed, so both primary and secondary infections are attributed to conidial infection 14 . Conidia are produced within a sticky extracellular mucilaginous matrix that cannot be dispersed by wind alone, and thus the kinetic energy of rain or vectors are required to disperse conidia 15 . Long-distance spread is associated with human-mediated transport of infected plant material 3 . An infection cycle can be completed in less than a week under ideal conditions, with conidial germination beginning 3 h after inoculation, penetration occurring 5 h after inoculation, and new conidia developing within 7 days of inoculation 16 – 18 . There have been no systematic studies on the temporal patterns of boxwood blight development. Knowledge of the temporal progression of diseases is needed for understanding disease dynamics and developing disease management guidelines, such as fungicide applications 19 . Additionally, this knowledge can be used to develop a theoretical basis for determining whether an epidemic will occur as well as predicting the intensity of the disease 19 . There are significant knowledge gaps in the of epidemiology of boxwood blight. A few studies have been conducted in controlled environments to investigate the effect of relative humidity, temperature, and leaf wetness duration on boxwood blight development 14 , 20 – 22 . Studies investigating the effect of weather variables on boxwood blight development under field conditions are rare. Establishing the relationship between weather variables and disease development is the key component of developing disease forecasting systems 23 . Knowledge of the weather conditions that favour disease development is also useful in developing strategies for limiting fungicide applications as fungicides are applied according to the risk of infection 24 . Additionally, this knowledge can be used to identify areas at risk where the pathogen is either present or has not (yet) been introduced 9 , 25 . There is no epidemic without inoculum, so the design of forecasting systems is greatly influenced by the source and density of inoculum 23 . The source of inoculum also plays a key role in the rain-splash dispersal of fungal spores. For instance, plant canopies affect rainfall characteristics by reducing the kinetic energy of falling rain drops, intercepting spore-carrying droplets, and creating a series of successive splashes often known as the relay effect of the canopy 26 . There are no systematic studies on how weather conditions, topographic factors (e.g., distance) and the source of inoculum interact to affect boxwood blight development. Information on the distance spores disperse from the inoculum source is valuable information for disease forecasting systems 27 . The role of microsclerotia and infected debris as the primary sources of inoculum is well known 13 . However, the role of infected boxwood canopies in boxwood blight epidemiology was not known. This information is crucial in understanding the main sources of inoculum and the routes of the pathogen’s dissemination to inform policy makers where to put limited resources to protect the protectable. The boxwood blight infection risk model 28 is the only publicly available model in the United States that predicts relative infection risk based on location-specific weather data. The model accumulates infection risk units (degree hours) during periods of leaf wetness. Leaf wetness is estimated by leaf wetness sensors connected to weather stations, where available. However, unlike other weather variables, leaf wetness is not a standard meteorological measurement 29 , and thus leaf wetness data are typically not available from official weather station networks 30 . Many boxwood cultivars have very dense canopies that tend to retain leaf wetness, at least in the interior of the plant, which could be rather different from that predicted from off-site weather stations. In the absence of leaf wetness sensors, the model estimates leaf wetness from other weather variables using fuzzy logic leaf wetness (FLLW) and precipitation-drying leaf wetness (FoxLW) algorithms 28 , 31 . However, there have been no studies comparing the predicted disease risks based on leaf wetness data estimated by leaf wetness sensors and those derived from FLLW or FoxLW heuristic algorithms. The current set of rules used in calculating boxwood blight infection risk index, such as temperature thresholds for boxwood blight infection and the number of dry hours required to stop the infection process 28 was primarily derived from controlled environment experiments, and needs to be validated with field data. To address the above key knowledge gaps, boxwood blight was monitored in disease hotspots in western North Carolina and Virginia using detector boxwood plants that were rotated in and out of the field at weekly intervals from 2014 to 2017. Specific objectives were to: 1) investigate the temporal patterns of boxwood blight development on detector plants; 2) identify key weather variables affecting boxwood blight development under field conditions; and 3) validate the boxwood blight infection risk model using the actual blight disease readings on detector plants. Results Temporal patterns of boxwood blight development The highest infected leaf count per week was recorded in late summer or early fall from 2014 to 2016 and during the last monitoring week in the late summer of 2017. At Lambsburg (2014), the highest infected leaf count per week was recorded in the week starting August 18 (422 leaves), a week for which we have no rainfall data, followed by September 8 (396 leaves) (Supplementary Table S1 , Fig. 2 ). At Lowgap, the highest infected leaf count per week was recorded in the week starting September 23 (971 leaves) in 2015, and the week starting September 28 (658 leaves) in 2016. The temporal patterns of boxwood blight development in 2017 were not directly comparable with those observed during 2015 and 2016 because detector plants were placed in the field earlier, in March and April, but not after mid-August, whereas in 2015 and 2016, detector plants were not placed in the field as early, and were exposed much later through the fall (Supplementary Table S1 , Fig. 2 ). Nonetheless, the highest infected leaf count per week was recorded in the last monitoring week, from August 9 to 16 (160 leaves) (Supplementary Table S1 , Fig. 2 ). Weather data from the trial area for the past decade showed that late summer and early fall tended to have optimum monthly mean temperature range (18–22°C) whereas monthly mean relative humidity tended to be high (> 75%) all year, and monthly total rainfall varied across years (Supplementary Fig. S1 ). Key weather variables affecting boxwood blight development DHARMa diagnostics showed that the data met the model assumptions. The R 2 value for the model was 0.69. The infected leaf count per plant significantly increased (p = 0.0001) with increasing relative humidity outside of rainy periods (Fig. 3 , Table 1 ). Daily relative humidity outside rainy periods ranged from 21–100%. Outside rainy periods, the weekly number of hours relative humidity exceeding 90% ranged from 12.7 to 122 h (daily range: 1–17.4 h) and the mean weekly temperature during those hours ranged from − 1.2–21.4°C (Supplementary Table S1 ). High infected leaf counts were recorded in weeks when humidity exceeded 90% for more than 34 hours, accompanied by a mean temperature above 10°C during those hours (Supplementary Table S1 , Supplementary Fig. S2). Lower infected leaf counts were recorded in weeks when mean weekly relative humidity outside of rainy periods remained below 75%, while the highest infected leaf count was recorded when mean weekly relative humidity outside of rainy periods exceeded 90% (Supplementary Fig. S3). In six weeks during which boxwood blight developed despite the absence of rainfall, the mean weekly relative humidity was 76.6% (with 46.8 hours of relative humidity exceeding 90%) in week 4 of 2015, and 75.1% (with 48 hours of relative humidity exceeding 90%) in week 15 of 2015. In week 22 of 2015, it was 69.1% (with 54 hours of relative humidity exceeding 90%) and in week 26 (a 6-day week) of 2015 it was 56.1% (with 18.3 hours of relative humidity exceeding 90%). In week 11 and 21 (a 6-day week) of 2016, it was 78.3% (with 73.8 hours of relative humidity exceeding 90%), and 57.6% (with 22 hours of relative humidity exceeding 90%), respectively (Supplementary Table S1 ). The weekly number of hours with relative humidity exceeding 90%, inside plus outside rainy periods, ranged from 15.2 to 141.7 h (daily range: 1.4–20.2 h) and the mean weekly temperature when relative humidity exceeded 90% ranged from − 1.2–21.5°C (Supplementary Table S2, Supplementary Fig. S4). High infected leaf counts were recorded in weeks when humidity exceeded 90% for more than 40 hours, accompanied by a mean temperature above 10°C when relative humidity remained above 90% (Supplementary Table S2, Supplementary Fig. S4). There was a significant positive interaction effect (p < 0.0001) of leaf wetness duration and temperature during wet periods on the infected leaf count per plant (Table 1 , Fig. 4 ). That is, the effect of leaf wetness duration on the infected leaf count depended on temperature. Higher infected leaf count was recorded in optimal temperatures (13.6–22.7°C) during periods of prolonged leaf wetness, and vice versa. The weekly total leaf wetness duration ranged from 1.87 to 160 h, and the average daily leaf wetness duration ranged from 0.27–22.9 h (Supplementary Table S1 ). The infected leaf count was very low when the weekly total leaf wetness duration was below 49 h and temperatures during wet periods were below 9°C (Supplementary Table S1 , Supplementary Fig. S5). The highest infected leaf counts were recorded in weeks when temperatures during wet periods were between 13.6°C and 22.7°C, and the weekly total leaf wetness duration exceeded 65 hours (Supplementary Fig. S5). Rainfall had a significant (p = 0.0049) positive effect on the infected leaf count (Fig. 3 , Table 1 ). The effect of rainfall on infected leaf count was especially pronounced when temperatures during rainy periods were between 16 and 22.7°C (Supplementary Fig. S6, Supplementary Fig. S7). Over the course of the 86-week experiment conducted across four years, there were seven weeks in which no rainfall was recorded (Supplementary Table S1 , Supplementary Fig. S8). Boxwood blight was recorded in six of those seven weeks, with infected leaf counts ranging from 1–157 (Supplementary Table S1 ). Conversely, boxwood blight was not recorded in 24 out of 86 weeks (28% of weeks) despite recording rainfall (Supplementary Table S1 ). These included 14 weeks recording between 0.4 to 10 mm rainfall, eight weeks recording between 11 to 35.8 mm rainfall, and two weeks recording 52.6 and 66 mm rainfall (Supplementary Table S1 ). Validation of the boxwood blight infection risk model Boxwood blight was recorded on detector plants in 61 of the 86 monitoring weeks (Supplementary Table S1 ). Using the 250 degree-hours criterion for disease prediction, the SENSOR, ESTIM-LW and SENSOR-ADJ models predicted disease for 67, 76 and 65 weeks, respectively (Supplementary Table S1 ). The accuracy, precision and F-score values were almost similar for SENSOR, ESTIM-LW and SENSOR-ADJ models (Table 2 , Supplementary Table S4). The SENSOR-ADJ model was 4% more specific than the SENSOR model and 20% more specific than the ESTIM-LW model. The ESTIM-LW model achieved the highest recall score (89%), followed by the SENSOR model (80%) and the SENSOR-ADJ model (79%) (Table 2 , Supplementary Table S4). Using the 160-degree-hours criterion for disease prediction, the SENSOR, ESTIM-LW and SENSOR-ADJ model predicted disease for 73, 77 and 74 weeks, respectively (Supplementary Table S1 ). The accuracy, precision and F-score values were almost similar for SENSOR, ESTIM-LW and SENSOR-ADJ models (Table 2 , Supplementary Table S4). The SENSOR model was 8% more specific than ESTIM-LW model and 4% more specific than the SENSOR-ADJ model. The ESTIM-LW model had higher recall score (89%) than both SENSOR and SENSOR-ADJ models (85% each) (Table 2 , Supplementary Table S4). Using the 56-degree hours criterion for disease prediction, the SENSOR, ESTIM-LW and SENSOR-ADJ model predicted disease for 79, 80 and 79 weeks, respectively (Supplementary Table S1 ). Almost similar accuracy, precision, recall and F-scores values were achieved (Table 2 , Supplementary Table S4). The SENSOR and SENSOR-ADJ models showed 4% specificity, whereas the ESTIM-LW model failed to show any specificity (Table 2 , Supplementary Table S4). Overall, the model using the 250 degree-hours criterion for boxwood blight prediction achieved higher accuracy, AUC, and specificity scores than those using 160 and 56 degree-hours criteria. The ability of the model to predict true negatives (disease-free weeks or specificity) decreased as the degree hours criteria for boxwood blight prediction decreased (Table 2 , Supplementary Table S4). Likewise, both the SENSOR and SENSOR-ADJ models exhibited a greater ability to predict true negatives and had fewer false positives compared to the ESTIM-LW model, especially when the 250 or 160 degree-hours criterion for boxwood blight prediction was used. The highest range of McFadden’s pseudo-R 2 value was recorded for the SENSOR-ADJ model (0.02–0.48), followed by SENSOR model (0.02–0.47), and ESTIM-LW model (0.03–0.25). The highest range for weekly total accumulated blight risk index was observed for the ESTIM-LW model (29–3950), followed by the SENSOR model (9–3692), and the SENSOR-ADJ model (4–3559) (Supplementary Fig. S9). Generalised linear models showed a significant association between the weekly total accumulated blight risk index and infected leaf count for three years (2014, 2015 and 2017) in the case of the SENSOR-ADJ and SENSOR models, and just a year (2015) in the case of ESTIM-LW model (Supplementary Fig. S10). Table 1 Generalised linear mixed model’s coefficients for the effect of weather variables on boxwood blight development. Predictor Estimate Std. Error z value Pr(>|z|) Lambsburg -6.75 2.49 -2.71 0.0068 Relative humidity † 0.18 0.04 4.00 0.0001 Total rain 0.03 0.01 2.82 0.0049 Lowgap 1.35 0.47 2.86 0.0042 Wind direction x Wind speed †† 0.01 0.01 0.60 0.5511 Temperature § x Leaf wetness duration 0.01 0.00 5.14 < 0.0001 † = Relative humidity was considered outside of rainy periods. †† = Wind direction and wind speed were considered during rainy periods. § = temperatures during wet periods were considered. Table 2 Model evaluation metrics for the boxwood blight infection risk model using leaf wetness estimated by leaf wetness sensor (SENSOR and SENSOR-ADJ) and algorithms based on energy principles (ESTIM-LW). Model Degree hours for disease prediction Model evaluation metrics Accuracy (%) Precision (%) Recall (%) F-score (%) AUC (%) Specificity (%) SENSOR 56 65 69 90 79 41 4 ESTIM-LW 56 64 69 90 78 34 0 SENSOR-ADJ 56 65 69 90 79 42 4 SENSOR 160 65 71 85 78 51 16 ESTIM-LW 160 65 70 89 78 46 8 SENSOR-ADJ 160 64 70 85 77 47 12 SENSOR 250 65 73 80 77 55 28 ESTIM-LW 250 66 71 89 77 50 12 SENSOR-ADJ 250 65 74 79 76 56 32 Discussion The observation of the highest weekly infected leaf counts on detector plants exposed to natural inoculum in August in 2014 and 2017 and September in 2015 to 2016 suggests that blight disease pressure in these areas is likely higher in late summer and early fall than other seasons. The highest infected leaf counts in early fall can mainly be attributed to prolonged leaf wetness duration resulting from high rainfall and/or greater number of hours with relative humidity exceeding 90% both outside and inside rainy periods. Hours with relative humidity exceeding 90% were especially pronounced during early fall (Supplementary Table S3), which may have contributed to inoculum build up preceding rainfall events and disease development following rainfall events. A significant positive association between rainfall and infected leaf count on detector plants was not surprising. Rainfall not only provides the kinetic energy to release conidia from sticky mucilaginous matrix 15 , but also facilitates sporulation, germination, and host penetration by keeping host tissues wet. However, boxwood blight cannot be predicted solely on a forecasted rainfall event, as boxwood blight was not recorded in 24 of 86 weeks during which rainfall occurred, with the model using different leaf wetness criteria predicting disease for over 20 of those 24 weeks. Our findings also suggest that small and scattered rainfall (0.2–1.2 mm) events occurring on different days are unlikely to cause high disease outbreaks. Conversely, small or large rainfall events occurring frequently throughout the day/week lead to high disease outbreaks, as the latter provide sufficient kinetic energy to dislodge more conidia and provide adequate leaf wetness to facilitate conidial germination, penetration, and subsequent infection. Given that C. pseudonaviculata conidia cannot be dispersed by wind alone, and the absence of boxwood blight development in some weeks with high relative humidity but no rainfall, we propose including rainfall as a condition for spore dispersal in the model. It is difficult to determine an exact rainfall threshold for spore dispersal, as rainfall threshold can vary with rain intensity, the time of the day, inoculum density, and the level of resistance of the cultivar 19 . In the present study, two infected leaves were recorded in the week starting October 19 of 2016, which received only 0.2 mm rainfall over 0.25 h, followed by a continuous 6-h leaf wetness due to high relative humidity. The hourly rainfall threshold for spore dispersal in forecasting models can perhaps be set at values greater than 0 or 0.2 mm. High relative humidity outside rainy periods was found to be very important for boxwood blight development. This is not surprising because high relative humidity facilitates survival and infection by conidia laying on plants after dispersal by keeping host tissues moist 14 . Our results are consistent with findings from a study conducted in a controlled environment, where less than 65% humidity halted C. pseudonaviculata infection on detached leaves 20 . However, high relative humidity-induced leaf wetness, when rainfall is absent or lower, is unlikely to trigger high disease outbreaks-even if temperatures are within the optimum range and high disease risk is predicted. This was particularly evident in the weeks starting September 15 and 22, 2014, during which boxwood blight did not develop although high relative humidity provided continuous leaf wetness for most parts of these weeks. The most likely reason for the lack of disease development was the low rainfall in these weeks: 0.6 mm and 2.8 mm, respectively. Rainfall was recorded on two separate days in the former week and on three separate days in the latter week. The significant positive interaction effect of optimum temperatures during periods of prolonged leaf wetness on infected leaf count was not unexpected because disease outbreaks are mostly associated with optimal temperatures during wet periods 14 . Their significant interaction effect in the present study was especially evident in 2017, where more than 70% (586 of 830 mm) of rainfall occurred between April and May, but negligible blight developed during these months because weekly mean temperatures during wet periods were mostly below 15.3°C. These results are aligned with those of Avenot, et al. 21 who observed significantly higher sporulation, on detached leaves, with longer leaf wetness duration and temperatures near the optimum. The finding that temperature during periods with measurable rain or some wetness exceeded 30°C only a single and 26 times, respectively (based on 15-minute data points collected across four years) suggests that high temperatures are unlikely to impede boxwood blight infection in the study area as temperature drops quickly during rainfall. The non-significant effect of wind speed and wind direction during rainy periods on infected leaf count in the present study as well as recording the lowest infected leaf counts on detector plants placed beyond the dripline of canopy suggests that wind driven rain may not often play a big role in dispersing C. pseudonaviculata conidia under the local field conditions, although this will require further testing as wind gusts remained below 3 m/s and wind was not omni-directional during the present study. However, we found that the blight development on detector plants placed beyond the dripline of canopies was associated with at least 27 mm rainfall per week or 3.86 mm mean rainfall per day, except for a single week starting June 9. Overall, the model showed high accuracy and precision. The absence of disease, despite the model predicting disease in some weeks, can be attributed to light or scattered rainfall and decreased inoculum viability. Specifically, during the years 2014 to 2016, boxwood blight was predicted due to wet conditions resulting from light or scattered rainfall (mostly ranging from 0.2 to 1.2 mm on different days), high relative humidity outside rainy periods, and optimal temperatures during wet periods. However, these light and scattered rainfall events were insufficient to trigger the blight development. Using the 250 degree-hours criterion for boxwood blight prediction, the reasons for underpredicting disease for 6, 10 and 13 weeks by ESTIM-LW, SENSOR, and SENSOR-ADJ models, respectively can mainly be attributed to the ability of the pathogen to cause disease below the lower temperature threshold (6.7°C) for infection during wet periods. For instance, in the weeks starting November 9 (8-day exposure period) and November 17 (6-day exposure period) in 2016; the hourly mean temperatures during wet periods remained below the lower threshold for infection (6.7°C), except for 5 and 4 hours, respectively. The lower temperature threshold for infection can perhaps be set at 5.6°C when 250 degree-hour threshold is used. The increased false positives at 160 and 56-degree hours criteria for boxwood blight prediction suggests that the current 250 degree-hours criterion is close to optimal. The observation of significant association between the weekly total accumulated blight risk index and infected leaf counts for three out of four years for the SENSOR and SENSOR-ADJ models compared to a significant association in just a single year in the ESTIM-LW model suggests that the model using leaf wetness estimated by leaf wetness sensor should be chosen if leaf wetness sensors are available. Finally, the reasons for under-prediction or over-prediction are not known for certain weeks. For instance, the reasons for the absence of disease in three weeks (week 1 in 2014 and week 10 and 12 in 2015) for which boxwood blight was predicted are not known, as these weeks recorded 6.2, 18.6, and 11.8 mm rainfall, respectively. The individual continuous rain spells in these weeks reached up to 3, 8.4 and 4.8 mm, respectively, while temperatures during wet periods were also within the optimum range for infection. Likewise, the reasons for recording boxwood blight, despite the model not predicting disease, in the week starting November 11, 2015 are not clear as no rainfall was recorded and the weekly total leaf wetness duration was 2.1 hour; therefore, the minimum leaf wetness duration requirement for infection was not met. Conducting this study in boxwood blight hotspots improved our understanding of the pathogen’s “behaviour” and informs policy makers and managers where to place limited resources to protect the iconic evergreen shrub and key stone forest species from further invasion by the pathogen. Although boxwood blight is highly weather dependent and high disease outbreaks are expected whenever weather conditions are favourable for boxwood blight development, boxwood growers are advised to especially stay vigilant in early fall in western North Carolina and Virginia. Both infected canopies and infested leaf debris are important sources of inoculum for disease epidemics. The chances of high disease outbreaks greatly decreased with increasing distance from the inoculum source, emphasizing the need for extensive hygiene, adequate spacing between plants, and mulching 32 . Recording different infected leaf counts on detector plants exposed to different sources of inoculum and/or distance from the source of inoculum suggests that disease forecasting models should also consider inoculum source and spatial factors (e.g. distance from the inoculum source). Although desirable, such improvements in prediction models will be challenging. Even for diseases like apple scab, which has been the subject of extensive modelling efforts, the quantification of risk has been predominantly focused on predicting the release of primary inoculum based on weather factors 47 , 48 . Materials and methods Study areas description This study was conducted in two US locations: Lambsburg (approx. 36°35’N, 80°46’W), Virginia (VA) in 2014, and Lowgap (approx. 36°31’N, 80°50’W), North Carolina (NC), from 2015 to 2017. The Lambsburg trial was conducted in a rural residential landscape near a site where boxwood blight was first detected in the fall of 2011. No boxwood blight control activities were carried out at the site since at least 2013. The Lowgap trial was conducted in a former commercial field nursery where boxwood blight had become established by 2013 32 . No management activities had been carried out to control boxwood blight since its establishment, but insecticides were applied in the spring to control boxwood leafminer ( Monarthropalpus flavus ). Both locations had loamy soil and were surrounded by deciduous forest. Boxwood blight monitoring using detector plants Two-year-old blight-free Buxus sempervirens ‘Suffruticosa’ and ‘Justin Brouwers’ plants, donated by Saunders Brothers Inc. (Piney River, VA), were kept in a greenhouse until used as detector plants. Plants were grown in 3.79-L containers containing potting mix that consisted of 56% composted pine bark, 32% coarse Perlite, and 12% peat moss. Plants were watered daily with tap water and fertilized with Osmocote® Plus 15-9-12 (Everris NA Inc., Dublin, OH) when required. In Lambsburg, boxwood blight monitoring was performed for a total of 20 weeks from May to November 2014 (Supplementary Table S1 ). Each week, a total of nine potted B. sempervirens ‘Suffruticosa’ detector plants were placed at the site to expose to natural inoculum and prevailing environmental conditions. Of the nine detector plants, three were randomly placed under already infected boxwood canopies (Fig. 1 A). Another set of three plants was placed between two rows of infected boxwood, with approximately 6–8 feet distance between the rows: henceforth called “between-row” treatment (Fig. 1 B). The between-row treatment was included only in 2014 since it experienced only limited blight development in that year. The remaining three plants were placed in three plantainers (round plastic pans about 60 cm in diameter, MacCourt Products Inc, Denver, CO) containing infested leaf debris collected from surrounding boxwood (Fig. 1 C). One detector plant was installed per plantainer. In Lowgap, boxwood blight monitoring was conducted for a total of 26 weeks from May to November in 2015, 24 weeks from June to December in 2016, and 16 weeks from March to August in 2017, using eight B. sempervirens ‘Justin Brouwers’ detector plants each week (Table S1 ). Of the eight detector plants, four were randomly placed under already infected boxwood canopies. To prevent the pathogen from infested soil spreading onto detector plants by rain splash, a polyethylene landscape fabric (Vigoro Corporation, Lake Forest, IL) was installed in sections of 2 x 1 m around the base of those detector plants (Fig. 1 A). The remaining plants were placed within four plantainers containing infested debris collected from surrounding boxwood. In 2017, the early-season exposure period of detector plants was 28 days in “week 1” and 21 days in both “week 2” and “week 3”: these three longer-than-a-week exposure periods are referred to as “weeks” for uniformity. At both locations, it was ensured that adequate blighted leaves were present on canopies, and completely defoliated plants were not used. In the case of detector plants placed within plantainers, efforts were made to use the similar amount of inoculum each week by distributing infested leaf debris in a 5 cm layer around each detector plant. After a week of field exposure, a new set of detector plants was introduced into the field, while the previously exposed set was transferred to a lighted growth chamber (14 h photoperiod) set at 21°C for 24 h. The next day, the infected leaf count per plant was recorded. The detector plants were then kept in the lighted growth chamber at 21°C for another week to allow maximum lesion development, and the infected leaf count per plant was recorded again. Each week, the date and time detector plants were placed and removed from the field were recorded (Supplementary Table S1 ). Infested leaf debris was replenished regularly to account for inoculum depletion. Weather monitoring and data processing For the entire duration of boxwood blight monitoring on detector plants in the field, weather data were recorded at 15-minute intervals using an Em50G data logger (Decagon Devices Inc. (now METER Group), Pullman, WA) equipped with the following instruments: anemometer (model Davis Cup), pyranometer (model PYR), rain gauge (model ECRN-100), leaf wetness sensor (model LWS), and temperature, humidity, and vapor pressure sensor (model VP-3). The weather station was placed next to infected canopies at both locations. To identify key weather variables affecting boxwood blight development, weather data recorded at 15-minute intervals were summarized to weekly means for temperature during wet periods, relative humidity outside rainy periods, wind speed, and wind direction during rainy periods, and to weekly totals for rainfall, rain duration, and leaf wetness duration. Data for three weeks in 2014, from August 18 to noon on September 8, were removed from the analyses because the rain gauge became clogged with leaves during these weeks. To ensure weather data were comparable among monitoring weeks, rainfall and leaf wetness duration data collected during the first three extended monitoring periods in early 2017 were converted into per-week data, but no adjustments were made for 5-, 6- or 8-day weeks (Supplementary Table S1 ). The infected leaf count data, recorded on detector plants placed between infected boxwood rows, were only included in the descriptive analyses on how weather conditions and distance may affect disease dynamics on detector plants exposed to each inoculum source. Data were processed and visualised in the R programming language using packages ‘tidyverse’ 33 , ‘lubridate’ 34 , and ‘ggpubr’ 35 . Mean wind direction was calculated using the ‘circular.averaging()’ function of the ‘SDMTools’ package 36 . Calendar heatmaps were created using the custom function `theme_calendar()` function 37 . Temporal patterns of boxwood blight development Infected leaf count data recorded on detector plants placed under infected canopies and those exposed to leaf debris inoculum in plantainers were used to investigate the temporal patterns of boxwood blight development since these two treatments were used in all four years of this study. To help understand and interpret the observed temporal patterns of boxwood blight development, additional weather data were obtained from the NASA POWER website ( https://power.larc.nasa.gov/ ) for the trial sites over a period of 12 years from 2010 to 2021 using the nasapower R package 38 . Hourly weather observations were summarised into monthly means for relative humidity, temperature, and monthly total for rainfall (Supplementary Fig. S1 ) to visualize the seasonal weather patterns and identify time of year when temperature, relative humidity and rainfall were all in the conducive range for boxwood blight. Key weather variables affecting boxwood blight development Only infected leaf count data recorded on detector plants placed under infected canopies and those exposed to leaf debris inoculum were used to identify key weather variables affecting boxwood blight development. The `cor()` function was used to generate a correlation matrix to identify pairs of predictors with high correlation coefficients (> 0.7). The correlation matrix identified a very high correlation (r = 0.86) between rain amount and rain duration, so rain duration was removed from the multivariate analysis to avoid multicollinearity. The effect of weather variables and weekly total leaf wetness duration on infected leaf count per plant was investigated by fitting generalised linear mixed models (glmms) using the `glmmTMB` function of the ‘glmmTMB’ package 39 in the R programming language 40 . Rainfall and relative humidity were included as additive terms, while leaf wetness duration and temperature, and wind speed and wind direction were included as interactive terms. A negative binomial family was used because the preliminary analysis showed the data to be overdispersed, i.e., variance was greater than mean. The overdispersion was tested using Bolker’s custom function ‘overdisp_fun()` 39 . Weather variables and leaf wetness duration were included as fixed effects, and the predictor location was included as a covariate. The model failed to converge when both ‘cultivar’ and ‘week’ detector plants were placed in the field were included as random effects. Additionally, ‘cultivar’ explained very little variation (variance p < 0.0001) when included as a random effect in the model. Subsequently, only the predictor/independent variable ‘week’ was included as a random effect. Model diagnostics was performed using the DHARMa package, which uses a simulation-based approach via the function `simulateResiduals` to create readily interpretable scaled residuals for glmms 41 . The marginal coefficient of determination (pseudo- R 2 ) was calculated using the function `r.squaredGLMM` of the `MuMIn` R package 42 . The autocorrelation between residuals was checked using `check_autocorrelation()` function of the performance package 43 . Analysis was based on pooled data across all years for both locations. Validation of the boxwood blight infection risk model The boxwood blight infection risk model is a degree-hour based model that predicts hourly infection risk during wet periods 28 . Degree-hours for the model are calculated, using piecewise regression, based on the current knowledge about the lower, optimum, and upper temperature thresholds for infection, and the number of dry hours required to stop the infection process. For each hour, if leaf wetness is not recorded, or the temperature is less than 44°F (6.7°C) or greater than 86°F (30°C), the infection risk index is set to 0 for that hour. Otherwise, the infection risk index is calculated from a lookup table with information on degree-hours for corresponding temperatures 28 . The infection process is halted, and the index is reset to zero when a dry period exceeds 5 hours. Where a leaf wetness sensor is not available, leaf wetness is estimated using the FLLW algorithm, which is based on energy principles, from temperature, dewpoint, and wind speed, or the FoxLW heuristic algorithm from rainfall, wind speed, time of day, and season of the year 28 , 31 , 44 . Leaf wetness estimated by both algorithms ranges from 0–10, with 0 denoting dry and 1 + denoting wetter conditions. Both algorithms run in parallel and independent of each other; their values are not added together. The algorithm estimating the maximum value for leaf wetness was chosen to estimate leaf wetness for this study, which is also the default setting of the model. Readers are referred to the model’s documentation 28 for more detailed information on the model. The “Ignore Leaf Wetness Sensor Data” option was unchecked or checked to run the model, depending on whether data recorded by leaf wetness sensors connected to on-site weather stations (henceforth referred to as SENSOR and SENSOR-ADJ models) or that estimated by FLLW or FoxLW heuristic algorithms (henceforth referred to as ESTIM-LW model) were used. All models were reset when detector plants were exchanged after each monitoring week. In the SENSOR-ADJ model, the increments from the lookup table (degree-hours) for corresponding temperatures were made proportional to the duration of leaf wetness recorded by the leaf wetness sensors during each hour. This was achieved by multiplying each degree-hour by 'x/60', where 'x’ represents the duration (in minutes) of leaf wetness, and 60 represents the total number of minutes in that hour. The threshold for predicting disease in each monitoring week was the continuous accumulation of 250 degree-hours (°F) before the infection process stopped, which corresponds to the prediction of 6 lesions in a highly susceptible cultivar 28 . Additionally, we investigated the continuous accumulation of 56 and 160 degree-hours (which corresponds to the prediction of first infection in highly susceptible and susceptible cultivars, respectively) as a criterion for disease prediction to compare their prediction accuracy to the currently used degree-hours threshold (250). Binomial data were generated for each monitoring week, where weeks during which boxwood blight was recorded on detector plants were labelled as 1, and 0 otherwise. For disease prediction, weeks in which the continuously accumulated degree hours reached 56, 160, and 250-degree hours at least once were labelled as 1 (indicating boxwood blight was predicted) and 0 otherwise (indicating boxwood blight was not predicted) for respective degree hour thresholds, To compare the disease risk predicted by the model using three leaf wetness criteria (recorded, recorded with degree-hours made proportional to the duration of leaf wetness recorded, and estimated) and three degree-hour thresholds (56, 160, 250) with disease observed on detector plants, confusion matrices for the model were obtained, and metrics such as area under the receiver operating curve (AUC), accuracy, specificity, precision, recall, specificity, and F1-scores were calculated 45 . AUC indicates the ability of the model to distinguish between the classes 0 and 1. Accuracy is the ratio of correct predictions (true positive and true negative) among total number of predictions. Specificity is proportion of true negatives correctly identified by the model. Precision is the ratio of correctly classified positive cases (true positive) to a total number of classified positive cases (either correctly or incorrectly, i.e., precision = true positives/true positives + false positives). Recall is the ratio between the numbers of positive cases correctly classified as positive to the total number of positive samples (recall = true positives/true positives + false negatives). The F-score provides a harmonic mean between precision and recall to describe the overall accuracy of a model. The F-score is calculated by the following formula: F1 = 2 × (precision × recall) / (precision + recall). The values for these metrics were expressed in percentages, with values closer to 100 indicating a better model. Finally, the association between weekly total accumulated blight risk index (degree hours) and the infected leaf count on detector recorded during each monitoring week was investigated by fitting generalised linear models with a quasi-Poisson family. The model’s fit was evaluated using McFadden’s pseudo-R 2 46 . Declarations Competing interests The authors declare no competing interests. Correspondence All requests for materials should be addressed to CH. Author Contribution HA and AB conducted field trials. IK, AB and LC analysed the data. I.K wrote the initial draft of the manuscript. CH and AB secured funding. All authors reviewed the manuscript and gave final approval for publication. Acknowledgement We gratefully acknowledge Sasha Marine, Jordan Craddock, and Tyler Edwards (Virginia Tech) for their contributions to field work and data collection. We also thank Ping Kong (Virginia Tech) for her invaluable comments on the manuscript. Data collection was supported in part by the 2013 Farm Bill through USDA Animal and Plant Health Inspection Service. Data analyses and manuscript preparation were supported by the US Department of Agriculture—National Institute of Food and Agriculture, under award number 2020-51181-32135. Data Availability All raw and generated data used in the statistical analyses and data visualization have been made available as a part of a research compendium for reproducibility. Please see raw and generated data and the fully reproducible code at https://github.com/IhsanKhaliq/epiboxwoodblight References Khaliq, I., Burgess, T. I., Hardy, G. E. S. J., White, D. & McDougall, K. L. Phytophthora and vascular plant species distributions along a steep elevation gradient. Biol. Invasions 23, 1443–1459 (2021). Fisher, M. C. et al. Emerging fungal threats to animal, plant and ecosystem health. Nature 484, 186–194 (2012). Daughtrey, M. L. Boxwood blight: Threat to ornamentals. Annu. Rev. Phytopathol. 57, 189–209 (2019). Henricot, B., Sierra, A. P. & Prior, C. A new blight disease on Buxus in the UK caused by the fungus Cylindrocladium . Plant Pathol. 49, 805 (2000). Ridley, G. New plant fungus found in Auckland box hedges ( Buxus ). For. health news 77, 1–2 (1998). Ivors, K. et al. First report of boxwood blight caused by Cylindrocladium pseudonaviculatum in the United States. Plant Dis. 96, 1070 (2012). Hong, C. United States map of boxwood blight by the time of its first confirmed invasion. Available at: https://irp.cdn-website.com/217658e5/files/uploaded/US%20Boxwood%20Blight%20Map%207-2023.png . (2023). Hall, C. R., Hong, C., Gouker, F. E. & Daughtrey, M. Analyzing the structural shifts in US boxwood production due to boxwood blight. J. Environ. Hortic. 39, 91–99 (2021). Barker, B. S., Coop, L. & Hong, C. Potential distribution of invasive boxwood blight pathogen ( Calonectria pseudonaviculata ) as predicted by process-based and correlative models. Biology 11, 849 (2022). LaMondia, J. Fungicide efficacy against Calonectria pseudonaviculata , causal agent of boxwood blight. Plant Dis. 98, 99–102 (2014). Lombard, L., Crous, P. W., Wingfield, B. D. & Wingfield, M. J. Systematics of Calonectria : a genus of root, shoot and foliar pathogens. Stud. Mycol. 66, 31–69 (2010). Gehesquière, B. et al. Characterization and taxonomic reassessment of the box blight pathogen Calonectria pseudonaviculata , introducing Calonectria henricotiae sp. nov. Plant Pathol. 65, 37–52 (2016). Dart, N., Hong, C., Craig, C. A., Fry, J. & Hu, X. Soil inoculum production, survival, and infectivity of the boxwood blight pathogen, Calonectria pseudonaviculata . Plant Dis. 99, 1689–1694 (2015). Gehesquière, B. Cylindrocladium buxicola nom. cons. prop .(syn. Calonectria pseudonaviculata) on Buxus: molecular characterization, epidemiology, host resistance and fungicide control. PhD thesis, Ghent, Ghent University. (2014). LaMondia, J. A. & Maurer, K. Calonectria pseudonaviculata conidia dispersal and implications for boxwood blight management. Plant Health Progress 21, 232–237 (2020). Henricot, B. Box blight rampages onwards: The latest news on the spread and control of a devastating disease. Plantsman 5, 153–157 (2006). Henricot, B., Gorton, C., Denton, G. & Denton, J. Studies on the control of Cylindrocladium buxicola using fungicides and host resistance. Plant Dis. 92, 1273–1279 (2008). Kong, P. & Hong, C. Host responses and impact on the boxwood blight pathogen, Calonectria pseudonaviculata . Planta 249, 831–838 (2019). Madden, L. V., Hughes, G. & Van Den Bosch, F. The Study of Plant Disease Epidemics . (2007). Avenot, H., King, C., Edwards, T., Baudoin, A. & Hong, C. Effects of inoculum dose, temperature, cultivar, and interrupted leaf wetness period on infection of boxwood by Calonectria pseudonaviculata . Plant Dis. 101, 866–873 (2017). Avenot, H. F., Baudoin, A. & Hong, C. Conidial production and viability of Calonectria pseudonaviculata on infected boxwood leaves as affected by temperature, wetness, and dryness periods. Plant Pathol. 71, 696–701 (2022). Kodati, S., Allan-Perkins, E., Cowles, R. & LaMondia, J. Effect of temperature, leaf wetness period, and cultivar susceptibility on boxwood blight disease development and sporulation. Plant Dis. 107, 142–148 (2023). Van Maanen, A. & Xu, X. Modelling plant disease epidemics. Eur. J. Plant Pathol. 109, 669–682 (2003). Roubal, C., Regis, S. & Nicot, P. C. Field models for the prediction of leaf infection and latent period of Fusicladium oleagineum on olive based on rain, temperature and relative humidity. Plant Pathol. 62, 657–666 (2013). Cruz, C. D. et al. Climate suitability for Magnaporthe oryzae Triticum pathotype in the United States. Plant Dis. 100, 1979–1987 (2016). Schoeny, A. et al. Effect of pea canopy architecture on splash dispersal of Mycosphaerella pinodes conidia. Plant Pathol. 57, 1073–1085 (2008). Khaliq, I. et al. The role of conidia in the dispersal of Ascochyta rabiei . Eur. J. Plant Pathol. 158, 911–924 (2020). Coop, L. Brief documentation for boxwood blight infection risk model. Available at: https://uspest.org/wea/Boxwood_blight_risk_model_summaryV3.pdf [Accessed 14 September 2023]. (2023). Madeira, A., Kim, K., Taylor, S. & Gleason, M. A simple cloud-based energy balance model to estimate dew. Agric For Meteorol. 111, 55–63 (2002). Rowlandson, T. et al. Reconsidering leaf wetness duration determination for plant disease management. Plant Dis. 99, 310–319 (2015). Kim, K. S. et al. Spatial portability of numerical models of leaf wetness duration based on empirical approaches. Agric For Meteorol. 150, 871–880 (2010). Likins, T. et al. Preventing soil inoculum of Calonectria pseudonaviculata from splashing onto healthy boxwood foliage by mulching. Plant Dis. 103, 357–363 (2019). Wickham, H. et al. Welcome to the Tidyverse. J. Open Source Softw. 4, 1686 (2019). Grolemund, G. & Wickham, H. Dates and times made easy with lubridate. J. Stat. Softw. 40, 1–25 (2011). Kassambara, A. & Kassambara, M. Package `ggpubr`. R package version 0.1. Available at: https://rpkgs.datanovia.com/ggpubr/ [Accessed 20 September 2023]. (2020). VanDerWal, J., Falconi, L., Januchowski, S., Shoo, L. & Storlie, C. SDMTools: Species Distribution Modelling Tools: Tools for processing data associated with species distribution modelling exercises. R Package Version 1.1–221. Available at https://cran.r-project.org/src/contrib/Archive/SDMTools/ 1, 1 (2014). Roye, D. A heatmap as calendar: Available at: https://dominicroye.github.io/en/2020/a-heatmap-as-calendar/ . (2020). Sparks, A. nasapower: NASA-POWER data from R. R package version 4.0.10. Available at https://CRAN.R-project.org/package=nasapower . (2023). Bolker, B. M. GLMM FAQS. Available at https://bbolker.github.io/mixedmodels-misc/glmmFAQ.html . (2023). R: A language and environment for statistical computing (R Foundation for Statistical Computing. Available at: https://www.R-project.org/ [Accessed 20 February 2024], Vienna, Austria, 2021). Hartig, F. DHARMa: residual diagnostics for hierarchical (multi-level/mixed) regression models. Available at: https://cran.r-project.org/web/packages/DHARMa/vignettes/DHARMa.html [Accessed 15 September 2023]. (2019). Barton, K. & Barton, M. K. Package ‘mumin’. Version 1, 439 (2015). Lüdecke, D., Ben-Shachar, M. S., Patil, I., Waggoner, P. & Makowski, D. performance: An R package for assessment, comparison and testing of statistical models. J. Open Source Softw. 6 (2021). Kim, K. S., Taylor, S. E., Gleason, M., Villalobos, R. & Arauz, L. Estimation of leaf wetness duration using empirical models in northwestern Costa Rica. Agric For Meteorol. 129, 53–67 (2005). Hand, D. J. Assessing the performance of classification methods. International Statistical Review 80, 400–414 (2012). McFadden, D. Conditional Logit Analysis of Qualitative Choice Behavior . (New York: Academic Press, 1974). Gadoury, D. M. & MacHardy, W. E. A model to estimate the maturity of ascospores of Venturia inaequalis . Phytopathology 72, 901–904 (1982). Giosuè, S., Rossi, V., Ponti, I. & Bugiani, R. Estimating the dynamics of airborne ascospores of Venturia inaequalis . EPPO Bulletin 30, 137–142 (2000). Additional Declarations No competing interests reported. Supplementary Files Supplementaryinformation.docx Cite Share Download PDF Status: Published Journal Publication published 05 Nov, 2024 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 06 Aug, 2024 Reviews received at journal 29 Jul, 2024 Reviews received at journal 23 Jul, 2024 Reviewers agreed at journal 16 Jul, 2024 Reviewers agreed at journal 11 Jul, 2024 Reviewers invited by journal 11 Jul, 2024 Editor assigned by journal 11 Jul, 2024 Editor invited by journal 02 Jul, 2024 Submission checks completed at journal 01 Jul, 2024 First submitted to journal 27 Jun, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4651076","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":329779190,"identity":"bf5a1b38-0baa-459d-87f4-fd31900c092b","order_by":0,"name":"Ihsanul Khaliq","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA2UlEQVRIiWNgGAWjYBACAwYGZmYGBgseBvYGENeCaC0SPAw8B0BcCeK1AFECiE+EFnP+s4eNC2okZPhuPr+64UeBBAN/e3cCXi2WM/KSk2cck+CRvJ1TdrMH6DCJM2c34HfYDR7jwzxsEjwGt3PSbvAAtRhI5BLQcv4MUMs/oJabZ9Ju/iFKy4Ec42TeNqCWG+zHbhNny40cY2PePqBfzuSw3ZYxkOAh7Begw6R5vtnY8x0//uzmmz82cvztvfi1IMABHgMQxUOkcrAW9gckqB4Fo2AUjIKRBADU+UNbzFx5IgAAAABJRU5ErkJggg==","orcid":"","institution":"Virginia Tech","correspondingAuthor":true,"prefix":"","firstName":"Ihsanul","middleName":"","lastName":"Khaliq","suffix":""},{"id":329779196,"identity":"ea1664c0-f5ba-4807-a1ec-a2038900ab77","order_by":1,"name":"Herve F. Avenot","email":"","orcid":"","institution":"Virginia Tech","correspondingAuthor":false,"prefix":"","firstName":"Herve","middleName":"F.","lastName":"Avenot","suffix":""},{"id":329779198,"identity":"b585984a-a58f-4395-b75a-b100b379a1c7","order_by":2,"name":"Anton Baudoin","email":"","orcid":"","institution":"Virginia Tech","correspondingAuthor":false,"prefix":"","firstName":"Anton","middleName":"","lastName":"Baudoin","suffix":""},{"id":329779200,"identity":"4498de63-6086-47b6-9953-8190698ce412","order_by":3,"name":"Leonard Coop","email":"","orcid":"","institution":"Oregon State University","correspondingAuthor":false,"prefix":"","firstName":"Leonard","middleName":"","lastName":"Coop","suffix":""},{"id":329779202,"identity":"f21f8845-f1eb-4888-93c7-1f636c9fa3e1","order_by":4,"name":"Chuanxue Hong","email":"","orcid":"","institution":"Virginia Tech","correspondingAuthor":false,"prefix":"","firstName":"Chuanxue","middleName":"","lastName":"Hong","suffix":""}],"badges":[],"createdAt":"2024-06-27 22:38:19","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4651076/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4651076/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-024-76443-5","type":"published","date":"2024-11-05T15:58:05+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":60904661,"identity":"d4b57141-e02f-48d2-bd97-8f57d50df0d2","added_by":"auto","created_at":"2024-07-23 11:35:54","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":570943,"visible":true,"origin":"","legend":"\u003cp\u003eWeekly monitoring of boxwood blight using potted two-year-old detector plants \u003cem\u003eBuxus sempervirens\u003c/em\u003e ‘Suffruticosa’ in 2014 and ‘Justin Brouwers’ from 2015 to 2017. Detector plants were placed: (a) under infected boxwood canopies; (b) between two rows of infected boxwood; and (c) within plantainers containing infested leaf debris collected from infected boxwood nearby.\u003c/p\u003e","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-4651076/v1/82115c2c98ed8d7ae5911e43.png"},{"id":60904659,"identity":"85f1d30f-3726-46df-b559-01ac00dd3253","added_by":"auto","created_at":"2024-07-23 11:35:54","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":8415,"visible":true,"origin":"","legend":"\u003cp\u003eInfected leaf count recorded on two-year-old detector plants \u003cem\u003eBuxus sempervirens\u003c/em\u003e ‘Suffruticosa` (2014) and ‘Justin Brouwers’ (2015–2017) during each week. Data for three weeks in 2014, from August 18 to noon on September 8, were removed from the analyses because the rain gauge became clogged with leaves.\u003c/p\u003e","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-4651076/v1/0769a405d497fd9de4dabdd4.png"},{"id":60905185,"identity":"d6ddd4c9-0690-46b8-bc30-0869a124f1d6","added_by":"auto","created_at":"2024-07-23 11:43:54","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":63065,"visible":true,"origin":"","legend":"\u003cp\u003eA\u003cstrong\u003e \u003c/strong\u003egeneralized linear mixed model fit showing the significant positive effect of: (a) weekly mean relative humidity outside rainy periods, and (b) total rain on infected leaf count.\u003c/p\u003e","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-4651076/v1/7ed50cc1f7bd9091c3b84766.png"},{"id":60904662,"identity":"f4af722c-7967-4915-abee-5e6f256092c8","added_by":"auto","created_at":"2024-07-23 11:35:54","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":51284,"visible":true,"origin":"","legend":"\u003cp\u003eA generalized linear mixed model fit showing a significant positive interaction effect of weekly mean temperature during wet periods and weekly total leaf wetness duration on infected leaf count.\u003c/p\u003e","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-4651076/v1/0b1fe65611a8a771f7787c0d.png"},{"id":68750018,"identity":"5bf34fb2-48c8-45f4-8a5f-547e925a9efc","added_by":"auto","created_at":"2024-11-11 16:08:33","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1570123,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4651076/v1/e1c45120-a57f-49a5-8b99-b1b066d558b7.pdf"},{"id":60904663,"identity":"8e3a3c1e-e176-4611-a034-c198701a4ddc","added_by":"auto","created_at":"2024-07-23 11:35:54","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":12335091,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementaryinformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-4651076/v1/df1d3d6c5faaa56809ada0e9.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Epidemiology of boxwood blight in hotspots of western North Carolina and Virginia and validation of the boxwood blight infection risk model","fulltext":[{"header":"Introduction","content":"\u003cp\u003e Invasive pathogens are a major threat to plant health, diversity, and distribution in both natural and managed ecosystems worldwide \u003csup\u003e \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e \u003c/sup\u003e. Boxwood blight is a highly invasive emerging disease that causes leaf spots, stem lesions, and defoliation on all species of boxwood, one of the most iconic evergreen shrub and keystone forest species \u003csup\u003e \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e \u003c/sup\u003e. Boxwood blight was first discovered in the United Kingdom \u003csup\u003e \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e \u003c/sup\u003e and New Zealand in the mid-1990s \u003csup\u003e \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e \u003c/sup\u003e, and has now spread widely across Europe, Asia and North America. In the United States, boxwood blight was initially identified in North Carolina and Connecticut in 2011 \u003csup\u003e6\u003c/sup\u003e, and has now spread to 30 states \u003csup\u003e \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e \u003c/sup\u003e, risking more than 90% of boxwood production in the United States\u003csup\u003e \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e \u003c/sup\u003e. Total losses from boxwood blight have been significant, with Connecticut state alone incurring losses exceeding \u003cspan\u003e$\u003c/span\u003e3\u0026nbsp;million within the first year of detection \u003csup\u003e \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e \u003c/sup\u003e. Boxwood blight is caused by two invasive ascomycete fungi: \u003cem\u003eCalonectria pseudonaviculata\u003c/em\u003e \u003csup\u003e \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e \u003c/sup\u003e and \u003cem\u003eC. henricotiae\u003c/em\u003e \u003csup\u003e \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e \u003c/sup\u003e, but only \u003cem\u003eC. pseudonaviculata\u003c/em\u003e has been detected in the United States. Infested soil serves as an important source of inoculum \u003csup\u003e \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e \u003c/sup\u003e. The sexual stage/spores of the pathogen has not been observed, so both primary and secondary infections are attributed to conidial infection \u003csup\u003e \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e \u003c/sup\u003e. Conidia are produced within a sticky extracellular mucilaginous matrix that cannot be dispersed by wind alone, and thus the kinetic energy of rain or vectors are required to disperse conidia \u003csup\u003e \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e \u003c/sup\u003e. Long-distance spread is associated with human-mediated transport of infected plant material \u003csup\u003e \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e \u003c/sup\u003e. An infection cycle can be completed in less than a week under ideal conditions, with conidial germination beginning 3 h after inoculation, penetration occurring 5 h after inoculation, and new conidia developing within 7 days of inoculation \u003csup\u003e \u003cspan additionalcitationids=\"CR17\" citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e \u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThere have been no systematic studies on the temporal patterns of boxwood blight development. Knowledge of the temporal progression of diseases is needed for understanding disease dynamics and developing disease management guidelines, such as fungicide applications\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. Additionally, this knowledge can be used to develop a theoretical basis for determining whether an epidemic will occur as well as predicting the intensity of the disease \u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. There are significant knowledge gaps in the of epidemiology of boxwood blight. A few studies have been conducted in controlled environments to investigate the effect of relative humidity, temperature, and leaf wetness duration on boxwood blight development \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e,\u003cspan additionalcitationids=\"CR21\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. Studies investigating the effect of weather variables on boxwood blight development under field conditions are rare. Establishing the relationship between weather variables and disease development is the key component of developing disease forecasting systems \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. Knowledge of the weather conditions that favour disease development is also useful in developing strategies for limiting fungicide applications as fungicides are applied according to the risk of infection \u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. Additionally, this knowledge can be used to identify areas at risk where the pathogen is either present or has not (yet) been introduced\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThere is no epidemic without inoculum, so the design of forecasting systems is greatly influenced by the source and density of inoculum \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. The source of inoculum also plays a key role in the rain-splash dispersal of fungal spores. For instance, plant canopies affect rainfall characteristics by reducing the kinetic energy of falling rain drops, intercepting spore-carrying droplets, and creating a series of successive splashes often known as the relay effect of the canopy \u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. There are no systematic studies on how weather conditions, topographic factors (e.g., distance) and the source of inoculum interact to affect boxwood blight development. Information on the distance spores disperse from the inoculum source is valuable information for disease forecasting systems \u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. The role of microsclerotia and infected debris as the primary sources of inoculum is well known \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e. However, the role of infected boxwood canopies in boxwood blight epidemiology was not known. This information is crucial in understanding the main sources of inoculum and the routes of the pathogen\u0026rsquo;s dissemination to inform policy makers where to put limited resources to protect the protectable.\u003c/p\u003e \u003cp\u003eThe boxwood blight infection risk model \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e is the only publicly available model in the United States that predicts relative infection risk based on location-specific weather data. The model accumulates infection risk units (degree hours) during periods of leaf wetness. Leaf wetness is estimated by leaf wetness sensors connected to weather stations, where available. However, unlike other weather variables, leaf wetness is not a standard meteorological measurement \u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u003c/sup\u003e, and thus leaf wetness data are typically not available from official weather station networks \u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. Many boxwood cultivars have very dense canopies that tend to retain leaf wetness, at least in the interior of the plant, which could be rather different from that predicted from off-site weather stations. In the absence of leaf wetness sensors, the model estimates leaf wetness from other weather variables using fuzzy logic leaf wetness (FLLW) and precipitation-drying leaf wetness (FoxLW) algorithms \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. However, there have been no studies comparing the predicted disease risks based on leaf wetness data estimated by leaf wetness sensors and those derived from FLLW or FoxLW heuristic algorithms. The current set of rules used in calculating boxwood blight infection risk index, such as temperature thresholds for boxwood blight infection and the number of dry hours required to stop the infection process \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e was primarily derived from controlled environment experiments, and needs to be validated with field data.\u003c/p\u003e \u003cp\u003eTo address the above key knowledge gaps, boxwood blight was monitored in disease hotspots in western North Carolina and Virginia using detector boxwood plants that were rotated in and out of the field at weekly intervals from 2014 to 2017. Specific objectives were to: 1) investigate the temporal patterns of boxwood blight development on detector plants; 2) identify key weather variables affecting boxwood blight development under field conditions; and 3) validate the boxwood blight infection risk model using the actual blight disease readings on detector plants.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eTemporal patterns of boxwood blight development\u003c/h2\u003e \u003cp\u003eThe highest infected leaf count per week was recorded in late summer or early fall from 2014 to 2016 and during the last monitoring week in the late summer of 2017. At Lambsburg (2014), the highest infected leaf count per week was recorded in the week starting August 18 (422 leaves), a week for which we have no rainfall data, followed by September 8 (396 leaves) (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e2\u003c/span\u003e). At Lowgap, the highest infected leaf count per week was recorded in the week starting September 23 (971 leaves) in 2015, and the week starting September 28 (658 leaves) in 2016. The temporal patterns of boxwood blight development in 2017 were not directly comparable with those observed during 2015 and 2016 because detector plants were placed in the field earlier, in March and April, but not after mid-August, whereas in 2015 and 2016, detector plants were not placed in the field as early, and were exposed much later through the fall (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Nonetheless, the highest infected leaf count per week was recorded in the last monitoring week, from August 9 to 16 (160 leaves) (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eWeather data from the trial area for the past decade showed that late summer and early fall tended to have optimum monthly mean temperature range (18\u0026ndash;22\u0026deg;C) whereas monthly mean relative humidity tended to be high (\u0026gt;\u0026thinsp;75%) all year, and monthly total rainfall varied across years (Supplementary Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eKey weather variables affecting boxwood blight development\u003c/h2\u003e \u003cp\u003eDHARMa diagnostics showed that the data met the model assumptions. The R\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e value for the model was 0.69. The infected leaf count per plant significantly increased (p\u0026thinsp;=\u0026thinsp;0.0001) with increasing relative humidity outside of rainy periods (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Daily relative humidity outside rainy periods ranged from 21\u0026ndash;100%. Outside rainy periods, the weekly number of hours relative humidity exceeding 90% ranged from 12.7 to 122 h (daily range: 1\u0026ndash;17.4 h) and the mean weekly temperature during those hours ranged from \u0026minus;\u0026thinsp;1.2\u0026ndash;21.4\u0026deg;C (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). High infected leaf counts were recorded in weeks when humidity exceeded 90% for more than 34 hours, accompanied by a mean temperature above 10\u0026deg;C during those hours (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, Supplementary Fig. S2). Lower infected leaf counts were recorded in weeks when mean weekly relative humidity outside of rainy periods remained below 75%, while the highest infected leaf count was recorded when mean weekly relative humidity outside of rainy periods exceeded 90% (Supplementary Fig. S3). In six weeks during which boxwood blight developed despite the absence of rainfall, the mean weekly relative humidity was 76.6% (with 46.8 hours of relative humidity exceeding 90%) in week 4 of 2015, and 75.1% (with 48 hours of relative humidity exceeding 90%) in week 15 of 2015. In week 22 of 2015, it was 69.1% (with 54 hours of relative humidity exceeding 90%) and in week 26 (a 6-day week) of 2015 it was 56.1% (with 18.3 hours of relative humidity exceeding 90%). In week 11 and 21 (a 6-day week) of 2016, it was 78.3% (with 73.8 hours of relative humidity exceeding 90%), and 57.6% (with 22 hours of relative humidity exceeding 90%), respectively (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe weekly number of hours with relative humidity exceeding 90%, inside plus outside rainy periods, ranged from 15.2 to 141.7 h (daily range: 1.4\u0026ndash;20.2 h) and the mean weekly temperature when relative humidity exceeded 90% ranged from \u0026minus;\u0026thinsp;1.2\u0026ndash;21.5\u0026deg;C (Supplementary Table S2, Supplementary Fig. S4). High infected leaf counts were recorded in weeks when humidity exceeded 90% for more than 40 hours, accompanied by a mean temperature above 10\u0026deg;C when relative humidity remained above 90% (Supplementary Table S2, Supplementary Fig. S4).\u003c/p\u003e \u003cp\u003eThere was a significant positive interaction effect (p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) of leaf wetness duration and temperature during wet periods on the infected leaf count per plant (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e4\u003c/span\u003e). That is, the effect of leaf wetness duration on the infected leaf count depended on temperature. Higher infected leaf count was recorded in optimal temperatures (13.6\u0026ndash;22.7\u0026deg;C) during periods of prolonged leaf wetness, and vice versa. The weekly total leaf wetness duration ranged from 1.87 to 160 h, and the average daily leaf wetness duration ranged from 0.27\u0026ndash;22.9 h (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). The infected leaf count was very low when the weekly total leaf wetness duration was below 49 h and temperatures during wet periods were below 9\u0026deg;C (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, Supplementary Fig. S5). The highest infected leaf counts were recorded in weeks when temperatures during wet periods were between 13.6\u0026deg;C and 22.7\u0026deg;C, and the weekly total leaf wetness duration exceeded 65 hours (Supplementary Fig. S5).\u003c/p\u003e \u003cp\u003eRainfall had a significant (p\u0026thinsp;=\u0026thinsp;0.0049) positive effect on the infected leaf count (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The effect of rainfall on infected leaf count was especially pronounced when temperatures during rainy periods were between 16 and 22.7\u0026deg;C (Supplementary Fig. S6, Supplementary Fig. S7). Over the course of the 86-week experiment conducted across four years, there were seven weeks in which no rainfall was recorded (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e, Supplementary Fig. S8). Boxwood blight was recorded in six of those seven weeks, with infected leaf counts ranging from 1\u0026ndash;157 (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Conversely, boxwood blight was not recorded in 24 out of 86 weeks (28% of weeks) despite recording rainfall (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). These included 14 weeks recording between 0.4 to 10 mm rainfall, eight weeks recording between 11 to 35.8 mm rainfall, and two weeks recording 52.6 and 66 mm rainfall (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eValidation of the boxwood blight infection risk model\u003c/h2\u003e \u003cp\u003eBoxwood blight was recorded on detector plants in 61 of the 86 monitoring weeks (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Using the 250 degree-hours criterion for disease prediction, the SENSOR, ESTIM-LW and SENSOR-ADJ models predicted disease for 67, 76 and 65 weeks, respectively (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). The accuracy, precision and F-score values were almost similar for SENSOR, ESTIM-LW and SENSOR-ADJ models (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Supplementary Table S4). The SENSOR-ADJ model was 4% more specific than the SENSOR model and 20% more specific than the ESTIM-LW model. The ESTIM-LW model achieved the highest recall score (89%), followed by the SENSOR model (80%) and the SENSOR-ADJ model (79%) (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Supplementary Table S4).\u003c/p\u003e \u003cp\u003eUsing the 160-degree-hours criterion for disease prediction, the SENSOR, ESTIM-LW and SENSOR-ADJ model predicted disease for 73, 77 and 74 weeks, respectively (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). The accuracy, precision and F-score values were almost similar for SENSOR, ESTIM-LW and SENSOR-ADJ models (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Supplementary Table S4). The SENSOR model was 8% more specific than ESTIM-LW model and 4% more specific than the SENSOR-ADJ model. The ESTIM-LW model had higher recall score (89%) than both SENSOR and SENSOR-ADJ models (85% each) (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Supplementary Table S4).\u003c/p\u003e \u003cp\u003eUsing the 56-degree hours criterion for disease prediction, the SENSOR, ESTIM-LW and SENSOR-ADJ model predicted disease for 79, 80 and 79 weeks, respectively (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Almost similar accuracy, precision, recall and F-scores values were achieved (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Supplementary Table S4). The SENSOR and SENSOR-ADJ models showed 4% specificity, whereas the ESTIM-LW model failed to show any specificity (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Supplementary Table S4).\u003c/p\u003e \u003cp\u003eOverall, the model using the 250 degree-hours criterion for boxwood blight prediction achieved higher accuracy, AUC, and specificity scores than those using 160 and 56 degree-hours criteria. The ability of the model to predict true negatives (disease-free weeks or specificity) decreased as the degree hours criteria for boxwood blight prediction decreased (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Supplementary Table S4). Likewise, both the SENSOR and SENSOR-ADJ models exhibited a greater ability to predict true negatives and had fewer false positives compared to the ESTIM-LW model, especially when the 250 or 160 degree-hours criterion for boxwood blight prediction was used. The highest range of McFadden\u0026rsquo;s pseudo-R\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e value was recorded for the SENSOR-ADJ model (0.02\u0026ndash;0.48), followed by SENSOR model (0.02\u0026ndash;0.47), and ESTIM-LW model (0.03\u0026ndash;0.25).\u003c/p\u003e \u003cp\u003eThe highest range for weekly total accumulated blight risk index was observed for the ESTIM-LW model (29\u0026ndash;3950), followed by the SENSOR model (9\u0026ndash;3692), and the SENSOR-ADJ model (4\u0026ndash;3559) (Supplementary Fig. S9).\u003c/p\u003e \u003cp\u003eGeneralised linear models showed a significant association between the weekly total accumulated blight risk index and infected leaf count for three years (2014, 2015 and 2017) in the case of the SENSOR-ADJ and SENSOR models, and just a year (2015) in the case of ESTIM-LW model (Supplementary Fig. S10).\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eGeneralised linear mixed model\u0026rsquo;s coefficients for the effect of weather variables on boxwood blight development.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePredictor\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eEstimate\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStd. Error\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ez value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003ePr(\u0026gt;|z|)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLambsburg\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e-6.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2.49\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-2.71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.0068\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRelative humidity\u003csup\u003e\u0026dagger;\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.04\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.0001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTotal rain\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.03\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.0049\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLowgap\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e1.35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.0042\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eWind direction x Wind speed\u003csup\u003e\u0026dagger;\u0026dagger;\u003c/sup\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.5511\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eTemperature\u003csup\u003e\u0026sect;\u003c/sup\u003e x Leaf wetness duration\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.00\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e5.14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026lt;\u0026thinsp;0.0001\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e\u0026dagger; = Relative humidity was considered outside of rainy periods.\u003c/p\u003e \u003cp\u003e\u0026dagger;\u0026dagger; = Wind direction and wind speed were considered during rainy periods.\u003c/p\u003e \u003cp\u003e\u0026sect; = temperatures during wet periods were considered.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eModel evaluation metrics for the boxwood blight infection risk model using leaf wetness estimated by leaf wetness sensor (SENSOR and SENSOR-ADJ) and algorithms based on energy principles (ESTIM-LW).\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eDegree hours for \u003c/p\u003e \u003cp\u003edisease prediction\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"6\" nameend=\"c8\" namest=\"c3\"\u003e \u003cp\u003eModel evaluation metrics\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAccuracy (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003ePrecision (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eRecall (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eF-score (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eAUC (%)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eSpecificity (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSENSOR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e69\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eESTIM-LW\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e64\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e69\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSENSOR-ADJ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e69\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e42\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSENSOR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e160\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e51\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eESTIM-LW\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e160\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSENSOR-ADJ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e160\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e64\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSENSOR\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e250\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e28\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eESTIM-LW\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e250\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e66\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e12\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSENSOR-ADJ\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e250\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e65\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e32\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe observation of the highest weekly infected leaf counts on detector plants exposed to natural inoculum in August in 2014 and 2017 and September in 2015 to 2016 suggests that blight disease pressure in these areas is likely higher in late summer and early fall than other seasons. The highest infected leaf counts in early fall can mainly be attributed to prolonged leaf wetness duration resulting from high rainfall and/or greater number of hours with relative humidity exceeding 90% both outside and inside rainy periods. Hours with relative humidity exceeding 90% were especially pronounced during early fall (Supplementary Table S3), which may have contributed to inoculum build up preceding rainfall events and disease development following rainfall events.\u003c/p\u003e \u003cp\u003eA significant positive association between rainfall and infected leaf count on detector plants was not surprising. Rainfall not only provides the kinetic energy to release conidia from sticky mucilaginous matrix \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e, but also facilitates sporulation, germination, and host penetration by keeping host tissues wet. However, boxwood blight cannot be predicted solely on a forecasted rainfall event, as boxwood blight was not recorded in 24 of 86 weeks during which rainfall occurred, with the model using different leaf wetness criteria predicting disease for over 20 of those 24 weeks. Our findings also suggest that small and scattered rainfall (0.2\u0026ndash;1.2 mm) events occurring on different days are unlikely to cause high disease outbreaks. Conversely, small or large rainfall events occurring frequently throughout the day/week lead to high disease outbreaks, as the latter provide sufficient kinetic energy to dislodge more conidia and provide adequate leaf wetness to facilitate conidial germination, penetration, and subsequent infection.\u003c/p\u003e \u003cp\u003eGiven that \u003cem\u003eC. pseudonaviculata\u003c/em\u003e conidia cannot be dispersed by wind alone, and the absence of boxwood blight development in some weeks with high relative humidity but no rainfall, we propose including rainfall as a condition for spore dispersal in the model. It is difficult to determine an exact rainfall threshold for spore dispersal, as rainfall threshold can vary with rain intensity, the time of the day, inoculum density, and the level of resistance of the cultivar\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. In the present study, two infected leaves were recorded in the week starting October 19 of 2016, which received only 0.2 mm rainfall over 0.25 h, followed by a continuous 6-h leaf wetness due to high relative humidity. The hourly rainfall threshold for spore dispersal in forecasting models can perhaps be set at values greater than 0 or 0.2 mm.\u003c/p\u003e \u003cp\u003eHigh relative humidity outside rainy periods was found to be very important for boxwood blight development. This is not surprising because high relative humidity facilitates survival and infection by conidia laying on plants after dispersal by keeping host tissues moist \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Our results are consistent with findings from a study conducted in a controlled environment, where less than 65% humidity halted \u003cem\u003eC. pseudonaviculata\u003c/em\u003e infection on detached leaves \u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. However, high relative humidity-induced leaf wetness, when rainfall is absent or lower, is unlikely to trigger high disease outbreaks-even if temperatures are within the optimum range and high disease risk is predicted. This was particularly evident in the weeks starting September 15 and 22, 2014, during which boxwood blight did not develop although high relative humidity provided continuous leaf wetness for most parts of these weeks. The most likely reason for the lack of disease development was the low rainfall in these weeks: 0.6 mm and 2.8 mm, respectively. Rainfall was recorded on two separate days in the former week and on three separate days in the latter week.\u003c/p\u003e \u003cp\u003eThe significant positive interaction effect of optimum temperatures during periods of prolonged leaf wetness on infected leaf count was not unexpected because disease outbreaks are mostly associated with optimal temperatures during wet periods \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Their significant interaction effect in the present study was especially evident in 2017, where more than 70% (586 of 830 mm) of rainfall occurred between April and May, but negligible blight developed during these months because weekly mean temperatures during wet periods were mostly below 15.3\u0026deg;C. These results are aligned with those of Avenot, et al. \u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e who observed significantly higher sporulation, on detached leaves, with longer leaf wetness duration and temperatures near the optimum. The finding that temperature during periods with measurable rain or some wetness exceeded 30\u0026deg;C only a single and 26 times, respectively (based on 15-minute data points collected across four years) suggests that high temperatures are unlikely to impede boxwood blight infection in the study area as temperature drops quickly during rainfall.\u003c/p\u003e \u003cp\u003eThe non-significant effect of wind speed and wind direction during rainy periods on infected leaf count in the present study as well as recording the lowest infected leaf counts on detector plants placed beyond the dripline of canopy suggests that wind driven rain may not often play a big role in dispersing \u003cem\u003eC. pseudonaviculata\u003c/em\u003e conidia under the local field conditions, although this will require further testing as wind gusts remained below 3 m/s and wind was not omni-directional during the present study. However, we found that the blight development on detector plants placed beyond the dripline of canopies was associated with at least 27 mm rainfall per week or 3.86 mm mean rainfall per day, except for a single week starting June 9.\u003c/p\u003e \u003cp\u003eOverall, the model showed high accuracy and precision. The absence of disease, despite the model predicting disease in some weeks, can be attributed to light or scattered rainfall and decreased inoculum viability. Specifically, during the years 2014 to 2016, boxwood blight was predicted due to wet conditions resulting from light or scattered rainfall (mostly ranging from 0.2 to 1.2 mm on different days), high relative humidity outside rainy periods, and optimal temperatures during wet periods. However, these light and scattered rainfall events were insufficient to trigger the blight development. Using the 250 degree-hours criterion for boxwood blight prediction, the reasons for underpredicting disease for 6, 10 and 13 weeks by ESTIM-LW, SENSOR, and SENSOR-ADJ models, respectively can mainly be attributed to the ability of the pathogen to cause disease below the lower temperature threshold (6.7\u0026deg;C) for infection during wet periods. For instance, in the weeks starting November 9 (8-day exposure period) and November 17 (6-day exposure period) in 2016; the hourly mean temperatures during wet periods remained below the lower threshold for infection (6.7\u0026deg;C), except for 5 and 4 hours, respectively. The lower temperature threshold for infection can perhaps be set at 5.6\u0026deg;C when 250 degree-hour threshold is used. The increased false positives at 160 and 56-degree hours criteria for boxwood blight prediction suggests that the current 250 degree-hours criterion is close to optimal. The observation of significant association between the weekly total accumulated blight risk index and infected leaf counts for three out of four years for the SENSOR and SENSOR-ADJ models compared to a significant association in just a single year in the ESTIM-LW model suggests that the model using leaf wetness estimated by leaf wetness sensor should be chosen if leaf wetness sensors are available. Finally, the reasons for under-prediction or over-prediction are not known for certain weeks. For instance, the reasons for the absence of disease in three weeks (week 1 in 2014 and week 10 and 12 in 2015) for which boxwood blight was predicted are not known, as these weeks recorded 6.2, 18.6, and 11.8 mm rainfall, respectively. The individual continuous rain spells in these weeks reached up to 3, 8.4 and 4.8 mm, respectively, while temperatures during wet periods were also within the optimum range for infection. Likewise, the reasons for recording boxwood blight, despite the model not predicting disease, in the week starting November 11, 2015 are not clear as no rainfall was recorded and the weekly total leaf wetness duration was 2.1 hour; therefore, the minimum leaf wetness duration requirement for infection was not met.\u003c/p\u003e \u003cp\u003eConducting this study in boxwood blight hotspots improved our understanding of the pathogen\u0026rsquo;s \u0026ldquo;behaviour\u0026rdquo; and informs policy makers and managers where to place limited resources to protect the iconic evergreen shrub and key stone forest species from further invasion by the pathogen. Although boxwood blight is highly weather dependent and high disease outbreaks are expected whenever weather conditions are favourable for boxwood blight development, boxwood growers are advised to especially stay vigilant in early fall in western North Carolina and Virginia. Both infected canopies and infested leaf debris are important sources of inoculum for disease epidemics. The chances of high disease outbreaks greatly decreased with increasing distance from the inoculum source, emphasizing the need for extensive hygiene, adequate spacing between plants, and mulching \u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. Recording different infected leaf counts on detector plants exposed to different sources of inoculum and/or distance from the source of inoculum suggests that disease forecasting models should also consider inoculum source and spatial factors (e.g. distance from the inoculum source). Although desirable, such improvements in prediction models will be challenging. Even for diseases like apple scab, which has been the subject of extensive modelling efforts, the quantification of risk has been predominantly focused on predicting the release of primary inoculum based on weather factors \u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e,\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eStudy areas description\u003c/h2\u003e \u003cp\u003eThis study was conducted in two US locations: Lambsburg (approx. 36\u0026deg;35\u0026rsquo;N, 80\u0026deg;46\u0026rsquo;W), Virginia (VA) in 2014, and Lowgap (approx. 36\u0026deg;31\u0026rsquo;N, 80\u0026deg;50\u0026rsquo;W), North Carolina (NC), from 2015 to 2017. The Lambsburg trial was conducted in a rural residential landscape near a site where boxwood blight was first detected in the fall of 2011. No boxwood blight control activities were carried out at the site since at least 2013. The Lowgap trial was conducted in a former commercial field nursery where boxwood blight had become established by 2013 \u003csup\u003e32\u003c/sup\u003e. No management activities had been carried out to control boxwood blight since its establishment, but insecticides were applied in the spring to control boxwood leafminer (\u003cem\u003eMonarthropalpus flavus\u003c/em\u003e). Both locations had loamy soil and were surrounded by deciduous forest.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eBoxwood blight monitoring using detector plants\u003c/h2\u003e \u003cp\u003eTwo-year-old blight-free \u003cem\u003eBuxus sempervirens\u003c/em\u003e \u0026lsquo;Suffruticosa\u0026rsquo; and \u0026lsquo;Justin Brouwers\u0026rsquo; plants, donated by Saunders Brothers Inc. (Piney River, VA), were kept in a greenhouse until used as detector plants. Plants were grown in 3.79-L containers containing potting mix that consisted of 56% composted pine bark, 32% coarse Perlite, and 12% peat moss. Plants were watered daily with tap water and fertilized with Osmocote\u0026reg; Plus 15-9-12 (Everris NA Inc., Dublin, OH) when required.\u003c/p\u003e \u003cp\u003eIn Lambsburg, boxwood blight monitoring was performed for a total of 20 weeks from May to November 2014 (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Each week, a total of nine potted \u003cem\u003eB. sempervirens\u003c/em\u003e \u0026lsquo;Suffruticosa\u0026rsquo; detector plants were placed at the site to expose to natural inoculum and prevailing environmental conditions. Of the nine detector plants, three were randomly placed under already infected boxwood canopies (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). Another set of three plants was placed between two rows of infected boxwood, with approximately 6\u0026ndash;8 feet distance between the rows: henceforth called \u0026ldquo;between-row\u0026rdquo; treatment (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e1\u003c/span\u003eB). The between-row treatment was included only in 2014 since it experienced only limited blight development in that year. The remaining three plants were placed in three plantainers (round plastic pans about 60 cm in diameter, MacCourt Products Inc, Denver, CO) containing infested leaf debris collected from surrounding boxwood (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). One detector plant was installed per plantainer.\u003c/p\u003e \u003cp\u003eIn Lowgap, boxwood blight monitoring was conducted for a total of 26 weeks from May to November in 2015, 24 weeks from June to December in 2016, and 16 weeks from March to August in 2017, using eight \u003cem\u003eB. sempervirens\u003c/em\u003e \u0026lsquo;Justin Brouwers\u0026rsquo; detector plants each week (Table\u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Of the eight detector plants, four were randomly placed under already infected boxwood canopies. To prevent the pathogen from infested soil spreading onto detector plants by rain splash, a polyethylene landscape fabric (Vigoro Corporation, Lake Forest, IL) was installed in sections of 2 x 1 m around the base of those detector plants (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). The remaining plants were placed within four plantainers containing infested debris collected from surrounding boxwood. In 2017, the early-season exposure period of detector plants was 28 days in \u0026ldquo;week 1\u0026rdquo; and 21 days in both \u0026ldquo;week 2\u0026rdquo; and \u0026ldquo;week 3\u0026rdquo;: these three longer-than-a-week exposure periods are referred to as \u0026ldquo;weeks\u0026rdquo; for uniformity.\u003c/p\u003e \u003cp\u003eAt both locations, it was ensured that adequate blighted leaves were present on canopies, and completely defoliated plants were not used. In the case of detector plants placed within plantainers, efforts were made to use the similar amount of inoculum each week by distributing infested leaf debris in a 5 cm layer around each detector plant. After a week of field exposure, a new set of detector plants was introduced into the field, while the previously exposed set was transferred to a lighted growth chamber (14 h photoperiod) set at 21\u0026deg;C for 24 h. The next day, the infected leaf count per plant was recorded. The detector plants were then kept in the lighted growth chamber at 21\u0026deg;C for another week to allow maximum lesion development, and the infected leaf count per plant was recorded again. Each week, the date and time detector plants were placed and removed from the field were recorded (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). Infested leaf debris was replenished regularly to account for inoculum depletion.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003eWeather monitoring and data processing\u003c/h2\u003e \u003cp\u003eFor the entire duration of boxwood blight monitoring on detector plants in the field, weather data were recorded at 15-minute intervals using an Em50G data logger (Decagon Devices Inc. (now METER Group), Pullman, WA) equipped with the following instruments: anemometer (model Davis Cup), pyranometer (model PYR), rain gauge (model ECRN-100), leaf wetness sensor (model LWS), and temperature, humidity, and vapor pressure sensor (model VP-3). The weather station was placed next to infected canopies at both locations.\u003c/p\u003e \u003cp\u003eTo identify key weather variables affecting boxwood blight development, weather data recorded at 15-minute intervals were summarized to weekly means for temperature during wet periods, relative humidity outside rainy periods, wind speed, and wind direction during rainy periods, and to weekly totals for rainfall, rain duration, and leaf wetness duration. Data for three weeks in 2014, from August 18 to noon on September 8, were removed from the analyses because the rain gauge became clogged with leaves during these weeks. To ensure weather data were comparable among monitoring weeks, rainfall and leaf wetness duration data collected during the first three extended monitoring periods in early 2017 were converted into per-week data, but no adjustments were made for 5-, 6- or 8-day weeks (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe infected leaf count data, recorded on detector plants placed between infected boxwood rows, were only included in the descriptive analyses on how weather conditions and distance may affect disease dynamics on detector plants exposed to each inoculum source.\u003c/p\u003e \u003cp\u003eData were processed and visualised in the R programming language using packages \u0026lsquo;tidyverse\u0026rsquo; \u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e, \u0026lsquo;lubridate\u0026rsquo; \u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e, and \u0026lsquo;ggpubr\u0026rsquo; \u003csup\u003e\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. Mean wind direction was calculated using the \u0026lsquo;circular.averaging()\u0026rsquo; function of the \u0026lsquo;SDMTools\u0026rsquo; package \u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. Calendar heatmaps were created using the custom function `theme_calendar()` function\u003csup\u003e\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eTemporal patterns of boxwood blight development\u003c/h2\u003e \u003cp\u003eInfected leaf count data recorded on detector plants placed under infected canopies and those exposed to leaf debris inoculum in plantainers were used to investigate the temporal patterns of boxwood blight development since these two treatments were used in all four years of this study. To help understand and interpret the observed temporal patterns of boxwood blight development, additional weather data were obtained from the NASA POWER website (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://power.larc.nasa.gov/\u003c/span\u003e\u003cspan address=\"https://power.larc.nasa.gov/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) for the trial sites over a period of 12 years from 2010 to 2021 using the nasapower R package \u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e. Hourly weather observations were summarised into monthly means for relative humidity, temperature, and monthly total for rainfall (Supplementary Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e) to visualize the seasonal weather patterns and identify time of year when temperature, relative humidity and rainfall were all in the conducive range for boxwood blight.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eKey weather variables affecting boxwood blight development\u003c/h2\u003e \u003cp\u003eOnly infected leaf count data recorded on detector plants placed under infected canopies and those exposed to leaf debris inoculum were used to identify key weather variables affecting boxwood blight development. The `cor()` function was used to generate a correlation matrix to identify pairs of predictors with high correlation coefficients (\u0026gt;\u0026thinsp;0.7). The correlation matrix identified a very high correlation (r\u0026thinsp;=\u0026thinsp;0.86) between rain amount and rain duration, so rain duration was removed from the multivariate analysis to avoid multicollinearity. The effect of weather variables and weekly total leaf wetness duration on infected leaf count per plant was investigated by fitting generalised linear mixed models (glmms) using the `glmmTMB` function of the \u0026lsquo;glmmTMB\u0026rsquo; package \u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e in the R programming language \u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. Rainfall and relative humidity were included as additive terms, while leaf wetness duration and temperature, and wind speed and wind direction were included as interactive terms. A negative binomial family was used because the preliminary analysis showed the data to be overdispersed, i.e., variance was greater than mean. The overdispersion was tested using Bolker\u0026rsquo;s custom function \u0026lsquo;overdisp_fun()` \u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. Weather variables and leaf wetness duration were included as fixed effects, and the predictor location was included as a covariate. The model failed to converge when both \u0026lsquo;cultivar\u0026rsquo; and \u0026lsquo;week\u0026rsquo; detector plants were placed in the field were included as random effects. Additionally, \u0026lsquo;cultivar\u0026rsquo; explained very little variation (variance p\u0026thinsp;\u0026lt;\u0026thinsp;0.0001) when included as a random effect in the model. Subsequently, only the predictor/independent variable \u0026lsquo;week\u0026rsquo; was included as a random effect.\u003c/p\u003e \u003cp\u003eModel diagnostics was performed using the DHARMa package, which uses a simulation-based approach via the function `simulateResiduals` to create readily interpretable scaled residuals for glmms \u003csup\u003e\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e. The marginal coefficient of determination (pseudo- R\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e) was calculated using the function `r.squaredGLMM` of the `MuMIn` R package \u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. The autocorrelation between residuals was checked using `check_autocorrelation()` function of the performance package \u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e. Analysis was based on pooled data across all years for both locations.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eValidation of the boxwood blight infection risk model\u003c/h2\u003e \u003cp\u003eThe boxwood blight infection risk model is a degree-hour based model that predicts hourly infection risk during wet periods \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. Degree-hours for the model are calculated, using piecewise regression, based on the current knowledge about the lower, optimum, and upper temperature thresholds for infection, and the number of dry hours required to stop the infection process. For each hour, if leaf wetness is not recorded, or the temperature is less than 44\u0026deg;F (6.7\u0026deg;C) or greater than 86\u0026deg;F (30\u0026deg;C), the infection risk index is set to 0 for that hour. Otherwise, the infection risk index is calculated from a lookup table with information on degree-hours for corresponding temperatures \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. The infection process is halted, and the index is reset to zero when a dry period exceeds 5 hours. Where a leaf wetness sensor is not available, leaf wetness is estimated using the FLLW algorithm, which is based on energy principles, from temperature, dewpoint, and wind speed, or the FoxLW heuristic algorithm from rainfall, wind speed, time of day, and season of the year \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e,\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e. Leaf wetness estimated by both algorithms ranges from 0\u0026ndash;10, with 0 denoting dry and 1\u0026thinsp;+\u0026thinsp;denoting wetter conditions. Both algorithms run in parallel and independent of each other; their values are not added together. The algorithm estimating the maximum value for leaf wetness was chosen to estimate leaf wetness for this study, which is also the default setting of the model. Readers are referred to the model\u0026rsquo;s documentation \u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e for more detailed information on the model.\u003c/p\u003e \u003cp\u003eThe \u0026ldquo;Ignore Leaf Wetness Sensor Data\u0026rdquo; option was unchecked or checked to run the model, depending on whether data recorded by leaf wetness sensors connected to on-site weather stations (henceforth referred to as SENSOR and SENSOR-ADJ models) or that estimated by FLLW or FoxLW heuristic algorithms (henceforth referred to as ESTIM-LW model) were used. All models were reset when detector plants were exchanged after each monitoring week. In the SENSOR-ADJ model, the increments from the lookup table (degree-hours) for corresponding temperatures were made proportional to the duration of leaf wetness recorded by the leaf wetness sensors during each hour. This was achieved by multiplying each degree-hour by 'x/60', where 'x\u0026rsquo; represents the duration (in minutes) of leaf wetness, and 60 represents the total number of minutes in that hour.\u003c/p\u003e \u003cp\u003eThe threshold for predicting disease in each monitoring week was the continuous accumulation of 250 degree-hours (\u0026deg;F) before the infection process stopped, which corresponds to the prediction of 6 lesions in a highly susceptible cultivar\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. Additionally, we investigated the continuous accumulation of 56 and 160 degree-hours (which corresponds to the prediction of first infection in highly susceptible and susceptible cultivars, respectively) as a criterion for disease prediction to compare their prediction accuracy to the currently used degree-hours threshold (250).\u003c/p\u003e \u003cp\u003eBinomial data were generated for each monitoring week, where weeks during which boxwood blight was recorded on detector plants were labelled as 1, and 0 otherwise. For disease prediction, weeks in which the continuously accumulated degree hours reached 56, 160, and 250-degree hours at least once were labelled as 1 (indicating boxwood blight was predicted) and 0 otherwise (indicating boxwood blight was not predicted) for respective degree hour thresholds,\u003c/p\u003e \u003cp\u003eTo compare the disease risk predicted by the model using three leaf wetness criteria (recorded, recorded with degree-hours made proportional to the duration of leaf wetness recorded, and estimated) and three degree-hour thresholds (56, 160, 250) with disease observed on detector plants, confusion matrices for the model were obtained, and metrics such as area under the receiver operating curve (AUC), accuracy, specificity, precision, recall, specificity, and F1-scores were calculated \u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e\u003c/sup\u003e. AUC indicates the ability of the model to distinguish between the classes 0 and 1. Accuracy is the ratio of correct predictions (true positive and true negative) among total number of predictions. Specificity is proportion of true negatives correctly identified by the model. Precision is the ratio of correctly classified positive cases (true positive) to a total number of classified positive cases (either correctly or incorrectly, i.e., precision\u0026thinsp;=\u0026thinsp;true positives/true positives\u0026thinsp;+\u0026thinsp;false positives). Recall is the ratio between the numbers of positive cases correctly classified as positive to the total number of positive samples (recall\u0026thinsp;=\u0026thinsp;true positives/true positives\u0026thinsp;+\u0026thinsp;false negatives). The F-score provides a harmonic mean between precision and recall to describe the overall accuracy of a model. The F-score is calculated by the following formula: F1\u0026thinsp;=\u0026thinsp;2 \u0026times; (precision \u0026times; recall) / (precision\u0026thinsp;+\u0026thinsp;recall). The values for these metrics were expressed in percentages, with values closer to 100 indicating a better model. Finally, the association between weekly total accumulated blight risk index (degree hours) and the infected leaf count on detector recorded during each monitoring week was investigated by fitting generalised linear models with a quasi-Poisson family. The model\u0026rsquo;s fit was evaluated using McFadden\u0026rsquo;s pseudo-R\u003csup\u003e2 46\u003c/sup\u003e.\u003c/p\u003e \u003c/div\u003e"},{"header":"Declarations","content":"\u003ch3\u003e\u003cstrong\u003e\u003cem\u003eCompeting interests\u003c/em\u003e\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003eThe authors declare no competing interests.\u0026nbsp;\u003c/p\u003e\n\u003ch3\u003e\u003cstrong\u003e\u003cem\u003eCorrespondence\u003c/em\u003e\u003c/strong\u003e\u003c/h3\u003e\n\u003cp\u003e\u0026nbsp;All requests for materials should be addressed to CH.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eHA and AB conducted field trials. IK, AB and LC analysed the data. I.K wrote the initial draft of the manuscript. CH and AB secured funding. All authors reviewed the manuscript and gave final approval for publication.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003eWe gratefully acknowledge Sasha Marine, Jordan Craddock, and Tyler Edwards (Virginia Tech) for their contributions to field work and data collection. We also thank Ping Kong (Virginia Tech) for her invaluable comments on the manuscript. Data collection was supported in part by the 2013 Farm Bill through USDA Animal and Plant Health Inspection Service. Data analyses and manuscript preparation were supported by the US Department of Agriculture\u0026mdash;National Institute of Food and Agriculture, under award number 2020-51181-32135.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eAll raw and generated data used in the statistical analyses and data visualization have been made available as a part of a research compendium for reproducibility. Please see raw and generated data and the fully reproducible code at https://github.com/IhsanKhaliq/epiboxwoodblight\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eKhaliq, I., Burgess, T. I., Hardy, G. E. S. J., White, D. \u0026amp; McDougall, K. L. \u003cem\u003ePhytophthora\u003c/em\u003e and vascular plant species distributions along a steep elevation gradient. Biol. Invasions 23, 1443\u0026ndash;1459 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFisher, M. C. \u003cem\u003eet al.\u003c/em\u003e Emerging fungal threats to animal, plant and ecosystem health. Nature 484, 186\u0026ndash;194 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDaughtrey, M. L. Boxwood blight: Threat to ornamentals. Annu. Rev. Phytopathol. 57, 189\u0026ndash;209 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHenricot, B., Sierra, A. P. \u0026amp; Prior, C. A new blight disease on \u003cem\u003eBuxus\u003c/em\u003e in the UK caused by the fungus \u003cem\u003eCylindrocladium\u003c/em\u003e. Plant Pathol. 49, 805 (2000).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRidley, G. New plant fungus found in Auckland box hedges (\u003cem\u003eBuxus\u003c/em\u003e). For. health news 77, 1\u0026ndash;2 (1998).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIvors, K. \u003cem\u003eet al.\u003c/em\u003e First report of boxwood blight caused by \u003cem\u003eCylindrocladium pseudonaviculatum\u003c/em\u003e in the United States. Plant Dis. 96, 1070 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHong, C. United States map of boxwood blight by the time of its first confirmed invasion. Available at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://irp.cdn-website.com/217658e5/files/uploaded/US%20Boxwood%20Blight%20Map%207-2023.png\u003c/span\u003e\u003cspan address=\"https://irp.cdn-website.com/217658e5/files/uploaded/US%20Boxwood%20Blight%20Map%207-2023.png\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHall, C. R., Hong, C., Gouker, F. E. \u0026amp; Daughtrey, M. Analyzing the structural shifts in US boxwood production due to boxwood blight. J. Environ. Hortic. 39, 91\u0026ndash;99 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarker, B. S., Coop, L. \u0026amp; Hong, C. Potential distribution of invasive boxwood blight pathogen (\u003cem\u003eCalonectria pseudonaviculata\u003c/em\u003e) as predicted by process-based and correlative models. Biology 11, 849 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLaMondia, J. Fungicide efficacy against \u003cem\u003eCalonectria pseudonaviculata\u003c/em\u003e, causal agent of boxwood blight. Plant Dis. 98, 99\u0026ndash;102 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLombard, L., Crous, P. W., Wingfield, B. D. \u0026amp; Wingfield, M. J. Systematics of \u003cem\u003eCalonectria\u003c/em\u003e: a genus of root, shoot and foliar pathogens. Stud. Mycol. 66, 31\u0026ndash;69 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGehesqui\u0026egrave;re, B. \u003cem\u003eet al.\u003c/em\u003e Characterization and taxonomic reassessment of the box blight pathogen \u003cem\u003eCalonectria pseudonaviculata\u003c/em\u003e, introducing \u003cem\u003eCalonectria henricotiae\u003c/em\u003e sp. nov. Plant Pathol. 65, 37\u0026ndash;52 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDart, N., Hong, C., Craig, C. A., Fry, J. \u0026amp; Hu, X. Soil inoculum production, survival, and infectivity of the boxwood blight pathogen, \u003cem\u003eCalonectria pseudonaviculata\u003c/em\u003e. Plant Dis. 99, 1689\u0026ndash;1694 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGehesqui\u0026egrave;re, B. \u003cem\u003eCylindrocladium buxicola nom. cons. prop\u003c/em\u003e.(syn. \u003cem\u003eCalonectria pseudonaviculata)\u003c/em\u003e on \u003cem\u003eBuxus: molecular characterization, epidemiology, host resistance and fungicide control.\u003c/em\u003e PhD thesis, Ghent, Ghent University. (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLaMondia, J. A. \u0026amp; Maurer, K. \u003cem\u003eCalonectria pseudonaviculata\u003c/em\u003e conidia dispersal and implications for boxwood blight management. Plant Health Progress 21, 232\u0026ndash;237 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHenricot, B. Box blight rampages onwards: The latest news on the spread and control of a devastating disease. Plantsman 5, 153\u0026ndash;157 (2006).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHenricot, B., Gorton, C., Denton, G. \u0026amp; Denton, J. Studies on the control of \u003cem\u003eCylindrocladium buxicola\u003c/em\u003e using fungicides and host resistance. Plant Dis. 92, 1273\u0026ndash;1279 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKong, P. \u0026amp; Hong, C. Host responses and impact on the boxwood blight pathogen, \u003cem\u003eCalonectria pseudonaviculata\u003c/em\u003e. Planta 249, 831\u0026ndash;838 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMadden, L. V., Hughes, G. \u0026amp; Van Den Bosch, F. \u003cem\u003eThe Study of Plant Disease Epidemics\u003c/em\u003e. (2007).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAvenot, H., King, C., Edwards, T., Baudoin, A. \u0026amp; Hong, C. Effects of inoculum dose, temperature, cultivar, and interrupted leaf wetness period on infection of boxwood by \u003cem\u003eCalonectria pseudonaviculata\u003c/em\u003e. Plant Dis. 101, 866\u0026ndash;873 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eAvenot, H. F., Baudoin, A. \u0026amp; Hong, C. Conidial production and viability of \u003cem\u003eCalonectria pseudonaviculata\u003c/em\u003e on infected boxwood leaves as affected by temperature, wetness, and dryness periods. Plant Pathol. 71, 696\u0026ndash;701 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKodati, S., Allan-Perkins, E., Cowles, R. \u0026amp; LaMondia, J. Effect of temperature, leaf wetness period, and cultivar susceptibility on boxwood blight disease development and sporulation. Plant Dis. 107, 142\u0026ndash;148 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Maanen, A. \u0026amp; Xu, X. Modelling plant disease epidemics. Eur. J. Plant Pathol. 109, 669\u0026ndash;682 (2003).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoubal, C., Regis, S. \u0026amp; Nicot, P. C. Field models for the prediction of leaf infection and latent period of \u003cem\u003eFusicladium oleagineum\u003c/em\u003e on olive based on rain, temperature and relative humidity. Plant Pathol. 62, 657\u0026ndash;666 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCruz, C. D. \u003cem\u003eet al.\u003c/em\u003e Climate suitability for \u003cem\u003eMagnaporthe oryzae\u003c/em\u003e Triticum pathotype in the United States. Plant Dis. 100, 1979\u0026ndash;1987 (2016).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchoeny, A. \u003cem\u003eet al.\u003c/em\u003e Effect of pea canopy architecture on splash dispersal of \u003cem\u003eMycosphaerella pinodes\u003c/em\u003e conidia. Plant Pathol. 57, 1073\u0026ndash;1085 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhaliq, I. \u003cem\u003eet al.\u003c/em\u003e The role of conidia in the dispersal of \u003cem\u003eAscochyta rabiei\u003c/em\u003e. Eur. J. Plant Pathol. 158, 911\u0026ndash;924 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCoop, L. Brief documentation for boxwood blight infection risk model. Available at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://uspest.org/wea/Boxwood_blight_risk_model_summaryV3.pdf\u003c/span\u003e\u003cspan address=\"https://uspest.org/wea/Boxwood_blight_risk_model_summaryV3.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e [Accessed 14 September 2023]. (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMadeira, A., Kim, K., Taylor, S. \u0026amp; Gleason, M. A simple cloud-based energy balance model to estimate dew. Agric For Meteorol. 111, 55\u0026ndash;63 (2002).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRowlandson, T. \u003cem\u003eet al.\u003c/em\u003e Reconsidering leaf wetness duration determination for plant disease management. Plant Dis. 99, 310\u0026ndash;319 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim, K. S. \u003cem\u003eet al.\u003c/em\u003e Spatial portability of numerical models of leaf wetness duration based on empirical approaches. Agric For Meteorol. 150, 871\u0026ndash;880 (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLikins, T. \u003cem\u003eet al.\u003c/em\u003e Preventing soil inoculum of \u003cem\u003eCalonectria pseudonaviculata\u003c/em\u003e from splashing onto healthy boxwood foliage by mulching. Plant Dis. 103, 357\u0026ndash;363 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWickham, H. \u003cem\u003eet al.\u003c/em\u003e Welcome to the Tidyverse. J. Open Source Softw. 4, 1686 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGrolemund, G. \u0026amp; Wickham, H. Dates and times made easy with lubridate. J. Stat. Softw. 40, 1\u0026ndash;25 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKassambara, A. \u0026amp; Kassambara, M. Package `ggpubr`. R package version 0.1. Available at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://rpkgs.datanovia.com/ggpubr/\u003c/span\u003e\u003cspan address=\"https://rpkgs.datanovia.com/ggpubr/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e [Accessed 20 September 2023]. (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVanDerWal, J., Falconi, L., Januchowski, S., Shoo, L. \u0026amp; Storlie, C. SDMTools: Species Distribution Modelling Tools: Tools for processing data associated with species distribution modelling exercises. \u003cem\u003eR Package Version 1.1\u0026ndash;221. Available at\u003c/em\u003e \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cran.r-project.org/src/contrib/Archive/SDMTools/\u003c/span\u003e\u003cspan address=\"https://cran.r-project.org/src/contrib/Archive/SDMTools/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e 1, 1 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRoye, D. A heatmap as calendar: Available at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://dominicroye.github.io/en/2020/a-heatmap-as-calendar/\u003c/span\u003e\u003cspan address=\"https://dominicroye.github.io/en/2020/a-heatmap-as-calendar/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSparks, A. nasapower: NASA-POWER data from R. R package version 4.0.10. Available at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://CRAN.R-project.org/package=nasapower\u003c/span\u003e\u003cspan address=\"https://CRAN.R-project.org/package=nasapower\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBolker, B. M. GLMM FAQS. Available at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://bbolker.github.io/mixedmodels-misc/glmmFAQ.html\u003c/span\u003e\u003cspan address=\"https://bbolker.github.io/mixedmodels-misc/glmmFAQ.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eR: A language and environment for statistical computing (R Foundation for Statistical Computing. Available at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.R-project.org/\u003c/span\u003e\u003cspan address=\"https://www.R-project.org/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e [Accessed 20 February 2024], Vienna, Austria, 2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHartig, F. DHARMa: residual diagnostics for hierarchical (multi-level/mixed) regression models. Available at: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://cran.r-project.org/web/packages/DHARMa/vignettes/DHARMa.html\u003c/span\u003e\u003cspan address=\"https://cran.r-project.org/web/packages/DHARMa/vignettes/DHARMa.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e [Accessed 15 September 2023]. (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarton, K. \u0026amp; Barton, M. K. Package \u0026lsquo;mumin\u0026rsquo;. Version 1, 439 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eL\u0026uuml;decke, D., Ben-Shachar, M. S., Patil, I., Waggoner, P. \u0026amp; Makowski, D. performance: An R package for assessment, comparison and testing of statistical models. J. Open Source Softw. 6 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKim, K. S., Taylor, S. E., Gleason, M., Villalobos, R. \u0026amp; Arauz, L. Estimation of leaf wetness duration using empirical models in northwestern Costa Rica. Agric For Meteorol. 129, 53\u0026ndash;67 (2005).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHand, D. J. Assessing the performance of classification methods. International Statistical Review 80, 400\u0026ndash;414 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcFadden, D. \u003cem\u003eConditional Logit Analysis of Qualitative Choice Behavior\u003c/em\u003e. (New York: Academic Press, 1974).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGadoury, D. M. \u0026amp; MacHardy, W. E. A model to estimate the maturity of ascospores of \u003cem\u003eVenturia inaequalis\u003c/em\u003e. Phytopathology 72, 901\u0026ndash;904 (1982).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGiosu\u0026egrave;, S., Rossi, V., Ponti, I. \u0026amp; Bugiani, R. Estimating the dynamics of airborne ascospores of \u003cem\u003eVenturia inaequalis\u003c/em\u003e. EPPO Bulletin 30, 137\u0026ndash;142 (2000).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Weather effect, forecasting, seasonal patterns, temporal patterns, trap plants, weekly exposure","lastPublishedDoi":"10.21203/rs.3.rs-4651076/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4651076/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBoxwood blight is a highly invasive emerging disease. Since the first US report in North Carolina and Connecticut in 2011, boxwood blight has spread to over 30 US states, risking more than 90% of boxwood production. A boxwood blight infection risk model was developed from limited studies in controlled environments. Our study investigated the disease field epidemiology and validated the model\u0026rsquo;s prediction, using leaf wetness estimated by leaf wetness sensor or algorithms, by analysing weekly blight monitoring data collected on detector plants exposed to the prevailing environmental conditions from spring through fall of 2014 to 2017. Boxwood blight was recorded in 61 of 86 weeks, with the highest infected leaf count recorded in late summer or early fall. Rainfall, high relative humidity outside rainy periods and optimal temperatures during prolonged leaf wetness had a significant positive effect on boxwood blight development. Classification analyses showed that disease predictions from the model using leaf wetness estimated by leaf wetness sensor were more closely aligned with observations from the field than predictions based on algorithms. This study improved our understanding of disease field epidemiology, provided leads to improve the existing model, and generated essential knowledge for formulating effective strategies for blight mitigation.\u003c/p\u003e","manuscriptTitle":"Epidemiology of boxwood blight in hotspots of western North Carolina and Virginia and validation of the boxwood blight infection risk model","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-07-23 11:35:49","doi":"10.21203/rs.3.rs-4651076/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-08-06T06:13:41+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-07-29T20:09:54+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-07-24T02:07:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"172557724161399037463719233450596840429","date":"2024-07-16T13:13:09+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"339370713338913135778419654851867162268","date":"2024-07-11T16:01:13+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-07-11T13:07:48+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-07-11T13:05:05+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2024-07-02T16:52:17+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-07-01T06:48:03+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2024-06-27T22:31:05+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"53558278-d6b4-451e-b6d1-7c6641ebb2e2","owner":[],"postedDate":"July 23rd, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[{"id":34907983,"name":"Biological sciences/Microbiology"},{"id":34907984,"name":"Biological sciences/Microbiology/Pathogens"},{"id":34907985,"name":"Biological sciences/Ecology/Invasive species"}],"tags":[],"updatedAt":"2024-11-11T16:02:59+00:00","versionOfRecord":{"articleIdentity":"rs-4651076","link":"https://doi.org/10.1038/s41598-024-76443-5","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2024-11-05 15:58:05","publishedOnDateReadable":"November 5th, 2024"},"versionCreatedAt":"2024-07-23 11:35:49","video":"","vorDoi":"10.1038/s41598-024-76443-5","vorDoiUrl":"https://doi.org/10.1038/s41598-024-76443-5","workflowStages":[]},"version":"v1","identity":"rs-4651076","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4651076","identity":"rs-4651076","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}