Leaf wetness duration in irrigated citrus orchards in the Mediterranean climate conditions

•Dry/wet thresholds for leaf wetness (LW) sensors and models were set by ROC analysis.•Sensors and models were good estimators of visual observations of LW. The RH model was recommended.•Leaf wetness durations (LWD) were similar throughout the citrus canopy and leaf sides.•LW periods were frequent in citrus orchards under Mediterranean conditions.•Flooding did not enhance LWD, though effects in the lower canopy cannot be excluded.

Leaf wetness (LW) is a key environmental variable for the development of foliar fungal diseases of citrus. However, little information of LW duration (LWD) in Mediterranean citrus-growing areas is available. LWD in six canopy positions and two leaf sides was studied with visual observations and Spectrum LW sensors in a citrus orchard in Spain. The performance of Spectrum and Hobo LW sensors, and CART/SLD and RH LW models was assessed by receiver operating characteristic (ROC) curve analysis. The dynamics of LWD in three citrus orchards with flood irrigation and three with drip irrigation were evaluated using time-series models. LWD data from all experiments and recordings in 14 commercial citrus orchards illustrated the high frequency of LW periods, particularly those associated with dew. No substantial differences in LWD were detected among citrus canopy positions and leaf sides. An optimal dry/wet threshold of 2.85 for the Spectrum LW sensor, 75.88 for the Hobo LW sensor, 15.99 for the CART/SLD model, and 87.45% for the RH model was selected by ROC curve analysis. All the sensors and models evaluated showed a substantial strength of agreement with visual observations of LW, with area under the ROC curve of 0.89-0.93 and Cohen's Kappa index of 0.61-0.77. The RH model was recommended because of easier implementation in citrus orchards in Spain. Time-series models did not detect any significant increase in LWD associated with flood irrigation, but only after some rain events. High mesoclimatic relative humidity in the study area outweighed possible microclimatic changes induced by flood irrigation. Thus, negligible effects of flood irrigation in the infection of foliar fungal pathogens would be expected, but longer LWD in the lower canopy as well as potential effects on inoculum dynamics in the leaf litter cannot be excluded.