Persistence under simulated rainfall
We carried out rainfall experiments in a controlled laboratory environment using a pressure sprayer filled with deionized water and an adjustable nozzle to alter droplet size from a direct stream to a fine mist. To simulate vegetation, we used six tomatoes and six peaches to provide a comparison between smooth and textured surfaces. Tomatoes and peaches were first washed and dried to remove any external contamination, before 3µl of slurry were deposited onto five of the six pieces of fruit, leaving one as a negative control. Once the slurry had dried, we placed four tomatoes and four peaches onto an ~35 x 25 cm metallic mesh grate, which was first cleaned with 10% bleach solution and rinsed. The last tomato and peach were kept as positive controls. Fruit were then sprayed to simulate rainfall using a handheld pressure sprayer from one meter away with a maximum working pressure of 36 PSI (2.48 Bar) and maximum water exit speed of approximately 10 meters per second; which would result in comparable speed and kinetic energy of heavy rainfall (van Boxel, 1997; Tilg, Vejen, Hasager, & Nielsen, 2020). We began by using a large nozzle opening to represent large droplets during heavy rain and sprayed 220 ml of water over the fruit. Using nitrile gloves, all pieces of fruit were then placed alone into two-liter buckets (cleaned with 1% bleach and double rinsed) filled with a liter of deionized water to remove any eDNA remaining on their surface (Valentin et al., 2018). Fruits were then removed from the buckets and the resultant solution filtered through a 10 µm PCTE filter to capture intracellular eDNA. We then rewashed the tomatoes and peaches in a 1% bleach solution then double rinsed in deionized water and randomly assigned one of each as a negative each time to control for decontamination errors. We repeated the above process with an intermediate nozzle opening to represent light rainfall and sprayed 220 ml of water, then once again after changing the nozzle to simulate a misty rain event and spraying 220 ml of water. We repeated the mist process an additional three times, halving the volume of mist applied each time (i.e. 110 ml, 55 ml, 27 ml). We then extracted the DNA from the filters using the HotSHOT method and tested for target intracellular eDNA via qPCR in an Applied Biosystems 7500 real-time PCR machine in duplicate following the above protocol.