UV degradation
We tested UV degradation of both intracellular and extracellular eDNA in four experimental treatments: 1) full sun, unshaded solar exposure from dawn until dusk; 2) half sun, unshaded solar exposure from dawn until 1pm and in full shade thereafter; 3) full shade, shaded from solar exposure for the full day; and, 4) no sun, where samples were kept within a completely dark workstation blocking all UV (e.g. CC-150 Cabinet, Spectroline, Westbury, NY, USA) to disentangle time from UV solar radiation. The in situ experimental site was located in a temperate environment (40.49º latitude) during mid-summer (25–31 July 2019). The three experiments were conducted simultaneously over seven days in an empty lot in New Brunswick, New Jersey, USA, in three locations no more than twenty meters from each other to provide three treatments: full sun, half sun, and full shade. The no sun treatment was completed in the laboratory from the 5th to the 11th of December 2019.
The full sun treatment was carried out at the top of a 3.15 m tall tree away from other structures so as to never be shaded, the half sun treatment along a fence-line facing the southern direction, and the full-shade treatment was conducted below the canopy of mature oak trees. We used living leaves from vegetation already present as our testing surfaces for the in situ experiments and Parafilm (Amcor, Neenah, Wisconsin, USA) for the no sun treatment. Leaves were not used in the no sun (completely dark) treatment as they would likely have died and / or desiccated, possibly affecting results. We added 15 µl of previously unused slurry to each of 35 surfaces in each treatment to provide five daily replicates throughout the experiment. Leaves that received slurry droplets were marked with lab tape on the stem for later identification. On day zero (i.e. before placing the slurry droplets for each experiment, and not exposed to the sun) we filled four 100 ml glass bottles with 80 ml of deionized water and added 15 µl of slurry. We then filtered three intracellular and one extracellular positive controls (captured in series using a 10 µm filter first then a 0.2 µm filter) to establish baseline concentrations. This was repeated with five bottles containing 80 ml of water and 15 µl of slurry to produce five intracellular and extracellular (captured in series) positive controls for the no sun treatment. Each day (in 24 h intervals), five samples were collected from each treatment and placed in labeled sterile 50 ml falcon tubes, in addition to a sixth leaf (or blank parafilm) that did not receive any slurry for a negative control. We collected all samples while wearing fresh nitrile gloves for each treatment and avoided the lab tape used in UV sun and shade treatments to prevent possible contamination.
We returned leaves from the in situ sun and shade treatments to the lab to be processed immediately after collection (Figure2 ). We added 40 ml of deionized water to each of the falcon tubes containing samples and shook them twice. We then filtered the solution from each treatment as follows: full sun was filtered in series, using first a 10 µm filter into a 0.2 µm filter, to capture intracellular and extracellular eDNA, respectively; and the remaining three treatments were filtered to collect only intracellular eDNA due to 0.2 µm filter limitations. A filter negative control was present for each treatment to ensure no contamination occurred during filtration. We then immediately extracted DNA from filter membranes using the HotSHOT method and stored extracts at -20 C until all samples had been collected and processed. We analyzed all samples via qPCR in duplicate for 40 cycles on an Applied Biosystems 7500 real-time PCR machine and converted quantified concentrations to copy number as outlined above.
We calculated the decay rate (r ) from \(r\ =\ 1-e^{k}\ \)for eDNA in each treatment by fitting the following model via non-linear least squares to estimate the decay constant k (nls, R v. 3.6.1; R Core Team 2019):
\begin{equation} \frac{\text{Copy\ number}}{10^{6}}=Ce^{-kX}\nonumber \\ \end{equation}
where X represented either time of collection (in hours) or cumulative UV exposure, depending on the model evaluated, and Crepresented the intercept (i.e., estimated copy number in millions whenX  = 0). Within each treatment group, we ran competing models estimating the rate of eDNA degradation between cumulative UV-A, UV-B, and UV-A+B exposure, as well as time continuous decay. UV exposure data were obtained from the USDA UV-B Monitoring and Research Program for Geneva, New York, Beltsville, Maryland, and Queenstown, Maryland. We used the extrapolation tool in ArcGIS version 10.7.1 to estimate daily UV totals for UV-A and UV-B over the course of our experiment. As cumulative UV exposure increases with time, we expected the curves for time and UV to show similar patterns. We evaluated the models using AICc and judged them to be similar in performance if within 2 AICc units. Thresholds were placed at 39 PCR cycles (0.0015 ng or 1.45 \(\times\)107 copies) to provide a conservative limit of detection, and 40 PCR cycles (0.00075 ng or 7.24 \(\times\)106 copies) for the absolute limit of detection for our qPCR runs. The time to reach each threshold copy number was calculated from the decay curves using the following equation:
\begin{equation} Time\ to\ threshold=\ln\left(\frac{\text{threshold}}{C{\ \times\ 10}^{6}}\right)\ \left(-\frac{1}{k}\right)\ \nonumber \\ \end{equation}