Results

Controls and limit of detection
The qPCR assay had typically a R2 of over 0.99 and an efficiency of 70%. Although PCR efficiency was quite low we could amplify even single molecules, indicating that the assay was rather sensitive. The negative PCR control reactions did not result in any amplification. However, we amplified low levels of eDNA from the samples collected prior to inoculation with DNA sources and in the samples collected from the control aquariums. For all samples analyzed of the control aquariums and samples collected prior to inoculation, we plotted a cumulative density function in R (R core team, 2014) on the CT-values. From this we estimated the 5% percentile to be 45.5 (Figure S1). Based on this result, we estimated our detection limit to be 45 cycles, which, given our standards corresponds to 8221 molecules of DNA. CT-values exceeding 45 in the non-control measurements were subsequently set to 45 as such values are likely to be caused by low levels of contamination or by double-stranded non-target DNA, such as primer dimers. Based on the melting curve of the positive controls, the typical melting temperature of target DNA (e.g. the temperature that the highest amount of DNA products dissociates and becomes single-stranded) was inferred to be in the range of 75.5 and 77.5 ºC. We assumed reactions showing a melting temperature outside this range to be non-target DNA such as primer dimers. Therefore, we set their CT-values to 45.

eDNA accumulation in aquariums with live shrimps

Significantly less eDNA was accumulated in the aquariums to which additional organic matter was added (P<0.05, Figures 1-2). CT-values were on average 5.4 higher in the aquariums to which organic matter was added, and on average 1.3 higher in the aquariums with low pH. However, the effect of pH on eDNA accumulation was not significant at the 5% level.

eDNA survival over time

For this analysis, we only used the aquariums that were monitored over time until the eDNA concentration dropped below the limit of detection. The spiked DNA was degraded beyond detectability within 2 to 60 hours, whereas the eDNA from the live shrimps was degraded in 0 to 1680 hours. We performed a survival analysis with the Cox’s proportional hazards model to estimate how the treatments affected the time needed for eDNA to degrade beyond the detection limit (45 cycles) (Therneau, 2014; Therneau et al., 2000). We used OM, pH and DNA source as treatment groups to estimate the effect of the treatments. eDNA degraded significantly faster in the treatments in which OM was added (p=0.003) whereas the pH did not significantly affect eDNA degradation (p=0.360). The survival analysis shows that spiked DNA was degraded significantly faster than the eDNA released by the living shrimps (p=0.023). The largest difference in eDNA survival versus survival of the spiked DNA was found in the treatment with high pH and no OM. The spiked DNA (treatment F) was degraded between 0 and 12 hours whereas it took between 1008 and 1680 hour for the eDNA to degrade (treatment B).