eDNA provides replicable and meaningful data on relative abundance of aquatic organisms, but not absolute abundance (except in some very specific cases where extensive calibration has taken place – see Levi et al., 2018 for an example of using eDNA to count salmon in Alaska).
Some behavioural factors affect the amount of DNA given off by a particular species at a particular time (e.g. spikes of DNA associated with breeding or high levels of activity), and there are some interspecific differences in DNA shedding – for instance, small active fish tend to give off more DNA than large, slow ones.
In rivers, if you detect a small trace of a species it is difficult to tell whether this means there are a small number of individuals close to the sampling point or a larger number some distance upstream. That said, overall the rank abundance of species based on eDNA data tends to be a good reflection of the community.