It's definitely intriguing considering your results. A couple notes,
- what was the volume of your container? A small container of media with 20gph flow is a lot different than a 10 gallon tank receiving the same flow.
- were you sure that the water going into the container was "oxygen deprived" ?
Before using the container, the De*Nitrate was sitting in a mesh bag in my sump. The sump has about ~3.5 gallons of water in it when the system is running (e.g. when return pump is running). And my system was turning over about 120 ghp through the sump.
Container was originally a DIY reactor made from a 32oz mayo jar. (Recently I replaced with a shortened 2 Little Fishes Phosban 15 reactor due some epoxy I used to assemble the DIY jar finally giving out.)
It was the same volume of De*Nitrate rubble in both instances -- I merely dumped the rubble from the mesh bag into the reactor.
It is also the ONLY change I made at the time of the NO3 drop. As a former mechanical and quality assurance engineer I know to avoid testing more than one variable or system change at any give time if at all possible.
Anaerobic bacteria generally only activates when they cannot perform aerobic functions which they prefer.
Kind of goes back to my initial comment: "In my experience, denitrification through media is immense -- but it has to be proper media implemented in the proper way."
Porosity of the media, too little flow, to much flow -- all can have an impact.
If any of them are not present to an appropriate degree for the given system, denitrification from media probably will be negligible.
If they are in proper balance though, again at least in my experience, the impact can be quite significant.
I'm in the process of setting up a 65g, and plan on repeating the set-up type. Same type of rock. And then, if NO3 is above 1-5 ppm, will follow up with De*Nitrate in a reactor fed of the return line. Will be a good opportunity for me to see if the results are repeatable.