Publishing a method lets someone else check your working. That is the point of publishing it.
The protocol
- 1
Baseline survey
Before any intervention, the stretch is walked on foot. Bank condition, existing vegetation, current land use and the households involved are recorded, with GPS points and photographs from fixed positions.
- 2
Fixed-point photography
The same frame, from the same marked position, every quarter. Nothing shows change to a non-specialist as clearly, and nothing is harder to argue with.
- 3
Quarterly water sampling
Six points. Turbidity, pH, dissolved oxygen and nitrate, recorded on paper in the field and entered the same week. Raw sheets are retained.
- 4
Survival counting at six and twelve months
Physical count, not estimate. Every planted seedling on a sample transect is checked. Losses are recorded by probable cause where it is evident.
- 5
Control comparison
RW-05 is a stretch we have deliberately not restored. Without it, an improvement at a restored site cannot be attributed to the restoration.
What this method cannot tell you
A method section that lists no limitations has not been thought about.
- Quarterly sampling at six points shows direction over years. It cannot support confident claims about a single season.
- One control stretch is not a controlled trial. It reduces the chance we are fooling ourselves; it does not eliminate it.
- We do not test for heavy metals, pathogens or pesticide residue. Those need laboratory capacity we do not have.
- Survival counts sample transects rather than every seedling on large sites, so site-level figures carry sampling error we do not currently quantify.
- Training figures rely on attendance registers, which are only as good as the person holding the pen that morning.
- Our data begins in 2023. Anything we say about longer-term change draws on other people's records, and we cite them.
When a figure changes
If we discover a published figure is wrong, we correct it and note that we did, with the date and what changed. We do not edit numbers quietly.
Figures on this site are labelled either verified or internal estimate. Verified means checked by someone outside this organisation. Internal estimate means our own records, kept carefully, and not independently examined.
Researchers: please pick holes in this
We would rather have the method improved by someone with better training than defend it.