Three years is short for this kind of work. What follows is a direction of travel, not a conclusion.
Seedling survival at twelve months, by planting cohort
The number that decides whether a planting day meant anything. We publish it beside the planting total, always — quoting one without the other overstates the work by roughly 60%.
Survival by site
| Site | Planted | Alive at 12 months | Rate | Notes |
|---|---|---|---|---|
| Biharwe upper | 2,400 | 1,730 | 72% | Best site. Established, no longer replanting. |
| Biharwe lower | 3,100 | 1,800 | 58% | Heavy dry-season losses in year one. |
| Rwentondo | 3,600 | 2,270 | 63% | Strong household agreements. Steady. |
| Rwampara wetland margin | 1,200 | 790 | 66% | Mostly natural papyrus recovery, low planting. |
| Nyakayojo | 900 | 0 | 0% | Total loss. Land use not agreed with the occupier. |
What the turbidity record suggests
At RW-04, downstream of our earliest Rwampara buffer work, dry-season turbidity has trended below the control stretch at RW-05 for two consecutive years. That is the closest thing we have to evidence that the buffer work does what we claim.
It is a weak signal from a small dataset and we treat it as such. Two years is not a trend, one control is not a controlled trial, and the two stretches differ in ways beyond our intervention.
At RW-02 and RW-03, in and below the city, turbidity worsened over the last dry season. Those points sit downstream of urban discharge and construction, which is beyond what buffer planting can influence.
The honest summary
In the rural middle catchment, where we work and where the pressure is agricultural, our sites are improving relative to a control. In the urban stretch, where the pressure is discharge and construction, nothing we do is moving the readings. Both of those are worth knowing.
The full records are available
Including the raw field sheets, if you want to check the entry rather than trust the summary.