Example
Forest conservation · Sabah, Malaysia · since 2024-03
Counted, not averaged. A project-wide index would hide which site moved. 1 site is still building a baseline.
Team
A project outlives the people monitoring it. Access is granted through a team and can be withdrawn without the record becoming unreadable — which is why the key belongs to the project rather than to one person's device.
01
Canopy vegetation index · 8 occasions
The shaded band is this site's threshold of ±0.04 NDVI, learned from 7 of its own occasions. Inside it, a reading is indistinguishable from this site's normal variation.
The verdict
Beyond ±0.04 NDVI, from 7 occasions.
Passive acoustic monitoring can show a trend across many occasions, but cannot carry a change claim
02
Canopy vegetation index · 8 occasions
The shaded band is this site's threshold of ±0.04 NDVI, learned from 7 of its own occasions. Inside it, a reading is indistinguishable from this site's normal variation.
The verdict
Inside ±0.04 NDVI, from 7 occasions.
Passive acoustic monitoring can show a trend across many occasions, but cannot carry a change claim
03
1 prior occasion — no published threshold exists for canopy vegetation index, so this site needs at least 5 before a change can be called real
—
Changing how you measure
This series was measured one way. A reading produced by a different laboratory or a different method is confounded with the difference between them — for eDNA, that difference can be larger than the difference between sites. The new method starts its own baseline.
Sequencing depth
Chosen because it is robust to sequencing depth. Richness inflates about twofold per tenfold increase in reads, so a deeper run looks like a richer site; inverse Simpson does not move that way. Richness is reported only alongside it, never alone.