What the platform calculates from verified data, and where the numbers stop telling you anything.
What the platform provides
| Section | What it shows |
|---|---|
| Dashboard | Numbers of photos, locations, working days, species; most frequent species; filters by date, institution, habitat |
| Relative abundance (RAI) | Detections per 100 camera-days with confidence intervals |
| Trends | Year-to-year change in a species' detection rate: Mann-Kendall test, confidence intervals |
| Daily activity | When during the day a species is active, several species on one chart |
| Maps | Detections of a species on a map |
| Area comparison | Species lists, Shannon, Simpson and Jaccard indices |
| Data gaps | Effort on the EEA grid, species detection sensitivity, GBIF data as an independent source of presence |
| Export | Darwin Core for GBIF, Excel, aggregation of observations |
Independence of detections
For abundance, all detections of a species at one location within 5 minutes count as one observation. You can choose 30 minutes or a day, but the interval must be the same in all data you compare.
Reporting
- Charts can be saved with a transparent background and pasted into reports, including the Chronicle of Nature (Litopys pryrody). Aim to finish verification by October to November, when annual reports are prepared.
- The Darwin Core export can be imported into GBIF directly.
Limits of inference
"Not detected" does not mean "absent"
How many camera-days are needed before a non-detection means something (95 % confidence, calculated at locations where the species occurred, so this is a lower bound):
| Species | Camera-days |
|---|---|
| Roe deer | ~24 |
| Wild boar | ~67 |
| Wolf | ~120 |
| Lynx | ~180 |
| Wildcat | ~230 |
| Marten | ~380 |
| Polecat | ~470 |
So "there is no lynx here" after one camera working for a month means nothing.
Neighbouring days are not independent
If an animal was recorded today, the chance of recording it tomorrow is three times higher than average. For detection estimates, a camera-week is a better unit than a camera-day.
Warning: the "Data gaps" page shows where effort is too low, but it does not replace occupancy models. Do not read "100 % coverage" as "the fauna is fully known".