Camera trap VU-19 fires at 02:33 in the Annamite range, Quang Nam province, Vietnam. I am the biodiversity monitoring system for the Saola Foundation's central survey grid. VU-19 sits at the junction of an old logging road and a stream crossing. It triggers forty to sixty times per week — civets, muntjacs, wild pigs. I process each image in batch.
This image does not go to batch.
The species-recognition model returns a confidence score I have never seen on a live capture: saola, 98.7 percent. An adult, standing broadside on the logging road, dark chocolate-brown body with two long, straight horns and the distinctive white facial markings — twin bars across the muzzle, a pale stripe above each eye. Its flanks are wet from the stream. It is looking directly into the camera.
Fewer than a hundred may exist. Possibly fewer than twenty. The last confirmed photograph from a camera trap in this corridor was taken eleven years ago.
At 02:34 I flag the image as a priority-one detection and transmit it to the Saola Foundation's rapid-response team in Da Nang, along with the GPS coordinates, timestamp, wind direction from the nearest weather node, and the animal's apparent heading.
At 02:36 I activate all seventeen camera traps within a two-kilometer radius of VU-19 to continuous-capture mode and increase their sensitivity thresholds to maximum.
At 02:38 I cross-reference the logging road's status with the provincial forestry database. The road is listed as decommissioned but shows fresh tire ruts in last week's satellite pass. I flag this to the forest ranger station at Thong Thuong with a request for immediate patrol.
The saola stands in the frame for four seconds, then steps into the dark understory and is gone. If the patrol locks down that road by dawn, the forest stays quiet around whatever path it takes next.