The alarm came on 19 October, when a camera at a fish pass on the River Teme registered 128,406 fish movements in 11 minutes. The River Renewal Office, which oversees restoration grants, halted verification of the figures after staff found that reported counts had surged across 17 English sites at once.
The programme has a budget of £38 million (about €44 million), with £6.4 million (about €7.4 million) in milestone payments awaiting confirmation. Officials said no payment tied to the suspect counts had been released, but contractors have been told that claims relying on the system will remain paused while the records are checked.
One shoal, 17 rivers
The anomaly was not a sudden wave of fish. Analysts comparing footage found the same 22-second sequence — 11 fish passing through a swirl of foam — attached to records from sites hundreds of kilometres apart. The software had logged that clip thousands of times, turning it into millions of supposed passage events over 10 days.
The contractor, Rillbyte Systems, said a software update caused its recovery queue to reuse a test clip under multiple camera identifiers. When the system retried failed uploads, it treated each replay as fresh footage. The firm has disabled the affected processing function and is rebuilding the count from stored camera recordings.
“The counter did not see a river full of fish. It saw the same 22 seconds many times,” said Antal O. Kovács, Rillbyte’s engineering director. He said the company had alerted the River Renewal Office on 23 October and was checking whether other wildlife records processed by the same system were affected.
The counter did not see a river full of fish. It saw the same 22 seconds many times.
Antal O. Kovács, Rillbyte engineering director
Payments wait for a recount
Genta Mayasari-Prasetyo, an aquatic ecologist who reviewed footage from the Teme, said the false totals had been easy to question because they exceeded what the channels could physically carry. “A camera can misread a fish; it cannot make one shoal appear in 17 rivers,” she said. She warned that the episode could weaken trust in monitoring data used to judge whether habitat work is succeeding.
The River Renewal Office is now comparing camera files with manual counts and records from local survey teams. Raisa Hidayat, its deputy director, said contractors would not be penalised for the software fault, but any payment linked to an inflated count would wait. She expects an initial review of the affected sites by 6 November.
Some restoration work is continuing, including repairs to fish-pass channels and the removal of obstructions. The office said those projects would be assessed on their own evidence, rather than the disputed dashboard scores. Ecologists say the camera records may still provide a useful baseline once the duplicated clips are removed.
The incident has prompted the office to require a human review of unusually large counts before future grant milestones are approved. Rillbyte said it would publish a technical account of the fault after the recount. For now, the question facing river managers is not how many fish returned, but which measurements can still be trusted.
