The first round of field observations on the main channel was completed in late February. What looked like a straightforward dataset on pike hunting positions turned out to be more layered once we cross-referenced it with flow velocity readings from the same afternoon. The initial review flagged several attack events that did not match the expected pattern — pikes striking in faster water than the literature suggests.

We went back to the raw footage and checked the time stamps against the telemetry log. It turned out that three of the five anomalous strikes happened within minutes after a brief drop in water level caused by an upstream gate adjustment. The fish had repositioned closer to the bank, where the current was slower, but the strike itself was recorded in the main stream because the camera angle covered a wider area. The coordinates were correct; the behavioural context was not.

This forced a change in how we tag events in the database. Instead of recording only the strike location, we now log the fish’s position for the ten minutes before the attack. That extra window gives a clearer picture of whether the pike moved into the current deliberately or was already holding there. The difference matters when you try to model energy expenditure per strike.

The same review also changed our approach to camera placement. The original plan put units at fixed intervals along a 400-metre stretch. After seeing how much the fish shifted position with small changes in discharge, we switched to a staggered setup — cameras closer together near known ambush spots, wider apart in straight runs. The data from the first week under the new layout already shows fewer missed events.

None of this was in the original protocol. It came from sitting down with the raw numbers and admitting that the first interpretation was too clean. The page exists because that kind of adjustment is worth documenting — not as a correction, but as a normal part of field research.