Flood flags gave negative catch estimates for beetles other than P. madidus in four UK samples in 1993-2015

We tested recorded catch in flooded and unflagged beetle traps from the same collection in England and Scotland in 1993-2003 and 2004-2015. We excluded Pterostichus madidus because its sex and colour entries can overlap. Four collection and fixed trap models used 6,299 events. Flood estimates for log(1 + catch) were -0.052, -0.061, -0.149 and -0.047. Only early Scotland had a four-test adjusted interval below zero. That interval crossed zero with two-year error blocks or any one Scottish site removed. All raw-count intervals crossed zero. Thus the evidence does not establish a stable catch reduction across regions and periods. Selected mixed collections, sparse catches, explicit zero codes and few error groups limit inference. Local biological response, capture loss and reporting rules cannot be separated. The package contains code, hashes, aggregate evidence, numerical checks and a figure. External records stay outside the bundle.

observation bias · flooding · carabid beetles · pitfall traps · weather extremes · regional replication