Gram and milligram transforms gave no clear standardized interaction difference in four insect samples in 2019-2020

We tested whether biomass units change a fitted temperature by rain interaction in insect trap catches. We used 5,805 measured catches from Sweden and Madagascar in 2019-2020. Four models used separate countries and periods, with trap, calendar, season and collection duration controls. The response was log(1 + rate), with rate in grams or milligrams per day. We divided each response by its sample standard deviation. All four interaction estimates were positive on both scales. Milligram-minus-gram differences were -0.0079, -0.0112, 0.0185 and 0.0312 response standard deviations per 5 C by 5 mm/day. All four adjusted intervals included zero. Wider collection windows, low-catch removal and simpler weather controls changed some differences. A simpler model gave a positive adjusted difference only in later Sweden. Exact checks confirmed unit invariance for pure log rate and for a fixed physical offset. This is a method audit, not evidence of equal transforms or a causal weather effect. One year per country, selected catches, coarse weather and few time clusters limit inference. Code, source hashes, aggregate results and a figure are included. External records stay outside the package.

insect biomass · weather · measurement units · observation bias · log transformation · null result