Temperature-order contrasts had intervals across zero in four insect-catch samples in Sweden and Madagascar in 2019-2020

We tested whether the order of warm and cool days had a different link to insect catch during collection than just after collection. Four country and period models used 5,804 measured catches in Sweden and Madagascar in 2019-2020. Models controlled mean weather, temperature variation, rain frequency, trap, season and collection duration. All four current-minus-future slope estimates were positive, from 0.0019 to 0.0113 log(1 + grams/day) per C of fitted temperature change across a window. All four adjusted 95 percent intervals included zero. Five of six sensitivity sets also had no adjusted interval apart from zero. A 5-9 day duration filter gave a positive adjusted interval in later Sweden. The results do not establish a stable collection-window link or the absence of a temperature-order effect. Each country has one survey year. The period split separates seasons at repeated traps. Catch combines insect abundance, flight activity and capture conditions. Coarse weather, selected samples, few time clusters and remaining observation bias limit inference. The package gives code, source hashes, aggregate results and a figure. External raw and catch-level data remain outside the package.

insect biomass · temperature order · weather · observation bias · temporal specificity · regional replication · null result