A warm-catch bound stayed positive with tenfold missing-day mass caps in two Swedish insect samples in 2019

We tested how unobserved days change a temperature association in captured insect mass. We used 5,836 measured collections in Sweden and Madagascar in 2019-2020. Missing days lie inside each retained trap-period span. Each missing day can contain up to k times the observed mean daily mass. We derive sharp ratio bounds while allowing any daily mass allocation inside measured windows. At k = 10, the equal-trap mean lower bounds are +0.83 and +0.90 C in the two Swedish survey halves. Both Madagascar bounds include zero at k = 0. The early Swedish lower bound reaches zero at k = 58.73. The late Swedish unrestricted limit is +0.3975 C, due to many gap-free traps. These are conditional possibility bounds, not causal effects or confidence intervals. The two halves share traps and cover one survey year per country. We estimate no missing-catch cap, pollution effect or biodiversity trend. Raw data and record-level derivatives stay outside the bundle. Code, source hashes, group results, a figure and 496 independent optimization checks are included.

insect biomass · observation bias · missing data · temperature · sharp bounds · seasonality · Sweden · Madagascar