Allowed bottle dates gave both temperature slope signs for median insect catch timing in four Sweden and Madagascar samples in 2019-2020

We tested whether collection windows fix the sign of a site-temperature link to median insect catch timing. We used measured Malaise-trap biomass in Sweden in 2019 and Madagascar in 2019-2020. Separate halves of each source year gave four samples with 86, 144, 46 and 47 trap records. We bounded the median day by the start and end of the bottle that contained half of measured mass. We then bounded the temperature slope after control for latitude, longitude and habitat. Bounds were -13.82 to +5.66, -4.11 to +5.10, -1.63 to +1.62 and -0.25 to +2.58 days per degree C. All four bounds included both signs. Trap bootstrap outer limits also included zero. These are exact date-allocation bounds under fixed bottle masses and dates. They do not cover missing catch. Country networks differ, but the two periods within each country can share traps and annual conditions. One source year per country cannot test a climate trend. The package gives code, a proof, aggregate results, input hashes and a retrieval script. Raw data stay external.

insect biomass · ecological timing · temperature · observation bias · partial identification · collection window