Globally, many insect populations are declining, prompting calls for action. Yet these findings have also prompted discussion about sampling methods and interpretation of long-term datasets. As insect monitoring and research efforts increase, it is critical to quantify the effectiveness of sampling methods. This is especially true if sampling biases of different methods covary with climate, which is also changing over time. We assess the effectiveness of two types of flight intercept traps commonly used for beetles, a diverse insect group responsible for numerous ecosystem services, under different climatic conditions in Norwegian boreal forest. One of these trap designs includes a device to prevent rainwater from entering the collection vial, diluting preservatives and flushing out beetles. This design is compared to a standard trap. We ask how beetle capture rates vary between these traps, and how these differences vary based on precipitation levels and beetle body size, an important species trait. Bayesian mixed models reveal that the standard and modified traps differ in their beetle capture rates, but that the magnitude and direction of these differences change with precipitation levels and beetle body size. At low rainfall levels standard traps catch more beetles, but as precipitation increases the catch rates of modified traps overtake those of standard traps. This effect is most pronounced for large-bodied beetles. Sampling methods are known to differ in their effectiveness. Here, we present evidence for a less well-known but likely common phenomenon - an interaction between climate and sampling, such that relative effectiveness of trap types for beetle sampling differs depending on precipitation levels and species traits. This highlights a challenge for long-term monitoring programs, where both climate and insect populations are changing. Sampling methods should be sought that eliminate climate interactions, any biases should be quantified, and all insect datasets should include detailed methodological metadata.

From manuscript text:

Insect traps and study sites

We trapped beetles at 20 sites in coniferous boreal forest in Norway from July to August in 2018 and from June to August in 2019. The sites were located in near-natural forest, and selected from databases of protected forests and/or forests with high occurrence of red-listed wood-living fungi (Naturbase https://www.miljodirektoratet.no/tjenester/naturbase/ and Artskart https://artskart.artsdatabanken.no/). Five triangular single-plane transparent window traps (90 x 76 cm), including one to two standard traps and three to four tras with water draining devices, were placed at each site. These Polish IBL-2 traps are produced by CHEMIPAN (Warsaw, Poland) (Pettersson, Stenbacka, Hjältén & Hilszczanski, 2007; Stenbacka et al., 2010). The rainwater draining device, designed by the Department of Forest Protection, Forest Research Institute (Sękocin Stary, Poland), contains an internal screen that prevents beetles from escaping through a funnel that diverts rainwater out of the trap. Insects must crawl over the edge of this screen in order to fall into the collection bottle. At each site traps were placed randomly within a 30 m radius and hung between trees.

Traps were emptied in the middle of the season (‘Period 1’; 2019 only) and again at the end (‘Period 2’; 2018 & 2019). We recorded the number of days traps were placed at each site in each trap period. Beetle species captured in each window trap were identified to species level by an expert taxonomist, and the number of individuals of each species was recorded. Beetles records were all uploaded to GBIF (gbif.org). Body size (length) of each species was taken from Seibold et al. (2015) when available (n = 420 of 452 species). Relevant species were also classified as natural-forest indicator species based on a classification system used by Dahlberg (2011).  

Precipitation estimates

To estimate rainfall at each of the trap sites during each trap period, we downloaded estimated daily total precipitation values for the study area from Lussana, Tveito, Dobler and Tunheim (2019). These values are estimated across Norway at a spatial resolution of 1 x 1 km and are based on a model-based interpolation. Precipitation values for each site were summed across the days that traps were placed during each trapping period. All analyses were conducted in R (R Core Team, 2019); climate data were extracted using the raster (Hijmans, 2019) package. Total precipitation values in our dataset ranged from 38 mm to 228 mm during a mean trap period of 38 days.

--all references are found in manuscript--

ReadMe file included with data.