For many elusive insect species, which are difficult to cover by standard monitoring schemes, innovative survey methods are needed to gain robust data on abundance and population trends. The recording of pre-imaginal butterfly life stages provides great potential for ecological studies and conservation monitoring. However, using counts of pre-imaginal stages for quantitative research requires detection probability to be determined.

We tested different removal and capture-mark-recapture (CMR) approaches to determine the detection probability for overwintering larvae of the endangered nymphalid butterfly Limenitis reducta. Classical removal and CMR studies require movement of the organisms under study but in our approach, we replaced movement of the study organisms by random movement of multiple different surveyors. The study was conducted in three plots within a spruce clear-cut in the 'Alb-Donau' region, Germany.

Our dataset provides detection data of nine/ten different surveyors per study plot. The surveyors differed in their experience ('experts' vs. 'novices'). In the R-scripts we present the analysis of the data using i) the removal method, ii) CMR approaches with varying personnel expenditure. In addition, we test the validity of our method by comparing observed and simulated detection frequencies.

The results of our study indicate that multi-surveyor removal/CMR techniques are highly suitable for estimating abundance of overwintering L. reducta larvae and that the proposed methodology has several strengths: long survey period, estimates of the absolute population size accompanied by uncertainty measures, estimates of overwinter mortality. The methods from our study can be adapted and used for several different butterfly species, other insect taxa with specific immobile life-stages, and some sessile organisms, e.g. elusive plants, fungi, or corals.

The data was collected by 13 surveyors of different experience on a spruce clear-cut near Merklingen (48.494° N, 9.788° E, 677 m a.s.l.). Within a time period of six weeks (5 February 2021 - 22 March 2021), the surveyors independently searched for hibernating larvae of L. reducta. Seven surveyors searched in all three plots, two surveyors searched in two plots, and four surveyors searched in only one plot. Consequently, the three plots were surveyed by ten, nine, and ten different surveyors, respectively. The surveyors were allowed to search for a maximum of two hours per plot. For most surveyors, this time budget allowed a very thorough search, aiming at finding all hibernacula in a plot. The surveyors conducted their searches, one after the other, without prior information and without seeing each other. The detected hibernacula were recorded with GPS coordinates and a photograph. After all surveyors had finished their searches, we counted the hibernacula detected by each individual surveyor and matched those with the records of the others.

The raw dataset contains the detection histories for all hibernacula detected by at least one surveyor. The different columns represent the different surveyors/occasions. No further processing of the data was necessary prior to the analysis.

The tabular data (.xlsx) can be read with standard software programs. The analysis scripts (.R) can be read with the open-source software package R. We recommend the use of the interface RStudio.