The optimisation of methods to evaluate the impact of insecticides on non-target arthropods in forests

Insect pests affect approximately 35 million hectares of forest globally each year. To mitigate tree mortality, insecticide treatments are employed as a last resort within the framework of Integrated Pest Management (IPM). Among these, large-scale aerial applications are a common strategy used to suppress insect-induced defoliation. However, public concerns remain—fuelled by negative historical experiences with insecticides, the global decline in insect populations, and discrepancies between the expected effects based on insecticide chemical properties and the effects observed in field studies. Additionally, many studies do not include statistical analyses at the species level, where unpredictable toxic effects are most likely to occur. This is because some species may exhibit ecological responses that deviate from patterns observed at higher taxonomic levels. Conflicting study outcomes may result from logistical constraints that lead to methodological shortcomings. In parallel, the limited detection of species-level effects might be attributed to insufficient taxonomic expertise across various arthropod taxa. This thesis aims to identify and address these methodological flaws, which result in reporting inconsistencies, and to develop approaches for the effective application of DNA barcoding in such field trials. Chapter 1 presents a systematic review evaluating study designs assessing the side effects of aerially applied insecticides on non-target forest arthropods. Specifically, we examined whether appropriate study designs yielded results consistent with the expected toxicological outcomes based on the insecticides’ chemical properties. We found that pseudoreplication was the primary cause of implausible results that contradicted these predictable toxic effects, and we provided guidance to ensure true replication in future studies. Chapters 2 to 4 focus on optimizing DNA barcoding to support taxonomic identification. Chapter 2 (Stein et al. 2022) introduces a non-destructive DNA barcoding method that preserves specimens for morphological identification in cases where DNA barcoding fails. Chapter 3 (Stein et al. 2024) outlines a method for curating species assignments in the Barcode of Life Database (BOLD) to enhance DNA barcoding accuracy using sequence data and metadata, including morphological references and sampling locations. Chapter 4 (Stein and Gailing 2025) employs the curated record set from Stein et al. (2024) to identify systematic deficiencies in BOLD and proposes approaches to address them. The results of this thesis indicate that pseudoreplication unduly fuels concerns about aerial insecticide applications in forests, while the optimized DNA barcoding methods developed here ensure the detection of unpredictable toxic effects at the species level. Moreover, sampling non-target arthropods plays a crucial role, as it bridges these two components. Without addressing both, field assessments of insecticides are likely to continue producing unreliable results, thereby compromising the scientific foundation of public debates surrounding aerially applied insecticides in forest ecosystems. Such public concerns may paradoxically delay the development and authorization of forthcoming, highly specific RNA silencing-based insecticides.