Advancing taxonomic identification of native bees through integrated morphometric analysis

This is the authors' abstract. We don't add key points for this paper.

The global biodiversity crisis highlights the urgent need for scalable, accurate, and accessible taxonomic tools, particularly for cryptic and understudied groups such as insects. Traditional species identification methods are time-consuming and require specialist expertise, limiting their utility in large-scale biodiversity monitoring. Here, we present AusWing Explorer, a web-based platform implementing geometric morphometric analysis for bee identification, designed to improve accessibility and standardization in morphometric taxonomy. Bees represent an ideal test case due to their ecological importance and taxonomic complexity. We compiled a dataset of 2200 forewing images from expert-verified museum and research specimens, representing 120 species across five families. Using landmark-based Procrustes alignment, t-distributed stochastic neighbour embedding (t-SNE), and Linear Discriminant Analysis (LDA), the tool achieved classification accuracies of 99.4% (family), 98.3% (genus), 94.9% (subgenus), and 87.5% (species). The approach demonstrates strong resilience to operator variability, making it suitable for both researchers and citizen scientists. By providing a transparent, interpretable, and resource-efficient method, AusWing Explorer complements molecular and deep learning techniques, offering a flexible framework that can integrate with automated imaging pipelines to enhance global taxonomic capacity. • Understudied and cryptic species are often time-consuming and costly to identify. • Pollinator declines make rapid and accessible insect identification critical. • Our new web-based tool makes bee identification simple, rapid, and accessible. • We accurately ID native bees to species level from high-quality photos. • Our method of ID has applications to other winged and cryptic insect species.