Advances in soil arthropod identification: Integrating morphological, molecular, and AI-based approaches

Abstract

Soil arthropods are key contributors to ecosystem functioning, yet their assessment remains limited by the complexity of species identification. This review explores the main methodologies used for identifying soil arthropods, tracing their evolution from morphological approaches to molecular and artificial intelligence (AI)–based systems. Morphological identification remains essential for understanding abundance, community structure, and functional traits, but it is time-consuming and relies heavily on specialized taxonomic expertise. Molecular tools such as DNA barcoding and metabarcoding have revolutionized biodiversity monitoring, providing rapid and standardized identification of taxa and uncovering cryptic diversity, although their accuracy depends on the completeness of reference databases and they often lack functional information. Recent advances in AI and machine learning have opened new perspectives, enabling automated identification from digital images with increasing accuracy and speed. However, these systems still require extensive, expert-validated training datasets and are currently limited to well-documented taxa. The comparative analysis presented here highlights that no single method is sufficient alone. Instead, an integrative approach combining morphological, molecular, and AI-based techniques offers the most comprehensive and reliable framework for soil biodiversity assessment. This synthesis emphasizes the need for cross-disciplinary collaboration to enhance monitoring efficiency and to support sustainable soil management and conservation.