Integrative insect taxonomy increasingly combines morphology, DNA barcodes, multilocus and genome-scale data, ecology, geography, and reproductive evidence, yet agreement among these sources is often treated more simply than the underlying inference permits. This critical review examines how evidential classes contribute to species delimitation and higher-level taxonomic interpretation when they converge, conflict, or share non-independent sources of error. The central problem is not whether molecular evidence should replace morphology, or whether several agreeing datasets automatically justify a taxonomic decision, but how observations generated at different biological and analytical levels are translated into species-level inference. Morphological characters may be diagnostic, convergent, life-stage restricted, or difficult to quantify; molecular signals may reflect population structure, introgression, incomplete lineage sorting, marker-specific histories, or model misspecification; ecological differences may provide independent context without proving reproductive separation; and reproductive evidence may directly test biological isolation but is unevenly available across insect systems. The review therefore distinguishes evidence acquisition from taxonomic interpretation and argues that evidential weight should depend on relevance, independence, error structure, inferential distance, and explicit consideration of competing explanations rather than on a fixed ranking of data classes. This dependence-aware view provides a basis for interpreting both concordance and conflict while preserving uncertainty, taxonomic scope, and the distinction between lineage detection and species recognition.
