A Bottleneck-First Framework for Diversity-Constrained Genetic Gain in Annual Self-Pollinated Crops

Breeding programs often adopt technologies before identifying the process that most limits genetic progress. This narrative review presents a bottleneck-first framework for annual self-pollinated crop breeding. It separates the diagnosis of a deficient pipeline domain from assessment of whether a specific intervention is actionable under uncertainty, capacity, and diversity constraints. Five domains are considered: usable variation, information for selection, biological progression and fixation, multi-environment validation, and delivery. Mandatory product gates and dependencies among domains determine priority. An intervention is actionable only when its lower uncertainty bound meets a predeclared minimum useful effect, downstream resources can absorb the added output, and diversity safeguards remain within declared limits. Three published cases provide retrospective stress tests: wheat high-throughput phenotyping, wheat pre-breeding, and SUBMERGENCE1 rice. Five additional cases examine realized gain, rapid generation advance, perennial selection, autotetraploid prediction, and promoter editing. A worked cereal example shows how capacity and diversity gates can reverse a technology-first priority. The framework complements rather than replaces selection indices, optimal contribution methods, genomic prediction, and breeding-scheme simulation. Its contribution is a transparent sequence for diagnosis, intervention testing, prioritization, and re-diagnosis. A prospective comparison with current breeding practice is still required.