Using Stochastic Simulations to Shed Light on How to Deploy Speed Breeding and Genomic Selection in Self‐Pollinated Recurrent Breeding Programs

ABSTRACT Speed breeding can shorten breeding cycles and, when combined with genomic selection, can accelerate genetic gain. Yet it remains unclear how different integration strategies affect long‐term response, genetic variance and cost‐efficiency in small public programs. Here, we use simulations of a 20‐year breeding pipeline to compare speed breeding strategies with traditional pedigree selection and to identify designs that balance genetic gain, sustainability and cost. We simulated a breeding program in AlphaSimR , and all schemes used single‐seed descent. Three speed breeding scenarios, differing in the generation of genomic selection and in whether population size was recovered after selection, were contrasted with two traditional pedigree schemes that advanced fewer lines per cycle. For each design, we monitored population mean, additive genetic variance and prediction accuracy over 20 years. Speed breeding schemes delivered faster short‐term gains than traditional schemes. The design with genomic selection in F 2 and recovery of population size produced the largest cumulative response (3.55 genetic standard deviations) than the other speed scenarios. However, this design required genotyping and phenotyping 16,000 individuals per cycle, compared with 1200 in the best traditional scheme, resulting in lower efficiency per unit cost under current prices. Sensitivity analyses further showed that reducing the F 2 recovery proportion substantially improves the cost–gain trade‐off while retaining much of the long‐term advantage of the recovered‐population design. Designs without population‐size recovery reached lower plateaus of response (~2.4) but demanded far fewer resources and showed a higher return on investment. Here, we demonstrate that coupling speed breeding and genomic selection with recovery of population size is the most powerful strategy for long‐term gain, but its immediate adoption may be constrained in resource‐limited programs. In the short term, simpler speed breeding designs with smaller populations may be more realistic, whereas falling genotyping costs will favour high‐population designs. These results provide guidance for redesigning breeding pipelines that accelerate cultivar development while preserving sustainable genetic gains.