Abstract The double haploid (DH) technique in maize ( Zea mays L.) relies on in vivo haploid induction, in which the identification of haploid seeds and seedlings represents a critical step in this technique that determines the efficiency of DH line production. This study aimed to evaluate the effects of endosperm texture, donor genotype, and inducer on putative haploid rates based on the R1-navajo ( R1-nj ) marker and the use of guard cell length and stomatal density as a tool to identify diploids among putative haploids. Eleven hybrids were assessed as donor genotypes and were pollinated with ten haploid inducers. The response was the number of putative haploid seeds based on R1-nj phenotype, evaluated across the total number of seeds. Previously classified seeds of two induced maize hybrids were grown, and epidermal impressions of the abaxial face of maize leaves were used to measure guard cell length (GCL) and stomatal density (SD). Plants were classified according to the real ploidy by the gold standard evaluation. The three-way cross inducer (UH400×UH401)×TAIL 9 achieved the highest putative haploid rate (PHR) with five donor hybrids. The magnitude of the differences presented for GCL and SD were sufficient for the efficiency in the classification of ploidy in maize by means of logistic regression. The AUC values obtained were 0.916 for GCL and 0.941 for SD, indicating that both tests under study can be integrated into maize breeding programs, which can be used for proper disposal of diploids among putative haploids.
Marcondes et al. (Wed,) studied this question.