Effect of ploidy level on guard cell length and use of stomata to discard diploids among putative haploids in maize

Guard cell (GC) length proved to be efficient in sorting diploids from haploids in doubled haploid development in maize (Zea mays L). It compensates for the weakness of widely used R1-nj marker approach that showed low reliability in haploids identification from tropical genotypes. Guard cell length differs between haploid and diploid plants, and these differences were evaluated in the progeny of three different induction crosses obtained using Krasnodar haploid inducer. Epidermal impressions of the abaxial surface of leaves removed from the second, third, and fourth nodes (from base to apex) were collected and measured using an optical microscope. This was also conducted on the flowering phenotype. Guard cell length varied according to germplasm source, leaf stage, and ploidy level. Mean GC length ranged from 23.67 to 33.82 mu m in haploids, and from 36.1 to 41.25 mu m in diploids. Based on these differences in GC length at any of the chosen leaf stages (second, third, or fourth), diploid and haploid maize plants were successfully classified. Classification efficiency was found to be more closely related to germplasm source than leaf stage. Comparing GC length according to phenotype (haploid or diploid), GC limits for classification as a diploid plant (threshold points) were estimated and ranged from 29.74 to 34.49 mu m, depending on germplasm source. The highest false discovery rate was 2.93% and false negative rate was 15.06%, indicating that classification based on GC length was reliable.