Figure 1.
Spring-driven force gauge built to measure maize rind penetrometer resistance (RPR).
A digital force gauge (model Z2S-DPU) measuring the kilograms of force imposed on a probe was adapted to accommodate a steel spike and fit to a triggered spring-driven track gliding on ball bearings. This apparatus was used to assess the force required to puncture maize stalk rinds with the spike mid-internode below the primary ear as a proxy for stalk strength. To ease data acquisition, custom Java code was developed to maintain measurements and provide users with audible commands denoting current inbred score and field plots remaining to be measured.
Figure 2.
Asymmetric transgressive segregation observed for RPR.
Comparisons of parent, mid-parent, and progeny BLUE line means for RPR revealed transgressive segregation. Parents of the NAM and IBM families were not chosen from populations divergently selected for RPR. Recombination of additive effects and novel mutations likely play a role in the transgressive variation among their inbred progeny. Furthermore, the asymmetry present in these distributions suggests a role for epistasis.
Table 1.
Broad-sense heritability of RIL families and the NCRPIS diversity panel.
Figure 3.
Variation in RPR, days to anthesis (DTA), and ear height (EHT).
About 15% of the total variation in RPR across all NAM and IBM families was attributable to genetic factors. This proportion of genetic variation was smaller than that observed for DTA or EHT. Despite this reduction in genetic variation of RPR, the proportion of genetic-by-environment variation were slightly larger for RPR than the other surveyed traits. Differences in the remaining RPR, DTA, and EHT variation were due to environmental factors or could not be attributed to known sources of variation.
Figure 4.
Trait correlations between RPR, DTA, and EHT variation.
Positive correlations between RPR and DTA were greater among plot means (A) than line means (B) across all RILs of the NAM families. However, the opposite relationship was observed between these traits among plot means (C) and line means (D) in the inbreds of the NCRPIS diversity panel. The relationship between RPR and EHT was less varied between the NAM and NCRPIS diversity panels.
Table 2.
Family-nested QTL for RPR in RIL families.
Figure 5.
Prediction of RPR in RIL families and NCRPIS diversity panel by GBLUP.
Cross-validation revealed steady gains in prediction accuracy with respect to the NAM family-stratified calibration set size (A). About 80% of the variation in BLUP line means for RPR was explained using an identity-by-state (IBS) genomic relationship matrix constructed from 1.6 million SNPs of the maize HapMapV1 (B). Using the same SNP set, prediction accuracy was not significant in the IBM family (C), and the variation explained by the entire family was only 56% (D). In the NCRPIS diversity panel, prediction accuracies were lower than in the NAM panel when using an IBS genomic relationship matrix of 681,257 SNPs constructed and fit in the same manner (E). In total, 71% of the variation in RPR was explained upon fitting the entire panel (F).