Validation of functional polymorphisms affecting maize plant height by unoccupied aerial systems discovers novel temporal phenotypes

Abstract

Plant height (PHT) in maize (Zea mays L.) has been scrutinized genetically and phenotypically due to relationship with other agronomically valuable traits (e.g., yield). Heritable variation of PHT is determined by many discovered quantitative trait loci; however, phenotypic effects of such loci often lack validation across environments and genetic backgrounds, especially in the hybrid state grown by farmers rather than the inbred state more often used by geneticists. A previous genome-wide association study using a topcrossed hybrid diversity panel identified two novel quantitative trait variants controlling both PHT and grain yield. Here, heterogeneous inbred families demonstrated that these two loci, characterized by two single nucleotide polymorphisms (SNPs), cause phenotypic variation in inbred lines, but that size of these effects were variable across four different genetic backgrounds, ranging from 1 to 10 cm. Weekly unoccupied aerial system flights demonstrated the two SNPs had larger effects, varying from 10 to 25 cm, in early growth while effects decreased toward the end of the season. These results show that allelic effect sizes of economically valuable loci are both dynamic in temporal growth and dynamic across genetic backgrounds, resulting in informative phenotypic variability overlooked following traditional phenotyping methods. Public genotyping data show recent favorable allele selection in elite temperate germplasm with little change across tropical backgrounds. As these loci remain rarer in tropical germplasm, with effects most visible early in growth, they are useful for breeding and selection to expand the genetic basis of maize.

Keywords: high-throughput phenotyping; temporal loci effects; unoccupied aerial system.

Figure 1

Breeding scheme of generating HIFs based on two SNP models and selection stages of pedigrees via KASP technology (http://www.kbioscience.co.uk/). Ten to 20 plants from each plot were randomly selected or aided by markers for multiple generations until obtaining NILs (BC₃F₂ or more recurrent parent crosses or selfs). Only those having heterozygous loci (XY) were selected each generation and their ears were grown as rows (ear-to-row selection). After obtaining NILs, homozygous calls from both SNPs were selected as both identical (XX:XX, YY:YY) and opposite (XX:YY, YY:XX) to generate HIFs. All parents were genotyped (left). Parents; Ki3, NC356, Tx740, and LH82, calls (SNP1: SNP2) are YY:YY, XX:XX, YY:YY, and XX:XX, respectively. No template controls, black color in KASP figure, were used in each plate as negative controls.

Figure 2

BLUEs of all three ruler measures of plant heights. This showed XX calls significantly increased all height measures in a consistent direction across populations. Population 1, 2, 3, and 4 are NILs of [LAMA (recurrent parent) × LH82], [Ki3 × NC356 (recurrent parent)], [Ki3 (recurrent parent) × NC356], and [Tx740 (recurrent parents) × NC356], respectively. BLUEs were calculated using Equation 1$({\mathrm{SNP}}_i \mathrm{term})$. Differences of BLUEs between XX and YY calls were statistically significant across all populations for TH which changed between 2.0 cm and 8.9 cm for SNP1 and between 3.0 cm and 11.9 cm for SNP2. *, **, and *** indicate significance levels at 0.05, 0.01 and 0.001, respectively, while ns indicates not significant. Whiskers represent the standard error. TH, tip of tassel height; FH, flag leaf collar height; and EH, height of the first ear shank from ground on the x-axis.

Figure 3

Temporal resolution of differences between SNP1 (left) and SNP2 (right) calls obtained by Equation 2$({\mathrm{SNP}}_j \mathrm{term})$ during UAS flights across all populations. Whiskers represent the standard error. BLUEs of calls (XX vs YY) were orthogonally contrasted for each SNP at each time point and statistically significant differences were placed above the effects. *** indicates significance level at 0.001, while ns indicates not significant.

Figure 4

Temporal resolution of interactions of ${[\mathrm{Pop}*\mathrm{SNP}]}_{ij}$ obtained by Equation 2 during UAS flights. Modeling interactions showed that there were large differences between how the SNPs behaved on different genetic backgrounds. Whiskers represent the standard error. BLUEs of calls (XX vs YY) were orthogonally contrasted for each SNP in each population at each time point and statistically significant differences were placed above the effects for each time points. *, **, and *** indicate significance levels at 0.05, 0.01, and 0.001 respectively, while ns indicates not significant.

Figure 5

Temporal resolution of differences among SNP1-SNP2 interactions during UAS flights. The interaction${[\mathrm{SNP}1*\mathrm{SNP}2]}_{jk}$was obtained from Equation 3 and shows that the two loci had a synergistic effect on increasing height. Whiskers represent the standard error. BLUEs of XX:XX (SNP1:SNP2) and other call combinations (XX:YY, YY:XX, and YY:YY) were contrasted for SNP1 and SNP2 interactions at each time point and statistically significant differences were placed above the effects for each time points. and *** indicate significance levels at 0.05, 0.01, and 0.001 respectively, while ns indicates not significant.

Figure 6

Temporal resolution of differences for two populations among SNP1-SNP2 interactions during UAS flights. Interactions ${[\mathrm{Pop}*\mathrm{SNP}1*\mathrm{SNP}2]}_{\mathit{ijk}}$ obtained from Equation 3 showed the SNP combinations had different effects across different populations genetic backgrounds, especially early in the season. Whiskers represent the standard error. BLUEs of XX:XX (SNP1:SNP2) and other call combinations (XX:YY, YY:XX, and YY:YY) were contrasted for SNPs and population interactions at each time point and statistically significant differences were placed above the effects for each time points. and *** indicate significance levels at 0.05, 0.01, and 0.001 respectively, while ns indicates not significant.

Figure 7

The allelic frequency combinations of SNP1 and SNP2 over years for five germplasm categories. The favorable C (SNP1) and A (SNP 2), referred to as XX, XX in this study, are both increasing in frequency in newer germplasm and are essentially fixed in US temperate Ex-PVP and public germplasm. The 989 subset of genotyped lines contained 448 public inbred lines, 87 GEM-like lines, 215 GEM lines, 118 Ex-PVP lines, and 121 CIMMYT germplasm lines.