Induction crosses
All the populations used in this study were lowland tropical or subtropical germplasm, sourced from different breeding programs of CIMMYT in Mexico. These populations were planted in a haploid induction nursery in the summer cycle of 2015 at CIMMYT’s Agua Fria experimental station (20.26°N, 97.38°W), as described in Chaikam et al. (2012). The source populations was crossed with bulked pollen from TAIL9 × TAIL8 hybrid inducer three consecutive days after silk emergence as described in Chaikam et al. (2012).
Investigating the differences between haploid and diploid seedling traits
Seed produced from the induction crosses of five populations (population 1: CL106728/LH212Ht//CML451, population 2: ((CML491/LAPOSTASEQ-C7-F64-2-6-2-1-B-B)/CML491)-B-50-1-2-1-1-B/(CLG2312/CML9)-B-80-1-1-1-B, population 3: (CLG2312/CML505)-B-43-1-1-1-B/(CLRCW79/CLRCW98)-B-22-3-1-1-B-B, population 4: (LaPostaSeqC7F1021311BBBBCL04934)DH1 × CLWN247, and population 5: (LaPostaSeqC7F1021311BBBBCL04934)DH1 × CML551)) were used in this experiment as all these induction crosses showed excellent expression of Navajo marker (data not shown). Haploid and diploid seeds were separated based on Navajo phenotype, where the diploid seeds have purple coloration on the endosperm and embryo while the haploid seeds have purple coloration only on the endosperm (Chaikam and Prasanna 2012). Twenty seeds each were randomly chosen among the haploid and diploid fractions from each population and were germinated in paper towels for 72, 94 and 104 h in a growth chamber maintained at 28–30 °C, based on protocol described by Chaikam and Mahuku (2012). RL and CL were measured from the point of visible emergence to the tip using a meter scale. NLSR was counted at each time point. After measurements were done, the seedlings were transplanted in the field in rows spaced at 75 cm with a plant-to-plant spacing of 20 cm. After 30 days in the field, true ploidy status of each plant was confirmed for each surviving plant based on adult plant characteristics (plant vigor, erectness and paleness of leaves that can typically differentiate haploids from diploids). Haploids are generally weak with erect, narrow and pale leaves compared as to their diploid counterparts (Chase 1964; Prigge et al. 2011; Melchinger et al. 2013; Xu et al. 2013). Data collected on the confirmed haploids and the diploids was used for further analysis. A linear model was used to test the significance of differences between haploids versus diploids, as well as among different populations, and different time points, assuming normal distribution of the errors for RL and CL. The statistical model was:
$$y_{ijkl} = \mu + \alpha_{i} + \beta_{j} + \alpha \beta_{ij} + \delta_{k} + \alpha \delta_{ik} + \beta \delta_{jk} + \alpha \beta \delta_{ijk} + \varepsilon_{ijkl}$$
where y
ijkl
response variable, µ overall mean, α
i
effect of the ith population, β
j
effect of the jth ploidy type, δ
k
effect of the kth hour, αβ
ij
, βδ
jk
and αβδ
ijk
are the interactions of the respective effects, \(\varepsilon \sim N\left( {0,\;\sigma^{2} } \right)\).
Least squares method was used to perform estimation and testing, F test was used for overall effect testing and t test to compare levels of the factors. Results are presented as mean squares, F values and significance values.
For the NLSR a generalized linear model with poisson distribution and log link was used. The used model was similar as the model described for RL and CL with the difference being that the response variable follows a poisson distribution and that the response variable is the log of NLSR. Maximum likelihood method was used to perform estimation and testing, as above; F and t tests were used to compare the overall effects and levels of the effects. Due to the method used, mean squares are not available, and only F values and related P values are shown.
Use of seedling traits for haploid/diploid classification in induction crosses with inhibition of R1-nj
Induced seed from six populations that showed complete inhibition of Navajo phenotype were selected for this experiment. From each of the population, 1500 seeds were germinated in paper towels for 96 h, as described above. Seedlings were separated into haploids, diploids and undetermined classes based on visual assessment of seedling traits, primarily CL and RL. Additional traits that were used include NLSR, thickness of coleoptiles, thickness of radicles, and extent of root hairs. High proportion of germinated seedlings were vigorous with long coleoptiles, long radicles, and higher NLSR—traits typical of putative diploids. However, a smaller proportion of seedlings showed significantly lesser CL, RL and NLSR. Such seedlings also showed relatively thin coleoptiles, thin radicles and low number of root hairs. These were classified as putative haploids. Seedlings that cannot be easily classified into either putative haploids or putative diploids based on seedling traits were classified into an undetermined class. Separated putative haploids, putative diploids and undetermined seedlings were transplanted in the field to establish true ploidy status based on adult plant characteristics through the gold standard test based on plant vigor, erectness and paleness of the leaves, as described in Melchinger et al. (2013) and Chaikam et al. (2016). False discovery rate (FDR), false negative rate (FNR) and Matthews correlation coefficient (MCC) were also calculated (Melchinger et al. 2013; Chaikam et al. 2016). FDR is the proportion of diploids in the fraction classified as putative haploids; FNR is the proportion of true haploids misclassified as diploids; and MCC indicates the correlation between the real and predicted ploidy status based on the seedling traits.
Use of seedling traits for identification of false positives resulting from R1-nj marker based haploid identification
To determine if false positives can be reduced by early detection of diploids in haploid fraction (identified based on Navajo marker), 500 putative haploid seeds each identified using Navajo marker expression were germinated from ten populations in paper towels, as described above. After 96 h of germination, seedlings were separated into haploids and diploids based on seedling traits as described above. Separated putative haploids and putative diploids were transplanted in the field and true ploidy status of each surviving plant was ascertained based on the gold standard test using adult plant characteristics, as described earlier.