Hybrid breeding for autogamous crops was less successful because of the lower amount of heterosis, high seed densities coupled with difficulties to implement a cost-effective system for hybrid seed production, lack of high yielding heterotic patterns and lower selection gain for hybrid compared to line breeding (Gu et al., 2016). Heterosis in selfing crops is often driven by additive and additive × additive gene action, the molecular basis of which is increasingly being revealed. Unlike non-additive heterosis, additive forms can be relatively easily fixed in homozygous lines meaning that their seed can simply be resown to express the same heterosis (Ginkel and Ortiz, 2018). An effort to fix heterosis due to additive forms in homozygous lines has been made by Ginkel and Ortiz who proposed a new integrated breeding strategy called Hybrid Enabled Line Profiling (HELP) which integrates modern high-throughput versions of existing and new concepts and methodologies into a breeding system, strategy which enables identifying a high frequency of superior lines fixed with heterosis due to additive gene action in self pollinated crops. This strategy emerged from experiments in the Bread Wheat Program of the International Center for the Improvement of Maize and Wheat (CIMMYT) (Ginkel and Ortiz, 2018). The HELP strategy enables the performance of F1 hybrids to guide the profiling and development of F1 derived lines or doubled haploids (DHs). It shows how hybrids can help to identify the best performing crosses with additive and additive × additive gene action which predicts high progeny performance. Based on F1 hybrid testing, only the very top hybrids are promoted for line or DH derivation which further dramatically reduces the number of crosses being promoted while increasing the likelihood of obtaining superior new cultivars. In HELP, the F1 hybrids themselves are not the end product of commercial cultivars, rather hybrids help to identify near homozygous lines or DHs which are then released as varieties. In brief, the various integrated steps involved in HELP are selection of best parents, predicting heterosis and making crosses, assessing F1 hybrids rigorously, fixing homozygosity, assessing derived line performance and releasing cultivar (Ginkel and Ortiz, 2018). Gu et al. (2016) detected and analysed the expression pattern of differentially expressed genes (DEGs) responsible for the establishment of growth heterosis in superhybrid rice (LYP9) under nutrient-deficient and nutrient-sufficient condition. They identified 336 nutrient-sufficient specific DEGs related to heterosis and expression analysis of these genes between a hybrid and its parents revealed that majority of DEGs (88.9–89.8%) expressed in an additive manner. Moderate numbers of DEGs (10.06-10.58%) were dominantly expressed. Surprisingly, only 0.6% of the DEGs were expressed in an over-dominant manner. They concluded that gene expression pattern that avoids extremes seems beneficial to enhance the phenotypic performance of superhybrid rice (LYP9). Friedrichs et al. (2016) investigated a possible relationship between heterosis in the F2 generation of four biparental crosses (F, G, H and I) and genetic variation among random F5:6 lines derived from the crosses in soybean. They found that crosses F and H exhibited positive yield heterosis, while crosses G and I exhibited negative yield heterosis. They observed significant differences in estimates of genetic variance for yield among the 46 or 47 random F5:6 lines derived from heterotic crosses (F and H). Whereas, genetic variance estimates were zero and non-significant for F5:6 lines derived from crosses I and G respectively. They concluded that heterosis in the F2 bulk of a cross can be used to predict genetic variation for yield in lines derived from that cross. Xu et al. (2016) predicted heterosis in hybrid rice based on genomic, transcriptomic and metabolomic data using six different prediction methods (LASSO, BLUP, SSVS, PLS, SVM-RBF and SVP-POLY). They found that among six prediction methods, LASSO and BLUP are the most efficient methods for yield prediction. Further, they elucidated that genomic data remain the most efficient predictors for high heritability traits. The predictability of hybrid yield was almost doubled when metabolomic data was used compared to genomic and transcriptomic data. Hence, they concluded that metabolomic prediction for hybrid yield is more effective than genomic prediction. Lado et al. (2017) compared strategies for predicting crosses based on mid-parent value and progeny variance using genomic data of grain yield and baking quality traits in wheat. Their results indicated that predicted mean progeny performance and progeny variance was the strongest driver for selecting superior crosses for grain yield and quality traits respectively. They concluded that selecting crosses based on mid-parent value and progeny variance were best for yield and quality traits respectively. Hybrid breeding in selfing crops has not been a major and consistent commercial success in many crops for several reasons. The known obstacles tend to frustrate attempts to establishing a hybrid industry for selfing crops. However, there is ample scope to exploit heterosis in selfing crops by integrating genomic, biotechnological and bioinformatic tools. HELP is such a new integrated breeding strategy which focuses on the most superior crosses (less than 10 percent of all crosses). This focus results in significant increases in efficiency and can reverse the edible yield plateauing seen or feared in some of our major selfing food crops.

References:

Friedrichs, M. R., Burton, J. W. and Brownie, C., 2016, Heterosis and genetic variance in soybean recombinant inbred line populations. Crop Sci., 56: 2072-2079.

Ginkel, M. V. and Ortiz, R., 2018, Cross the best with the best, and select the best: HELP in breeding selfing crops. Crop Sci., 58: 17-30.

Gu, L., Wu, Y., Jiang, M., Si, W., Zhang, X., Tian, D. and Yang, S., 2016, Dissimilar manifestation of heterosis in superhybrid rice at early-tillering stage under nutrient-deficient and nutrient-sufficient condition. Plant Physiol., 172: 1142–1153.

Lado, B., Battenfield, S., Guzman, C., Quincke, M., Singh, R. P., Dreisigacker, S., Peña, R. J., Fritz, A., Silva, P., Poland, J. and Gutiérrez, L., 2017, Strategies for selecting crosses using genomic prediction in two wheat breeding programs. Plant Genome, 10(2): 1-12.

Xu, S., Xu, Y., Gong, L. and Zhang, Q., 2016, Metabolomic prediction of yield in hybrid rice. Plant J., 88: 219–227.