Machine learning for genomic and pedigree prediction in sugarcane.

Sugarcane (Saccharum spp.) plays a crucial role in global sugar production; however, the efficiency of breeding programs has been hindered by its heterozygous polyploid genomes. Considering non-additive genetic effects is essential in genome prediction (GP) models of crops with highly heterozygous polyploid genomes. This study incorporates non-additive genetic effects and pedigree information using machine learning methods to track sugarcane breeding lines and enhance the prediction by assessing the degree of association between genotypes. This study measured the stalk biomass and sugar content of 297 clones from 87 families within a breeding population used in the Japanese sugarcane breeding program. Subsequently, we conducted analyses based on the marker genotypes of 33,149 single-nucleotide polymorphisms. To validate the accuracy of GP in the population, we first predicted the prediction accuracy of the best linear unbiased prediction (BLUP) based on a genomic relationship matrix. Prediction accuracy was assessed using two different cross-validation methods: repeated 10-fold cross-validation and leave-one-family-out cross-validation. The accuracy of GP of the first and second methods ranged from 0.36 to 0.74 and 0.15 to 0.63, respectively. Next, we compared the prediction accuracy of BLUP and two machine learning methods: random forests and simulation annealing ensemble (SAE), a newly developed machine learning method that explicitly models the interaction between variables. Both pedigree and genomic information were utilized as input in these methods. Through repeated 10-fold cross-validation, we found that the accuracy of the machine learning methods consistently surpassed that of BLUP in most cases. In leave-one-family-out cross-validation, SAE demonstrated the highest accuracy among the methods. These results underscore the effectiveness of GP in Japanese sugarcane breeding and highlight the significant potential of machine learning methods.

Detection of a major QTL related to smut disease resistance inherited from a Japanese wild sugarcane using GRAS-Di technology.

Smut disease of sugarcane causes considerable yield losses and the use of resistant varieties is the best control practice. Our group identified a Japanese wild sugarcane with highly smut disease resistance named 'Iriomote8'. In this study, we conducted QTL analysis for smut disease resistance using a mapping population derived from a resistant variety 'Yaenoushie', in which resistance is inherited from 'Iriomote8'. We identified 4813 non-redundant markers using GRAS-Di technology and developed a linkage map of mapping parents. We evaluated smut disease resistance of the mapping population by the inoculation test. Consequently, a large number of clones did not show the disease symptoms and the distribution of smut disease incidence tended to be "L shaped". Composite interval mapping detected an identical QTL for indices of smut disease incidence with a markedly high LOD score (26.6~45.6) at the end of linkage group 8 of 'Yaenoushie'. This QTL explained approximately 50% of the cases of smut disease incidence. In the mapping population, there were no correlations between the indices of smut disease incidence and other agronomic traits. In conclusion, this QTL could be used for marker-assisted selection to significantly improve smut disease resistance without negative effects on other agronomic traits.

Estimation of total CH4 emission from Japanese rice paddies using a new estimation method based on the DNDC-Rice simulation model.

Methane (CH4) is a greenhouse gas, and paddy fields are one of its main anthropogenic sources. In Japan, country-specific emission factors (EFs) have been applied since 2003 to estimate national-scale CH4 emission from paddy field. However, these EFs did not consider the effects of factors that influence CH4 emission (e.g., amount of organic C inputs, field drainage rate, climate) and can therefore produce estimates with high uncertainty. To improve the reliability of national-scale estimates, we revised the EFs based on simulations by the DeNitrification-DeComposition-Rice (DNDC-Rice) model in a previous study. Here, we estimated total CH4 emission from paddy fields in Japan from 1990 to 2010 using these revised EFs and databases on independent variables that influence emission (organic C application rate, paddy area, proportions of paddy area for each drainage rate class and water management regime). CH4 emission ranged from 323 to 455ktCyr-1 (1.1 to 2.2 times the range of 206 to 285ktCyr-1 calculated using previous EFs). Although our method may have overestimated CH4 emissions, most of the abovementioned differences were presumably caused by underestimation by the previous method due to a lack of emission data from slow-drainage fields, lower organic C inputs than recent levels, neglect of regional climatic differences, and underestimation of the area of continuously flooded paddies. Our estimate (406ktC in 2000) was higher than that by the IPCC Tier 1 method (305ktC in 2000), presumably because regional variations in CH4 emission rates are not accounted for by the Tier 1 method.

Development of a method for estimating total CH4 emission from rice paddies in Japan using the DNDC-Rice model.

Methane (CH4) is a greenhouse gas, and paddy fields are one of its main anthropogenic emission sources. To mitigate this emission based on effective management measures, CH4 emission from paddy fields must be quantified at a national scale. In Japan, country-specific emission factors have been applied since 2003 to estimate national CH4 emission from paddy fields. However, this method cannot account for the effects of weather conditions and temporal variability of nitrogen fertilizer and organic matter application rates; thus, the estimated emission is highly uncertain. To improve the accuracy of national-scale estimates, we calculated country-specific emission factors using the DeNitrification-DeComposition-Rice (DNDC-Rice) model. First, we calculated CH4 emission from 1981 to 2010 using 986 datasets that included soil properties, meteorological data, and field management data. Using the simulated site-specific emission, we calculated annual mean emission for each of Japan's seven administrative regions, two water management regimes (continuous flooding and conventional mid-season drainage), and three soil drainage rates (slow, moderate, and fast). The mean emission was positively correlated with organic carbon input to the field, and we developed linear regressions for the relationships among the regions, water management regimes, and drainage rates. The regression results were within the range of published observation values for site-specific relationships between CH4 emission and organic carbon input rates. This suggests that the regressions provide a simplified method for estimating CH4 emission from Japanese paddy fields, though some modifications can further improve the estimation accuracy.