Multi-task Gaussian process for imputing missing data in multi-trait and multi-environment trials.

KEY MESSAGE: A method based on a multi-task Gaussian process using self-measuring similarity gave increased accuracy for imputing missing phenotypic data in multi-trait and multi-environment trials. Multi-environmental trial (MET) data often encounter the problem of missing data. Accurate imputation of missing data makes subsequent analysis more effective and the results easier to understand. Moreover, accurate imputation may help to reduce the cost of phenotyping for thinned-out lines tested in METs. METs are generally performed for multiple traits that are correlated to each other. Correlation among traits can be useful information for imputation, but single-trait-based methods cannot utilize information shared by traits that are correlated. In this paper, we propose imputation methods based on a multi-task Gaussian process (MTGP) using self-measuring similarity kernels reflecting relationships among traits, genotypes, and environments. This framework allows us to use genetic correlation among multi-trait multi-environment data and also to combine MET data and marker genotype data. We compared the accuracy of three MTGP methods and iterative regularized PCA using rice MET data. Two scenarios for the generation of missing data at various missing rates were considered. The MTGP performed a better imputation accuracy than regularized PCA, especially at high missing rates. Under the 'uniform' scenario, in which missing data arise randomly, inclusion of marker genotype data in the imputation increased the imputation accuracy at high missing rates. Under the 'fiber' scenario, in which missing data arise in all traits for some combinations between genotypes and environments, the inclusion of marker genotype data decreased the imputation accuracy for most traits while increasing the accuracy in a few traits remarkably. The proposed methods will be useful for solving the missing data problem in MET data.

Moving toward rice self-sufficiency in sub-Saharan Africa by 2030: Lessons learned from 10 years of the Coalition for African Rice Development.

The Coalition for African Rice Development (CARD) was initiated as a policy framework with the aim of doubling rice production in sub-Saharan Africa (SSA) over the period from 2008 to 2018. This paper assesses the contribution of the CARD policy to rice production and forecasts the local rice supply and demand to provide a better understanding of the policies needed to attain rice self-sufficiency by 2030. A combination of the autoregressive integrated moving average method and counterfactual approach was adopted using rice statistical data from 23 countries in SSA. The results showed that the contribution of CARD to paddy rice production in 2018 was 10.2 million tons, equivalent to 74% of the target. This contribution resulted from increases in area and yield of 23% and 19%, respectively. However, the yield growth rate was not sustainable in almost two-thirds of countries. Investments in supply-push factors such as fertilizer and irrigation development, which were the focus in the past, have limited effects on rice production. We conclude that sustainable investments in demand-pull factors such as the private-led modern milling sector and contract farming development should be prioritized to achieve rice self-sufficiency in SSA.

Mycotoxin concentrations in rice from three climatic locations in Africa as affected by grain quality, production site, and storage duration.

Information on the mycotoxin contamination of rice in Africa is limited although the risk of contamination is high. In this study, domestic milled rice processed by actors using suboptimal methods was purchased and total fumonisin (FUM), zearalenone, and aflatoxin concentrations determined at 0, 90, and 180 days after storage. Three different climatic locations, Cotonou (Benin) in the Guinea savanna, Yaoundé (Cameroon) in the Tropical forest, and N'diaye (Senegal) in the Sahel, were selected as storage sites. Subsets of the samples collected from Glazoue (Benin), Ndop (Cameroon), and Dagana (Senegal) were stored in plastic woven bags under room conditions in the respective sites with or without calcium oxide (burnt scallop shell-BSS, 0.1% w/w) treatment. Multivariance analysis showed that FUM concentration was positively influenced by the duration of storage only while zearalenone concentration was negatively influenced by relative humidity and head rice but positively by impurities. Zearalenone concentration was also influenced by sample collection/storage location, processing type, and duration of storage. Aflatoxin concentration was influenced negatively by storage room temperature and head rice but positively by impurities and chalky grains. In addition, aflatoxin concentration was influenced by collection/storage location and processing type. BSS treatment followed by storage for 6 months had no effect on the concentration of the three assessed mycotoxins. Strategies to reduce the risk of mycotoxin contamination in study sites will include the improvement of physical rice quality through better pre- and postharvest practices and proper packaging of both treated rice and untreated rice in hermetic systems before marketing and storage.

Physicochemical and nutritional properties of rice as affected by parboiling steaming time at atmospheric pressure and variety.

To elucidate the effect of different parboiling steaming time on the physicochemical and nutritional quality of rice, four varieties, NERICA1, NERICA7, IR841, and WITA4, were soaked at the same initial temperature (85°C) and steamed for 5, 15, 25, 35, and 45 min. NERICA7 steamed for 25 min recorded the highest head rice yield (71.9%). Nonparboiled IR841 recorded the shortest cooking time (17.0 min), while NERICA1 steamed for 35 min recorded the longest cooking time (26.1 min). NERICA1 steamed for 45 min was the hardest (63.2 N), while nonparboiled IR841 was the softest (28.7 N). NERICA7 recorded higher peak and final viscosities across all steaming times compared to the other varieties. NERICA7 steamed for 35 and 45 min recorded the lowest total starch (77.3%) and the highest protein (13.2%) content, respectively. NERICA7 steamed for 25 and 45 min recorded the highest phosphorus (0.166%), magnesium (572 mg/kg), and potassium (2290 mg/kg) content, respectively. We conclude that, depending on desired physicochemical and nutritional properties, specific varieties and steaming times can be selected to achieve those outcomes.

Appropriate parboiling steaming time at atmospheric pressure and variety to produce rice with weak digestive properties.

Consumers with diabetes mellitus have shown interest in products with low postprandial glucose. To produce rice for this group of consumers, the effect of parboiling steaming time (0, 5, 15, 25, 35, and 45 min) and variety (NERICA1, NERICA7, WITA4, and IR841) on resistant, damaged starch fractions and glycemic response in rats was investigated. Resistant and damaged starch fractions were influenced by variety and steaming time but this was not the case for glycemic index. Nonparboiled NERICA7 and NERICA7 steamed for 25 min recorded the highest (10.07%) and lowest (2.49%) resistant starch fraction, respectively. Resistant starch correlated negatively with protein and sodium and positively with lipids. Damaged starch was high for WITA4 steamed for 45 min (26.80%) and low for nonparboiled NERICA1 (6.59%). Damaged starch correlated positively with lipid content and negatively with ash and total starch content. NERICA7 steamed for 35 min recorded the lowest postprandial glucose level 30 min after feeding (0.16 g/L), while WITA4 steamed for 15, 25, and 35 min and nonparboiled NERICA7 recorded higher levels (0.76, 0.91, 0.84, and 0.76 g/L, respectively). NERICA7 steamed for 35 min recorded both low glycemic and weak digestive properties because the glycemic index was lowest 120 min and increased steadily up to 180 min after feeding. We conclude that the digestive properties of rice depend both on the intrinsic properties of the variety and the parboiling steaming time.

Upgrading the quality of Africa's rice: a novel artisanal parboiling technology for rice processors in sub-Saharan Africa.

In order to increase the quality of locally produced rice, the artisanal parboiling process in West and Central Africa was reconceptualized. A novel parboiling unit was constructed using stainless steel (Inox 304) and fitted directly on an improved stove made from fired bricks. The heat profile at different locations in the unit, the physicochemical properties, cooking properties of the parboiled rice, and the fuel efficiency of the stove were evaluated and compared with that of the traditional system. The heat flow in the new unit was from the top to the bottom while the reverse occurred in the traditional unit. The percent impurities and heat-damaged grains, swelling and water uptake ratios, amylose content, stickiness, and cohesiveness were lower for rice produced using the improved technology (IT) compared to the traditional technology (TT). Whole grains (%), lightness (L*), yellowness (b*), cooking time, viscosity were higher for rice produced using the IT compared to the TT. Most of physicochemical and cooking properties of rice produced using the IT were not different from that of premium quality imported rice and this was achieved when steaming time was between 20-25 min. The improved stove recorded a lower time to boil water and specific fuel consumption and a higher burning rate and firepower at the hot-start high-power phase compared to the traditional stove. Most end users rated the IT as easy and safe to use compared to the TT. The new technology was code-named "Grain quality enhancer, Energy-efficient and durable Material (GEM) parboiling technology."