Sorghum mutation breeding for tolerance to water deficit under climate change

1 Institut de l’Environnement et de Recherches Agricoles (INERA), 04 BP 8645 Ouagadougou 04, Burkina Faso. 2 Université Pr Joseph Ki Zerbo, O6 BP 9499 Ouagadougou 06, Burkina Faso. 3 Joint FAO/IAEA Division of Nuclear Techniques in Food and Agriculture Department of Nuclear Sciences and Applications, International Atomic Energy Agency, Vienna International Centre, P. O. Box 100, 1400 Vienna, Austria.


INTRODUCTION
Sorghum (Sorghum bicolor (L.) Moench) is one of the major cereal crops in the world. It is the fifth most cultivated dry cereal after wheat, maize, rice and barley and the second most cultivated in Africa after maize (Ng'uni et al., 2011). It is a staple food crop for millions of African farmers living in the semi-arid tropics (Dora et al., 2014). However, sorghum cultivation is affected by drought, a situation which could become severe in sub-Saharan Africa in the context of climate change. Water deficit caused by drought is the most severe environmental limitation to sorghum grain yield during the entire crop production period ( nche -Blanco et al., 2002). Due to population growth (3%) in Africa, the core challenge for agriculture in Africa would be to increase food production under changing climatic conditions. Drought events can occur at any stage of sorghum growth but three stages are identified as critical phases sensitive to water deficit (Menezes et al., 2015). The growth stage 1 (GS1) corresponds to the vegetative phase, the growth stage 2 (GS2) corresponds to the preflowering phase with panicle initiation at flowering and the growth stage 3 (GS3) corresponds to the post-floral phase with filling and physiological maturity of the grains. Stay-green in sorghum is one of the reliable traits related to drought tolerance. Traits associated with pre-or postfloral water deficit resistance in sorghum also involve relative water content (RWC) and leaf senescence (Sakhi et al., 2014). The most sustainable ways to mitigate adverse effects of drought on sorghum production are field irrigation and provision of drought-tolerant varieties to farmers. Unfortunately, farmers in developing countries cannot afford irrigation facilities. Therefore, the development of drought-tolerant sorghum varieties is the most promising option to assist African farmers in adapting to drought. The strategy for development of new crop genotypes for drought tolerance could be to create variation within the gene pool. Genetic variability in traditional sorghum varieties is very low in Burkina Faso, around 4.5% of the genetic variability between agroecological zones and 5.8% between villages in the same zone (Kondombo-Barro, 2010). Mutation induction has been proven to be an effective method to increase genetic variability in crops. Induced mutagenesis in crop varieties preferred by farmers is a promising strategy to improve agronomic traits such as tolerance to water stress. Genetic variability created through mutagenesis is important for sustainable agriculture (Griggs et al., 2013). According to International Atomic Energy Agency (IAEA) database (http://mvgs.iaea.org), there are more than 3,300 officially released mutant varieties of 170 different species in more than 60 countries around the world that not only increase biodiversity but also provide material for plant breeding (Jankowicz-Cieslak et al., 2017). Mutation induction can be carried out using chemical or physical mutagens (Shahab et al., 2018). Some of the agronomic traits generated as result of mutation induction are: increasing 3-Deoxyanthocyanidin accumulation in sorghum leaves (Petti et al., 2014), dwarfism, early flowering, high protein digestibility and high lysine content which have been widely used in sorghum breeding (Oria et al., 2000). The aim of this study was to develop drought-tolerant mutants in a farmer-preferred sorghum variety (ICSV1049) for adaptation to water deficit that limits cereal production in sub-Saharan Africa.

Study sites and genetic materials
A survey on adoption and dissemination of sorghum varieties from participatory breeding in Burkina Faso was conducted in partnership between the Institute for the Environment and

Generation of mutant progenies and selection of potential drought-tolerant M4 mutant lines
The irradiated seeds and control were sown and M1 panicles were harvested and planted as M2 panicle-to-one progeny. Forty of the M2 seeds from each M1 panicle were planted as a head row. Three panicles from each row were bagged before anthesis. To prevent redundancy of mutations, only one fertile plant from each M2 head row were selected to produce M3 seeds according to Xin et al. (2008). The M3 families were repeatedly evaluated for phenotypes distinctive from wild-type ICSV1049. Thus, the phenotypes were organized into tillering types, plant height, leaves vigor, panicle shapes and seeds size from ascend stage to grains physiological maturity. 394 lines were selected on the basis of the phenotypes described earlier and confirmed in the next generation. A field-trial was conducted during the cropping season in 2017 at Kouaré Research Station to evaluate 394 putative mutant lines at the M4 generation and their parent for tolerance to the end of season drought. The planting was done on 18th August so that the bloom stage coincided with the end of rainfall. Each line was planted on row 2.7 m length, 0.3 m between planting hills and 0.7 m between rows. The experimental design was an alpha lattice design plot using 15 blocks and each block contained 20 genotypes with two replications.
The field was weeded three times. Mineral fertilizer of 100 kg ha -1 of NPK (12-24-12) was applied to the plots at sowing and 50 kg ha -1 of urea was applied at the booting growth stage. The amount of rainfall recorded from planting to the harvest (18th August to 15th December, 2017) was 201.5 mm corresponding to 10 rain events or 23.4% of total rainfall (860 mm) recorded in 2017 on Kouaré Station.
The selection of drought-tolerant lines was made in under field conditions based on productivity per line including phenotypic traits such as tiller number, panicle filling, grains quality, number of leaves per plant and leaf vigour. A total of 143 mutant progenies M5 were selected for screening under water deficit in controlled conditions.

Screening of M5 mutant lines under soil water deficit
The potential drought tolerant lines which exhibited different morphological traits of the parent were selected and confirmed by screening under water deficit in controlled conditions during the dry season of 2018. The experiment was conducted at Kamboinsé research station (143 mutant lines and one control were screened). The experimental design used was an alpha lattice square with 12 blocks and 12 genotypes per block using three replications. Each genotype was sown on row of 1.5 m length. The spacing of planting hills within single and between rows, plot fertilization and weeding were carried out as described previously. After planting, watering was performed every three days with tap water installed around the field. Sorghum seedlings were thinned at 14 days after sowing to get one plant per hill. Water deficit stress was applied to sorghum plants by cessation of irrigation 65 days after the sowing (DAS) until harvest.
Statistical analyses were carried out using Statistical Analysis System (SAS, 9.1, 2 Institute, Cary, NC). Analysis of variance (ANOVA) was used to determine significance based on P-value.
Means were separated using Newman Keuls Multiple Range test and differences between sorghum lines traits were considered significant levels of 5% (P < 0.05). The correlation coefficient between traits and genotypes clustering were analysed using R x 64 3.5.2 software.

Induced traits in M5 mutant lines
Selection of sorghum mutants is based on phenotypes observed by comparison of putative mutants with the parent variety (ICSV1049). Upon exposition of 143 M5 lines to soil water deficit, 118 lines and the parent survived and the semi-erect leaves, late maturity and single stalked were the most frequently observed phenotypes (Table 1). Some agronomic traits such as tiller number, plant height, leaf aspect and grain maturity were affected by gamma radiation (Table 1). As expected, variation was higher in the M5 population screened under water deficit compared to control population.

Correlations between measured parameters
Panicle weight (g) had high significance and positive correlation with grain weight (Table 3). Leaf senescence and relative water content were significantly and positively correlated (R 2 =0.57%). Panicle and grain weight were also correlated with leaf senescence, SPAD I, SPAD II and RWC. The correlation between weight of panicle, weight of grains and leaf senescence is relevant to the leaf senescence effect on grain yield. However, there was a negative correlation between LS, DaFl and DaMa (Table 3). RWC is negatively correlated to SPAD I and positively correlated with leaf senescence while plant height was positively correlated to all the characters except SPAD I which indicates that these traits do not evolve in the same direction as the SPAD I. There is a negative correlation between DaFl, MaDa and grain yield (Table  3).

Clustering and selecting of best mutant lines for tolerance to water deficit
Analysis of variance of LS showed that there was a significant difference (P < 0.0001) between the mutant lines. Therefore, genotypes were grouped with statistically identical leaf senescence values. Thus, sorghum mutants were classified into 4 clusters (Table  4). Mutant lines with average leaf senescence between 64.3 and 100% (C1) were 105 including the parent. The next cluster made up of 6 lines with average leaf senescence between 52.6 and 61.7% (C2) followed by cluster (C3) with 5 lines and an average leaf senescence around 25.2 to 46.4%. The last cluster (C4) consisting of 3 lines had an average leaf senescence around 11.3 to 20%. The lowest percentages of LS were recorded in C3 (LS average~36%) and C4 (LS average~16.38%) corresponding to the scale 4 and 2, respectively. The highest average SPAD I value was recorded with mutant lines found in cluster C3 (44.6 µmol/mg) and C1 (44.2 µmol/mg) compared to the others clusters. However, the coefficient of variation (<15%) indicates that there is low variability of chlorophyll content at the beginning of the application of soil water deficit. The highest values of SPAD II were recorded at 33.8 µmol/mg in C4 and there was high variability in chlorophyll content (SPAD II) after the application of soil water deficit (CV>15%). The highest grain and panicle weights were recorded with clusters C3 and C4 clusters and the lowest were recorded with C1 and C2. There was a significant difference (P<0.0001) between the traits within the cluster Ht= Plant height, SPAD I and SPAD II, chlorophyll content at initial day of water deficit application and 13 days after water deficit application, DaFl= flowering delay, DaMa= physiological maturity delay, RWC= relative water content, GrWt= grains weight, PaWt= panicle weight, LS= leaf senescence, CV = coefficient of variation. C1. But no significant difference was observed within the other clusters (P>0.05) except C3 where PaWt exhibited a significant statistical difference (P<0.05) ( Table 5). Tables 4 and 5 indicated that sorghum tolerant mutants to water deficit could be selected inside C3 and C4. According to LS, RWC, SPADII and GrWt which are the best parameters of tolerance to water deficit, seven promising water deficit tolerant mutants were selected. This selecting was based on the mutants with high values of RWC, SPAD II, GrWt and low values of LS (Figure 1). Based on the analysis outputs, the best performing mutants under water deficit conditions were ICM5_6, ICM5_104, ICM5_76, ICM5_3, ICM5_30, ICM5_15 and ICM5_105. They are distinguished from other mutants and parent by high relative water content (between 61 -83%), high SPAD II (21 -37 µmol/mg) and GrWt (28-54 g) with the lowest LS (10.6-39.9%) while RWC, SPAD II, GrWt and LS of the parent were 34%, 18 µmol/mg, 18.9 g and 94%, respectively (Figure 2).

DISCUSSION
Mutation induction is a powerful tool that plant breeders use to create genetic variability. That variability can be exploited to select desired traits. Mutagens can affect all parts of the plants by either decreasing plants height or increasing it relative to the parent. Their effect can shorten or extend the plant cycle. In plants exposed to mutagens, morphological abnormalities and reduced growth have been observed and attributed to oxidative stress (Singh, 2003) or deleterious mutational effects (Valluru et al., 2019). The mutagenesis affected some agronomic traits in millet such as reduction in plant height compared to the control (Ambli and Mullainathan, 2014). The results of this study showed that the overall height of plant mutant and grain yield were higher than that of the parent. These results support findings by previous studies (Burow et al., 2014) in which mutation increased sorghum plant height and grain yield. However, reduction in plant height was observed in mutagenized rice , rapeseed and mustard (Javed et al., 2003). The present results together with those cited confirm that induced mutation using gamma rays can play an important role in the genetic variability induction within plant architecture.
The positive correlation between weights of panicle and grains to SPAD I, SPAD II and RWC suggests that these traits can be simultaneously selected and may be used as selection criteria for tolerance to soil water deficit. Negative correlation between RWC and SPAD I suggest that it would be difficult to select drought tolerant plants at the beginning of water deficit application based on these two parameters. The positive correlation between RWC and LS indicates that mutants which accumulate sufficient water in their leaves are those which have a slower leaf age. Therefore, genetic improvement of RWC also implies genetic improvement of LS. RWC is a useful trait for plants to mitigate the effects of drought at the reproductive stage. Negative correlation between delay flowering, delay maturity and grain yield implies that early mutants have no significant difference in yield compared to late mutants. These results disagree with those obtained by previous studies (Menezes et al., 2015) reporting that there is a positive correlation between productivity and maturity in sorghum grain under water deficit. Water deficit tolerance is the capacity of plants to support water deficit while keeping suitable physiological activities to safeguard cellular and metabolic integrity at tissue and cellular level (Xiong et al., 2006). Plant leaf senescence is considered as a post-flowering drought stress symptom (Burke et al., 2013). Green plants such as sorghum have two options for maintaining a high tissue water status during periods of soil moisture deficit, either by decreasing water loss due to transpiration or by increasing water uptake (Devnarain et al., 2016). Leaf senescence reduces seriously the source-sink translocation from leaves to grain (Krupa et al., 2017). In the present study, drought scoring based on LS was significantly higher for the wild type accession. Based on some reports (Ji et al., 2010) the results of this study on PaWt and GrWt revealed that soil water deficit affects grain number and weight. The decrease in quantitative traits such as yield of some mutant lines may be attributed to the physiological disruption or chromosomal deterioration caused to plant cells by the mutagen gamma ray (Thilagavathi and Mullainathan, 2011). Relative water content designates the metabolic activity in tissues and used as the most meaningful index for dehydration tolerance. RWC ranged between 85 and 95% and a critical reduction of less than 50% could cause tissue death (Vinodhana and Ganesamurthy, 2010). From the results of this study, some sorghum mutant lines were able to maintain RWC above 60% for 14 days during soil water deficit. Previous studies showed that maintenance of a relatively high RWC during mild drought is indicative of drought tolerance (Colom and Vazzana, 2003).
Sorbitol treatments to simulate drought-induced osmotic stress in sorghum cell suspension cultures showed that sorbitol raised an overall increase in secretion of 92 proteins that were differentially expressed in response to sorbitol-induced osmotic stress (Ngara et al., 2018). So, additional molecular studies on developed sorghum mutant lines would allow identifying protein or genes of interest via biotechnological or marker assisted breeding strategies with the prospect to combine them in one line for more performance.

Conclusion
Late drought is the most limiting factor in sorghum production. The results showed that induced mutation is a suitable tool to create genetic variability for selecting drought tolerant mutants. Further evaluation of those mutants to confirm their tolerance and stability under water deficit conditions would be useful. So, multi-local tests on other experimental sites will be conducted in the coming years to evaluate the agronomic performance of the best lines, taking into account genotype-environment interaction. In addition to leaf senescence and relative water content already recommended in phenotyping for drought-tolerance, the chlorophyll content 13 days after water stress application should also be considered as phenotypic trait in similar studies.