Introduction

Millets are small-seeded edible cereals belonging to the Gramineae Family, grown under tropical and arid climates (Maharajan et al. 2019). Millets include several genera (Eleusine, Setaria, Echinochloa, Cenchrus, Panicum and Paspalum) of annual grasses that produce small seeds; they are categorized into several main types which are finger millet (Eleusine coracana), foxtail millet (Setaria italica), pearl millet (Cenchrus americanus), little millet (Panicum sumatrense), proso millet (Panicum miliaceum), kodo millet (Paspalum scrobiculatum), and barnyard millet (Echinochloa esculenta) (Arya et al. 2014). They were one of the first cultivated foods of humans and are an important staple food in many parts of the developing world (Krishna et al. 2018). Millets are nutritionally superior to other major non-millet cereal crops like rice and wheat (Ragaee et al. 2006). Millets are exploited for valuable food products in the semi-arid tropics of developing countries; therefore, they are considered as crops of poor people (Maharajan et al. 2021). Millets are rich sources of minerals, particularly iron, calcium, magnesium, phosphorus, zinc, and potassium, and also contain a rich source of carbohydrate, protein, fatty acids, vitamins, dietary fiber, and polyphenols, which contribute to several health benefits (Swaminaidu et al. 2015). They are termed as nutria-millets/nutria-cereals/nutraceuticals. The grains of millets have a longer storage life and can be termed as famine reserve. They can fit well in any multiple cropping systems due to their short life cycle and wide range of adaptability to diverse agro-ecosystems.

Despite possessing climate-resilient and nutrient-rich properties, millets also face multiple constraints that can influence their production. Among biotic constraints, millets suffer more from fungal diseases than from viral, bacterial, and nematode diseases (Das and Rakshit 2016). Newly emerging problems with disease and insect pests may pose a threat to millet production as a result of climate change. Abiotic constraints that limit millet production include soil moisture incidence (drought), low soil fertility (nutrient deficiency), and elevated soil salinity (Das and Rakshit 2016). Nutrient deficiency is another important issue that affects the productivity and quality of millets (Ceasar et al. 2020; Ceasar et al. 2014). Nutrient deficiency was observed in a wide range of soil types in major crops including millets (Maharajan et al. 2018). Mainly, these soils are low in organic carbon, nitrogen, phosphorus and zinc. Soil salinity is also an important constraint that reduces plant growth, water content, shoot and root biomass, delayed flowering and finally impaired millet production (Parvathi and Nataraja 2017).

A wide range of biotic and abiotic constraints are known to reduce the productivity of millets, leading to low efficiencies of input use, suppressed output, and ultimately reduced yield (Saha et al. 2016). Despite years of improvement in agriculture and food production, people fighting hunger around the world exceed one billion. The world’s population may increase upto 9.8 billion by 2050, which will drive up global food demand (Vetriventhan et al. 2020). Therefore, the growing demand for food becomes a serious challenge in the meeting of food security. From a global perspective, rice, wheat and maize are major cereals in terms of food security, followed by millets. However, the importance of millets in conserving food and nutritional security cannot be judged due to their poor share in total food grain production. The projected food demand will require the yield of cereals, including millets, to rise to 70% (Vetriventhan et al. 2020). Farmers will need to increase crop cultivation, either by increasing the amount of agricultural land to grow more crops or by increasing the productivity on existing agricultural lands through fertilizer, irrigation, and implementing new molecular and biotechnological approaches.

Crop improvement has been one of the major priority areas of research in millets. Genome-assisted breeding (GAB) plays a crucial role in modern breeding and enhancing crop productivity in other crops. The rapid advances in molecular marker technology followed by genomics and next-generation sequencing (NGS) in major cereal crops such as rice, wheat and maize have provided opportunities to carry out genome-assisted crop improvement (Krishna et al. 2021). Genome sequencing provides information on the genes encoded within the genome. It helps to identify gene diversity in genomes of different accessions in the same species (Wu et al. 2019). The genome sequence information allows the development of new molecular markers, development of genetic maps, identification of valuable genes, alleles, and quantitative trait loci (QTL), etc., for key traits (Takeda and Matsuoka, 2008; Edwards and Batley, 2010). The transcriptome analysis also provides information on genes involved in the trait of interest. Analysis of genes differentially expressed under stress conditions can provide information on the functions of genes. For example, a fully sequenced and annotated transcriptome sequence of rice provided approximately 3000 expressed genes (Ohyanagi et al. 2006). The genome sequence is available for many major cereal crops such as rice, wheat and maize. The genomes of some millets are yet to be sequenced. The complete and annotated genome sequences are available only for foxtail millet and finger millet, and only draft genome sequences are available for proso millet, pearl millet, and barnyard millet. Very few genomic resources such as DNA markers and genetic maps are available in millets compared to other major cereals. It limits crop improvement via GAB. In this review, we discuss the details on the genome sequence data available for millets and highlight the importance and implications of such resources in the post-genomic era of millet improvement.

Genome sequences of millets

The genome sequence information is very useful for tracing key genes controlling agronomically important traits and identifying genetic variability among crop plants. The NGS technology provides insights into many important crop genomes. The genome sequencing techniques can be used for several strategies such as single nucleotide polymorphism (SNP) identification, allele mining, development of mapping population, identification and functional characterization of candidate genes, comparative genome studies, and GAB programs. Genome sequences of several cereals like rice, maize, barley, wheat, and soybean were released long back with complete annotations. Especially the draft genome of rice was released in 2005 (Matsumoto et al. 2005) and annotations were completed in 2013 (Kawahara et al. 2013). But sequencing and annotation of millet genomes are lagging far behind. In millets, genome sequences are available for foxtail millet, pearl millet, finger millet, proso millet, and barnyard millet so far. The genome sequence of millets is a good resource for an efficient improvement program based on genomics (Fig. 1). But the genome sequences of proso millet, pearl millet, and barnyard millet are not yet annotated. Among the millets, the genome size of foxtail millet is 510 Mb (smallest genome) and pearl millet is ∼1.79 Gb (largest genome). Foxtail millet is the first millet to have its whole genome sequenced for two varieties. It serves as a model crop (C4 plant) for many functional genomics studies in millets. The completely annotated genome sequence of foxtail millet helped to conduct comparative genomic studies in other millets (Ramakrishnan et al. 2016; Ramakrishnan et al. 2017; David et al. 2021). The comparative genomic study could help to give genomic insight into other millets that have very little genomic information. The comparison of whole-genome sequence data of millets is presented in Table 1 and discussed in detail below.

Table 1 Comparison of whole genome sequence data of various millets. The details such as the name of the millet, genotype, the platform used for genome sequencing, % of genome coverage, genome size, and the total number of genes are provided with the respective citation
Full size table
Fig. 1
figure 1

Milestones in the genome sequencing of millets. The year of release of millet genome sequence is depicted with different colors in hexagon boxes. The release of genome sequences is a key milestone and provides an opportunity to conduct genome-assisted studies in millets and other cereal crops

Full size image

Foxtail millet

Foxtail millet is an important model crop for improving other cereal crops due to its diploid genome, easy cultivation, short life cycle, and use of C4 photosynthesis. It is considered the best model system for genetic and genomic studies of millets and other cereals (Ceasar 2019). It belongs to the genus Setaria, under the subfamily Panicoideae. Foxtail millet shares significant genome-level synteny with other closely related grasses, including pearl millet, switchgrass, and Napier grass (Doust et al. 2009). Genome sequences are available for two different genotypes (cv. Zhang gu × A10 and cv. Zhang gu × A2) of foxtail millet (Zhang et al. 2012; Bennetzen et al. 2012) and it has relatively a small genome when compared to other millets. Zhang et al. (2012) sequenced foxtail millet using a whole-genome shotgun combined with NGS technology. DNA libraries of various sizes were constructed and sequenced using Illumina second-generation sequencers, which generated raw data of 63.5 Gb. After filtering of raw data, ⁓40 Gb clean reads were used for genome assembly by the SOAPdenovo method (Zhang et al. 2012). After gap filling, contig N50 was 25.4 kb and it contains 16,903 contigs. The scaffold N50 is 1.0 Mb and it contains 439 scaffolds. In total genome, 432 Mb scaffold (in total length) was found and it showed 6.6% gaps (28 Mb). Excluding the gaps, the scaffold covered ~ 86% of the total genome (Zhang et al. 2012). Based on the cytogenetic analysis, the genome size was estimated to be ⁓490 Mb. The repeat element annotation of the foxtail millet genome shows that 46% of the genome is composed of different transposable elements (TEs) such as retroelements and DNA transposons (Zhang et al. 2012). Bennetzen et al. (2012) also sequenced the genome of foxtail millet using Sanger sequence analysis on ABI3730xl capillary sequencing system. Foxtail millet genome sequence assembly contains 396.7 Mb of sequence and 4.2 Mb each in 327 scaffolds and covers ⁓80% genome. This genome sequence contains 40% TEs and 24,000 to 29,000 protein-encoding genes. The complete and annotated genome sequence of the foxtail millet is available in Phytozome (https://phytozome-next.jgi.doe.gov/info/Sitalica_v2_2). The genome sequence is also available for a wild relative of the foxtail millet, the green foxtail (Setaria viridis) which has a relatively small and well annotated genome (~ 510 Mb) (Brutnell et al. 2010). The two species, foxtail millet (domesticated) and green foxtail (wild variety) are excellent model systems to determine several aspects of physiological, evolutionary, and architectural aspects in the grass families (Ceasar, 2019). The annotated genome sequence of these millets provides the opportunity for comparative genome studies across all millets and closely related crops. Among millets, even the draft genome sequence is not yet available for little millet and kodo millet. Genome sequence could help to understand the genome organization of little millet and kodo millet for genomic studies and genomic-assisted improvements.

Finger millet

Whole genome assembly of two different finger millet genotypes, ML-365 (Hittalmani et al. 2017) and PR 202 (Hatakeyama et al. 2017) was reported using Illumina and SOLiD sequencing platforms. The WGS of finger millet genotype ML-365 has around 45 Gb paired-end, 21 Gb mate-pair data, and 52,5759 scaffolds (> 200 bp) with N50 length of 23.73 Kb and an average scaffold length of 2275 bp. Numerous genes for disease resistance (1766), drought-response (2866) and calcium transport and accumulation (330) were identified (Hatakeyama et al. 2017). The average DNA content (2 C) and genome size of finger millet genotype ML-365 were 3.01 pg and 1453 Mb respectively. The WGS covered approximately 82% of the total estimated genome size of finger millet. The WGS analysis showed the presence of 85,243 genes. Finger millet genotype PR-202 has the assembled genome of 1189 Mb covering only 78.2% genome (Hatakeyama et al. 2017). The genome assembly and annotation of finger millet genome pose a big challenge due to homologous genes and polyploidy nature of genome. It leads to splitting the finger millet assembly into shorter contigs. The complete and annotated genome sequence of finger millet is now available in Phytozome (https://phytozome-next.jgi.doe.gov/info/Ecoracana_v1_1). The annotated genome sequence was deposited by School of Plant Sciences, Ecology & Evolutionary Biology, Arizonz Genomics Institute in July, 2020. The final assembly (V1) of finger millet contains 1,110.3 Mb sequence, consisting of 674 contigs with a contig N50 of 15.3 Mb and a total of 97.2% of assembled bases in chromosomes. The annotated genome sequence of finger millet provides insight into the genome organization. It will enable the identification of valuable candidate genes for agro-morphologically and nutritionally important traits. The valuable candidate gene/QTL can helpful for GAB in finger millet. Genome assembly will also help to apply genome editing (GE) in the future, which may provide the opportunity for finger millet improvement.

Pearl millet

Pearl millet is a diploid (2n = 2x = 14) and cross-pollinated warm-seasonal crop. It has a short life cycle and large genome size (Varshney et al. 2017). The draft genome sequence of pearl millet was released by International Crops Research Institute for the Semi-Arid Tropics (ICRISAT), Hyderabad, India using Illumina HiSeq 2000 platform and generated 520 Gb of sequence data, it representing 296× genome coverage. It covered approximately 90% of the total estimated genome size (Varshney et al. 2017). The size of the draft genome of the pearl millet is ~ 1.79 Gb, which contains 38,579 genes (Varshney et al. 2017). A total of 27,893 (72.30%) genes’ functions were annotated. The sequenced genome shows contig N50 of 18,180 bp and scaffold N50 of 884,945 bp. A total of 1.22 Gb of repeat elements were identified indicating that 77.2% of the pearl millet assembled genome is repetitive (Varshney et al. 2017). The draft genome of the pearl millet provides a genetic blueprint of the species and it facilitates for development of a genomic tool for pearl millet improvement. Also, genomic information could be used for comparative genomic studies to validate the QTL/candidate genes.

Proso millet

Proso millet is an allotetraploid (2n = 4x = 36). The proso millet genome of a landrace (accession number: 00000390) was sequenced by Dr. Zhang Heng and Dr. Zhu Jiankang groups of the Chinese Academy of Sciences, China. The genome assembly was obtained using Illumina short-read coupled with the PacBio long-read sequencing platform (Zou et al. 2019). The assembled genome sequence of proso millet contains 5541 contigs and has a contig N50 of 369 kb. The genome size of the proso millet was estimated to be 923 Mb. Totally 55,930 protein-coding genes, 55,527 genes in pseudochromosomes, and 9643 noncoding RNAs were identified with a sequenced genome. It contains 339 microRNAs (miRNAs), 1420 tRNAs, 1640 rRNAs, and 2302 small nuclear RNAs. The genome assembly showed 58.2% of repetitive elements, of which 92.1% consist of TEs. Comparative genome studies of proso millet genome with other four closely related Poaceae Family plants such as rice, maize, sorghum, and foxtail millet revealed that 74. 5% of the gene families in proso millet genes are shared among all five Poaceae Family plants (Zou et al. 2019). Remarkably, only 4.2% (862) of gene families were specific to proso millet (Zou et al. 2019). The release of proso millet genome sequence is a milestone for comparative genome studies among the millets. Identification and validation of genes in proso millet provide the opportunity for the improvement of this millet. It also gives an insight into the genome organization of millets. It could be helpful for millet improvement through functional and comparative genomics studies in the future.

Barnyard millet

Barnyard millet or barnyardgrass is (2n = 6x = 54) an allohexaploid in nature. It has the ability for quick germination, quick growth, mimicking the character of rice, wide-ranging ecological tolerance, and huge seed production (Barrett 1983). Recently, the whole genome assembly of barnyardgrass (Echinochloa crus-galli) was done using Illumina HiSeq technology (Guo et al. 2017). A total of 207.4 Gb of genome data was generated from the Illumina HiSeq platform. The draft genome size of barnyardgrass was estimated to be 1.27 Gb, representing 90.7% of the genome coverage with a scaffold N50 length of 1.8 Mb. Sequenced data of barnyardgrass is predicted to have 108,771 protein-coding genes. A total of 758 miRNAs, 2306 transfer RNAs 1890 ribosomal RNAs, and other non-coding RNAs were also identified. The genome sequence of barnyardgrass is not yet annotated. Like a weed, the barnyardgrass can tolerate various biotic and abiotic stresses. The genome sequence of barnyardgrass could provide insight into the tolerance mechanisms related to biotic and biotic stresses. It may help to understand the genes involved in climate resilience and their beneficial effects on crop improvement programs. The release of the barnyardgrass genome sequence is a milestone in this millet improvement. The genome annotation in the barnyardgrass is still in its early stage. This is due to the complex nature (allohexaploid) of the genome. Previously, advanced sequencing technologies solved several genome sequencing issues and facilitated to identify the genome sequence of wild and transcriptome sequences in cultivated Echinochloa species (Yang et al. 2013; Li et al. 2013). We hope that the high-quality annotated genome sequence of barnyardgrass will be out soon. With the availability of the annotated genome sequence, it will be possible to improve genomic selection and genome prediction approaches. This will speed up the breeding process to improve this millet.

Millets research in the post-genomic era

Farmers and scientists have been facing unpredictable global climatic changes with multiple stresses which adversely affect agriculture productivity. In the coming decades, global climatic change will significantly affect crop productivity and it may directly influence food demand in the future. Millet production is threatened by both biotic and abiotic stresses associated with climatic change (Ceasar et al. 2018). Millets have many morphological, physiological, biochemical, and molecular characteristics which confer better adaptation (tolerance mechanism) to various environmental stresses (Bandyopadhyay et al. 2017). In millets, many genes that contribute to biotic and abiotic stress tolerance have been identified. For example, a total of 586 genes have been identified from the WGS of foxtail millet related to drought stress, which could play a key role in drought tolerance (Zhang et al. 2012). These genes may contribute to improving adaptation to arid and semi-arid regions. The short life cycle of millets assists in escaping from any stresses as they require only 12–14 weeks to complete their life cycle (Bandyopadhyay et al. 2017). Understanding the adaptive mechanisms and gene mining of millets could provide a potential resource to explore climate-resilience properties. More genomic information on millets could be helpful for understanding the features of climate-resilience and further for the improvement of millets and other major cereals. Efficient crop improvement may help to improve crop traits and adapt to stresses. Recent achievements in molecular biology and biotechnological areas, especially rapid development of high throughput genomics and phenomics (computer-based analyzing) tools, functional genomics and proteomics, molecular breeding, and genome-editing (GE) in agriculture have brought about an extraordinary revolution for crop improvement era. Genome-assisted breeding has been considered a “third-generation” approach for crop improvement and increasing crop productivity (Krishna et al. 2021). Various strategies and approaches were used to improve millet traits with the help of genomic resources. In the following section we discuss the strategies and approaches used to improve millet growth and development. This will be very useful for researchers to understand the scope and implications of WGS in millet.

Identification of QTL and marker-assisted breeding for millet improvement

Genome-based or marker-assisted selection (MAS) approaches have reduced the time and effort involved in the selection of breeding material for crop improvement. In MAS, the QTL mapping helps to understand the genetic and physiological basis of stress tolerance (biotic and abiotic) in millets and provides an opportunity to improve stress tolerance in millets. The identification of valuable QTL and candidate genes related to biotic and abiotic stress tolerance is the first step in MAB, which will help in the development of improved millet varieties. The effective utilization of genetic resources could help crop improvement and it is very crucial to increase the genetic gain to address challenges in both biotic and abiotic stresses. Molecular markers, genetic maps, and genome sequence information are some of the essential genomic resources required for efficient crop improvement in millets (Saha et al. 2016; Krishna et al. 2020b). The molecular marker is essential for detecting the specific genomic regions associated with tolerance in crops under biotic and abiotic stresses. The accuracy of MAS will depend upon high-throughput phenotyping technologies for discovering valuable genetic information from complex traits. In phenomics, characterizing the available germplasm collection can serve a dual purpose in MAB, first predicting a better population and the second mining the desirable genetic traits (governed by the QTL) to incorporate into elite cultivars. Many promising new technologies are available in phenomic and genomics studies. Availability of automation, accurate imaging system, efficient software, and sequence technology has helped in high-throughput data collections for millet improvement (Krishna et al. 2021). The high-throughput phenotypic and genotypic data will be helpful for MAS for an efficient breeding program in millet (Fig. 2). The availability of WGS for millets will enhance the efficiency and accuracy of MAS.

Fig. 2
figure 2

Genomic-assisted breeding of millets for crop improvement. The high-throughput phenotypic and genotypic data will be helpful for GS/MAS for an efficient breeding program in millet

Full size image

In millets, the majority of the agronomically important traits conferred tolerance to biotic and abiotic stresses such as disease, insect-pest, drought, salinity, and nutrient starvation tolerance are governed by the QTL (Ramakrishnan et al. 2017; Wang et al. 2017; Ramakrishnan et al. 2016). Millets improvement for such traits could be used for MAB once the QTL are identified and mapped. But only a little effort was taken for the identification of valuable QTL (linked to resistance genes) in millets. In most of the studies, the QTL was identified by low-throughput “first generation” molecular markers. For example, Jones et al. (2002) identified the QTL for resistance to downy mildew in pearl millet using restriction fragment length polymorphisms (RFLPs) markers. In pearl millet, the sequence characterized amplified region (SCAR) marker was developed by Jogaiah et al. (2014) and it was used for detecting the QTL (SCARISSR 863) associated with downy mildew disease resistance in recombinant inbred lines (RILs). In the same study, they further analyzed the identified QTL (SCARISSR 863) in 12 diverse pearl millet lines (seven disease resistance and five susceptible lines). The SCARISSR 863 was amplified in seven disease-resistant lines and absent in five susceptible lines (Jogaiah et al. 2014). Babu et al. (2014b) identified the QTL for finger blast (RM262, FMBLEST32, UGEP81, UGEP53, and UGEP24) leaf blast (FMBLEST35, FMBLEST15 and RM23842) and neck blast (UGEP18) through association mapping using 104 simple sequence repeats (SSRs) marker in 190 finger millet genotypes. Ramakrishnan et al. (2017) found four QTL in finger millet viz., qLRDW.1, qLRDW.2, qHSDW.1 and qHRL.1, which are associated with low phosphorus tolerance, and they are linked to the SSR markers UGEP19, UGEP68, UGEP13, and UGEP90, respectively. Similarly, two leaf blast resistance QTL, UGEP95 and UGEP101 were identified using 87 SSR markers in 128 genotypes of finger millet (Ramakrishnan et al. 2016). Kannan et al. (2014) identified several QTL associated with agro-morphological and yield traits of pearl millet through association mapping. The association studies identified the QTL associated with flowering time (Xicmp3027_202 Xpsmp2076_160, Xpsmp2077_136, Xpsmp2088_136, Xpsmp2227_194 Xpsmp2248_162 and Xpsmp2248_166), plant height (Xicmp3027_202, Xpsmp2069_214, Xpsmp2077_136, Xpsmp2085_175, Xpsmp2088_136, Xpsmp2224_157, Xpsmp2224_159, Xpsmp2233_256, Xpsmp2233_260, Xpsmp2233_262, Xpsmp2248_162, and Xpsmp2248_166), panicle length (Xpsmp2077_136, Xpsmp2224_157, and Xpsmp2233_260), panicle diameter (Xicmp3002_204 and Xpsmp2248_164), grain yield (Xpsmp2224_159), stover dry matter yield (Xpsmp2220_115, Xpsmp2224_157, Xpsmp2233_262, Xpsmp2248_162, and Xpsmp2248_166), biomass yield (Xpsmp2220_115 and Xpsmp2224_159), panicle threshing percentage (Xpsmp2085_175 and Xpsmp2227_196) and harvest index (Xicmp3027_200, Xicmp3027_202, Xpsmp2086_116, Xpsmp2248_162, and Xpsmp2248_166). Many yield related traits have been identified in foxtail millet (Fang et al. 2016; Liu et al. 2020; Zhi et al. 2021), finger millet (Ramakrishnan et al. 2016; Kumar et al. 2016a) and pearl millet (Yadav et al. 2003; Nepolean et al. 2006; Sehgal et al. 2012; Tharanya et al. 2018). Many QTL have been found to be associated with various agro-morphological traits. Babu et al. (2014a) identified four SSR markers associated with various agro-morphological traits such as basal tiller number (UGEP81), days of 50% flowering (UGEP77 and UGEP90), flag leaf blade width (FM9), and plant height (FM9) in finger millet. So, agro-morphological traits may be involved in biotic and abiotic stress tolerance (Pandey et al. 2017). The information on markers associated with QTL for tolerance to biotic and biotic stress could be useful for the development of new millet varieties.

Before the emergence of genome sequencing approaches, the important QTLs/genes were identified in millet using low-throughput “first generation” molecular markers. It slows down the efficiency of crop improvement through MAS. NGS techniques provide a new opportunity for high-throughput genotyping by sequencing (GBS), which could help to detect the genetic basis of phenotypic variation in crops (Maharajan et al. 2021). The GBS helps to increase the efficiency and accuracy of the GAB. So, researchers need to pay more attention to identifying QTL associated with various traits under both biotic and abiotic stresses using GBS. The genome sequence of millets helps for the validation of identified QTL for further GAB. ICRISAT, India, in collaboration with Chaudhary Charan Singh Haryana Agricultural University (CCSHAU), India and ICAR-all India Coordinated Research Project on pearl millet (AICRP-PM) developed a new pearl millet variety, HHB 67 using SSR markers, which is tolerant to downy mildew disease (Hash et al. 2006). It was the first product of MAB to be delivered to Indian farmers. The hybrid has helped to increase the grain and stover yields by conferring a high level of resistance to downy mildew disease. This study demonstrates that MAB could greatly contribute to the development of new variety with improved yield and resistance to disease. Such studies could be further augmented by the NGS which will help to find SNP markers and contribute to developing new varieties to increase the growth and yield of small millets under biotic and abiotic stresses.

Genome-wide association studies (GWAS) and genome-assisted breeding for millet improvement

Genome-wide assessment of the genetic variation of millet germplasm may improve the efficiency of GAB. Genome sequencing plays a vital role in GWAS for millet improvement through GBA. In GWAS, many valuable QTLs were identified for biotic and abiotic stress tolerance in many crops. High-throughput GBS data of millets might facilitate functional and comparative genome studies in millet once the QTL are identified. It could be helpful for the identification of candidate genes related to biotic and abiotic stress tolerance.

Several research groups have successfully employed GWAS in many cereal crops to identify salinity, drought, and nutrient stress-responsive QTL/genes and their alleles. Notably, little effort was paid on GWAS in millet compared to other major cereals (Table 2). Most of the reported GWASs are related to nutritional and agro-morphological trait aspects in millets. Tiwari et al. (2020) discovered QTL and genes governing seed protein, day to maturity, and grain yield in finger millet using 2977 SNP markers via GBS technology. In-silico analysis revealed that the identified SNP marker was linked to the candidate gene responsible for seed protein content in finger millet. The aspartyl protease has been found to be the most valuable candidate gene for seed protein content in finger millet (Tiwari et al. 2020). Similarly, GBS technology was used to discover genomic regions governing the grain nutritional traits in foxtail millet. The identified SNP markers linked to various candidate genes which are involved in several metabolic processes and encode protein including succinate dehydrogenase assembly factor 1, MFS transporter, F-box domain, zinc finger FYVE domain-containing protein leucine-rich repeat, X-box transcription factor-related, raffinose synthase, protein-tyrosine kinase, anthocyanidin reductase, ionotropic glutamate receptor, NADH dehydrogenase flavoprotein, PPR repeat family, etc. (Jaiswal et al. 2019a). Sharma et al. (2018) evaluated agro-morphological traits of 113 accessions by GBS and identified valuable QTL for day to 50% flowering, day to maturity, ear length, flag leaf blade width and grain yield. In this study, QTL sequence showed homology to candidate genes of rice, foxtail millet which might play an important role in plant growth (ATP synthase subunit beta, mitochondrial and photosystem I iron-sulfar centre), flowering (photosystem II system protein (PS-II)-cytocrome b6/f complex) and maturity and grain yield (ribosomal protein and GRF zinc finger) (Sharma et al. 2018). Similarly, Jia et al. (2013) analyzed 916 diverse foxtail millet accessions and identified 2.58 million SNP by GBS, and constructed a haplotype map using 0.8 million common SNPs. This study identified 512 loci associated with 47 agronomic traits (Jia et al. 2013). Such studies are needed to move on to the identification of stress tolerance traits/QTL/genes in millets. In GWAS, the annotated genome sequence of millet is very helpful for identification and validation of the candidate genes. The GBS in GWAS could be helpful for the mining of novel or valuable genetic information in millets, which is essential for millet improvement through GAB. Li et al. (2021) analyzed the blast resistance in 888 foxtail millet accession through GWAS and detected 154 genomic regions responsible for blast resistance in foxtail millet. So far, little information is available on GWAS related to biotic and abiotic stress tolerance millets. Further research is needed to identify SNP and candidate genes related to biotic and abiotic stress tolerance by GWAS, which may help to improve the growth and yield of small millets under biotic and abiotic stress.

Table 2 Details of genome-wide association studies (GWAS) conducted in millets using genotyping by sequencing (GBS) method. The details such as the name of the millet, number of genotypes used for GWAS, type of mapping population, name of trait, findings, number of identified candidate genes are provided with reference
Full size table

Genomic selection (GS) for millet improvement

GS is a promising strategy with great potential to explore and increase genetic gain per selection in a breeding scheme per unit timeline (Spindel et al. 2015). It speeds up breeding work and improves the efficiency of breeding programs. In GS, genome-wide distributed DNA markers are used to predict the genomic estimated breeding values (GEBV) for the breeding materials (Varshney et al. 2013). The GS information lacks in millets except for pearl millet (Srivastava et al. 2020). The GBS approaches have been used to predict the GEBV of inbred lines of pearl millet (Liang et al. 2018). This study shows that GBS can enable GS in pearl millet. Varshney et al. (2017) applied whole-genome resequencing (WGRS) data for GS by ridge regression best linear unbiased prediction (RR-BLUP) to predict the grain yield of pearl millet. GS is an alternative approach to MAS and phenotypic selection in crop improvement programs. It reduces the cost and time required for the development of new crop varieties. There are many breeding populations available in the millet germplasm. The GS is a promising approach for the selection of breeding materials. It may help to improve the gene pool in the millet germplasm in the future. Plant breeders need to carry out the millet germplasm characterization via GEBV. GS is considered to have a greater advantage over pedigree breeding and MAS due to its ability to improve complex traits such as yield in short term by reducing the number of breeding cycles (Bhat et al. 2016). Recently, the rapid crop production system which is referred to as speed breeding has also been applied to many crop plants (Wanga et al. 2021). This reduces the time required for the growth of the crop through short breeding cycles. Integration of these promising breeding strategies can accelerate the genetic gains needed for the rapid improvement of complex traits in millets, which could improve the crop productivity. Identification of QTL/genes associated with biofortification traits of millets may help to increase micronutrient concentration in the seeds, which may help to reduce global micronutrient deficiency (Krishna et al. 2022). Genomic approaches such as MAS, QTL and genomic selection are seen as cost-effective approaches for selecting desired plants and are widely used for the biofortification of cereals. A few QTL associated with biofortification traits were identified in millets by MAB. Two QTL each for Zn and Fe contents were identified in 106 RILs of pearl millet derived from ICMB 841 and ICMB 863 through linkage mapping by 208 DArTs, 95 SSRs and 2 STSs markers (Kumar et al. 2016b). In another study, 210 RILs derived from PPMI 683 and PPMI 627 were used to identify 22 QTL (14 for grain Fe and 8 for grain Zn) by linkage mapping using 151 SSR markers (Singhal et al. 2021). No study has yet been conducted to identify biofortification traits in small millets using GS. Heffner et al. (2011a) compared the accuracy of phenotypic and marker based prediction of genetic value for various grain quality traits within two biparental wheat populations. They reported that the prediction accuracy for all traits was higher in GS compared to MAS. Many researchers have reported that the prediction accuracy for most agronomic traits in rice and wheat was higher in GS compared to phenotypic selection and MAS (Heffner et al. 2011a; b). Selection of genotypes for further developments of RILs through genomic selection will help improve the biofortification traits of millets and alleviate micronutrient deficiency.

Functional genomics and validation of genes for millet improvement

The advent of sequencing technologies provided the opportunity for transcriptome analysis. It is the primary technique used to identify candidate genes concerned with a biological process. For example, transcriptome analysis has identified 82 unique calcium sensor genes in finger millet spikes (Mirza et al. 2014). Overexpression of these calcium sensor genes in finger millet may improve the calcium content in the seed. Many drought-stress responsive genes [serine threonine protein phosphatase 2 A (PP2A), calcineurin B-like interacting protein kinase31 (CIPK31), farnesyl pyrophosphate synthase (FPS), signal recognition particle receptor α (SRPR α)] have been identified under drought stress conditions via transcriptome analysis in finger millet (Parvathi et al. 2019). Transcriptome analysis revealed that 6920 and 6484 genes are expressed under heat stress and drought stress, respectively, in pearl millet (Sun et al. 2020). Shivhare et al. (2020) identified genes and pathways involved in drought tolerance in pearl millet through comparative transcriptome analysis. So far, many transcriptome studies have been reported in millets (Puranik et al. 2011; Sun et al. 2020; Lata et al. 2010; Rahman et al. 2014; Hittalmani et al. 2017). These transcriptome data are related to various abiotic and biotic stress conditions, which may help to identify the key genes responsible for the particular stress tolerance in millets. The valuable key genes will help millet improvement through various approaches such as GE.

Identification and functional validation of key genes associated with biotic and abiotic stress tolerance may help to improve the growth and development of millets. The available genome sequence of millets will provide the opportunity to mine the important genes for stress tolerance in millets. Many genes have been identified in various tissues of millets under biotic and abiotic stresses. Only a little effort is made for the functional validation of genes in millets. Abiotic stress-responsive genes are identified and some genes are functionally characterized in millets. For example, dehydrin7 is one of the drought stress-tolerance genes in plants. In finger millet, the drought-responsive gene Ecdehydrin7 was identified and overexpressed in tobacco plants (Singh et al. 2015). Overexpression of EcDehydrin7 in tobacco plants played a significant role in drought stress tolerance (Singh et al. 2015). Similarly, many drought-responsive genes such as Metallothionein (MT), RISBZ4, Farnesylated Protein (ATFP6), Transcriptional Regulator (TR), Protein Phosphatase 2 A (PP2A), Early Light Inducible Protein (ELIP), Farnesyl Pyrophosphate Synthase (FPS) (Parvathi et al. 2013) and TATA-box Binding Protein Associated Factor 6 (TAF6) (Parvathi and Nataraja 2017) were identified in finger millet under different levels of drought stress. These genes were involved in the drought-stress tolerance mechanism under drought conditions in millets. The late embryogenesis abundant (LEA) family gene also contributes to abiotic stress tolerance in foxtail millet and pearl millet (Kummari et al. 2021). The overexpression of SiLEA in transgenic Arabidopsis seedlings showed tolerance to salinity and osmotic stress than the wild type plants (Wang et al. 2014a). Overexpression of PgLEA of pearl millet in transgenic tobacco plants also showed high tolerance to salinity, drought, heat, and cold stresses (Kummari et al. 2021). The LEA is considered as multiple stress-responsive gene in millets. Identification of LEA family gene in other millet also helpful for millet improvement. The transcription factor G-BOX BINDING FACTOR 3 (GBF3) was also shown to be tolerant to multiple stresses such as osmotic, drought, and salinity in finger millet (Ramegowda et al. 2017). Similarly, many abiotic stress responsive genes are identified in millet (Table 3). The functional validation of abiotic stress tolerant genes is very important for further research. Researchers need to pay more attention to the functional characterization of the abiotic stress tolerant genes in millet. Manipulation of the genes related to biotic and abiotic stress tolerance may help to improve crop quality. Transgenic and GE approaches might be helpful for millet improvement by targeting suitable genes.

Table 3 Details of abiotic stress tolerance genes identified in various millets. Details on the name of the millets, type of stress, name of the stress-responsive gene, and their functions are provided with reference
Full size table

Nutrient starvation is a major problem affecting agriculture production. The macro- and micro-nutrient transporter families such as zinc-regulated, iron-regulated transporter-like protein (ZIP), nitrate transporters (NRT), potassium transporters, boron transporter, phosphate transporters (PHT), cation diffusion facilitator (CDF), plasma membrane-type ATPase (P-type ATPase), and cation exchanger (CAX) have been identified and characterized in plants under various nutrient deficient conditions (Ceasar et al. 2014; Onuh and Miwa 2021; Deng et al. 2013; Nadeem et al. 2018; Krishna et al. 2020a). These nutrient transporters are responsible for nutrient uptake, transport, and homeostasis in plants (Krishna et al. 2017). Genes responsible for some nutrient transporters were also identified in millets (Table 3). The identification and characterization of novel genes associated with nutrient transport/nutrient use-efficiency are very important for millet improvement under low-nutrient soil. Ceasar et al. (2014) identified 12 PHT1 family genes in foxtail millet and analyzed the expression pattern of SiPHT1;1 to SiPHT1;12 transporters in leaf, root, and shoot samples under low and high phosphorus conditions and to colonization with the arbuscular mycorrhizal fungus (Funneliformis mosseae). In the same group, the foxtail millet SiPHT1 transporters were functionally characterized through yeast complementation assay and RNA interference technology (Ceasar et al. 2017). The down regulation of SiPHT1;2, SiPHT1;3 and SiPHT1;4 transporter genes in the transgenic foxtail millet revealed a significant reduction of phosphate content (total and inorganic) in root and shoot tissue (Ceasar et al. 2017). This study showed that the PHT1 transporters play important role in phosphate uptake under phosphorus starvation in foxtail millet. Maharajan et al. (2019) analyzed the expression pattern of PHT1 family genes in all millets under low and high phosphorus conditions. This study reveals the different expression patterns of SiPHT1 family transporter genes among millets. Roch et al. (2020) evaluate the expression pattern of SiPHT1;1–1;12 family transporter genes in 20 foxtail millet genotypes and revealed that the expression patterns of 12 PHT1 transporter genes varies from genotype to genotype under both low and high phosphorus conditions. The functional characterization of PHT1 transporters is essential for understanding their role. A huge number of HAK family transporter member genes (SiHAK1-SiHAK29) were identified in the foxtail millet (Zhang et al. 2018). Only a few SiHAK genes were used for expression analysis and functional characterization so far (Zhang et al. 2018). The SiHAK1 gene was found to be highly expressed in root and shoot tissue and predominantly up-regulated in shoot tissue under low-potassium conditions (Zhang et al. 2018). The SiHAK2 gene showed less sensitivity to low-potassium conditions (Zhang et al. 2018). Further functional characterization revealed that expression of SiHAK1 restores the growth of potassium uptake-deficient strain of yeast under low-potassium conditions (Zhang et al. 2018). The growth of mutant yeast indicates that the SiHAK1 is responsible for potassium uptake in foxtail millet under low-potassium conditions (Zhang et al. 2018). In plants, calcium transporters are classified into three types such as Ca2+ channel, Ca2+ ATPase, and, Ca2+ CAX, all of which are actively involved in calcium-sensing, binding, transport, and grain filling. Among the millets, some calcium transporters have been analyzed in various tissues of finger millet (reviewed in Maharajan et al. 2021). These calcium transporter genes were identified in finger millet before the release of whole genome sequence. Also, no transporters have yet been functionally characterized. Further genome-wide identification (from the annotated genome sequence of finger millet) and functional characterization of calcium transporter may help to understand the actual role of calcium transporter in finger millet. Many other nutrient transporters are identified and their expression patterns are analyzed in millets. But most of these genes were not functionally characterized in yeast or in planta, and all these studies are restricted to gene expression analysis. Further functional characterization of nutrient transporters in millets will help to understand the role and functions of nutrient transporters under low-nutrient conditions. It could help improve nutrient use-efficiency in millet through biotechnological approaches.

Genome-editing for millet improvement

The availability of WGS in millet provides the scope for GE approaches and the possibility to insert desirable traits (Ceasar, 2021). The GE technology is a recent tool used for crop improvement. The GE approach helps to enhance the growth and yield of crops under both biotic and abiotic stresses. GE tools like meganucleases, zinc finger nucleases (ZFNs), transcriptional activator-like effector nuclease (TALENs), and recently invented, clustered, regularly interspaced short palindromic repeat (CRISPR)/CRISPR-associated protein 9 (Cas9) precisely target any gene of interest (Jaganathan et al. 2018). CRISPR/Cas9 system requires only a piece of RNA (gRNA) to target any gene of interest. Scientists have adapted the CRISPR/Cas9 system as an efficient GE system in targeting the gene of several organisms. The CRISPR/Cas9 system has been applied in many plants including rice (Zhou et al. 2015; Liu et al. 2012), maize (Shi et al. 2017) and wheat (Wang et al. 2014b; Shan et al. 2014). Application of the CRISPR/Cas9 GE system in cereal and other crops was reviewed by many researchers (Hillary and Ceasar 2019; Jaganathan et al. 2018). In rice, the CRISPR/Cas9 targeted mutation in ethylene responsive factor 922 (OsERF922) gene increases the rice blast disease caused by Magnaporthe oryzae (Liu et al. 2012). The Knockout of OsSWEET13 susceptible gene (bacterial blight in rice by infection of Xanthomonas oryzae pv. Oryzae) using CRISPR/Cas9 system showed improved resistance to bacterial blight disease in rice (Zhou et al. 2015). CRISPR/Cas9 system was also successfully demonstrated by the knockout of downey mildew-resistant locus gene (TaMLO) in wheat (Shan et al. 2014). Many abiotic stress tolerance and susceptible genes have been identified in crops. The knockout or knockin of abiotic stress responsive genes through the CRISPR/Cas9 system could be helpful for abiotic stress management and crop improvement in millets. CRISPR/Cas9 system could be helpful for millet improvement. Application CRISPR/Cas system mediated genome editing in millets is presented in Fig. 3.

Fig. 3
figure 3

Application CRISPR/Cas system mediated genome editing in millets. Prudent use of CRISPR based system could help to understand the mechanism of climate resilience and nutrition fortification in millets by knock-in or knock-out studies. Millets could be improved by CRISPR/Cas system to overcome drought, salinity and fungal stresses. CRISPR/Cas system also helps to alter the expression of genes and tools like base editing may be applied to improve the nutrient transport and enzyme activity in millets

Full size image

Only a little effort has been made for GE in millet using the CRISPR/Cas9 system. In millets, several genes have been identified that respond to abiotic (drought, salinity, heat, cold, oxidative, and nutrient starvation) stress tolerance (Ceasar et al. 2014; Parvathi et al. 2013; Jadhav et al. 2018; Nagarjuna et al. 2016; Gupta et al. 2013; Cao et al. 2019). Researchers need to dissect the role of biotic and abiotic stress-responsive genes identified in millets through the GE system, which may help to develop desirable traits. Function of some abiotic stress responsive genes has already been characterized by CRISPR/Cas9 system. Knockout of stress activated protein kinase (SAPK) gene in rice through CRISPR/Cas9 system reduced water loss, promote stomatal closure and upregulate the expression levels of abiotic stress responsive genes (OsRab16, OsRab21, OsbZIP23, OsLEA3, OsOREB1) and slow anion channel (SLAC)-associated genes (OsSLAC1 and 7) under drought stress (Lou et al. 2017). These findings indicate that SAPK2 might be a potential candidate gene for future crop development. Knockout of drought and salt tolerance (DST) gene in rice through CRISPR/Cas9 enhanced flag leaf growth, leaf water retention and reduced stomatal density under drought stress (Kumar et al. 2020). The same CRISPR/Cas9 system was used to knockout enhanced response to abscisic acid (ERA), abscisic acid receptor of pyrabactin resistance like (PYL), no apical meristem (NAC) genes in rice, which enhanced the growth and yield under drought stress (Usman et al. 2020; Ogata et al. 2020; Wang et al. 2020). As like rice, knockin of auxin regulated gene involved in organ size (ARGOS8) gene in maize increased the growth and yield under drought stress (Shi et al. 2017). Some salinity stress responsive genes such as response regulator (RR) genes (OsRR9, 10 and 22) and overly tolerant to salt1 (OsOTS1) were also knocked out in rice through CRISPR/Cas9system and the resultant mutants have enhanced shoot and root biomass under salt stress (Wang et al. 2019; Zhang et al. 2019). The growth and yield of small millets are severely affected by various abiotic stresses such as drought, salinity and low nutrient stress (Ceasar et al. 2014; Maharajan et al. 2019; David et al. 2021; Roch et al. 2020; Parvathi et al. 2013). As with rice and maize, initiation of both knockout and knockin experiments in small millets through CRISPR/Cas9 system with already identified abiotic stress responsive genes may help to identify the function of each gene. However, the role of very few genes was identified through CRISPR/Cas9 system in foxtail millet and green foxtail. Lin et al. (2018) targeted the phytoene desaturase (PDS) gene of foxtail millet using the CRISPR/Cas9 system. They observed the albino phenotype of the foxtail millet and reported that CRISPR/Cas9 system is applicable for mutating the foxtail millet (Lin et al. 2018). In another study, the Less Shattering1 (Les1) gene of green foxtail was targeted using the CRISPR/Cas9 system, which resulted in the reduced shattering of green foxtail (Mamidi et al. 2020). Leaf rolling is one of the major traits which reduced crops yields in many plants including small millets under abiotic stress, particularly under drought. Leaf rolled mutant rice was developed by insertion of semi-rolled leaf 1 (SRL1) and SRL2 genes through CRISPR/Cas9 system (Liao et al. 2019). The developed mutant rice enhanced survival rate, abscisic acid content, activities of superoxide dismutase (SOD) and catalase (CAT) and grain filling percentage under drought stress compared to the wild type (Liao et al. 2019). This report is an important example for controlling leaf rolling in cereals including small millets by genome editing. The genome sequencing of millets provides the opportunity to carry out high-throughput research in future to understand and improve the millets. In millets, the CRISPR/Cas9 system might be helpful in targeting the gene associated with desirable traits to develop stress tolerance variety (Fig. 3). Various CRISPR/Cas mediated genome editing tools (CRISPR interference (CRISPRi), CRISPR activation (CRISPRa), base editors, epigenetic engineering, chromatin imaging and prime editors) have been discovered and are all used for crop improvement. Identifying the role of each biotic and abiotic stress responsive gene through CRISPR/Cas mediated tool may help to improve the growth and yield of small millets under both biotic and abiotic stresses.

Conclusion and future prospects

Millets are important food and nutritional crops and are considered as the crops of future food security. Millet production is threatened by various biotic and abiotic stresses. Crop improvement has been one of the major priority areas of research in millets. The availability of WGS provides the opportunity to improve the millet through various biotechnological approaches. The WGS is helpful to mine and validate genes through functional and comparative genomics studies. First-generation molecular marker technology followed by high throughput GBS technology provided opportunities to carry out efficient crop improvement through GAB. In the future, GE systems like base editing and prime editing also would offer several possibilities to improve the millets to overcome the adverse effects of biotic and abiotic stresses. It may help to improve millet production and strengthen food security in developing countries.