Genetic insights in pearl millet breeding in the genomic era: challenges and prospects.

Pearl millet, a vital staple food and an important cereal, is emerging as crop having various end-uses as feed, food as well as fodder. Advancement in high-throughput sequencing technology has boosted up pearl millet genomic research in past few years. The available draft genome of pearl millet providing an insight into the advancement of several breeding lines. Comparative and functional genomics have untangled several loci and genes regulating adaptive and agronomic traits in pearl millet. Additionally, the knowledge achieved has far away from being applicable in real breeding practices. We believe that the best path ahead is to adopt genome-based approaches for tailored designing of pearl millet as multi-functional crop with outstanding agronomic traits for various end uses. Presently review highlight several novel concepts and techniques in crop breeding, and summarize the recent advances in pearl millet genomic research, peculiarly genome-wide association dissections of several novel alleles and genes for agronomically important traits. © Korean Society for Plant Biotechnology 2022.

Introduction

The rising global population, integrated with climate changes striking natural disasters more frequently impose remarkable challenges and pressure on global food security. To meet the food demands, we need to produce 70% more food as the world population is exceed to 9.7 billion by 2050 (Tripathi et al. 2019). Due to the COVID-19 pandemic, additionally, 83–132 million people in the world will become undernourished, and that further exacerbates insecurity of global food production and poses a hindrance in achieving the goal of Zero Hunger, indicated by a recent FAO report (FAO 2020). To avoid global food insecurity, it is necessary to manage important crop ideotypes to enhance crop production.

Pearl millet (Pennisetum glaucum), is an important cereal crop in the arid and semiarid ecologies of Sub-Saharan Africa and South Asia where the temperature is high and also challenged by erratic rainfall. It has the capacity to withstand extreme conditions and also built-in adaptation to low fertility soils, and is an excellent crop for shorter growing season because of its short development stages, high growth rate, and high photosynthetic efficiency (Yadav and Rai 2013; Serba et al. 2020). Pearl millet is emerging as an important crop for feed, relay crop, fodder, and food in Canada, North Africa, United States, Mexico, Central Asia, and Brazil. Pearl millet has significant importance because of its nutrient-rich grains for human consumption, and dry Stover as well as green fodder for livestock (Rai et al. 2008).

Pearl millet breeding has made enormous progress and is often referred to as one of the considerable success stories in agriculture but the average grain yield of pearl millet is still very low as compared to other cereals (Yadav et al. 2019). Globally, the average yield of millet in 2018/2019 was only ~ 0.94 MT/ha, which is much lower than that of wheat (~ 3.39 MT/ha), maize (~ 5.86 MT/ha), and rice (~ 4.58 MT/ha) (FAOSTAT 2020; Tang and Cheng 2018). Yield levels of pearl millet are greatly influenced by genotype and environment and its yield varies significantly among different countries. In this review, we first introduce molecular breeding approaches and systems. We also review the current advances in pearl millet genomic and genetic research. Main emphasis is on important genes and allelic variations focusing agronomic traits with immense potential in pearl millet breeding. Moreover, novel cutting-edges breeding using modern breeding techniques are discussed along with future development directions of pearl millet breeding.

Cutting edge molecular breeding approaches and systems

Breeding and genetic improvement of crops aim at enhancing genetic gain, defined as the enhancement in performance achieved via artificial selection over generations. In the era of molecular breeding, conventional breeding approaches including phenotypic selection, have transformed into molecular approaches to enhance genetic gain (Moose and Mumm 2008; Xu et al. 2017). Crop breeding and genetic improvement have been elaborated by exploiting and integrating the concepts of DNA markers, genetic engineering, and marker-assisted breeding (MAB) as anticipated by more accurate background and foreground selection, shortening of breeding cycle (MARS), and utilization and discovery of diverse genetic resources (Bruce 2012). We know that agronomically important traits are governed by multiple genes (polygenic), which restrict breeding on a molecular basis. So, the success of molecular plant breeding of traits having genes with major effect is limited due to: unavoidable linkage drag linked with larger chromosomal regions selection, the underlying hindrance of lower genome coverage of molecular markers, and difficulties in utilizing a wider range of crop genetic resources. Hence, advancements in theoretical configuration and technologies are in need to boost the breeding system (Gupta et al. 2010). Since time immemorial, the emergence of novel molecular approaches such as next-generation sequencing techniques, genome editing, molecular modules, genome-wide association studies, genomic selection, marker-assisted breeding, and high throughput phenomics, has reformed the scope of crop breeding, authorize much more productive utilization of artificial and naturally created variation, and phenotyping (Zhou et al. 2018; Chen et al. 2019; Xu et al. 2020). Marker assisted selection (MAS) stand out to be an intensive approach providing the breeders an efficient tool to use his skill. In pearl millet, it has been utilized for development of lines tolerant to drought stress, thus improving overall genetic gain (Rani et al. 2021). Here we review the concepts of genomic selection and molecular modules and up-to-date reviews can be found for marker-assisted recurrent selection (Singh et al. 2021a).

Genomic selection (GS)

Genomic selection (GS) has been authenticated with immense potential to increase genetic gain in breeding of crop plants but also used extensively in animal breeding (Weller et al. 2017; Mehrban et al. 2017). GS breeding programs require adequate and affordable genotyping platforms, availability of pedigree records, and a less structured population. One of the major aims of using GS is enabling genomic selection per se with improved predictive accuracy. Prediction accuracy is highly affected by population size, marker density, population structure, genetic models, TP-BP (Training Population-Breeding Population) relationship, and heritability. It can be expressed using the formula: where y1 is related to marker density. y2 is population size, y3 is TP-BP relationship, y4 is heritability, and y5 is a genetic model and a to e are the constants related with variables y1 to y5. Marker density is one of the important factors in prediction of accuracy. The number of markers should be high to ensure that the maximum number of QTLs associated with trait should be in strong LD with at least one marker (Daetwyler et al. 2010) whereas low-density markers may be helpful in predicting genomic estimated breeding value (GEBVs) with less accuracy (Singh and Singh 2015; Xu et al. 2020). Unlike marker-assisted selection, genomic selection uses information from whole genome-wide marker data whether they are related with a marker of interest or not, showing advantage of genomic selection over marker-assisted selection in reducing cost of genotyping and phenotyping. It is noteworthy that GS does not need any data related to phenotyping of the breeding population and no need to identify the QTLs associated with target trait (Xu et al. 2020). At whole genome level, major and minor genes effect can be estimated by exploiting the genotype-to-phenotype relationship thus helping in reducing time cycle and cost thereby enhancing genetic gain of every cycle (Singh and Singh 2015; Guo et al. 2019). In recent years, genomic selection has been utilized in breeding of major crops such as maize, rice, and wheat (Crossa et al. 2014; Schrag et al. 2019; Spindel et al. 2015). In a study, Liang et al. (2018), characterize the inbred pearl millet lines, developed by ICRISAT, and evaluated the utility of genomic selection using two genotypic strategies viz., tGBS and RAD-seq. It has been reported that genomic prediction scheme (RR-BLUP) generated median predictions that ranges for different traits viz., grain yield (0.48–0.51), 1000 grain weight (0.73–0.74), plant height (0.72–0.73), and days to flowering (0.87–0.89), when used hybrid data but can also be improved moderately by incorporating inbred phenotypic data sets (Liang et al. 2018). In pearl millet, GS was mainly emphasized model training using several training populations such as inbred for GCA and SCA, and test cross hybrids. By combining the GCA and SCA concepts, we can enhance our power for identification of superior cultivars (Jarquin et al. 2020). In ICRISAT, efforts are being made to exploit the available whole-genome resequencing (WGRS) data of PMiGAP lines along with phenotyping data for different traits for genomic selection.

Over the past years, various machine learning methods and statistical models have been advocated for genomic selection such as random forest, rrBLUP, RKHS, and BayesA/B. The selection of the model is entirely relying on the purposes for which they are used and it may be different for different traits. Although the models may be different, the values of prediction accuracy were substantially alike and rrBLUP method appears to be more popular (Yu et al. 2016).

Pearl millet genetic resources and populations

Pearl millet germplasm

The United States Department of Agriculture’s National Plant Germplasm System (USDA-NPGS) and International Crops Research Institute for Semi-Arid Tropics (ICRISAT) are the two major pearl millet germplasm banks, housing more than 26,000 accessions including landraces, wild relatives, historic accessions, and breeding lines. In addition, the National Bureau of Plant Genetic Resources (NBPGR) in India has got 8291collections, The Institut de Recherche Pour le Développement (IRD, France) holds about 3968 accessions, and the Canadian Genetic Resources Program, Saskatoon holds about 3821 accessions.

Genus Pennisetum represents 55 taxa, out of which 53 are wild species and 2 are intra-specific taxa. These taxa are classified according to their relatedness with P. glaucum using the model given by Maxted and Kell (2009). They are classified into primary, secondary, and tertiary wild relatives, in which P. glaucum cultivars and landraces P. glaucum subsp. stenostachyum and P. glaucum subsp. monodii and are included in primary wild relatives, P. purpureum and P. squamulatum come under secondary wild relatives and all other species in the genus are included in the tertiary wild relatives.

The center of domestication and geographical origin of pearl millet is located in Western Africa. In India, it was introduced dating back to 2000 B.C., and records reveal cultivation of pearl millet into Brazil in 1960s and into the United States in 1850s. The earliest finding of wild and domesticated pearl millet was recorded at about 1459 BC in Birimi in northern Ghana (D'Andrea et al. 2001), and the domesticated pearl millets was catalogued at about 3500 B.C. (Amblard and Pernes 1989). Whether pearl millet has a single center of the origin or more than one center of origin, called “non-centers”, there was a dispute among the scholars in several regions. According to latter hypothesis, these non-centers of pearl millet include whole Sahel from Mauritania to western Sudan (Clark 1962; Harlan 1971).

Diversity panels

Developing a core collection to represents diversity is an efficient pathway to increase the utilization of germplasm in crop improvement. The core collection of pearl millet consisting of 1600 accessions which are selected from 16,000 accessions characterized in 1998 at the ICRISAT Gene bank. This was supplemented by addition of 501 accessions that represent 4717 accessions and the revised core collection consisting of 2094 accessions (Updhyaya 2009). The core collection was far too huge, making crop inefficient and expensive. To overcome these difficulties, a mini core collection of pearl millet consisting of 238 accessions (Table 1) was organized by evaluating core collection of 2094 accessions of pearl millet for 18 morpho agronomic traits. The mini core collection was designed in a way that represents majority of the variation present in core collection and its reduced size provides an insight in a more economical point of view for exploitation of genetic resources of pearl millet for crop improvement (Upadhyaya 2011). Based on genotypic data, a subset of 300 diverse accessions were selected from composite collection to reduce the redundancy between units thus limits diversity losses. This reference subset of pearl millet accession, comprising 230 accessions from composite collection, were evaluated for drought stress and other agronomically important traits to enhance yield potential. These reference sets are available from ICRISAT gene bank by signing on SMTA (Standard Material Transfer Agreement). A set of germplasm material from Africa and India having diverse phenotypic characters such as grain size, panicle size, tillering, etc. was exploited for broadening genetic base of restorer and seed parents (Yadav et al. 2012; Patil et al. 2020).

Table 1

A name of mini core collection having 238 accessions

Source: ICRISAT Hyderabad India

Sr. No.	Accession ID	Accession name	Material type/country	Sr. No.	Accession ID	Accession name	Material type/country	Sr. No.	Accession ID	Accession name	Material type/country
1	IP 196	B 90	L/Kenya	81	IP 8155	GS 138	B/India	161	IP 13623		L/India
2	IP 277	No. 8	B/Congo	82	IP 8205	P 7-1	L/Nigeria	162	IP 13624		L/India
3	IP 446	IC 5462	L/ India	83	IP 8220	GS 163	B/India	163	IP 13,636	Etah type	L/India
4	IP 869	Georgia inbred 37	B/U. S. A	84	IP 8245	IP 228-2	B/India	164	IP 13760	PEN 5/78-1-1	B/Germany
5	IP 952	–	L/India	85	IP 8276	IP 2130-1/CG 51	B/India	165	IP 13875	60	L/Burkina Faso
6	IP 1060	Moti Mehrana	L/India	86	IP 8288	IP 5335-1	B/India	166	IP 13991	–	L/Zimbabwe
7	IP 1098	Bajra Tadpang	L/India	87	IP 8350	Acc No. 21	L/India	167	IP 14294	–	L/Cameroon
8	IP 1405	Bajra Chakpakki	L/India	88	IP 8418	E 127	L/Nigeria	168	IP 14428	–	L/Cameroon
9	IP 1536	Bajri Rampahari	L/India	89	IP 8472	D 172 C-2	L/Niger	169	IP 14522	–	L/Cameroon
10	IP 1566	Bajra Hirasda	L/India	90	IP 8529	–	L/India	170	IP 14537	Yadiri	L/Cameroon
11	IP 1625		L/India	91	IP 8540	–	L/India	171	IP 14542	–	L/Cameroon
12	IP 1834	Anand No. 5	L/India	92	IP 8562	–	L/India	172	IP 14599	–	L/Cameroon
13	IP 1917	Anand No. 77	L/India	93	IP 8672	30 K	L/Sudan	173	IP 14753	–	L/Cameroon
14	IP 2083	No. 13	L/South Africa	94	IP 8707	49 KA Darfur-2	L/Sudan	174	IP 14776	–	L/Cameroon
15	IP 2167	CPI 26307	B/Australia	95	IP 8818	5 LUP	L/Zimbabwe	175	IP 14787	–	L/Cameroon
16	IP 2246	N 6	L/Nigeria	96	IP 8863	–	L/Zambia	176	IP 15095	–	L/India
17	IP 2322	SZ 36	L/Nigeria	97	IP 8913	Sanio	L/Gambia	177	IP 15119	–	L/India
18	IP 2704	BAM-C 130	L/Chad	98	IP 9000	P3 Kolo P-1/1/1	B/Niger	178	IP 15256	–	L/India
19	IP 2761	Antia	L/ Burkina Faso	99	IP 9026	Sanio	L/Gambia	179	IP 15273	Kattu cumbu	L/India
20	IP 2789	Mousse	Mauritania	100	IP 9157	–	L/India	180	IP 15372	–	L/India
21	IP 3110		L/India	101	IP 9198	SB 21	L/India	181	IP 15448	–	L/India
22	IP 3329	Karauli	L/India	102	IP 9449	–	L/Ghana	182	IP 15556	162	L/Burkina Faso
23	IP 3432	Chadi local	L/India	103	IP 9464	–	L/Ghana	183	IP 15829	–	L/Tanzania
24	IP 3489	Oong	L/India	104	IP 9492	–	L/Ghana	184	IP 15836	–	L/Tanzania
25	IP 3525	Podi cumbu	L/India	105	IP 9527	–	L/Ghana	185	IP 15953	Sajjalu	L/India
26	IP 3626	Peria cumbu	L/India	106	IP 9596	126	L/Yemen	186	IP 16402	Sifumbata	L/Zimbabwe
27	IP 3642	Kullan	L/India	107	IP 9617	PI 337492	B/Brazil	187	IP 16489	–	L/Zimbabwe
28	IP 3646		L/India	108	IP 9645	PI 286842	L/Nigeria	188	IP 16540	–	L/Zimbabwe
29	IP 3706		L/India	109	IP 9692	PI 286979	L/Nigeria	189	IP 16754	–	L/Zimbabwe
30	IP 3852		L/India	110	IP 9795	UI 1639	L/United states	190	IP 16863	–	L/Zimbabwe
31	IP 4177		L/India	111	IP 9813	Mexioera	L/Mozambique	191	IP 17396	–	L/Central African Republic
32	IP 4291		L/India	112	IP 9934	Acc 229	L/Sudan	192	IP 17465	–	L/Algeria
33	IP 4363		L/India	113	IP 10085	P 5439	L/Mali	193	IP 17490	–	L/Tunisia
34	IP 4488		L/India	114	IP 10151	P 5521	L/Mali	194	IP 17532	–	L/Togo
35	IP 4747	Meerut 30	L/India	115	IP 10263	P 5684	L/Mali	195	IP 17775	–	L/Togo
36	IP 4903	NEP 218-5081	B/Lebanon	116	IP 10371	EC 134848	L/Nigeria	196	IP 18040	–	L/Pakistan
37	IP 4979	700164	L/Nigeria	117	IP 10399	Purple dwarf	B/India	197	IP 18353	–	L/Namibia
38	IP 5085	700706	L/Nigeria	118	IP 10437	P 3908	L/Benin	198	IP 18545	–	L/Namibia
39	IP 5153	D 70	L/Niger	119	IP 10467	Bikita	L/Zimbabwe	199	IP 18579	–	L/Namibia
40	IP 5185	D 46	L/Niger	120	IP 10601	M'Boneri	L/Mali	200	IP 18657	–	L/Namibia
41	IP 5261	D 222	L/Niger	121	IP 10632	CMM 465	L/Mali	201	IP 18824	–	L/Namibia
42	IP 5298	D 219-1	L/Niger	122	IP 10665	Haini	L/Mali	202	IP 18854	–	L/Namibia
43	IP 5389	P 2672	L/Niger	123	IP 10713	Acc 447	L/Sudan	203	IP 18900	–	L/Namibia
44	IP 5407	P 2693	L/Niger	124	IP 10729	Acc 473	L/Sudan	204	IP 19072	–	L/Namibia
45	IP 5438	P 2727	L/Niger	125	IP 10761	Acc 516	L/Sudan	205	IP 19141	–	L/Namibia
46	IP 5455	P 2745	L/Niger	126	IP 10925	Acc 772	L/Sudan	206	IP 19305	–	L/Namibia
47	IP 5581	P 2877	L/Niger	127	IP 10953	BM 8	L/Kenya	207	IP 19415	TCD 115	L/Namibia
48	IP 5711	45-347	L/Nigeria	128	IP 11010	–	L/India	208	IP 19425	–	L/Republic of the Congo
49	IP 5719	45-359	L/Nigeria	129	IP 11036	–	L/India	209	IP 19448	–	L/Namibia
50	IP 5793	WJR 8/28530	B/Russia	130	IP 11044	Gullisita	L/India	210	IP 19629	C 90–136	L/Niger
51	IP 5869	P 1471/Sl.165	L/Senegal	131	IP 11113	-	L/India	211	IP 19722	ICMSR 45	B/India
52	IP 5957	P 1565/Sl.395	L/Senegal	132	IP 11247	Rushambo	L/Zimbabwe	212	IP 19816	ICMSR 147	B/India
53	IP 5964	P 1572/Sl.419	L/Senegal	133	IP 11268	-	L/Zimbabwe	213	IP 19851	ICMSR 183	B/India
54	IP 6057	P 172	L/Central African Republic	134	IP 11405	CVP 508	L/Burkina Faso	214	IP 19913	ICMSR 247	B/India
55	IP 6113	P 965	L/Niger	135	IP 11428	CVP 610	L/Burkina Faso	215	IP 19964	ICMSR 299	B/India
56	IP 6193	P 89	L/Cameroon	136	IP 11546	P 6008	L/Burkina Faso	216	IP 20249	–	L/Yemen
57	IP 6275	P 246	L/Mali	137	IP 11666	Millet 181	L/United Kingdom	217	IP 20274	–	L/Yemen
58	IP 6278	P 249	L/Mali	138	IP 11799	Pitta ganti	L/India	218	IP 20409	Dauro	L/Nigeria
59	IP 6324	P 304	L/Mali	139	IP 11811	Pedda sajja	L/India	219	IP 20576	Dauro	L/Nigeria
60	IP 6340	P 333	L/Mali	140	IP 11930	Belenguine	L/Sierra Leone	220	IP 20577	Dauro	L/Nigeria
61	IP 6517	P 552	L/Mali	141	IP 11943	Tasur	L/Sierra Leone	221	IP 20611	Maiwa	L/Nigeria
62	IP 6769	–	L/Malawi	142	IP 12221	Maiwa	L/Nigeria	222	IP 20715	–	L/Nigeria
63	IP 6798	–	L/Malawi	143	IP 12364	–	L/Nigeria	223	IP 20768	–	L/Nigeria
64	IP 6805	–	L/Malawi	144	IP 12374	–	L/Nigeria	224	IP 20929	Maiwa	L/Nigeria
65	IP 7118	Haram Type	L/India	145	IP 12418	Fes MAR	L/Morocco	225	IP 20955	Gero	L/Nigeria
66	IP 7259	–	L/India	146	IP 12431	SA. Anter 11004	B/Cabo Verde	226	IP 21066	62	L/Myanmar
67	IP 7358	–	L/India	147	IP 12498	C 3-4	L/India	227	IP 21093	PI 331692	L/Ethiopia
68	IP 7522	Uwele	L/Tanzania	148	IP 12533	MB 6939	L/India	228	IP 21127	90-20-DVP-Dwarf PO	B/United States
69	IP 7497	–	L/Somalia	149	IP 12546	Akola from ausan	L/India	229	IP 21187	IP 8695-1	B/India
70	IP 7537	K 47	L/India	150	IP 12644	SM 28 (7042)	B/India	230	IP 21201	IP 8998-2	B/India
71	IP 7675	D 1941	L/India	151	IP 12650	SM 62 (7042)	B/India	231	IP 21244	ICMER 038	B/India
72	IP 7829	Nabhu	L/India	152	IP 12669	SM 568	B/India	232	IP 21283	ICMER 081	B/India
73	IP 7846	IP 4821-1	B/India	153	IP 12731	SM 2214	B/India	233	IP 21312	ICMER 111	B/India
74	IP 7860	J 856-1	B/India	154	IP 12805	NC 59759	L/India	234	IP 2138	ICMER 237	B/India
75	IP 7886	J 1357-1	B/India	155	IP 12839	–	L/Botswana	235	IP 21503	M 108-38	B/France
76	IP 7915	D 135 I-1-1	L/Niger	156	IP 12993	P 354-1	L/Mali	236	IP 22269	High tillering genepool	B/India
77	IP 7978	IP 6455-1	B/India	157	IP 13261	P 1381-1/Sl.10	L/Senegal	237	IP 22449	ICMB 92666
78	IP 8022	–	L/Zambia	158	IP 13345	Acc 754-1	L/Sudan	238	IP 22489	ARD 218 (325)	L/Niger
79	IP 8051	OZAC 183/79-1	B/Mexico	159	IP 13387	Serere 6A-1	L/Uganda
80	IP 8074	GS 8	B/India	160	IP 13523	–	L/India

B breeding material, L land races

Mutagenized populations

After successful exploitation of Targeting Induced Local Lesions IN Genomes (TILLING) approach in Arabidopsis plant, this approach has been extended in cereals such as maize (Till et al. 2004), wheat (Slade et al. 2005), and rice (Horst et al. 2007). In a study,the seeds of pearl millet (Pennisetum typhoides (Burn) Stapf. Var.Co (CU)-9 were mutated for chlorophyll and morphological mutation using ethyl methane sulfate (EMS). The results revealed EMS as an effective chemical mutagen which induce mutation that can be used in breeding programmes. (Ambli and Mullainathan 2015). Application of TILLING in mutation breeding plays a significant role in improvement of particular traits using radiation or chemical treatment (Szurman-Zubrzycka et al. 2018). Some of the key applications of TILLING to some important traits include starch synthesis in wheat (Slade et al. 2012); plant architecture in rice for salt-tolerance (Hwang et al. 2017); high number of tillers in barley (Marzec et al. 2016); powdery mildew resistance in wheat (Acevedo-Garcia et al. 2017); resistance to northern corn leaf blight in maize (Severune et al. 2015); for other yield-related parameters such as waterlogging tolerance and DNA repair in barley (Mendiondo et al. 2016). Maryono et al. (2020) studied the M3 population of pearl millet for performance and estimation of genetic variability. Population treated with different doses of gamma rays showed high heritability for the traits such as panicle diameter, number of nodes per plant and stem diameter (Maryono et al. 2020).

Although ICRISAT and other institutes collected thousands of accessions of pearl millet, the majority of the accessions lack genotyping and phenotypic data. With the advent of new sequencing techniques, all accessions can be genotyped and phenotyped at low cost both effectively and efficiently. Additionally, wild relatives of pearl millet offer excellent sources of genes for agronomic traits and various abiotic stresses (Yadav et al. 2021).

MAGIC and NAM populations

Molecular breeding for second generation platforms brought about a shift in population linkage mapping from bi-parental to the multi-parental population, such as multi-parent advanced generation inter-cross (MAGIC) population (Kover et al. 2009) and nested association mapping (NAM) population, to exploit allelic diversity in the fine mapping of QTLs for traits of interest. These populations utilize the supremacy of both association mapping and linkage analysis thereby reducing the limitations of both and facilitating high-resolution mapping using a multi-parental population (Tibbs et al. 2021). NAM population was first reported in Zea mays and MAGIC population was first reported in Arabidopsis and both offer potential for exploiting the structure of the genome and improving breeding populations. MAGIC populations have been utilized in various crops including maize (Dell'Acqua et al. 2015), wheat (Huang et al. 2012), sorghum (Ongom and Ejeta 2018), and rice (Bandillo et al. 2013). The information about the NAM and MAGIC population in pearl millet to decode allelic variations for stress tolerance and yield-related traits is lacking. Accessions of pearl millet exhibit novel alleles and can be conquered through the exploitation of a multi-parental populations. MAGIC and NAM populations should be developed in such a way that they consolidate diverse parents crossed with a common parent (Fig. 1).

Fig. 1

Development of populations such as NAM and MAGIC population for functional genomic would be helpful in the introgression of superior genes into cultivars

Genomic research in pearl millet

Pearl millet domestication

Sequencing of pearl millet genome provides an insight to assess the domestication origin of pearl millet. Moreover, it enables breeders and researchers to improve this staple food for agronomically important traits as well as against various biotic and abiotic stresses. Pearl millet domestication was related to the modification of plant architecture and spike morphology like that observed in maize crop (Schnable et al. 2009). Lakis et al. (2012), studied three flowering genes namely PgHd3a, PgDwarf8 and PgPHYC that involved in domestication of pearl millet. PgDwarf8 and PgPHYC genes were found to show significant differentiation between wild and domestic populations but such differentiations were not found for PgPHYC gene. Results revealed lack of differentiation between early and late landraces on the basis of three candidate genes (Lakis et al. 2012). For pearl millet domestication, 221 accessions of traditional varieties and wild forms were structured into 3 major geographic groups. Several genomic regions have been identified that showed reduced diversity in cultivated species only. A total of 140 genomic regions have been identified with values above 95% threshold for loss of differentiation and diversity (Gaut et al. 2018). Out of 24 genomic regions, 8 were located on pg7,6 on pg6 and 5 on pg1. Linkage group 6 and 7 carrying QTLs have been identified previously which explained the majority of phenotypic differences between cultivated and wild germplasm (Poncet 2000; Poncet 2002).

Domestication syndrome can be defined as set of traits that blemishes a divergence of crop from its wild relatives (Purugganam 2019). A linkage map has been obtained which revealed genes that are involved in qualitative traits of spikelet namely abscission layer (AL); pedicel length (PL); seed coating (Ct); Length of involucre bristle (BL); and presence of longer bristles (PB). Additionally, a QTL for length of glumes (GL), tillering, and spike morphology (WeS) and these genes have been identified on segment II (Poncet et al. 1998). However, only a few genes for pearl millet domestication have been reported. Domestication genes can also be uncovered using parallel selection theory (Purugganan 2019; Rendon-Anaya and Herrera-Estrella 2018).

Pearl millet draft genome

Pearl millet is a diploid (2n = 2x = 14) crop species having relatively non-duplicated and large genome (1.76 Gb). The genome of pearl millet was sequenced using whole-genome shotgun (WGS) and bacterial artificial chromosome (BAC) technique. Tift 23D2B1-P1-P5 genotype was used to construct 13 large inserts (~ 2, 5, 10, 20 and 40 kb) and 10 small inserts (~ 170, 250, 500, 800 bp) of WGS libraries and these libraries were sequenced on the Illumina HiSeq 2000 and 520 Gb of sequence data. From Tift 23D2B1-P1-P5, genotype two BAC libraries were constructed using HindIII and EcoR1 having an average insert size of ~ 120 kb. It has been estimated that the genome of pearl millet was about 1.76 Gb indicating that about 90% of the genome was assembled along with 50% of scaffolds (Varshney et al. 2017). About 77.2% of the repetitive sequences were found in the assembled genome. The true percentage of the repetitive sequence was about 80% (0.18 Gb) which is similar to the proportion of repetitive DNA found in the 466-Mb rice genome (~ 42%) (Yu 2002), ~ 400-Mb foxtail millet (~ 46%) (Bennetzen 2012), ~ 730-Mb sorghum (~ 61%) (Paterson 2009) and ~ 2.3 Gb maize (> 85%) (Schnable et al. 2009). Another study revealed a total of 69,398 transcriptome assembled contigs (TACs) in pearl millet using transcriptome sequence (Rajaram et al. 2013). The length of the CDS, mRNA, exon and introns in pearl millet were similar with those reported for other cereal crops genome. CEGMA analysis revealed that among 458 of the conserved genes, 437 genes were complete, 8 genes were not found in the genome sequence, 8 genes were not included in the gene set, and 5 genes has more than 1 copy (possibly fragmented genes). In addition, BUSCO analysis for 956 genes revealed that 96.7% genes were annotated and 95.4% of these were complete. Gene models of rice were chosen to investigate the completeness of pearl millet genes because of closely relatedness showing 90.86% homology with rice gene model than Arabidopsis gene model.

Population structure and genetic diversity

Population genomics is a widespread approach to understand the linkage disequilibrium, population structure, migration, genetic basis of adaptation, and relatedness at genome level. Pearl millet is an important cereal crop with wider adaptability and long cultivation history. Agronomic adaptation and speciation of pearl millet could facilitate breeding at molecular level and is also helpful in understanding this concept for other crops as well. Availability of diverse resources for pearl millet including geographical and genetic resources offers material worth for genomic research.

In pearl millet breeding, genetic diversity is an important basis for exploitation of complex traits. For better elucidation of genetic diversity and population structure, 994 lines of pearl millet were re-sequenced, including 260 inbred male sterility maintainer (B-) and 320 male fertility restorer (R-) lines, 345 PMiGAP (Pearl Millet Inbred Germplasm Association Panel) lines, 31 wild accessions and 38 inbred parents of mapping populations (Varshney et al. 2017). In this, a total of 1.16 Tb whole-genome resequencing (WRGS) on PMiGAP lines and 116 Gb WGRS on parental lines of mapping populations were performed. Additionally, 78.9 Gb data for PMiGAP lines were generated using GBS (genotyping by sequencing) (Elshire 2011) and R- and B- lines were generated using RAD sequencing (Miller et al. 2007). In the pearl millet genome sequence, 88,256 simple sequence repeats (SSR) were identified using MIcroSAtellite program (Thiel et al. 2003) and the primers were designed for 74,891 SSR-containing sequences which can be utilized for breeding and genetic applications. A total of 29,542,173 single nucleotide polymorphisms (SNPs) were identified in PMiGAP lines, including 3,844,446 insertions and deletions, and 423,118 genome-wide structural variations (Varshney et al. 2017).

Genome-wide single-Nucleotide polymorphism discovery identified 82,112 SNPs markers distributed over all 7 chromosomes. From identified SNPs, majority of the SNPs were found on chromosome 1 (38,71) and 2 (36,854) whereas 35,714 SNPs were mapped to the scaffolds. Genome-wide Linkage disequilibrium (LD) in west African population was shorter than in all other subpopulations (South Africa, Middle East, East Africa, USA and India) (Serba et al. 2019).

To explore the diversity of different pearl millet races, Kanfany et al. (2020) analyzed the 309 genotypes of pearl millet. They found lowest genetic distance (0.09) between ICML197458 and ICML197279 which were developed from landraces of Nigeria and India, respectively whereas highest genetic distance (0.33) was observed between ICML197390 and ICML197314.

GWAS in pearl millet

GWAS is a powerful tool to dissect the architecture of complex agronomically important traits of crops using genome-wide single nucleotide polymorphism (SNPs) markers. The discovery of pearl millet draft genome has unlocked enormous possibilities to dissect various QTLs along with the functions of its related genes having diverse traits. Association mapping analysis and QTL-mapping/interval mapping are two techniques that can be utilized for construction of genetic maps. Association mapping greatly reduces time and labor requirement as breeders can skip the hectic task of generating a mapping population through hybridization, continuous selection, and recurrent crossing. Instead, diverse germplasm accessions are employed as mapping panel to identify relationship between markers and trait under study. Since, these diverse accessions were the results of the plethora of random meiotic events amongst these accessions, relatedness for recombination events are uncontrolled (Verdeprado et al. 2018). Due to uncontrolled meiotic recombination within diverse accessions, association mapping approach can be suited well for high resolution identification of genes or QTLs with high resolution (100–1000 kb) which are tightly linked to diverse phenotypic traits (Mackay et al. 2009). Pearl millet germplasm accessions have very high levels of heterozygosity and heterogeneity which poses difficulty in association mapping, and hence, limited strategies were delivered to dissect genetic diversity (Kannan et al. 2014). Association study conducted on pearl millet reveals factors responsible for flowering time variation at phytochrome C (PHYC) locus (866 bp). A significant association between genetic variation and phenotypic traits was observed using a linear mixed model (Saidou et al. 2009). Further, accessions were explored using an association study that revealed an extra 100 bp region adjoining the PHYC genes using MCMC (Markov chain Monte Carlo) method for identification of tightly linked markers (75 SNPs and INDELS) adjoining 6 kb (PHYC) genomic region (Saidou et al. 2014).

GWAS study on three germplasm sets of pearl millet was conducted by using Genome-wide SNP data to compute linkage disequilibrium decay (LDD). GWAS on 288 testcross progenies of PMiGAP lines were carried out for 20 traits, and about 1054 marker-trait associations (MTAs) have been identified for 15 traits including panicle yield (9), stover dry matter yield (5), tillers per plant (147), grain number per panicle (91), fresh stover yield (38), grains per square meter (75) panicle diameter (1), panicle length (3), panicle harvest index 1), plant population (68), panicle number (246), 1000 grain weight (10), plant height (344), grain harvest index (5), and grain yield (11). Moreover, these MTAs explained 9–27% of the total phenotypic variation, and the selected markers were found on pg1 and pg5 commonly for yield-related parameters and stresses (Varshney et al. 2017). MTAs between 250 SSR and 17 genetic markers with grain zinc and iron content was developed using 130 diversified lines of pearl millet revealed that markers Xipes0224, Apsmp2213 and Xpsmp2086 showed significant association with zinc content in grain on LG6 and LG4, and marker Xicmp3092 had a strong association with iron content on LG7 (Anuradha et al. 2017; Gemenet et al. 2015). Moreover, GWAS analysis using 34 SSR markers and 250 full-sib progenies revealed that marker allele Xpsmp2237_230 was associated with grain yield on LG7, Xpsm2224_157 was linked with plant height on LG7, Xicmp3058_193 was strongly associated with stover dry matter yield on LG6, and marker alleles Xpsmp2224_157, Xpsmp2233_260, and Xpsmp2077_136 were associated with panicle length on LG7, LG5, and LG2, respectively (Kannan et al. 2014). Another association analysis study was conducted under high and low phosphorus conditions in West Africa with the available 285 DArT markers using phenotypic data of 151 PMiGAP lines. The results indicated that the pgpb12954 marker showed a strong association with grain yield and the pgpb11603 DArT marker showed a significant association with time of flowering (Gemenet et al. 2015).

Comparative genomics in pearl millet

The first published map of pearl millet had a genetic length of only 303 cM (Liu et al. 1994). Comparative genetic maps were constructed of pearl millet genome with foxtail millet and used to describe homoeology between genomes of pearl millet, foxtail millet and rice (Devos et al. 2000). Pearl millet genome is differentiated from rice genome by several structural rearrangements while strong relatedness has been observed between foxtail millet and pearl millet. Pearl millet genome carried one or probably two duplications in linkage group 1 (LG1) and group 4 (LG5). Endogenous gibberellic acid levels suggested that genes d1, d2, and d4 are recessive dwarfing genes which may be similar to the rye dwarfing genes ct1 and ct2 (Devi et al. 1994) and these genes were mapped in the centromeric regions of rye chromosomes 7R and 5R (long arm), respectively. Similarly, these regions are homoeologous to LG4 and LG2 segments of pearl millet, respectively (Devos et al. 2000). Comparative analysis of global accessions and landraces of pearl millet was carried out using 500 pearl millet accessions consisting of 252 global accessions, and 248 Senegalese landraces using genotyping by sequencing (GBS) technique of Pstl-Mspl reduced representation libraries. A total of 83,875 SNPs were identified as a genomic resource for population improvement for pearl millet. Comparative genomics for population improvement will provide an insight for the improvement of these climate-resilient crops (Hu et al. 2015). The genomic analysis of elephant grass (Cenchrus purpureus) provides an insight in discovery of enzyme-coding gene families responsible for biosynthesis of anthocyanidins and flavonoids content. Evolutionary analysis revealed that the subgenome A of elephant grass and pearl millet may have originated from common ancestor (Yan et al. 2021).

Transcriptome analysis in pearl millet

The draft genome of pearl millet has been sequenced in 2017, but short reads cannot be mapped due to incomplete genome annotations. Next-generation sequencing such as PacBio sequencing (single-molecule real-time sequencing), enables the production of full-length transcript making it ideal for transcript recovery (Abdelghany et al. 2016; Wang et al. 2017) but this has the limitations of low throughput (Rhoads and Au 2015). Transcription profiling in pearl millet identified 10 differentially expressed genes that validated for drought tolerance, namely Calmodulin-like proteins, Aspartic proteinase Oryzasin, DnaJ-like protein, Rab7, Glyoxalase, Putative beta-1,3-glucanase, Inosine-5'-monophosphate, Ascorbate peroxidase, and Abscisic stress ripening protein (Choudhary and Padaria 2015). Two sequencing techniques were used to study the differences and similarities in response of pearl millet under drought and heat stress. A total of 63,090 new transcripts and 26,299 new genes were identified and functional annotations were boosted by 20%. The results revealed regulation of 5039 DEGs and 4603 DEGs under drought and heat stress, respectively. Under drought and heat stress, 6484 and 6920 genes were expressed differentially and 1881 genes were expressed under both stresses (Sun et al. 2020). Likewise, RNA-Seq approach was used to understand the pathways involved in response to drought stress in pearl millet using two inbred lines ICMB843 (drought tolerant) and ICMB 863 (less tolerant), which were procured from ICRISAT. About 25 up-regulated genes in ICMB843 and 8 genes in ICMB 863 were found to involved in photosynthesis and its related pathways. Pathway and gene function analysis revealed that drought response in pearl millet was mainly regulated by pathways related to photosynthesis, plant hormone signal transduction and mitogen-activated protein kinase signaling. Results obtained from the analysis revealed molecular mechanisms dealing with drought stress for genetic improvement of pearl millet crop (Dudhate et al. 2018).

Genetic analysis of important adaptive and agronomic traits in pearl millet

With the advancement of sequencing and phenotyping techniques, various important genetic loci and genes which are agronomically important have in the past few decades been identified in pearl millet using several technologies like GWAS, QTL mapping, genotyping by sequencing in past few decades (Table 2). The regulatory mechanisms of the genes are still unknown and our understanding of genetic resources provides an opportunity for designing of super pearl millet for various purposes (Fig. 2). Below we summarized in detail the genome-wide dissection of agronomically important traits and their related genes.

Table 2

Major QTLs/genes for important input and output traits in Pearl millet

Trait	Markers	Gene name	Linkage group	Phenotype/pathways	References
Flowering time	Xibmsp27	Alanine glyoxylate aminotransferase	LG2	Causes variation in the flowering time	Sehgal et al. (2012)
	Xibmsp9	Uridylate kinase	LG2
	Xibmsp12	Acyl CoA oxidase	LG2
	Xibmsp60	MADS-box	LG2
	Xibmsp14	Serine-threonine protein kinase	LG2
	Xibmsp24	Ubiquitin conjugating enzyme	LG2
	Xibmsp31	HD3	LG2
	Xibmsp15	Zinc finger C- × 8-C × 5-C × 3-H type	LG2
	SNP	PgMADS11		Characterization of early and late flowering in landraces	Diack et al. (2017)
	InDel	PgPHYC
	Pgpb6981-Xipes226		LG1	Responsible for flowering time variation	Kumar et al. (2017)
	head5		LG5
	head7		LG7
Grain	Xibmsp27	Alanine glyoxylate aminotransferase	LG2	Responsible for enhanced grain yield	Sehgal et al. (2012)
	Xibmsp9	Uridylate kinase	LG2
	Xibmsp12	Acyl CoA oxidase	LG2
	Xibmsp60	Dipeptidyl peptidase IV	LG2
	Xibmsp34	MADS-box	LG2
	Xpsm592-Xpsm356		LG2 (6.91)	Enhanced grain yield	Yadav et al. (2004)
	Xpsm464-Xpsm716		LG4 (2.32)
	Xpsm588-Xpsm514		LG6(2.21)
Grain quality biofortification)	Zinc (Zn)			Increased Zinc level in the accessions	Kumar et al. (2016, 2017)
	Xpsmp2214-Xipes142		LG3 (4.68)
	pgpb10531-pgpb9130		LG1 (25.36)
	Iron (Fe)			Enhanced Iron content level	Kumar et al. (2016, 2017)
	pgpb9502-pgpb6039		LG 4B
	pgpb6825-Xipes195		LG7
Plant Height	D1/d1		LG1	Responsible for the plant height of pearl millet	Kumar et al. (2017); Kannan et al .(2014)
	D2/d2		LG4
	Pgpb6112-pgpb9106		LG1
	Pgpb9498-Xipes017		LG1
	Xipes203-Xpsmp2273		LG1
Biomass	Xpsm592-Xpsm443		LG2	Increased biomass production	Yadav et al. (2003)
	Xpsm716-Xpsm265		LG4
	Xpsm87.1-Xpsm514		LG6
Leaf rolling	Xibmsp60	Dipeptidyl peptidase IV	LG2	Identified genes conferring leaf rolling in pearl millet	Sehgal et al. (2012)
	Xibmsp34	Serine-threonine protein kinase	LG2
	Xibmsp14	Ubiquitin conjugating enzyme	LG2
	Xibmsp24	HD3	LG2
	Xibmsp31	Acetyl CoA carboxylase	LG2
	Xibmsp11	Zinc finger C- × 8-C × 5-C × 3-H type	LG2
	Xibmsp15	Photolyase	LG2
Delayed leaf senescence	Xpsmp2066		LG2	Identified locus in the linkage group conferred delayed leaf senescence	Sehgal et al. (2012)
	Xibmsp23		LG2
	Xipes0117		LG2
	Xicmp3056		LG2
	Xibmsp62		LG2
Panicle Length	los1		LG1		Poncet et al. (2000); Kumar et al. (2017)
	los2		LG2
	pgpb9647-Xicmp3027		LG5
	Xipes200-Xicmp3002		LG6B
	Xipes198-Xipes82		LG7
Drought tolerance	Xpsmp2237		LG2	Conferring drought tolerance	Sehgal et al. (2012); Choudhary and Padaria (2015)
	Xctm03		LG2
	Xibmsp53		LG2
	Xibmsp55		LG2
	Xibmsp51		LG2
	Xpsmp2072		LG2
		Aux/IAA, GH3 & SAUR		Auxin biosynthesis	Dudhate et al. (2018)
		GIDI, PIF		Diterpenoid biosynthesis
		PP2C, SnRK2, ABF		Carotenoid biosynthesis
		ETR, TRE, MKK6 &ChiB		Ethylene biosynthesis
		PVR, MAP3K17/18, CAT1, MAPK1 & MAPK6		Abscisic acids
	S3_216179591		LG7	Aerial Biomass	Debieu et al. (2018)
	S6_64437341		LG6
	S3_13047576		LG6
		Pgl_GLEAN_10002188	LG2	Salt stress protein	Shivhare et al. (2020)
		Pgl_GLEAN_10005403	LG2	Glutamine synthetase root isozymes
		Pgl_GLEAN_10011050	LG2	Lateral root formation
		GAPDH and EF-1a	LG3	Involved in primary metabolism and cellular processes	Shivhare and Lata (2016)
Leaf spot or blast	S2_7773-S2_8331		LG2		Punnuri et al.(2016)
	S3_0019-S3_4763		LG3
	BRP8, BRP11, & BRP7		LG1	Tolerance against blast and improved several other agronomic traits	Maganlal (2018)
	BRP3, BRP37 & BRP21		LG2
	BRP12, BRP23, BRP18, & BRP27		LG3
	BRP84, BRP87 & BRP 94		LG7
Rust	Rr1		LG3		Morgan et al. (1998)
	Rust Res (ICMP83506)		LG4

Biotic and abiotic stress traits come under the input traits while the agronomically important traits comes under the output traits

Fig. 2

Designing of super pearl millet with important genes controlling the complex traits. Super pearl millet can be designed by pyramiding of superior alleles controlling agronomically important traits, and biotic and abiotic stress tolerance

Grain yield and grain quality

A three major QTLs with less QTL * environment interaction for grain yield were identified in a post-flowering moisture environment (Bidinger et al. 2007). Results obtained from association mapping in pearl millet accessions indicated that the SNP101 of PHYC gene showed a significant association for grain yield-related parameters (Saidou et al. 2009). Association studies on pearl millet revealed that Xibmsp11/AP6.1, a known SNP marker that is present on an acetyl CoA carboxylase gene, is highly associated with yield traits (Grain harvest index and grain yield). InDels markers, Xibmcp09/AP10.2 and Xibmcp09/AP10.1, present on chlorophyll a/b binding protein genes, are strongly associated with stay green and grain yield traits (Gemenet et al. 2015). Much emphasis should be given on grain quality and grain yield-related traits to dissect the variability present among pearl millet accessions.

Plant height and flowering

A significant association between PHYC gene and flowering time has been reported in pearl millet inbred lines which is having a major role in pearl millet adaptation in different regions (Saidou et al. 2009). Further analysis of PHYC locus revealed association of Pg7830 and Pg7840 genes with flowering time in pearl millet but none of them were found to be associated with plant height (Saidou et al. 2014). Polymorphism of PgMADS11 associated with phenotypic variation has been identified and it might be possible that polymorphism can also be located in the neighbouring region that is in linkage disequilibrium with the revealed polymorphism. Actual sequenced data revealed that INDEL polymorphism is located in the intron of PgMADS11, although it is highly similar to a MADS-box gene family (Mariac et al. 2011). Pearl millet accessions collected in different years were evaluated for early flowering for PHYC gene (Vigouroux et al. 2011). Fifteen flowering genes and 20 random genes were amplified in 33 cultivated and 13 wild relatives of pearl millet. The results indicated that all flowering genes showed high density which was identified using BLASTn. Flowering genes in pearl millet accessions include: PgEMF2, PgFY, PgGI, PgHD1, PgHD3a, PgHD6, PgLFL1, PgMADS11, PgPHYA, PgPHYB, PgPHYC, PgPIPK1, PgPRR73, PgPRR95 and PgTFL1 (Clotault et al. 2012). Another study was conducted on three flowering candidate genes namely, PgPHYC, PgDwarf8, and PgHd3a, and results suggested that PgDwarf8, and PgHd3a were targeted through selection in the course of domestication. The gene, PgHd3a has been the target of selection in domestic population because this gene plays a very well-known function in flower transition (Lakis et al. 2012). Seeds from 16 early-flowering, 13 late-flowering, and 16 wild relatives of pearl millet were used to characterize the early and late-flowering in accessions using microsatellite loci (Dussert et al. 2015). Landraces of pearl millet from Senegal were used for characterization of early- and late-flowering using SSRs markers. Allelic diversity of PgPHYC and PgMADS11 genes were assessed in landraces of pearl millet and the results revealed that the early flowering landraces carried allele for early flowering at PgPHYC locus and PgMADS11 locus showed significant differences in genotype frequencies (Diack et al. 2017). In another study, Diack et al. (2020) identified one SNP that is located on PgPPR gene encodes a pentatricopeptide repeat protein belonging to ATP DNA-binding cassette family involved in plant resistance and defense. Additionally, they identified one SNP on PgAAO1 gene that encodes an indole-3-acetaldehyde oxidase.

Tillering, spike length, and other agronomic traits

Tillering and spike length are agronomically important traits in pearl millet breeding. A study showed that the SNP101 of PHYC gene was significantly associated with spike length and basal diameter. Moreover, most of the PHYCSNPs were tightly linked, so the same association was found for the whole PHYC amplified region (Saidou et al. 2009). Association mapping in pearl millet revealed that there was a significant marker traits association for various agronomic traits such as stover dry matter yield, fresh stover yield, panicle harvest index, spike length, and so on (Varshney et al. 2017). For tillering, one SNP on PgHK4 gene was identified which encodes a histidine kinase but no SNPs were identified for biomass (Diack et al. 2020).

Biotic stress tolerance

Pearl millet is a robust crop and has a low vulnerability to insect-pests and diseases in comparison to other crops. Downy mildew, blast, smut, and ergot are common pearl millet diseases which causes severe damage and ultimately reduces crop yield. To this end, host plant resistance (HPR) is an effective and efficient strategy to cope up with biotic stresses and it does not incur an additional cost.

Downy mildew disease of pearl millet, caused by Sclerospora graminicola is causing economic damage in Africa and India. Screening of breeding lines and germplasm accessions provides an insight for the identification of various sources of resistance (Singh et al. 1997). QTLs from the resistant parent were mapped to linkage groups 1, 2, and 4 and it has been observed that these QTLs are effective against pathogens isolates from Sudan, Mali, Nigeria and India. After the mapping of QTLs, marker-assisted selection is supposed to be an effective approach for breeding against pathogens (Breese et al. 2003). SCAR marker was used for screening of downy mildew in parents (ICMR-01004 and ICMR-01007), F1, F2, and F3 progenies. Primer pair ISSR-22 was found to be polymorphic with a fragment of 1.4 kb band in both parents and F2 populations, and it was then cloned, catalogued and sequenced and linkage map was established on linkage group 4 (LG4) (Jogaiah et al. 2014).

The leaf spot or blast disease, caused by Pyricularia grisea, has arisen as a serious disease of pearl millet (Rai et al. 2012). Screening of breeding and germplasm accessions of pearl millet led to the identification of resistant lines which can be further used for development of blast-resistance hybrids (Sharma et al. 2013; Goud et al. 2016). Six blast resistance genotypes (ICMB 97222, ICMB93333, ICMR 11003, IP 21187-P1 and ICMR 06222) of pearl millet were crossed with two susceptible genotypes (ICMB 89111 and ICMB95444) for inheritance study and their generations as well as backcrosses were screened against Magnoporthe grisea isolates Pg53 and Pg45 (Singh et al. 2018). Molecular markers are also being utilized for QTL identification against blast disease pathotypes. Two major QTL for blast resistance have been identified on linkage group 1 (LG1) and linkage group 6 (LG6) using SSR markers (Sanghani et al. 2018). Another study was conducted using Pennisetum species for improving abiotic and biotic stress tolerance in Pearl millet and the accessions were screened against different isolates of blast disease (Sharma et al. 2020a). Seeds of 305 accession of pearl millet from 13 countries were collected and screened for Magnoporthe grisea isolates Pg45, Pg53, Pg56, Pg118 and Pg119 and the accessions were classified based on level of resistance shown by genotypes. It has been observed that 182 accessions of pearl millet showed resistance against pathotype-isolates (Sharma et al. 2020b).

Rust (Puccinia substriata var. indica), smut (Moesziomyces penicillariae Bref. Vanky), and Ergot (Claviceps fusiformis Lov.) are important diseases causing grain yield losses that are proportional to their severity. Ergot and smut diseases are more severe to flowering stage of crop and are soil-borne. Several lines have been evaluated for screening of ergot disease (Abraham et al. 2019) but source of resistance against this disease was limited. Smut disease causes severe damage to crop and numerous lines were screened and identified, which showed resistance against this disease (Thakur et al. 2011). The advent of molecular markers emerged as a tool for dissecting QTL associated with resistance. A mapping population of 168 F7 RILs was used for the construction of DArT- and SSR-based linkage maps and screened for rust resistance. Three QTL on linkage groups 1, 4 and 7 were identified for pearl millet rust resistance explaining 58% of the observed phenotypic variation in rust reaction. Linkage group 1 (LG1), was novel QTL identified for rust resistance and is thought to confer a durable slow-rusting phenotype (Ambawat et al. 2016).

Abiotic stress tolerance

Drought tolerance

Drought, caused by low rainfall and its erratic distribution, adversely affect the growth and development of crop plants. In pearl millet drought tolerance has remained a strategic research issue, so pearl millet response to drought has been studied exhaustively. QTL for drought tolerance has been found to contribute to differences in photosynthetic pigments and ROS scavenging enzymes in pearl millet accessions. From studied QTL, APX activity was found to be increased in tolerant genotypes but the SOD and CAT activity remain unchanged. The presence or absence of drought-related QTL did not contribute to photosynthetic pigment molecules (Kholova et al. 2011).

Grain filling stage in pearl millet is the most sensitive stage to drought stress leading to reduction in grain size and grain number (Fussell et al. 1991). Pearl millet germplasm association panel was established recently and exploited for association mapping of drought tolerance traits. SNP in CoA carboxylase genes showed a significant association with panicle yield, grain harvest index and grain yield whereas an InDel was found to be significantly associated with grain yield and stay green traits under drought conditions (Sehgal et al. 2015). Four QTLs contributing to enhanced transpiration have been identified in accessions of pearl millet. Of these four QTLs, one QTL, contributing the majority of variation, was mapped on linkage group 6 (LG 6) whereas the QTLs for specific leaf weight and biomass was mapped on the linkage group 7 (LG 7) (Aparna et al. 2015). Under drought condition, Debieu et al. (2018) evaluated 188 inbred lines for the identification of QTLs associated with agronomic traits using genotyping by sequencing (GBS). After filtering of 3,168,971 unfiltered SNPs, 392,493 have been identified with an average density of 2.5 per 10 kb. Four marker-trait associations were identified on chromosome 6 for stay greens trait and two SNPs were found to be significantly associated with biomass production under early drought stress conditions. Out of the two SNPs identified for biomass production, one SNP was mapped between two predicted genes Pgl_GLEAN_10037359 and Pgl_GLEAN_10037360 whereas the second SNP was mapped being located between two predicted genes Pgl_GLEAN_10036946 and Pgl_GLEAN_10036945. It has been observed that early drought stress in lines led to reduction in grain and biomass production but limited changes were observed in grain weight (Debieu et al. 2018). The first report on validation of reference genes in pearl millet was given by Shivhare and Lata (2016), and result revealed two best reference genes whose specificity was confirmed by relative expression of PgAP2 like-ERF gene. This study can facilitate fastidious discovery of genes related to stress-tolerance (Shivhare and Lata 2016).

Transcriptomic analysis of pearl millet identified 6799 and 1253 differentially expressed genes (DEGs) in ICMB 843 and ICMB 863 respectively, and RNA sequencing for drought-responsive genes confirmed 7 genes using reverse-transcription PCR (Dudhate et al. 2018). Transcriptomic analysis conducted by Jaiswal et al. (2018) revealed 19,983 differentially expressed genes, 7595 transcription factors, and a hub of 45 genes having a regulatory gene network. Moreover, 34,652 putative markers, 4192 SSRs, 12111SNPs and 6249 InDels were reported and the results were validated using qPCR for 13 selected genes. Shivhare et al. (2020), identified 1129 DEGs on all the seven chromosomes of pearl millet except chromosome 4. The majority of genes were found to be present and mapped on chromosome 2 (196) followed by chromosome 3 (171), chromosome 5 (168), chromosome 6 (164), chromosome 7 (140) and chromosome 4 (108). A recent report on transcriptome analysis identified 2792 transcription factors and, 1223 transcriptional regulators, from which 315 transcription factors and 128 transcriptional regulators were expressed under drought condition and a total of 6484 genes for drought stress were identified using RNA sequencing and Pacbio-sequencing. The data were used as a reference sequence to examine the Illumina data of pearl millet (Sun et al. 2020).

In a recent study, Zhang et al. (2021) explored the mechanism of drought tolerance of pearl millet by comparing physiological and transcriptomic data under drought and controlled condition. It has been reported that during stress, a total of 12 genes were upregulated in which some genes are associated with drought stress in other species such as ADH1, FtsH, and CCCH. Additionally, genes namely SnRK2 and PP2C were found to have changes in their expression level that participate in ABA Signaling pathways (Zhang et al. 2021).

Heat tolerance

Reproductive and seedling stages are greatly affected by high temperature as the optimum temperature for normal crop growth is 33–34 °C. In southern and western Africa and India, the temperature of soil surfaces often exceeds 45 °C and may sometimes reaches up to 60 °C, causing poor plant growth because the pearl millet seedlings are more prone to high temperature during their first ten days of seedling (Peacock et al. 1993). Pearl millet has emerged as a remunerative and productive crop in western and northern parts of India (Yadav and Rai 2013). High temperature during flowering stage of pearl millet causes flower sterility, and ultimately leading to extreme reduction in seed set, lowering of grain yield (Gupta et al. 2015; Djanaguiraman et al. 2018). A clone from pearl millet heat stress-responsive EST database was used as a DNA probe, as it showed maximum homology to PgHsc70 gene, for screening of Pennisetum heat stress cDNA library using plaque hybridization method. Nucleotide sequencing of the 5′ flanking promoter region of the gene identified a heat-shock element and a protective activity was observed against the damage caused by heat stress (Reddy et al. 2010). Another gene, PgHsp90 consisting of three exons and three introns, was identified, characterized, cloned and the sequence was analysed for heat stress in pearl millet (Reddy et al. 2011). Nitnavare et al. (2016) reported a gene, PgHsp10 from pearl millet and characterized for heat stress using qRT-PCR analysis for gene expression in response to abiotic stresses with special reference to heat stress and it has been revealed that this gene consists of two introns and three exons. A recent report on transcriptome analysis identified 2792 transcription factors, 1223 transcriptional regulators, from which 318 transcription factors and 149 transcriptional regulators were expressed under heat stress condition and a total of 6920 genes for heat stress were identified using RNA sequencing and Pacbio-sequencing data as a reference sequence to examine the Illumina data of pearl millet (Sun et al. 2020). Recently, a study was conducted in pearl millet roots to explored the changes both at physiological and transcriptional level under heat stress. It has been observed that trehalose was accumulated in roots at 3 h to 7 h of heat stress. Additionally, POD activity increased gradually from 3 to 7 h of heat stress. At transcriptional level, HSFs, bZIP and bHLHs were main identified transcription factors expressed under heat stress. A total of 16 bZIPs, 7 HSFs and 18 bHLH genes were identified which were expressed differentially under heat stress condition (Sun et al. 2021).

Salinity tolerance

Salinity stress severely limit the crop production. The adverse effects of salinity on plants includes osmotic stress, oxidative stress, nutrient constraints and ion toxicity (Singh et al. 2021c; Shrivastava and Kumar 2015). According to a study, the reduced shoot nitrogen content and enhanced sodium and potassium content are related with salinity stress tolerance in pearl millet (Dwivedi et al. 2011). Various salinity stress-related genes have been identified in pearl millet but function of only some salinity-responsive genes such as PgDHN (dehydrin), PgNHX1 (Na + /H + antiporter), PgVDAC (voltage-dependent anion channel), and PgLEA (late embryogenesis abundant) have been studied (Agarwal et al. 2010; Reddy et al. 2012; Singh et al. 2015; Verma et al. 2007). To understand the mechanism of salinity tolerance at physiological and molecular level, pearl millet salinity tolerant (ICMB 01222) and susceptible (ICMB 081) line were subjected to de novo transcriptomic profiling. A total of 11,627 DGEs have been identified in both lines. In the tolerant line, 2965 unigenes were found to be upregulated whereas 2964 were downregulated whereas in susceptible line, 2243 unigenes were upregulated and 3473 were downregulated. Physiological analysis showed that soluble sugar content was higher in tolerant line under salt stress (Shinde et al. 2018).

Hybrid development in pearl millet

Hybrid development in pearl millet was initiated after recognition of cytoplasmic male sterile lines viz., Tift18A and Tift23A, and these lines being released as male sterile lines led to the development of hybrid breeding. The first pearl millet hybrid, HB1, was released by PAU in India in 1965 (Burton 1907, 1965; Burton and Athwal 1967). Genetic male sterility, caused by nuclear genes, has been identified in several crops like maize, rice, soybean and others, but little information is available in pearl millet (Yadav et al. 2010; Gupta et al. 2012). In pearl millet, genetic male sterility has been studied in male sterile lines, Vg 272 and IP 482, and the variation in the expression of sterility has been observed when crossed with different isogenic and non-isogenic lines, leading to an alteration in ms2 allele (Rao and Devi 1983).

Cytoplasmic male sterility (CMS) has been studied well in pearl millet and is used commonly for hybrid seed production (Yadav and Rai 2013). CMS systems in pearl millet were developed using genetic crosses, not protoplast fusion. For commercial hybrid production, this system requires male sterile (A-), maintainer (B-), and restorer (R-) lines in which male sterility is controlled by sterility factors and fertility restorer (rf) allele (Islam et al. 2015). Restriction fragment length polymorphism (RFLP) analysis in pearl millet identified five CMS cytoplasm that was different from each other because of the rearrangement of mitochondrial genes. The formation of A4, A1, A5, and Aegp CMS system was due to rearrangement of cox1 gene whereas the atp6 and cox3 gene alterations led to formation of A4 CMS system (Delorme et al. 1997). Additionally, two CMS sources 66A and 67A were identified as a genetic stock at PAU which were later named as A2 and A3 CMS sources (Athwal 1965). Subsequently, various other CMS sources have been reported by different centers in different genetic stocks. Gero, a CMS source, was identified by Ibadan Nigeria (Aken'Ova and Chheda 1981), and two novel sources, PT732A and A5, Aegp, were discovered from genetic pools and genetic sources of ICRISAT (Appadurai et al. 1982; Rai 1995). Moreover, a new CMS source ex-Bornu, a cross between wild relatives of pearl millet and landrace from Senegal and A4, was identified which is different from the existing CMS sources and is utilized for hybrid seed production (Govindaraj et al. 2019).

Conclusion and future prospects of pearl millet breeding

With an increasing population, global demand for feed, energy and food is increasing day by day thereby posing an opportunity for exploitation and development of sustainable food for various end uses. Pearl millet is an important cereal crop with immense potential but breeding has been lagging, compared to other cereals. Genomic approaches should be encouraged in pearl millet breeding programs. Genomics could be helpful for a better understanding of genetic and genomic insights of pearl millet by investigating the genetic diversity present in the wild species or germplasm accessions (Fig. 3). Moreover, populations for functional genomics, such as mapping populations, natural diverse panels, molecular modules, GWAS and genetic engineering could help dissect the useful variations present in the population (Singh et al. 2021b). Targeted genes/genomic regions/alleles, through introgression and breeding selection, can be replaced in the elite varieties via screening of breeding lines or genetic stocks with the help of genomic selection and functional genomics. During the past decades, considerable progress in pearl millet molecular breeding and genomics has been made, buy still more work is needed to comprehensively design pearl millet as a multi-purpose crop.

Fig. 3

Breeding scheme for pearl millet improvement using the Cutting-edge breeding technologies and systems. Identification of changes during diversification and domestication, characterization of genomic variation, pre-breeding material selection, and genomic assistant introgression of traits in the improved lines, are the four major components in a breeding programme

Firstly, attention should be paid to the validation and identification of genes controlling important agronomic traits, especially molecular modules, which are emerging topics in the post-genomic era. For Example, flowering time and plant height are the two important traits of pearl millet and major genes and QTL have been identified controlling flowering time and plant height but their association between stay green and grain size is poorly understood. Brown midrib in pearl millet is associated with decreased lignin content and this trait offers greater palatability and digestibility (Sattler et al. 2010) but QTL associated with brown midrib are unknown yet. Moreover, purple foliage of pearl millet is controlled by three alleles Rp1 Rp2, and rp (Hanna and Burton 1992) and the association analysis of pearl millet genotypes revealed that the foliar color locus was mapped in the linkage group 4 (LG4) (Azhaguvel et al. 2003). More QTL controlling brown color foliage should be mapped from the wild relatives, elite varieties or germplasms of pearl millet. The long bristle of panicle is an important trait providing an advantage in deterring birds feeding on grain. It has been reported that the bristle panicle trait is controlled by the dominant genes but none of the genes have been identified thereby proving an opportunity to identify the genes and map them in the linkage groups. In future, in-depth examination should be conducted for unravelling the genetic variations, phenotypic variations, and characterization of novel alleles for agronomically important genes.

Secondly, more pearl millet genome sequences are needed because their utilization and exploitation in pearl millet are far from enough when compared to wheat, maize and rice genomes. Sequencing of landraces, wild relatives and improved cultivars of pearl millet will provide novel genomic variability for the study of pearl millet diversification and domestication of various end -uses. As a multipurpose crop, pearl millet has a unique form of evolution. Therefore, domestication is a good model for evolutionary study purposes. The genetic basis of complex agronomic traits is essential to understand various end-use pearl millet improvements. Single genome sequence does not represent the whole genomic structure of a species. But pan-genome analysis, collecting genes at clad level, provides an opportunity to identify the genetic variability present in the whole genome with the help of sequencing of multiple individuals of a species.

Finally, it is necessary to consolidate the genome-based tools and technologies in pearl millet breeding programs thereby provide an opportunity for the development of “Super pearl millet” with key traits. The most common input traits are to improve the resistance against abiotic (drought, heat, and salt) stresses and biotic (diseases and insect pests) whereas output traits include grain yield, grain quality, flowering time, and plant height. Precision molecular breeding of pearl millet should be implemented based on different traits viz., pearl millet grain is characterized by low lignin content for easy digestibility and palatability and biofortification of pearl millet for high nutrient content. The future breeding goal is to enhance the protein and starch content and reduce lignin content. Improvement in the low crude protein, high starch content, and reduced tannin content is the major breeding objective of forage pearl millet breeding whereas increasing the tillering, multi-cut, and rapid growth should also be considered in the breeding program.

Moreover, modern emerging approaches can accelerate the breeding cycle of pearl millet and encourage breeders to develop superior varieties for various end uses but still, there is a long way to go. Identification and pyramiding of supper alleles controlling agronomically important traits using molecular modules at genome level is still a task for pearl millet breeding. Genomic selection should be carried out for genotyping and phenotyping of pearl millet accessions and the model should be selected according to the type of population and gene-environment interactions. Genome editing, an emerging tool in the genomic era, plays a significant role in the identification of genes and their function for crop improvement but there is lack of published work in pearl millet breeding. The application of this system is greatly relying on the transformation efficiency and is still much lower in crops like sorghum than those of other major crops. Optimization of the transformation system is essential to advance the CRISPR/Cas9 application in pearl millet breeding. CRISPR/Cas9 can be used to control the expression of miRNA/genes through genetic modifications for crop improvement. Thus, there is a great scope for research on pearl millet in reducing anti-nutrients, thus making it further enriched cereal.

The journey of plant breeding has significantly transcended from the large resource-intensive field trials to the molecular level. With the dawn of molecular markers, several techniques have found their way towards population improvement. Although tremendous progress has been achieved in the identification of genetic loci underlying various agronomically important traits, the epigenomes, pan-genomes, and other disciplines should be integrated for dissection of genetic diversity and the superior alleles underlying complex agronomic traits. Moreover, efficient breeding techniques should be incorporated with the novel breeding approaches to bring beneficial changes to the genetic breeding of pearl millet