Global nitrogen (N) demand, one of the most expensive farm inputs currently stands at about 117 million metric tons with a projected annual increase of ∼1.5% in the future (FAO, 2019). Indian agriculture consumes over 17 million tons of N fertilizer per year. However, plants can utilize only 30–40% of the applied N for food production and the remaining (up to 60%) is lost to the environment by leaching, de-nitrification, and runoff. Surplus nitrogen pollutes freshwater streams and air which is hazardous to the majority of living species (Hickman et al., 2014; Russo et al., 2017).

Besides, excess usage of N fertilizer not only decreases the efficiency of nutrient use but also affects the rate of economic returns per unit of chemical fertilizer applied. The effect of negative environmental and economic impacts could be reduced through better agronomic practices and also by utilizing N efficient lines with improved nitrogen use efficiency (NUE) (Raun et al., 1999). Hence, improving the NUE of crop plants could help in reducing fertilizer input, increased productivity and profitability coupled with a reduced negative impact on the environment.

Globally, pearl millet (Pennisetum glaucum (L.) R. Br) widely grown for food and fodder is considered as the sixth most important cereal crop in terms of area. It is one of the oldest cultivated cereal crops, originated in Africa but later spread to many countries (D’Andrea and Casey, 2002; Manning et al., 2011). This millet crop is vital for food and nutritional security for the world’s poorest people living in different agro-ecological zones. Besides, pearl millet has a rich nutritional profile, wide genetic diversity, high photosynthetic rates being a C₄ and also provides a healthy balanced diet, thus contributes to the economic security of poor farmers (Srivastava et al., 2019). In the Indian agriculture scenario, pearl millet is grown by poor and marginal farmers under low fertile and rainfall dependent areas often facing drought during different stages of crop growth. From the past few years, pearl millet demand has been increased for its nutritional characteristics and its adaptability to a wide range of climatic conditions (Tako et al., 2015). Since then farmers started applying a high dose of N fertilizer for maximizing grain yield (GY), but excessive usage of N fertilizer leads to low NUE of the crop along with various environmental hazards. In some parts of India, poor farmers lack knowledge about the ideal dose of fertilizer to harvest the real yield potential. Hence, yield improvement of pearl millet under low nitrogen input is indeed beneficial for economic and environmentally sustainable cultivation.

Nitrogen use efficiency is a complex trait, which is associated with various morphological, physiological, molecular, and biochemical changes in plants throughout the life cycle. For a clear understanding of this complex nature, studies of various physiological traits and their close correlation with one or more economically important traits like GY is foremost critical. Nevertheless, it will help in selecting low N tolerant/high yielding lines at different N conditions (Monostori et al., 2016). In general, plant function is always associated with chlorophyll content, which directly indicates the N status of the leaf (Yang et al., 2014). Leaf chlorophyll content and photosynthetic capacity are appropriate benchmarks for identifying high NUE (HNUE) genotypes under low N conditions during field trials (Vijayalakshmi et al., 2015; Kiran et al., 2016). Leaf N status was usually measured by using a hand-held optical chlorophyll meter to monitor the leaf nitrogen status and chlorophyll content. Moreover, leaf area (LA) is also one of the important physiological traits, plays a critical role in enhancing plant biomass (Gianquinto et al., 2004; Monostori et al., 2016).

Nitrogen use efficiency is defined as the ability to produce GY per unit N available in the soil. NUE mainly depends on the results of two main processes, such as N uptake efficiency (NupE), and nitrogen utilization efficiency (NutE) (Good et al., 2004; Hakeem et al., 2012; Vijayalakshmi et al., 2013). NupE is the ability of the plant to take up N from the soil and NutE is the ability to use N to produce gain yield (Ladha et al., 1998; Hirel et al., 2007). Studies on genotypic variations under low and recommended N for NUE traits at both seedling and maturity stage of plants under controlled and field conditions in various crops resulted in the identification of high NUE genotypes possessing high yield sustainability under low N condition (Muchow, 1998; Inthapanya et al., 2000; Le Gouis et al., 2000; Presterl et al., 2003; Anbessa et al., 2009; Namai et al., 2009; Vijayalakshmi et al., 2015). The high NUE/N-insensitive genotypes (NIS-top grain yielders) are defined as genotypes that give more or equal GY with minimal application of N fertilizer when compared to recommended or standard N fertilizer conditions (Hawkesford, 2017). The understanding of GY and its associated NUE traits performance is still largely lagging in the pearl millet diversity panel. However, few studies had explored genotypic variations for NUE at different N levels. In one such report, 20 diverse pearl millet genotypes and few high-yielding hybrids were screened under field conditions at two N levels (Alagarswamy and Bidinger, 1987). In another study, pearl millet hybrids were grown over 3 years in western and eastern Nebraska at four different N levels to determine the optimum N rate for cultivation and concluded that the NUtE trait was less responsive than the N uptake with increased N levels (Maman et al., 2006).

Nitrogen use efficiency, which is dependent on soil or external N supply, is an output of available N uptake, its efficient utilization and remobilization to grain at end of the season. Several studies revealed that improved NUtE might lead to enhanced NUE under low N conditions, particularly in cereals (Moll et al., 1982; Good et al., 2004). NUtE is defined as the genotype ability to assimilate and remobilize N which ultimately convert into GY. It is an essential physiological parameter that unveiled the positive relationship with GY. Whereas in pearl millet comprehensive studies are required to pinpoint the critical factors underlying NUtE under different N levels. GY is a complex trait controlled by a network of multiple traits and their associations. The GY in pearl millet is a result of many yield components, such as grain number, grain weight, tiller number, and panicle number (Rai et al., 2012; Basava et al., 2019). Hence, uncovering the genetic basis of GY, and other related NUE traits under low and high N conditions is a prerequisite to understand the mechanism and also to identify ideal NUE associated traits for selection. Thus, the current study was aimed to evaluate a diverse set of pearl millet lines for their response to GY and NUE related traits under different N levels and also to determine the suitable NUE traits for selection. This study provides useful information toward uncovering the physiological and genetic basis of GY and its related traits under low N which further facilitates the development of low N stress-resilient pearl millet cultivars.

According to the fertilizer map, the available nitrogen (N) content of the soil sample was divided in to low, medium, and high. Therefore, low N conditions were observed when soil nitrogen content is below 280 kg/ha (critical limits of the soil maintained as per ICRISAT guide limits). The selected experimental field was not used for any cultivation from the past 5–6 years and before starting the trial, soil samples were collected across the field and undergone soil nutrient analysis. Nutrient analysis (Supplementary Table 1) revealed 0.5% organic carbon (OC), indicating a low available N in the soil and the basal N availability was common across all the N regimes before the initiation of the experiment. In addition, the soil micronutrient availability was in the medium range. The basal N availability in the soil is common for all the treatments but coming to the N₅₀ and N₁₀₀ regimes, an additional N source in the form of urea was applied as per the protocol mentioned in the methodology section. Overall, the N₀ regime was maintained strictly without any additional supply of nitrogen.

A total of 380 pearl millet genotypes including the diversity panel, parents of mapping populations, and checks were screened for 11 physio-agronomic and NUE traits at three nitrogen levels (0, 50, and 100% of the recommended N doses) with two replications. Data recorded were averaged across the three seasons for each trait in each N level. Results revealed the native variation across the genotypes toward N response which had given the scope to identify nitrogen use efficient lines under low and high N conditions for the marginal and favorable ecologies, respectively. ANOVA results revealed significant variations among the genotypes, treatments and their interactions (season × treatment (N levels), season × genotype, treatment × genotypes, and season × treatment × genotype). The pooled means over three seasons in each genotype in each N level for five physio agronomic traits were provided in Supplementary Table 2. The pooled mean of over two seasons for six N related traits in each genotypes were provided in Supplementary Table 3.

Nitrogen (N) is an essential macronutrient required for plant growth and is often a limiting factor for crop yield. From the past 50 years of agriculture, augmented food production was attained by the extensive application of N fertilizer in combination with N-responsive cultivars across the globe. Excessive application of N fertilizer is becoming expensive which accounts for the loss of economic profit to the farmers along with negative impacts on the environment (Raun et al., 2002; Hawkesford and Griffiths, 2019). Hence to overcome this problem, a clear understanding of genotype behavior, identification and development of genotypes (without compromising GY) with high NUE under low N conditions is a paramount need for improving NUE.

Therefore, the present study aimed to identify the N-Sensitive (NS) and N-insensitive (NIS) genotypes and also to determine the traits regulating low N tolerance. However, accurate and reliable phenotyping under low N input is challenging and influenced by the genotype (G), environment (E), and G × E interactions (Chen et al., 2014; Rao et al., 2018). Moreover, there are limited studies available in pearl millet toward the identification of nitrogen insensitive genotypes under low N conditions at field level. Based on the current knowledge, this study presents the first report of genetic variations for NUE in the global association panel, Pearl Millet inbred Germplasm Association Panel (PMiGAP) (PMiGAP collected from a large set of diverse germplasm and breeding lines) and from mapping population parents. Notably, very few pearl millet breeding programs are targeting the development of low N tolerance traits which are a must for sustainable agriculture with minimal negative impacts on the environment. Components of NUE were studied in pearl millet hybrids at two N levels found that the efficient genotype (Souna B) had a 32% of higher NUE value than the N inefficient Indian genotype (BJ104) (Alagarswamy and Bidinger, 1987). Another study revealed that NUtE contributes more to genetic variation in NUE (Maman et al., 2006).

In this study, a total of 380 diverse pearl millet genotypes were characterized for SPAD chlorophyll content, LN, LA, GY, DSY, and NUE related traits at three different N levels under field conditions. Notably, diverse responses have been observed among the genotypes across N levels, despite a similar growth conditions and an equal amount of N fertilizer in a given N level. These observed genotypic variations purely reveal the genotype plasticity toward traits. This set of lines used in this study which are collected from different regions across the globe and these lines were used for various abiotic stress studies to identify the tolerant genotypes. Previous studies in rice, maize, wheat, and oilseed rape, etc., have established significant genetic variation for NUE-related traits with large germplasm panels, hybrids, open-pollinated and recombinant inbred line populations (Chen et al., 2014; Li et al., 2015; Vijayalakshmi et al., 2015; Ertiro et al., 2017; He et al., 2017; Rao et al., 2018). Generally, the results of ANOVA explained statistical differences among the treatments and genotypes and their interactions. In the current study, ANOVA results revealed significant variations among the genotypes and treatments for all measured traits except a few traits (Tables 1, 3). These results were indicating that the environment under different N levels was a crucial factor in explaining the genotypic variance of GY and its related traits, which are strongly influenced by the relative contribution of G × E interactions under field evaluation. The results concur with other reported field experiments in rice (Srikanth et al., 2016) and wheat (Sial et al., 2005; Belete et al., 2018).

In the present study, SPAD chlorophyll content significantly increased with nitrogen application, which might be sufficient availability of N in the leaf. Similar results of increased SPAD chlorophyll content by increasing N application was reported by Kitajima and Hogan (2003), Pramanik and Bera (2013). The relation between the slow or controlled release of N fertilizers supporting more N absorption and related physiological mechanisms has been well studied in different crops (Li et al., 2003; Long et al., 2013). LA of all the pearl millet genotypes was significantly increased with N application which might be attributed to translocation of N to leaves, which brings variation in plant architecture and leaf internal structure (Singh et al., 2004; Mhaskar et al., 2005). Reduction in LN under N₀ condition was observed, due to deprivation of N, thereby reduced source size and hindering the plant development.

Across the year on average, GY ranged from 9.64 g to 159.71 g (N₀), 20.72 g to 236.43 g (N₅₀), and 20.48 g to 242.12 g (N₁₀₀), indicating the genotypic variability in GY at different N regimes. In general, GY increased was correlated with the N application rate, which might be due to sufficient nitrogen availability. Recently similar kind of results was reported in a diverse set of foxtail millet genotypes (Bandyopadhyay et al., 2020). In wheat, GY and quality traits were evaluated by Šarčević et al. (2014) at low and normal N and found that GY was reduced by 10% under low N condition compared to normal condition. The current results also are in tune with the earlier reports as there was an increased GY with increase N levels. Hawkesford and Griffiths (2019) reported that split application of N is the best option to recover the maximum applied N in the form of harvested grain/increased GY and found a strong significant relationship between GY and physiological traits.

Remarkably, 25 high yielding genotypes were identified under the N₀ condition and considered as N-insensitive (NIS). Out of 25, nine genotypes (IP10820, IP17720, ICMB01222-P1, IP10379, ICMB89111-P2, IP8069, ICMB90111-P2, ICMV-IS89305, and ICMV221 = ICMV88904) are common in the top performing 25 lines under N₁₀₀ condition, which shows the genotype plasticity toward varying N levels in these genotypes. These selected genotypes are good genetic resources to breed for tolerance to low N conditions. Top 25 high yielding genotypes in N₁₀₀ can be considered to have the best acceptances for cultivation where soils are fertile and when followed the ideal N levels for cultivation. These identified genotypes could be used to improve GY, along with higher NUE. Numerous studies have reported utilization of N efficient lines with enhanced GY in the farmer fields which may help to reduce fertilizer input as well as increase the profitability of farm operations (Würschum, 2012; Vijayalakshmi et al., 2015; He et al., 2017). Likewise, selected NIS lines with high NUE will certainly play a role in reducing environmental pollution and could increase economic profit to farmers.

Pearl millet is an ideal fodder crop for feeding livestock and has unique features like dry season crop, production environments having low nitrogen, high photosynthetic efficiency, and high dry forage capacity (Govintharaj et al., 2018). Several reports depict the dry fodder demand and it would require approximately 568 million tonnes across the globe by the year 2030 (Govintharaj et al., 2018). To tackle this problem, high fodder yielding genotypes need to be identified. Generally, most of the genotypes have lower dry matter production under low N conditions, hence identification of genotypes with high dry matter production under low N conditions is important for sustainable fodder production. In this framework, the top and least performing 25 genotypes for DSY were identified under N₀ and N₁₀₀ conditions (Table 4).

These contrasting genotypes can be further exploited in the breeding program and molecular studies to develop varieties with high NUE for biomass. Irrespective of N, the common nine genotypes (IP14398, IP10579, IP15857, IP10394, IP17125, IP13608, IP6098, IP12020, and IP11584) were identified and can produce more dry matter production under both N₀ and N₁₀₀ conditions. These genotypes need further physiological and molecular characterization under low nitrogen conditions to understand the molecular basis for low nitrogen tolerance and biomass production. Overall, observed genetic variability in pearl millet genotypes are suited well for improving DSY, can be utilized in the breeding program to enhance livestock productivity. Studies on various crops to determine the relationship between nitrogen levels and biomass traits have shown that N fertilizer application increased the biomass related traits, including DSY (Chan-Navarrete et al., 2014). The present results were tuned with these observations as DSY was more in N₁₀₀ condition than in N₀ condition; the increasing quantity was up to the significant levels. Another study revealed wide genetic variations for biomass traits and stover nitrogen in pearl millet germplasm (Gupta et al., 2015). Few studies revealed that significant genetic variation present in pearl millet for biomass traits in dual-purpose hybrids and open-pollinated varieties (OPVs), populations, and top cross hybrids (Bidinger et al., 2010; Blümmel et al., 2010; Rai et al., 2012). More importantly from this study, dual-purpose genotypes were identified and these can be used as a reference set for the developing high grain and stover yielding lines.

In the present study, a total of 11 traits were recorded for all three seasons. However, due to heavy rain, we are unable to generate quality data for the rainy season. Hence to avoid misinterpretation of results/findings, data for only six NUE traits were pooled from two summer seasons and presented. Moreover, the NPG was slightly increased with the nitrogen application and genetic differences for the traits were also observed. These results are in tune with the report of Belete et al. (2018), who found the higher NPG with increased N levels. NPG in N₀ was reduced by 3.6 and 11.4% compared to N₅₀ and N₁₀₀, respectively. Whereas, in N₅₀ condition, the mean NPG decreased by 8.24% as compared to N₁₀₀. These results are similar to the finding of Lopez-Bellido et al. (2004), Arduini et al. (2006) who found the genetic variations among wheat genotypes for grain N concentration and increased NPG with increased N levels. In the case of the NPS, the trend was contrasting to that of NPG. Herein, NPS was slightly more in N₀ as compared with N₅₀ and N₁₀₀ but was not up to a significant level. These results concur with the earlier reports of Belete et al. (2018) who found higher NPS with increasing N levels.

Nitrogen utilization efficiency is defined as the genotype ability to assimilate and remobilize N ultimately to produce the GY (Rengel and Graham, 1995; Fageria et al., 2008). Determination of the genetic variations in NUtE is essential for the selection of efficient genotypes and can be used further in breeding programs to develop low N tolerant material. The concept of genotypes grouping is used widely in nutrient use efficiency (Fageria and Baligar, 2003). Based on NUtE efficiency data in N₀ versus N₁₀₀, genotypes are classified into four groups viz., N efficient non-responsive, N efficient responsive (NER), N responsive inefficient, and N inefficient non-responsive under N₀ and N₁₀₀ condition, respectively. In N₀ NUtE data, a total of 198 efficient and 176 inefficient genotypes were identified and out of 198 efficient genotypes, 58 genotypes were falling in the first desirable group which is N efficient non-responsive. These genotypes are exhibiting a progressive performance under low N. This may enable breeders to develop efficient genotypes under low input environments for pearl millet breeding activities. The remaining 140 genotypes were falling in the next most desirable group, NER and these genotypes were exhibiting a progressive response to increased N availability. The NER genotypes identified from the present investigation could be the prospective targets for selection toward the genetic improvement of pearl millet for N utilization. Interestingly, in the identified 25 NIS genotypes, 20 genotypes which are falling under N efficient responsive group were showed progressive performance in terms of efficient and responsive use of nitrogen. Genotypes with more NUtE could produce high GY per unit of N consumption (Moll et al., 1982). Various studies on NUtE have already reported that breeding for efficient genotypes under low N could be achievable with high NUtE (Fageria, 2014; He et al., 2017). The third group, responsive inefficient genotypes, can be used in breeding programs. The rest of the genotypes fall into the fourth group and these are less desirable from the NUE point of view. Interestingly, few common genotypes were identified from the non-responsive and inefficient genotypes. Overall, efficient genotypes are higher in the utilization of absorbed nutrients than inefficient genotypes. Vast genotypic variations in NUtE have been reported under field/pot screening in various wheat genotypes and other crops by several researchers (Bouchet et al., 2014; Ma et al., 2015).

The nitrogen harvest index indicates the level of efficiency of plants to use acquired total N for grain formation (He et al., 2017). In the present study, significant genotypic variation was observed for NHI at both N₀ and N₁₀₀ conditions. NHI was positively correlated with NUtE and GY in all three N levels. These findings concur with previous studies in oilseed rape (Ulas et al., 2013; Stahl et al., 2016) and wheat (Monostori et al., 2017).

In the current study, the Pearson correlation coefficient analysis for the GY and NUE related traits indicated that NUtE trait has the largest contribution to the GY followed by NHI and TNUp across N levels, implying that GY improvement in pearl millet could be possible by selecting genotypes that attain a higher NUtE and NHI across N levels. Similarly, a significant correlation between GY and NUtE was reported in maize, wheat, and oilseed rape (He et al., 2017; Belete et al., 2018). Fageria et al. (2010) reported that strong associations between GY and NUE related traits could be a better option for GY improvements under limited N condition. Furthermore, GY was negatively correlated with NPG and NPS across N levels. Previous studies also reported a similar significant negative correlation between grain N concentration and GY (Sinebo et al., 2004). In the present study, a strong positive correlation of DSY with NUpS and TNUp across N levels was observed indicating that more DSY was with higher NUpS and TNUp, where it was negatively correlated with NPS, NUtE, and NHI. The inverse relationship between NPS and DSY was reported in rice (Subudhi et al., 2020). Identified genetic stocks will be further utilized to carry an in-depth investigation to understand the genes and associated pathways related to NUE. Overall, important correlations identified from this study, will certainly help for the future pearl millet NUE breeding programs.

Improving NUE of pearl millet is pivotal for sustainable crop growth and yield especially under low nitrogen soils. Likewise, improving crop productivity using N fertilization is important for achieving climate resilience. Nevertheless, the genetic improvement of pearl millet NUE depends on the nature and extent of variation among the germplasm. As studies on pearl millet NUE are still nascent, this study was aimed to derive morphologic and agronomical traits associated with NUE in a set of 380 diverse lines under different N levels. The first step in this research revealed extensive genetic variations across the diversity panel and mapping population parents under N₀ and N₁₀₀ conditions. Also, large environmental variations (including N inputs) for GY and its related NUE traits were observed. Nitrogen limitation resulted in the reduction of GY and DSY in N₀, as compared with N₅₀ and N₁₀₀. Under low N, the best grain yielding genotypes were identified and considered as nitrogen-insensitive (NIS). Genotypes IP10820, IP17720, ICMB01222-P1, IP10379, ICMB89111-P2, IP8069, ICMB90111-P2, ICMV-IS89305, and ICMV221 (ICMV88904) proved to be the most efficient genotypes in terms of GY at low and high N levels, and indeed shows their inherent genotypic plasticity toward N application. Furthermore, the use of NIS genotypes will help in breeding N-efficient genotypes for arid and marginal agro-ecologies. Overall results suggest that genotypes with more yield and high to moderate NUtE can be used as parents for the breeding of N efficient genotypes. The lines identified from the present study, coupled with multi-omics technologies can help to identify candidate genes for NUE in pearl millet. The available genetic stocks will be useful to carry an in-depth dissection of the genes and pathways related to NUE. These contrasting genetic stocks may help to map the genome, transcriptome, proteome, and metabolome signatures of the traits leading to a better understanding of NUE in pearl millet. These current findings may also help the farmers for optimizing the use of fertilizer inputs for economic and environmentally sustainable food production.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

RKS and RG planned and designed this research. VP, MP, SB, RR, and VT performed the field experiments, phenotyping, and analyzed the data. AR, RD, and GA carried out statistical analysis. RKS and RG interpreted the data. VP, MP, RKS, and RG wrote the manuscript. All the authors have made their contribution in editing the manuscript for publication.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.