Identification of yield contributing traits and genotypes to drought tolerance in finger millet (Eleusine coracana L. Gaertn.)

Abstract

Screening of germplasm for specific traits is a continuous pre-breeding process in deriving the drought-tolerant donors required for crop improvement. The study evaluated 17 medium-late duration finger millet genotypes under drought stress (DS) for 28 days during the reproductive stage to identify the traits and genotypes for drought tolerance using different statistical analysis. The photosynthetic rate (by 26.3%), stomatal conductance (by 26.4%), transpiration rate (by 24.8%) and grain yield (by 13.2%) were decreased and found sensitive to DS, but the leaf temperature was increased (4.7%). From the path analysis and multiple linear regression analysis, the mean ear weight and productive tillers were found to contribute to the grain yield significantly under well-watered conditions. While under DS conditions, the mean ear weight, productive tillers and threshing percentage equally contributed to grain yield. Based on the ranking of traits significantly contributing to grain yield, the genotype GE-4683 with a higher mean ear weight (10.65 g) was found superior to the popular variety, GPU-28. The Multiple linear regression equation predicts the possibility to increase the yield of GPU-28 under DS from the existing 360.0 to 459.5 g per square metre (by 29.1%) by the incorporation of three productive tillers instead of the existing two tillers per plant in the MLR equation. An additional 1.0 g of mean ear weight will be able to predict an increased grain yield from 360.0 to 392.0 gm⁻², equivalent to 3.60 to 3.92 t/ha (by 9.4%).

Introduction

Amongst the abiotic stresses, water deficit stress during crop growth is one of the major constraints for crop productivity under rainfed conditions in arid and semi-arid regions (Zougmore, 2018; Krishna et al., 2021). Besides, in the changing climate scenario, the frequency of drought episodes is increasing and expected to increase further in the semi-arid regions of India and Africa with an irregular distribution of rainfall and a lesser number of rainy days during the monsoon season (Dash et al., 2009; Anon, 2019; Jalihal et al., 2019). With climate change, finger millet yield reduction of 3.6% (2021–2050) in India is projected (Rao et al., 2022). For such a climate change scenario of water deficit environments, finger millet is relatively most suitable with the advantage of suitability for delayed sowings, poor soils, intercropping and maintaining the biodiversity of crops (Pande et al., 2016).

Millets include pearl millet, foxtail millet, proso millet, finger millet, kodo millet, little millet, barnyard millet, tef and brown-top millet. Of these, finger millet constitutes 12% of the global millet area and is cultivated in more than 25 countries in Africa and India (http://exploreit.icrisat.org/profile/smallmillets/187; Vetriventhan et al., 2016). In 2018, the global total millet area was 33.49 m ha with a production of 31.74 m t, and in India, it was 9.22 m ha and 11.63 m t (http://fao.org). In 2019–20, the area of finger millet in India was 0.959 million hectares, with a higher productivity (1390 kg ha⁻¹) than the major rainfed crops, sorghum (998 kg ha⁻¹) and pearl millet (1237 kg ha⁻¹); thus, finger millet is a suitable crop for rainfed areas (http://www.indiaagristat.com). Additionally, the finger millet grain is superior in protein, calcium, iron and fibre with a lower glycemic index than wheat and rice (Chandrashekar, 2010; Hiremath et al., 2018; Nanja Reddy et al., 2019a).

Finger millet, a C₄ NAD-ME species (Ueno et al., 2006), is known for its drought tolerance over rice, wheat and maize, and yields relatively better where other major crops fail to yield substantially (Tadele, 2016). It is cultivated as a rainfed crop in >90% of the finger millet area (Davis et al., 2019) with intermittent rain-free periods during crop growth, and the drought stress (DS) decreases the grain yield. For instance, in India, a 16.4% reduction in total finger millet production was observed in 2018–19 as compared to 2017–18, because the contribution of production from Karnataka towards Indian production was decreased during 2018–19 (http://www.indiaagristat.com). The reduction in grain yield (22.3%) in Karnataka during 2018–19 has coincided with rainy days of only 10 as compared to 25 rainy days in 2017–18 between flowering and grain filling stage (Anon, 2019). In this context, utilization of available genetic resources for drought adaptation could be one of the easiest approaches to develop new genotypes suitable for rain-fed areas (Bharathi et al., 2013). Among the different crop growth stages, the reproductive period is most sensitive to DS (Maqsood and Ali, 2007). Therefore, the objective of this study was to examine the performance of elite germplasm lines under DS for 28 days (18 days prior to flowering up to 10 days after flowering) for the identification of important traits contributing to grain yield and better genotypes than the popular variety, GPU-28.

Material and methods

Management of the experimental site and crops

A field experiment was conducted at the Project Coordinating Unit, All India Coordinated Small Millet Improvement Project, University of Agricultural Sciences, GKVK, Bengaluru, situated at latitude 12°58¹ North, longitude 77°35¹ East at an altitude of 930 metres above the mean sea level. The experiment was laid out in a split-plot design with control (WW: well-watered) and DS as main treatments, using 17 medium-late duration genotypes (75 to 80 days to 50% flowering) as sub treatments in three replications, and the treatments were imposed on the 54th day after sowing, DAS). Each plot had 3 rows of 3.0 metres length each. Direct seed sowing was done at a spacing of 30 cm between rows and 10 cm between plants on 17 August 2009. Thinning was performed twice within 15 days after sowing (DAS) to maintain a single seedling per hill. The organic matter in the form of farm yard manure (7.5 tonnes per hectare) was incorporated into the soil 15 days prior to sowing. The recommended dose of 50:40:25 NPK (kg ha⁻¹) in the form of urea, single super phosphate, and muriate of potash was applied in a split dose. Half of the recommended dose of nitrogen and a full dose of P and K were given as basal doses at the time of sowing, and the remaining 50% N in the form of urea was applied as top dressing at 40 DAS. Manual weeding was undertaken twice within 30 days of transplanting.

Treatment imposition

Up to 54 days after sowing, the crop was raised normally as per the recommended package of practices, with no differences between the treatments. The crop was raised with weekly protective irrigation to a field capacity of approximately 25% soil moisture to meet the average evaporation demands of 4–5 mm/day. Irrigation was stopped on one set of plots from 54th to 82nd day after sowing, and this was known as DS treatment. Another set of plots were continued with irrigation, and this was called the well-watered (WW) treatment. The DS period lasted from 18 days before flowering to 10 days after flowering, and there was no rain during this time.

Data gathering and analysis

At the time of 50% flowering, the leaf area and dry matter were recorded. Destructive sampling was done by harvesting five plants continuously in a 0.5-meter row length (MRL) to record the leaf area and dry matter. For leaf area determination, randomly selected 15 leaves in each replication were measured for length × width × 0.75 (factor for leaf shape) to arrive at the sample leaf area, and such leaves were oven-dried at 70°C till the constant dry weight was reached. Then, total leaf area per 0.5 MRL was computed (leaf area of sample/dry weight of leaf sample) × total leaf dry weight per 0.5 MRL. The leaves plus stem with emerging ear-heads from 0.5 MRL were dried to a constant dry weight for calculating the dry matter at flowering (DMF). The gas exchange parameters, namely photosynthetic rate (Pn), stomatal conductance (gs), transpiration (T) and internal CO₂ concentration (Ci), were measured in three randomly selected plants using an IRGA (LiCor-6400) between 11.00 AM and 11.45 AM during the DS period (18 and 19 days after DS imposition) in the second or third matured leaf from the apex in each plant. At crop maturity, yield and yield parameters were measured from the net plot area leaving the border plants, and values were converted to 1.0 m² area. The dataset was statistically analysed for ANOVA in split-plot design, and path-analysis using OPSTAT (Sheoran et al., 1998). Furthermore, using Microsoft Excel Toolpak, stepwise multiple linear regressions (MLR) was performed to identify significant traits contributing to grain yield. In stepwise regression, non-significant parameters were eliminated one by one to obtain the significant traits contributing to grain yield. Considering significant parameters, genotypes were ranked for their drought tolerance. The possible yield improvement of popular Cv. GPU-28 was calculated using the MLR equation by incorporating the values of productive tillers or mean ear weight of superior genotypes into the MLR equation.

Results

The effect of DS on dry matter accumulation during flowering

DS for 18–20 days prior to flowering, significantly reduced the leaf area index (LAI by 8.8%), leaf area duration (LAD by 8.57%), and dry matter production (DMF by 9.5%). The mean DMF/LAD (an indirect measure of net assimilation rate) was not affected by DS (Table 1; Fig. 1). However, significant genotypic differences were observed for LAI, DM/LAD and DMF (Table 1). The germplasm accession, GE-4995, had higher LAI (3.80 in WW and 3.79 in DS), and DMF (1150 gm⁻² in WW and 1005 gm⁻² in DS) than the popular variety GPU-28 (LAI of 2.65 in WW and 2.56 in DS; DMF of 561 gm⁻² in WW and 571 gm⁻² in DS). The LAI had a significant positive relationship with DMF (r = 0.774** in WW and 0.699** in DS, data not shown). The DMF/LAD also showed a positive relationship with DMF (r = 0.513*; 0.462^ns in WW and DS, respectively, data not shown), emphasizing the importance of source size followed by net assimilation rates on biomass production at flowering in finger millet.

Fig. 1. Effect of drought stress on physiological parameters and yield attributing traits in finger millet Legend: St. cond., Stomatal conductance; Pn, Photosynthetic rate; T, Transpiration rate; Grain yd., Grain yield; DMF, Dry matter at flowering; MEW, Mean ear weight; LAI, Leaf area index; LAD, Leaf area duration; TDM, Total dry matter at harvest; FL, Finger length; HI, Harvest index; PT, Productive tillers; Straw wt., Straw weight; Thresh %, Threshing percentage; LT, Leaf temperature and Ci, Internal CO2 concentration.

Table 1. Effect of drought stress on leaf area index, leaf area duration, DM/LAD, biomass at flowering and photosynthetic parameters in finger millet genotypes

Note: T, Treatment; G, Genotype; G @ T, Genotype at given treatment; T@G, Treatment at a given genotype; WW, Well watered; DS, Drought stress; %Red., Per cent reduction under drought stress over the WW condition; SEm ± , Standard error of mean; CD, Critical difference at 5% level of significance; CV (%), Coefficient of variation; LAI, Leaf area index; LAD, Leaf area duration (days) from sowing to flowering; DMF, Dry at flowering; Pn, Photosynthetic rate (uM m⁻²s⁻¹); gs, (Stomatal conductance, Mol m⁻²s⁻¹); Ci, Internal Co₂ Conc. (ppm); T, Transpiration rate(mmol, m⁻²s⁻¹); LT, Leaf temperature (^°C).

Effect of DS on gas exchange parameters

The gas exchange parameters, namely the photosynthetic rate (Pn), stomatal conductance (gs) and transpiration rate (T), were decreased significantly due to DS, whereas the internal CO₂ concentration was increased due to DS conditions (Ci; Table 1; Fig. 1). The Pn was high in Cv. GPU-28 and none of the genotypes were significantly superior to GPU-28, although the stomatal conductance and transpiration rates were significantly higher in GE-303, GE-1034 and GE-4683. The Pn was positively and significantly related to stomatal conductance (r = 0.751** in WW and r = 0.822** in DS, data not shown) and transpiration rate (r = 0.751** in WW and r = 0.893** in DS, data not shown). The DS significantly decreased the photosynthetic rate (26.3%), stomatal conductance (26.4%) and transpiration (24.8%) (Fig. 1), and the least affected traits due to DS were threshing per cent (1.2%) and DMF/LAD (0.4%). The mean leaf temperature (LT) was increased significantly, due to DS by 4.7%. However, a few genotypes, namely, GE-303, GE-144, GE-1013 and GE-27, maintained lower leaf temperatures by <1.5°C compared to that of the popular variety, GPU-28 (Table 1).

The effect of DS on yield attributes and grain yield

The mean grain yield and finger length decreased significantly due to DS (Table 2). The other associated yield attributes did not decrease significantly due to DS, but the genotypes differed significantly due to DS conditions (Table 2). The mean grain yield was decreased by 13.2%, followed by mean ear weight (MEW), dry matter production and other yield-associated parameters.

Table 2. Effect of drought stress on yield attributing traits and grain yield in finger millet genotypes

Note: T, Treatment; G, Genotype; G @ T, Genotype at given treatment; T@G, Treatment at a given genotype; WW, Well watered; DS, Drought stress; %Red., Per cent reduction under drought stress over the WW condition, SEm ± , Standard error of mean; CD, Critical difference at 5% level of significance, CV (%): Coefficient of variation.

Selection of genotypes for drought tolerance

The path effects showed that the productive tillers (PT; 1.556 in WW and 1.648 in DS) and mean ear weight (MEW; 1.531 in WW and 1.680 in DS) had a very high positive direct effect on grain yield (Table 3; Lenka and Mishra, 1973). Threshing per cent showed a moderate effect on grain yield under DS conditions. Among the physiological traits, the LAI and leaf temperature had a positive, moderate direct effect on grain yield. The DMF, Pn and T showed a direct negative effect on grain yield (Table 3). The LT had a significant negative correlation with the grain yield under DS (r = −0.556*, Table 3). Multiple linear regression (MLR) revealed that only productive tillers (52.9%) and mean ear weight (47.1%) contributed significantly to grain yield in the WW condition (Table 4). In DS, productive tillers (36.3%), mean ear weight (35.6%) and threshing per cent (33.2%) contributed equally and significantly to grain yield. In addition, the LAI contributed positively by 8.6%, but the transpiration (−5.2%) and DMF (−8.6%) negatively influenced the grain yield under DS conditions. Based on the mean ranking of traits contributing significantly to the grain yield, GE-4683, GE-144, GE-1013 and the other five genotypes were found to be better than GPU-28 under WW conditions, while only two genotypes (GE-4683 and GE-4995) were found superior to GPU-28 in DS conditions (Table 5).

Table 3. Direct and indirect effects of physiological and yield attributing traits on grain yield in finger millet under WW and DS conditions

Note: WW, Well watered; DS, drought stress; LAI, Leaf area index; DMF, Dry matter at flowering; Pn, Photosynthetic rate; T, transpiration rate; LT, Leaf temperature; Str., Straw weight; PT, Productive tillers/m2, Th%, Threshing per cent; MEW, Mean ear weight; GY, Grain yield and r, Pearson's correlation coefficient. Diagonal bold values are direct effects.

Table 4. Stepwise regression analysis by physiological and yield attributes towards grain yield in finger millet under WW and DS conditions

MLR, Multiple linear equation (stepwise backward elimination method); Res, Residual effect; P, Probability; r = Multiple linear correlation coefficient; Act. and Est. GY are actual and estimated grain yield by the regression equation (g m²); x, independent traits contributing to grain yield and values in parentheses are per cent contribution by respective traits to grain yield.

Table 5. Performance (ranking) of finger millet genotypes under WW and DS condition

WW, Well watered; DS, drought stress; GE, refers to Germplasm Entry maintained at AICRP (Small Millets), UAS, GKVK, Bangalore, Karnataka, India, Rank value: was computed considering traits contributed towards grain yield significantly as given in Table 4 including grain yield, % Change: provides better or poor ranking performance as compared to popular variety, GPU-28. Data in bold letters represents the check variety, GPU-28.

Discussion

Effect of DS on physiological traits and yield attributes

Finger millet is an important millet crop in the changing climate scenario both as drought resilient and nutraceuticular crop. Crop improvement in rainfed finger millet has been through selection per se for yield and blast resistance as in case of development of a blast resistant variety, GPU-28 (Krishne Gowda et al., 2009). For further improvement, trait specific selection along with yield would be appropriate as compared to yield per se (Nanja Reddy et al., 2021). In this direction, genetic resources play a key role in improving crop productivity. The available finger millet germplasm (10,507 Numbers) in India (Ceasar et al., 2018) can be exploited for the development of superior varieties suitable for rainfed conditions. Finger millet being a rainfed crop it is likely to experience DS stress during its crop growth stage reduce the grain yield (Kandel et al., 2019; Krishna et al., 2021). Reduction in grain yield depends on genotype, crop growth stage and duration of DS. Identification of genotypes with superior drought adaptation as compared to popular variety could be highly useful for future breeding programmes. The grain yield is the product of above-ground biomass and harvest index, that are complex and computed traits. Therefore, in the present study, emphasis was laid on independent traits comprising of the source size, biomass accumulation at flowering, assimilation rates, gas exchange traits at flowering and yield attributing traits towards grain yield under DS conditions.

Dry matter accumulation in finger millet by the time of flowering (DMF) has been reported to influence the biomass at harvest and grain yield positively under rainfed conditions but not exactly under the DS conditions (Nanja Reddy et al., 2019b). Probably, higher the DMF, higher the accumulation of carbohydrates and then remobilization of stored carbohydrates from the stem towards grain could be high (van Herwaarden et al., 1998 in case of wheat), and it was reported to contribute to grain yield increase of up to 40% under DS condition as compared to only 10% under normal conditions (Yang et al., 2000; Inoue et al., 2004 in case of wheat). However, in the present study, it was not proved, as the finger millet is a C ₄ species that has a hard and thick stem unlike the wheat or rice, C ₃ species. Therefore, it is difficult to conclude the importance of DMF on grain yield in finger millet, and further studies are required in this context.

The decreased DMF due to DS (Fig. 1) was associated with significant decrease in source size (LAI). A similar reduction in leaf area due to DS has been reported (Krishna et al., 2021) and it could be a drought adaptation strategy to reduce the water loss under DS conditions. Such a reduction in source size due to DS could be due to decreased leaf expansion associated with lower water potential and subsequent reduction in cell division and cell expansion under rainfed drought situations (Shao et al., 2008; Farooq et al., 2009). Therefore, the reduced leaf area with higher leaf thickness would be appropriate strategy for a rainfed variety, instead of large leaf area (Sastry et al., 1982). The mean DM/LAD (net assimilation rate) was not affected by DS for 18 days (Table 1), as finger millet being a C ₄ species maintain a higher carbon exchange rates (Uma, 1987).

With respect to gas exchange traits, the photosynthetic rate (Pn) by leaf has been reported to have a positive influence on the grain yield of finger millet (Subramanyam, 2000; Nanja Reddy, 2020), in addition, the ear photosynthesis also contributes to yield up to 40% (Tieszen and Imamba, 1978; Sashidhar et al., 1984). The reduction in Pn under DS was due to decreased stomatal conductance and transpiration rate (Fig. 1; Mohanabharathi et al., 2019; Nanja Reddy et al., 2020). The increased internal CO₂ concentration (Ci) under DS condition could be due to decreased carboxylation, and hence CO₂ got accumulated in mesophyll cells (Table 2; Maai et al., 2020). Although the Pn was sensitive to DS, it did not reflect on the biomass or grain yield (Table 3; Inoue et al., 2004) suggesting that a DS for 18 to 20 days during the rainy season is not a constraint in finger millet.

Leaf temperature (LT) under DS has been recognized as an indicator of plant water status (Hirayama et al., 2006). The increase in LT could go up to 5°C depending on the sensitivity of genotype (Ramya and Nanja Reddy, 2018; Mohanabharathi et al., 2019; Krishna et al., 2021). Higher leaf temperature affects the metabolic activity of the plant and therefore plant develops a strategy for increasing transpiration rate to lower the leaf temperature, which could be possible by higher water uptake through a deeper root system. In the present study, a significant increase in leaf temperature from 25.7 to 26.8°C by DS (Fig. 1), had a significant negative relationship with grain yield (r = −0.556**; Table 3; Ramya and Nanja Reddy, 2018). Therefore, LT can serve as selection criterion for drought tolerance.

Cumulative effects of all physiological processes would determine the yield attributes and yield. Finger millet is a rainfed crop and the DS during critical crop growth stages, especially the reproductive phase reduces the grain yield considerably (Talwar et al., 2020; Krishna et al., 2021). For instance, DS from 40 DAS till harvest in black soils decreased the yield to the up to 45% (Talwar et al., 2020) due to a reduction in total biomass and harvest index. A reduction in grain yield by 36.1% was reported when DS was imposed from 35 to 42 DAS at a vegetative period in pot culture due to reduction in ear length and mean ear weight (Mohanabharathi et al., 2019). Under field condition, 15 to 20 days of DS (with 3.5 to 4.5 mm PET per day) is not a significant limitation for grain yield (Nanja Reddy et al., 2020), in the present study, only 13.2% reduction in grain yield was observed with DS for 28 days during reproductive phase in the rainy season. The lower effect of DS in finger millet could be due to (1) under field condition, it takes 15 to 16 days to reach a 50% field capacity (Krishna and Nanja Reddy, 2021), and the remaining 12 days of actual stress during flowering might not be a threshold duration for reducing the yield in the present study and (2) it could be due to higher recovery after stress alleviation. However, if the DS duration is longer (40 days) during the reproductive period under field condition, the tolerant genotypes show lesser yield reduction (14.5%) as compared to sensitive genotypes (44.8%; Suma, 2014). These results suggest that although finger millet is a drought-resilient crop, DS decreases the grain yield depending on duration of the DS and; identification of drought tolerant lines could be a continuous process in crop improvement.

Selection of genotypes for drought tolerance

The reduction in grain yield due to DS (13.2%) was higher as compared to other yield attributing traits. Based on the path analysis and MLR, the most important yield contributing traits were productive tiller number per plant and mean ear weight under WW condition. While under DS condition, productive tillers, mean ear weight and threshing percentage were equally contributed to grain yield (Table 4; Nanja Reddy et al., 2021). The DS might affect the productive tiller by immature ear formation with poor grain filling (Ashok et al., 2018) and the reduction in mean ear weight could be due to decrease in finger number and finger length (Mohanabharathi et al., 2019; Krishna et al., 2021). The productive tillers, mean ear weight and threshing percentage found to contribute significantly towards grain yield (by MLR) under DS conditions (Table 4). Under DS, the LAI was also found contributing trait towards grain yield. Considering these traits GE-4683 was found to be significantly superior to the popular cultivated variety, GPU-28. The Cv. GPU-28 was shown to adapt better to rainfed conditions with 2 to 3 dry spells of 10 to 15 days each (Srinivasa Reddy et al., 2017). Future emphasis should be laid to increase the productive tillers or mean ear weight but care should be taken as these two parameters compensates with one another (Mujahid et al., 2020). The MLR predicts that, under DS condition, keeping all other traits constant, increase in the productive tiller number from 2 to 3 per plant has the potential to increase the grain yield of GPU-28 from 360.0 to 459.5 gm⁻² (29.1%, 3.6 to 4.59 t/ha). With 1.0 g extra mean ear weight, the grain yield was estimated to increase from 360 to 392 gm⁻² (9.4%, 3.60 to 3.92 t/ha). The management practices can be effective in manipulating the productive tillers, but the finger size is relatively genotype-dependent. Hence the better way of selection for higher grain yield under DS could be through the mean ear weight.

Conclusions

Although absolute grain yield under DS could be used as one of the important parameters in the identification of drought-tolerant lines, additional physiological traits would aid in precision and confirmative selection of stable genotypes. The root parameters were known to be important in drought adaptation, but the root growth will be almost ceased by the flowering time. Hence, the above ground parameters like gas exchange rates and yield attributes namely, productive tillers, mean ear weight and threshing percentage are important in determining the drought tolerance during the reproductive period. Finger millet has higher recovery after natural stress alleviation and it can sustain 2 to 3 dry spells of 10 to 15 days each with a minimal reduction in grain yield. Hence, finger millet is a relatively a drought adaptive and climate-resilient crop suitable to rainfed situations. However, crop improvement should be a continuous process as the frequencies of DS are increasing with changing climate for which identification of donor lines could be a continuum.