Introduction

Foxtail millet [Setaria italica (L.) P. Beauv.], regarded as a native of China, is one of the world’s oldest cultivated crops. It was originated in China, where it’s domestication took place about 8000 YBP (Li and Wu 1996) as well as in south west Asia and central Europe (Vavilov 1926; Zohary et al. 2012) where it’s cultivation recorded about 5000 YBP (Hammer et al. 1999). The Hindukush Himalayan region is rich in foxtail millet diversity (Scheibe 1943; Hammer and Khosbhakht 2007). The crop has moved from China to Northern India and Nepal during 5th to 2nd Millennium BC (Stevens et al. 2016), thus the Nepalese foxtail millet, mostly is Setaria italica subsp. italica race maxima (de Wet et al. 1979). It is the second most widely planted species among millets in the world and the most important millet in East Asia (Kumari et al. 2011; Ning et al. 2015; Sheikh and Singh 2013; Xiaomei et al. 2016; Zhang et al. 2014). It is known for its better tolerance to abiotic stresses compared to other cereal crops. It is also known as Italian millet, German millet or Hay millet but locally known as Kaguno in Nepal. Foxtail millet grain contains 12.3% protein, 4.3% fat, 60.9% carbohydrates, 14.0% dietary fibre and 3.3% minerals, with 31 g calcium, 290 mg phosphorus, 5 mg iron and vitamins than major staple crops rice and wheat (Saha et al. 2016; Saud 2010). Foxtail millet is the third important crop among group of millets with wide range of utility in Nepal. The cooked grain is used as bhat (cooked like rice), dhindo (porridge) and kheer (like rice pudding). Foxtail millet is valued by mountain farmers for its nutritional content and health promoting properties, ability to grow under low external input conditions and tolerance to extreme environmental stress, particularly drought. It is also recently appreciated because of medicinal benefits such as reducing blood glucose levels and cholesterol control in normal as well as diabetic patients. It is the crop of future in the context of changing climate with great potentiality to cope with food insecurity in remote areas of the country (Goron and Raizada 2015).

In Nepal, this crop is considered as traditional climate resilient and nutritionally dense crop but trends of cultivation and use are shrinking fast due to globalization, land use change, out migration, social values, change in food habit, depleting traditional knowledge, lack of research and policy support such as crop improvement and formal seed distribution system (Gurung et al. 2016; Parajuli et al. 2017; Palikhey et al. 2016; Sheikh and Singh 2013; Bisht et al. 2006). Precise data of area and production under foxtail millet is not known because the production statistics of this crop had often been grouped with other millets. Major foxtail millet growing districts in Nepal are Mugu, Kalikot, Humla, Jumla, Bajhang, Bajura, Dolpa, Lamjung, Gorkha, Ramechhap, Kavre, etc. where crop is grown sole as well as mixed with finger millet, proso millet, beans, amaranths, maize etc. Nepal is one of the centres of diversity of foxtail millet (Nakayama et al. 1999) and has high genetic diversity of this crop (Mo FSC 2002), however, very little research has been conducted and its status is still unexplored in Nepal thus called neglected and underutilized crop (Amgai et al. 2011). Plant genetic resources are conserved so that they can be used to improve crop production and in other ways. However, Hodgkin et al. (2003) asserted that use of ex situ conserved germplasm is inadequate globally and that genetic diversity maintained in genebanks is underutilized.

Poor utilization of local genetic resources conserved in genebank for foxtail millet improvement and development program is evident in Nepal due to (1) absence of public and private sector breeding priority, and (2) unavailability of characterization and evaluation information for wider use. GEF-supported Local Crop Project (www.himalayancrops.org), executed jointly by Nepal Agricultural Research Council, Department of Agriculture, LI-BIRD and Bioversity International, therefore aims to fill this gap and re-introduce genebank materials of foxtail millet landraces in traditional farming system for improving farmers access to diversity and increasing on-farm diversity. The objective of this study is to characterize diverse farmer’s varieties of foxtail millet using agro-morphological data and use information to deploy useful diversity by participatory methods, and also for the efficient use of these accessions in future crop improvement programme.

Materials and methods

A total of 41 foxtail millet accessions were collected from six districts of Nepal (Fig. 1). Among them, 28 accessions were from Humla through diversity fair, six accessions were from Jumla, four accessions were from Lamjung and one accession each from Bajhang, Gorkha and Kavre districts. The detailed sources of accessions are presented in Table 1. These accessions were characterized ex situ at Genebank, Khumaltar (N27.4°, E085.2°, 1360 m a.s.l.), Nepal during summer 2015. Accessions were planted on 17th May, 2015 with an individual plot size of 6 m2 following row to row spacing of 25 cm and plant to plant spacing of 10 cm. Fertilizers were applied at the rate of 20:30:20 kg/ha N:P2O5:K2O as basal dose during land preparation and 20 kg/ha N was top-dressed 25 days after seeding. Different quantitative traits like days to 50% heading, days to 80% maturity, plant height (cm), panicle length (cm), panicle width (mm), panicle exertion (cm) and grain yield (kg/ha) as well as qualitative traits like growth habit, flag leaf angle, sheath pubescence, leaf pubescence, grain colour, grain yield potential, leaf blast susceptibility, stem borer susceptibility and overall phenotypic acceptability were recorded as per descriptors for Setaria italica (L.) P. Beauv. and S. pumila (Poir.) Roem. et Schult. (IBPGR 1985). Observations on days to heading, maturity and grain yield were based on whole plot data whereas observations on rest of the traits were from 10 randomly selected plants.

Fig. 1
figure 1

Collection map of experimental materials and characterization location

Full size image
Table 1 Origin and sources of experimental materials
Full size table

Shannon–Weaver diversity indices (Shannon and Weaver 1949) were calculated for each trait with Microsoft Excel using the formula:  \(H^{\prime} = \left[ {\sum \left( {{n}/{N}} \right)^*\left\{ {log2\left( {{n}/{N}} \right)^*\left( { - 1} \right)} \right\}} \right]/log2k\); where, H′ is the standardized Shannon–Weaver diversity index, k is the number of phenotypic classes for a character, n is the frequency of a phenotypic class of that character and N is the total number of observations for that character. For the quantitative traits, accessions were divided into 10 phenotypic classes as < x − 2sd, x − 2sd, x − 1.5sd, x − sd, x − 0.5sd, x, x + 0.5sd, x + sd, x + 1.5sd, x + 2sd and > x+2sd are as the margins of the classes, where x is average and sd is standard deviation. The descriptive statistics, clustering, Principal Component Analysis (PCA) and correlation analysis is done by using Minitab-14.

Results and discussion

Descriptive statistics and Shannon–Weaver diversity

Nepalese foxtail millet germplasm found diverse in shape, size and colour of panicles as well as grains which has been illustrated in Fig. 2.

Fig. 2
figure 2

Diversity in panicle and grain types of Nepalese foxtail millet

Full size image

Range and average of observations as well as Shannon–Weaver diversity indices for different quantitative traits has been presented in Table 2. Flowering of accessions ranged from 45 to 71 days with the mean of 60 days from seeding. This range was wider than the flowering range of Nepalese accessions (45–60 days) but much narrower than the global collections (32–135 days) in ICRISAT, India (545 m) reported by Reddy et al. (2006). According to that report, the earliest maturity accession was from Russia and the longest maturity was from Sri Lanka. This is very important adaptive trait for mountain farmers in rainfed dryland conditions, and also to escape for extreme cold temperature. Women farmers prefer earliness because the food is available during lean period. Plant height ranged from 108 to 232 cm with an average of 159 cm which is quite taller than the Nepalese (70–150 cm) as well as global (20–215 cm) accessions (Reddy et al. 2006). Our Nepalese accessions showed narrower range of panicle exertion (7–30 cm) as compared to the range in global collection of 1–36 cm (Reddy et al. 2006). Panicle length ranged from 13 to 30 cm which is similar to the range in Nepalese collection (13–25 cm) but narrower than the range in Indian collection (2–39) as reported by Reddy et al. (2006). Panicle width in our result showed narrower range (10–33 mm) wider than the range in Nepalese (10–25 mm) and very much narrower than the global (5–120 mm) collection (Reddy et al. 2006). Grain yield in our study ranged from 89 to 3483 kg/ha with the average of 1247 kg/ha. This is the preliminary yield from small plot (6 m2) non-replicated trial in high fertility condition of Khumaltar. Some accessions were severely damaged by leaf blast (Pyricularia setariae Nisikado.) and false smut (Ustilago crameri Korn.) diseases and gave very poor yield.

Table 2 Descriptive statistics and Shannon–Weaver diversity index (H′) of 8 quantitative traits
Full size table

Shannon–Weaver diversity index (H′) considers both richness and evenness of the phenotypic classes of the traits. H′ for all quantitative traits is ranged from 0.689 to 0.899 (Table 2) that for qualitative traits ranged from 0.443 to 0.989 (Table 3), suggested that characterized 41 foxtail millet accessions were diverse for recorded traits.

Table 3 Shannon–Weaver diversity index, descriptor states and frequency of 9 qualitative traits
Full size table

Clustering observations

UPGMA clustering divided 41 accessions of foxtail millet into five clusters (Fig. 3). Two clusters, Cluster-1 and Cluster-2, comprised of 15 accessions in each cluster mainly from Humla (Table 4). Cluster-1 has accessions that were low to intermediate yielder, medium maturity and height. Accessions from Cluster-2 were very low yielder (89–500 kg/ha), shortest panicle width (11–19 mm) and mainly late in maturity (81–113 days). Cluster-4 and Cluster-5 jointly comprised of 6 accessions (3 from Lamjung, 1 each from Gorkha, Humla and Jumla districts) were of higher grain yield (2767–3473 kg/ha), taller plants (158–194 cm), greater panicle width (19–33 mm) and early maturity (86–99 days). Accessions from these two clusters can be further evaluated on-farm for their validation.

Fig. 3
figure 3

UPGMA clustering of 41 foxtail millet landraces based on Euclidean distance

Full size image
Table 4 Number of accessions with average of major agronomic traits in each cluster
Full size table

Principal component analysis

Table 5 shows the principal component analysis with eigen vectors, eigen values and variances due to first five principal components. In the first principal component, grain yield, panicle width and plant height with negative loading of − 0.581, − 0.486 and – 0.435, respectively were the most important traits. Similarly in the second principal component, days to heading with negative loading of − 0.639 and panicle length with positive loading of 0.496 were the most important traits. Panicle exertion was the most important trait in the third principal component followed by days to maturity with positive loadings of 0.871 and 0.374, respectively. Eigen analysis of the correlation matrix revealed that five principal components explained 95% of the total variance among the accessions. The first five principal components accounted for 32, 28, 15, 11 and 9%, respectively of the total phenotypic variance among accessions based on 7 quantitative traits. Scatter plot of the first two principal components, accounting 60% of cumulative variance, suggested that PCA is in support with the result of cluster analysis (Fig. 4).

Table 5 Principal component analysis and Eigen analysis of the correlation matrix
Full size table
Fig. 4
figure 4

Scatter plots of foxtail millet landraces based on first two principal components

Full size image

Correlation analysis

Pearson’s correlation coefficient between quantitative traits has been presented in Table 6. There was highly significant positive correlation between days to heading and days to maturity. Significant but negative correlation was observed between days to heading and panicle length. Similarly days to maturity was negatively correlated with grain yield suggested that early genotypes gave higher yield as an adaptive trait under semi-arid and cold environment, thus early maturing accessions can be selected for dry areas which can escape from terminal drought and cold stresses. Plant height showed highly significant positive correlations with grain yield and panicle width, respectively suggesting that taller height can be considered as a desirable character of high yielding accession. The strong association between grain yield and panicle width suggested that accessions with thicker panicles can be selected as a high yielding trait.

Table 6 Pearson’s correlation analysis between 7 quantitative traits
Full size table

Crop genetic diversity of traditional mountain crops such as foxtail millet has disappeared in many parts of Himalayan mountains (Sheikh and Singh 2013). Community richness of foxtail millet diversity in Lamjung, Humla and Jumla has shown declining trend in recent decades. Recent surveys indicated area under foxtail millet and the number of households growing the crop is shrinking in mountains of Nepal (Gurung et al. 2016; Parajuli et al. 2017; Palikhey et al. 2016). Local crop genetic diversity from these areas exists in National Genebank or in farmers’ fields but they are not characterized and evaluated. They are not easily accessible to majority of farmers. This is not surprising as globally the use of ex situ materials is limited by lack of characterization data and information (Hodgkin et al. 2003) and direct use of genebank materials by re-introduction by farmers may benefits farming community immediately amongst less researched crops. With increased awareness on nutritional and health benefits and climate resilient properties of the crop, foxtail millet research for development in low HDI index districts of Nepal is recently recognized. Farmers of these areas of Nepal suffer from food insecurity, malnutrition and hidden hunger and therefore, importance of traditional crops like foxtail millet remains important. PRA surveys suggest that farmers prefer varieties have set of traits such as higher yield, earliness, easy threshing and taste. Participatory methods proposed by Joshi and Sthapit (1990) and Witcombe et al. (2016) will be employed for further evaluation in diverse geographic environments. The application of this study will be used to distribute a set of IRD kits i.e. half kg seed per HH of six landraces (namely C5644, H252, C5647, C5808, C5643 and C4581) in a range of 500–1000 kits for deploying new diversity and provide farmers for participatory selection for their needs. The pre-breeding or grassroots breeding of the crop will help to provide better options than available in local market (Sthapit and Ramanatha Rao 2009) and the best way in which genebank diversity can be used as a strategy for such nutritionally dense and climate resilient neglected crops, will have considerable positive effect on use of under-utilized ex situ genetic resources (Goron and Raizada 2015).

Conclusion

Among the 41 accessions of foxtail millet collected from six districts of the country, high level of diversity was observed for different agro-morphological markers like growth habit, phenology, plant stature, length and width of inflorescence, flag leaf angle, leaf and sheath pubescence, leaf blast and stem borer susceptibility, grain yield and grain colour, etc. High grain yield was associated with earliness, tallness and thicker panicles. Accessions C5644, H252, C5647, C5808, C5643 and C4581 are the important landraces (three from Lamjung and one each from Gorkha, Humla and Jumla) needs further evaluation in farmers’ field of multiple locations to identify candidate genotype for direct cultivation and also for future use in crop improvement programmes. The project aims to test these materials in large number (500–1000 kits) in wide geographic region of Nepal using participatory diversity kits for deploying new diversity and make traditional seed system more resilient. Farmers now have a greater range of genetic diversity available from genebank to them than ever before and have flexibility for local selection for target environment.