Introduction
Megathyrsus, the cosmopolitan genus, is a heterogeneous assemblage of approximately 500 species (Hunt et al., Reference Hunt, Badakshi, Romanova, Howe, Jones and Heslop-Harrison2014). The genus with remarkably uniform floral characters exhibits a considerable variation in anatomical, physiological and cytological features (Aliscioni et al., Reference Aliscioni, Giussani, Zuloaga and Kellogg2003). Megathyrsus maximus (Jacq.) B.K.Simon & S.W.L.Jacobs (Simon and Jacobs, Reference Simon and Jacobs2003) (Guinea grass), earlier known as Panicum maximum Jacq., is a widely adapted perennial grass. The grass is native to Africa, particularly East Africa, Kenya and Tanzania. The species is mostly represented by apomictic tetraploid forms in nature with the occasional presence of the sexual diploids (Combes and Pernès, Reference Combes and Pernès1970); pentaploid and hexaploid forms (Combes, Reference Combes1975; Savidan et al., Reference Savidan, Carman and Dresselhaus2001; Kaushal et al., Reference Kaushal, Malaviya, Roy, Pathak, Agrawal, Khare and Siddiqui2008a).
Guinea grass is one of the most important forage species in the vast rangelands of the tropical world including India. Its ease of propagation, high biomass, good nutritional quality, shade tolerance and availability of annual as well as perennial types (Malaviya, Reference Malaviya1996) make it popular. Recent estimates indicate a wide gap between the demand and supply of green and dry fodder in India (Roy et al., Reference Roy, Agrawal, Bhardwaj, Mishra, Mahanta, Roy, Agrawal and Bhardwaj2019a). To bridge this gap, genetic improvement of grasses, especially Guinea grass, is important owing to its wide adaptation in orchards, wastelands, grasslands, etc., without competing for arable lands.
Although the species is native to Africa, it is now pantropical with introduction in most part of the world as a pasture forage crop. Megathyrsus maximus is a highly successful invader in the tropical and warm temperate areas. Because of its robust growth habit and fast spread through seeds, it competes well with native flora. It can quickly spread to invade the gaps left in natural vegetation after fire. It is reported to be an aggressive invader in Brazil (Alves and Xavier, Reference Alves and Xavier1986), Australia (Clements and Henzell, Reference Clements and Henzell2010); and is an important weed of tropical cereals in Africa, America and Asia (Baker and Terry, Reference Baker and Terry1991). Although it is considered as an invasive plant in places such as South Texas, Sri Lanka, Queensland and Hawaii’, it is not a prohibited or restricted invasive plant in Queensland (Anonymous, 2020). Besides all these facts, under managed conditions, the species is liked by farmers, both as mono-crop as well as in intensive forage production systems, because of its high biomass yield, protein content and palatability (Bogdan, Reference Bogdan1977; Sukhchain and Sidhu, Reference Sukhchain and Sidhu1992; BhagMal et al., Reference Bhagmal, Singh, Roy, Ahmad and Malaviya2009; Singh and Chaturvedi, Reference Singh, Chaturvedi, Panwar, Tiwari and Dadhwal2011; Wasnik, Reference Wasnik2014; Roy et al., Reference Roy, Malaviya and Kaushal2019b).
Breeding tropical forage grasses is extremely difficult owing to their reproductive behaviour, particularly apomixis and polyploidy. Hence, evaluation of existing natural variation and selection of desirable genotypes is one of the most plausible and successful breeding methods in such grasses. Agronomical evaluation programmes have been attempted earlier with a limited number of germplasm (Combes and Pernès, Reference Combes and Pernès1970; Nakajima et al., Reference Nakajima, Ochi and Mochizuki1978; Malaviya, Reference Malaviya1995, Reference Malaviya1996, Reference Malaviya1998; Kaushal et al., Reference Kaushal, Malaviya and Singh1999; Jain et al., Reference Jain, Roy, Kaushal, Malaviya and Zadoo2003a). These studies indicated the presence of considerable genetic variability among the germplasm; hence, it is pertinent to make a core germplasm set for the researchers to narrow down the number of germplasm required for future genetic studies and identification of donors for desirable traits.
Considering the heterozygosity and diversity available in tropical grasses, evaluation and characterization of the germplasm still remain a quick method to identify the lines with desirable traits, which can be utilized either directly or through breeding programmes. Establishing a core collection capturing maximum diversity and minimum repetitiveness is a desired and established practice for exploitation and utilization of the genetic resource (Frankel and Brown, Reference Frankel, Brown, Chopra, Joshi, Sharma and Bansal1984; Brown, Reference Brown1989a, Reference Brown, Brown, Frankel, Marshall and Williamsb). Development of a core set of germplasm is essential for the breeders to expedite their efforts by utilizing the limited number of germplasm which can provide the range of diversity required. For the evaluation and characterization of germplasm; and to develop core sub-set, morphological traits are still widely used. In this context, use of the morphological descriptors for diversity study and identification of the most discrepant genitors has played an important role (Carvalho et al., Reference Carvalho, Lanza, Fallieri and Santos2003). The present work was therefore undertaken to evaluate the germplasm collection of M. maximus being maintained at the IGFRI Gene Bank, Jhansi, India for its characterization and development of a core sub-set.
Materials and methods
Experimental site
The experiment was carried out at the experimental farm of ICAR-Indian Grassland and Fodder Research Institute (IGFRI), Jhansi, India (25.27N, 78.35E, 271 m a.s.l.) under rain-fed conditions.
Materials
A total of 152 genotypes representing collections from Africa (27 genotypes), Brazil (one genotype) and India (124 genotypes) were studied (Table 1). Among these genotypes, five (BG 1, BG 2, Hamil, Haritha and Maratha Kamal) were cultivars, whereas the rest were wild collections from the rangelands and the open areas. Indian collections represented the eastern Himalayan region with humid hills (seven genotypes), northern semi-arid region (11 genotypes), north-west region having extreme summer and winters (15 genotypes) and south Indian sub-humid tropical regions (91 genotypes). Seeds of these genotypes were procured from the gene bank of ICAR-IGFRI, Jhansi, India.
Table 1. Megathyrsus maximus genotypes used in the study (listed as per clustering) and the core sub-set (highlighted in grey)
SI, south India; Et, Ethiopia; WI, western India; NI, north India; NEH, north-eastern hill region India; Br, Brazil; Af, South Africa.
a Serial numbers are as used in clustering and dendrogram.
b Exotic accessions.
Raising of plants and observations
Nursery of the accessions was raised and the 6-week-old seedlings were transplanted in 3 m long paired rows spaced 75 cm apart and accommodating six tussocks spaced at 50 cm in the row. Gap between two pairs of lines was kept at 1 m. The plants were harvested above ground at 15 cm height once in the first year of establishment and thrice in the second year during the rainy season. Data on morphological observations were recorded on three plants; excluding one border plant in each row at both the ends.
Data on the 19 quantitative traits were recorded for plant height, tillers/tussock, stem diameter, internodal length, node extension, node hair length, node hair extension, leaf blade length, leaf sheath length, leaf blade width, leaf sheath width, ligule hair extension, flag leaf blade length, flag leaf sheath length, flag leaf blade width, flag leaf sheath width, length between the node to panicle base, panicle length and panicle branch length. Plant height was recorded as the total length of the longest tiller from the ground to the top, i.e. tip of the last emerged leaf. The leaf width and length were recorded on the third leaf from the top of the three randomly selected plants. Most of the traits were recorded at 50% flowering stage. Data were also recorded for the non-metric traits such as stem colour, node colour, leaf sheath hairiness, leaf hairiness, flowering, panicle hairiness and leaf colour.
One hundred and forty randomly selected genotypes, out of the total 152 genotypes, were also analysed for nine nutritional parameters – ash content (%); organic matter (OM %); crude protein content (CP %); neutral detergent fibre content (NDF %); acid detergent fibre content (ADF %); hemicellulose (%), cellulose (%) and lignin (%). The CP and ash were estimated following AOAC (1980) method, whereas NDF, ADF, cellulose, hemicellulose and lignin contents were estimated following Goering and Van Soest (Reference Goering and Van Soest1970). The genotypes were also evaluated for in vitro dry matter digestibility (IVDMD %) as per Tilley and Terry (Reference Tilley and Terry1963).
Statistical analysis:
Agglomerative Cluster Analysis using Euclidean Distance method was done using the Statistical Tool for Agricultural Research (STAR) computer software (STAR, 2013). For clustering purpose, all the 19 metric morphological traits were used. The number of accessions for core sub-set was kept at 15% of the total germplasm. The number of accessions for core sub-set from each cluster resulting from cluster analysis was decided following Brown (Reference Brown1989a) as per the formula given here under:
$$s = \lpar {{\rm log}p^i/_m\Sigma^{t = 1}{\rm log}p} \rpar \times n\comma \;$$where s = number of accessions selected in a character; p = size of cluster; pi = proportion of ith cluster; n = number of accessions to be selected for core (15% of the base collection in this study); m = total number of cluster.
Second level of cluster analysis, i.e. cluster within cluster, was done for further assortment of the accessions, and the selection of accessions from each second-level cluster was done at random following Pandiyan et al. (Reference Pandiyan, Senthil, Packiaraj and Jagadeesh2012) to constitute the core sub-set.
Pearson's product moment correlation and Spearman's rank correlation were studied to see the correlation among various quantitative traits. Principal Component Analysis (PCA) was done to identify the key characters that contribute to the diversity. Scree plot analysis was done to determine the number of principal components to be retained as per Cattell (Reference Cattell Raymond1966). The mean value for different traits as obtained in the core value was compared with that of total germplasm by Student's t-test using Microsoft Excel programme.
Results
The set of Guinea grass germplasm, representing different agro-ecological conditions, was evaluated for 35 traits, including nine nutritional quality traits. Clustering of the genotypes based on 19 numeric morphological traits resulted in eight clusters (Table 1, Fig. 1). Cluster 3 was largest with 65 genotypes followed the Cluster 1 with 25 genotypes. Clusters 7 and 8 were small clusters with two and five genotypes, respectively (Table 1). Most of the genotypes from southern India clustered in the Clusters 3, 5 and 6, which accounted for 64.4% of the total collections from the south. Eleven of the total 26 Ethiopian collections clustered in Cluster 3, whereas five and four genotypes from Ethiopia were in Clusters 1 and 2, respectively. Twelve out of the 13 genotypes from the northern semi-arid region clustered in Clusters 1, 2 and 3; thus, these were well separated from the five remaining clusters. The largest Cluster 3 was a mixture of genotypes from Ethiopia, north India, north-western India, south India and north-eastern hill region, thus representing highest diversity. A comparatively smaller Cluster 5 had a representation of genotypes from all the collection places, except those from north India.
Fig. 1. Dendrogram showing similarity and clustering of M. maximus genotypes.
The mean value for plant height among all the genotypes ranged from 45.67 to 176 cm whereas the cluster means ranged from 81.36 to 117.24 cm (Table 2). The tillers per tussock ranged from 6 to 82. The stem diameter varied from 0.24 to 0.93 cm. High degree of variation was noted for internodal length which ranged from 9.5 to 37.8 cm. The internodal length was positively correlated with plant height (r = 0.255, P ≤ 0.01). The genotypes exhibited variation from 10.5 to 77.8 cm for leaf blade length and 0.87 to 3.10 cm for leaf blade width. Leaf sheath length ranged from 9 to 23.25 cm whereas leaf sheath width ranged from 0.5 to 3.3 cm and both the traits appeared to be positively correlated with leaf blade length and width as per cluster means. Flag leaf length and width varied from 1.5 to 26 cm and 0.27 to 2.20 cm, respectively. Flag leaf sheath length ranged from 2.60 to 44.60 cm. Cluster 8 was noted for the shortest flag leaf sheath length. Distance from the last node to the beginning of the panicle varied among the genotypes from 1.2 to 56.5 cm. Similarly, the panicle length varied from 11.65 to 40.70 cm. Spread of the panicle as reflected by the panicle branch length varied from 5.6 to 27.0 cm.
Table 2. Cluster mean performance of morphological traits of genotypes of M. maximus and principal components of the traits
Ch1, plant height; Ch2, tillers/tussock; Ch3, stem diameter; Ch4, internodal length; Ch5, node extension; Ch6, node hair length; Ch7, node hair extension; Ch8, leaf blade length; Ch9, leaf sheath length; Ch10, leaf blade width; Ch11, leaf sheath width; Ch12, ligule hair extension; Ch13, flag leaf blade length; Ch14, flag leaf sheath length; Ch15, flag leaf blade width; Ch16, flag leaf sheath width; Ch17, length between node to panicle base; Ch18, panicle length; Ch19, panicle branch length.
PCA was done to identify the key characters that contribute to the diversity. The PCA revealed that the first 10 principal components accounted for a total of nearly 77% variability cumulatively (Table 2) and the first 14 principal components accounted for more than 90% variability. Scree plot analysis revealed that the first three principal coordinates or up to seven principal components can be retained for explaining most of the diversity (Fig. 2).
Fig. 2. Scree plot of M. maximus genotypes.
The nutritional parameters recorded among genotypes also showed a high degree of variation (Table 3). Ash content varied from 6.8 to 16%. Similarly, the organic matter content also showed substantial variation from 84.0 to 93.2%. Crude protein ranged from 3.7 to 7.4% on a dry matter basis among the genotypes. Fibre content also showed high genetic variability. NDF ranged from 72.4 to 86.8% whereas the ADF ranged from 38.8 to 64.38%. Hemicellulose, cellulose and lignin content also varied in the range of 11.5–41.5%, 30.0–49.8% and 4.0–12.8%, respectively. These nutritional parameters contributed to overall IVDMD, which ranged from 26.3 to 50.0% among the genotypes.
Table 3. Cluster mean performance of nutritive traits of genotypes of M. maximus
OM, organic matter; CP, crude protein; NDF, neutral detergent fibre; ADF, acid detergent fibre; Hemicell, hemicellulose; Cell, cellulose; Lig, lignin, IVDMD, in vitro dry matter digestibility.
Data were also recorded for a few non-metric traits such as hairiness of leaf and panicle, colour of stem, node and leaf as well as flowering behaviour (Table 4). Almost an equal number of genotypes were having green or green violet stem colour, whereas it was yellowish green in 11 genotypes (Table 4). Genotypes with green stem showed maximum tillering and the marginally higher average CP content and IVDMD% than the other two groups (Table 4). Violet green node was represented among 98 genotypes followed by green nodes among 35. Yellow nodes were present in two genotypes only. Genotypes with violet green and green nodes were noted for a high CP content whereas a high IVDMD was for genotypes with violet nodes. Dark green leaf colour, represented among 16 genotypes, was associated with a high CP content and a high IVDMD. Green leaf coloured genotypes were among those with the next high values for CP%, whereas the light green leaf coloured genotypes were those with the second highest value for IVDMD. Number of the genotypes with dense hairy, glabrous and hairy leaf sheath was 14, 81 and 57, respectively, whereas genotypes with dense hairy, glabrous and hairy leaf blades were 6, 46 and 100, respectively. CP content was higher among the genotypes with glabrous or hairy leaf and leaf sheath; however, IVDMD was high among the genotypes with glabrous leaf sheath and those with dense hairy leaves. Leaf sheath was mostly glabrous among the genotypes of the Clusters 4, 5 and 7. Among the germplasm collection, 86 genotypes flowered throughout the year after each harvest, whereas the 66 genotypes flowered once in a year, i.e. mostly in the November after the cessation of monsoon period. Among morphological clusters, Clusters 4 and 5 had round-the-year flowering genotypes, whereas Clusters 6, 7 and 8 had all the genotypes flowering once in a year. The other clusters consisted of both types of genotypes. Panicle hairiness is also an important distinguishing trait and the six genotypes showed hairy panicle.
Table 4. Non-numeric morphological trait groups of M. maximus and their average nutritive values
DH, dense hairy; H, hairy; GL, glabrous; DG, dark green; G, green; LG, light green; NH, non-hairy; VG, violet green; YG, yellow green; Y, yellow; V, violet.
Looking into the number of genotypes and the variability present, the size of the core was set at 15% of the total germplasm. For the formation of the core sub-set, the re-clustering was done for the Clusters 1–6, and from each sub-cluster, one random accession was selected for the core sub-set. Thus, four, three, five and two accessions were taken from Clusters 1, 2, 3 and 4, respectively (Table 1). Additionally, three accessions each from Clusters 5 and 6, and one and two accessions each from Clusters 7 and 8 were taken to form a sub-set of total 23 accessions. This set included 20 indigenous accessions (IG 01-115-1, IG 01-116, IG 01-102-1, IG 01-85, IG 01-89, IG 01-185, IG 01-196, PC 330, PC 313, IG 01-152, IG 01-225, PC 329, N 5/6 03-429, PGG 617, MS 4691, PGG 2005, LS, 5/2 N 03-428, IG 01-206, IG 01-214) and three exotic accessions (IG 97-7, IG 97-48, IG 97-35). Comparison of the mean value for all the 19 traits as obtained in the core value and that of all the accessions by Student's t-test was found to be non-significant, thus indicating a non-significant variation. The range of diversity for the 19 traits in total germplasm and that in the selected core germplasm is presented in Table 2. It indicated that most of the genetic diversity was captured in the selected core germplasm. The mean value of the core was almost equal to the mean value of total germplasm for several traits. Among the core sub-set, almost all the formats of the non-metric traits, such as stem colour, node colour, leaf sheath hairiness, leaf colour, leaf sheath hairiness and inflorescence hairiness, were represented. Similarly, the nutritive traits among the core germplasm were representative of the whole range of diversity for the nutritional parameters (Table 3). For geographical origin, it was found that out of 23 genotypes selected as core, 12 belonged to the sub-humid climate of south India; three to exotic origin from Ethiopia; five to western India and two from northern India and one from the north-eastern hills.
Discussion
An insight into the germplasm of Guinea grass evaluated through 35 traits revealed a high degree of genetic variability. Descriptive statistics was done to get the univariate summary statistics for variables. It showed that a wide range of variation existed for the traits among the genotypes collected from various sources, which is crucial for any crop improvement programme. Clustering of the genotypes, based on morphological traits, clearly distinguished the genotypes. Clustering of genotypes as per their place of collection was also apparent; for example, most of the southern India genotypes clustered into three clusters. Similarly, the genotypes from the northern semi-arid region clustered into three clusters, and thus, were well separated from the remaining five clusters. Genetic diversity, based on morphological markers, showed significant differences for various morphological traits among accessions of Guinea grass by Sudrik et al. (Reference Sudrik, Abraham, Thomas, Roy, Kumar, Agrawal, Mahanta, Singh, Das, Dwivedi, Prabhu and Shah2015) along with no parallelism between geographic and genetic diversity. Ramakrishnan et al. (Reference Ramakrishnan, Babu, Iyanar and Manivannan2019) also reported that a considerable amount of genetic diversity existed among the accessions along with some accessions having unique characters which may be effective in future breeding. However, in these earlier studies, a limited number of germplasm were included. Earlier studies involving tropical perennial grasses indicated a wide range of diversity for different characters, and the clustering pattern was observed to be independent of their geographical distribution in Dichanthium annulatum (Agarwal et al., Reference Agarwal, Gupta, Roy and Gupta1999; Chauhan et al., Reference Chauhan, Tiwari, Roy, Kaushal, Malaviya, Chandra and Mahanta2007), Sehima nervosum (Roy et al., Reference Roy, Agarwal and Gupta1999), Heteropogon contortus (Roy, Reference Roy2004; Bhat and Roy, Reference Bhat and Roy2007, Reference Bhat and Roy2014) and Guinea grass (Jain et al., Reference Jain, Roy, Kaushal, Malaviya and Zadoo2003a, Reference Jain, Zadoo, Roy, Kaushal and Malaviyab, Reference Jain, Roy, Kaushal, Malaviya and Zadoo2006).
Wide variation for plant height among the genotypes was established along with a high variation for tillers per tussock. These traits significantly contribute to biomass production. Stem diameter also contributes significantly to the dry matter production of the genotypes, and the genotypes showed variation for the trait. The genotypes differed profoundly for leaf size. The clusters with more plant height possessed longer and wider leaves. de Wouw et al. (Reference de Wouw, Jorge, Bierwirth and Hanson2008) considered the leaf width as a selection criterion in Guinea grass while extending their finding of the leaf width having high heritability and positive correlation with culm mass in P. coloratum. Owing to a long establishment period, the robust plant types of Guinea grass were not suitable for swards under grazing (Boonman, Reference Boonman1992), although such genotypes are suitable for cultivation. These robust types can also be good alternatives to Napier grass or Bajra × Napier hybrid in the cut-and-carry system. Such genotypes with upright, broad leaves and superior dry matter production, compared to popular cultivars, have also been recommended in Thailand (Hare, Reference Hare2020). Accessions with wider leaves were more common in areas with high precipitation (de Wouw et al., Reference de Wouw, Jorge, Bierwirth and Hanson2008). Longer and wider leaf sheath genotypes clustered with taller genotypes. Small fine-leaved accessions were reported to be present in drier areas (de Wouw et al., Reference de Wouw, Jorge, Bierwirth and Hanson2008). Thus, the accessions with such traits in this study can be evaluated for adoption in the drier areas of arid and semi-arid zones.
Although flag leaf length and width did not show a definite trend with plant height, many tall genotypes possessed short flag leaves. Flag leaf is an important trait with respect to the seed health/vigour because at the time of seed development, the flag leaf becomes the source for the required nutrition supply. Hence, it is possible that the genotypes with larger flag leaf areas may have better seed setting ability. Plant height, number of tillers and leaf length were considered as important traits while constructing selection criteria for forage yield in other tropical perennial grasses (Roy et al., Reference Roy, Agarwal and Gupta1999; Roy, Reference Roy2004). The genotypes possessed high genetic variation for panicle traits. Some genotypes possessed very long spreading type panicles, whereas others had small panicles. The long panicle with good spread has a low chance of self-pollination and may result in low ovule-to-seed ratio. PCA revealed the first 10 principal components to account for the major amount of variability cumulatively. Martuscello et al. (Reference Martuscello, dos S Braz, Jank, da Cunha D. de and da Fonseca2012) have also reported, in PCA of half-sib progenies of Guinea grass, that 84.3% variation accumulated out of four components was sufficient for the formation of morphological groups.
Nutritional parameters of the forage grasses are important selection criteria. The present study revealed a high genetic variability for the various nutritional parameters observed at the 50% flowering stage. Among the various parameters, crude protein is an important attribute. The genotypes in the present study showed high variation for the trait. In an earlier study on Guinea grass germplasm collections from north-eastern and southern India, the range of variation for crude protein was reported to be from 4.6 to 6.8% along with ADF% and NDF% to range from 30 to 45% and 60 to 68%, respectively (Malaviya, Reference Malaviya2001). Among the grasses, forage quality is adversely affected with flowering/maturity. The nutritional parameters contribute to overall IVDMD, which also varied among genotypes. Mahala et al. (Reference Mahala, Nsahlai, Basha and Mohammed2009) reported that with the advancing maturity of the plants, the NDF, ADF and lignin proportions increase while CP content decreases. Johnson et al. (Reference Johnson, Hardison and Castillo1967) observed that fibre content in Guinea grass increased by 7.3% from 2½ weeks to 2½ months after planting/harvesting. The variations in the nutritional attributes could also be because of the differences in ecotype, herbage growth stage (i.e. days after sowing/planting/harvesting) or the geographical zone (Ismail et al., Reference Ismail, Fatur, Ahmed, Ahmed and Ahmed2014). High nutritive value, annual leaf yield and dry season leaf yield were considered as the important traits for identifying the cultivars of Guinea grass in Brazil (Fernandes et al., Reference Fernandes, Ramos, Jank, Carvalho, Martha and Braga2014). High intake rate indicates its higher digestibility (Fernandes et al., Reference Fernandes, Ramos, Jank, Carvalho, Martha and Braga2014) and leafy swards provide more suitable intake conditions (Benvenutti et al., Reference Benvenutti, Gordon, Poppi, Crowther and Spinks2008). Hence, for developing grazing-type Guinea grass, leafy types should be preferred. Considering these facts, Stabile et al. (Reference Stabile, Bodini, Jank, Renno, Santos and Silva2012) suggested breeding for higher dry matter digestibility without affecting dry matter yield.
The non-metric traits such as hairiness of leaf, panicle and nodes also effectively differentiated the genotypes. Additionally, these traits were also sometimes associated with the nutritional quality and tolerance to abiotic and biotic stresses. For example, shade induces high chlorophyll content (Malaviya et al., Reference Malaviya, Roy and Kaushal2020a) and crude protein content (Kaushal et al., Reference Kaushal, Malaviya and Singh2000). The genotypes with dark green and green colour can be preferred over light green coloured leaves for high nutritional quality because of their association with a high CP and IVDMD. Violet green and green nodes were indicative of a high CP content whereas the high IVDMD was found among the genotypes with violet nodes in the present study. In general, the genotypes with dense hairy leaf and leaf sheath possessed a low CP content. However, Martuscello et al. (Reference Martuscello, dos S Braz, Jank, da Cunha D. de and da Fonseca2012) reported that the pubescence characteristics were the least important in the discrimination of the genotypes. Among the non-metric quantitative traits, stem colour and node were found to be represented in all variants among all the major clusters. In the clustering pattern, the genotypes with glabrous leaf sheath were mostly present in the Clusters 4, 5 and 7; however, such distinction was not observed for leaf hairiness, and among most of the clusters, the genotypes with glabrous, hairy, less hairy leaf were mixed. Flowering initiation in general is linked to deterioration in forage quality. Guinea grass genotypes usually exhibit two distinct flowering behaviours. Some genotypes flower throughout the year, whereas the others flower once in a year. The genotypes flowering once in a year showed slightly better CP and IVDMD values. The genotypes which flower once in a year can also prove to be less invasive because of restricted spread through seeds.
Breeding of any crop depends upon identification of suitable genotypes as parents or for direct utilization. Such genotypes must be genetically diverse. Hence, the present set of germplasm was systematically characterized encompassing both quantitative and nutritional traits, and a core sub-set developed. Size of the cluster has been suggested to be 10% of the total germplasm (Brown, Reference Brown, Brown, Frankel, Marshall and Williams1989b) and 5–10% by Charmet and Balfourier (Reference Charmet and Balfourier1995) and Bisht et al. (Reference Bisht, Mahajan, Lokknathan and Agrawal1998) which allows capturing 75–90% of the diversity. Noirot et al. (Reference Noirot, Messager, Dubos, Miqucl and Lavorez1986) have suggested 15% of the base collection for core because higher percentage is needed particularly when the objective is to capture the genetic diversity of quantitatively inherited characters. Hence, looking into the number and variability of the accessions (mostly quantitative traits studied), size of the core was set at 15% of the total germplasm. The members of core were selected based on the group size of primary clusters as per the logarithm strategy of Brown (Reference Brown1989a) which proposed to allocate entries to each cluster in proportion to the logarithm of the number of accessions in that particular cluster. For the formation of the core sub-set, the re-clustering was done for the Clusters 1–6 and from each sub-cluster; one random accession was selected for the core sub-set following Pandiyan et al. (Reference Pandiyan, Senthil, Packiaraj and Jagadeesh2012). A core sub-set of 23 germplasm thus formed possessed 20 indigenous accessions and three exotic accessions. The non-significant variation between the variability present in the core and that of total germplasm indicated successful capturing of maximum diversity in the core sub-set. Similarly, t-test comparison of the mean value for all nutritive traits as obtained in the core value with that of total germplasm indicated non-significant variation. Thus, most formats of the non-metric traits, whole range of nutritive traits and geographical origin and almost the whole range of metric traits were represented in the core sub-set.
The present study, thus, confirmed a high degree of genetic variability. Roy et al. (Reference Roy, Malaviya and Kaushal2019b) reported that Guinea grass germplasm is highly heterozygous and heterogeneous, suggesting the existence of a large amount of genetic variability within the species. High levels of genetic diversity for various traits including biomass yield have been reported from India in earlier studies (Malaviya, Reference Malaviya1995, Reference Malaviya1996, Reference Malaviya1998, Reference Malaviya2001; Malaviya et al., Reference Malaviya, Kaushal and Kumar2006). The species diversity is also reported from East Africa (Combes and Pernès, Reference Combes and Pernès1970). The species shows a high degree of variation for its leaf anatomy and the genotypes exhibit variations for stomatal characters and photosynthetic pathways (Malaviya et al., Reference Malaviya, Roy and Kaushal2020a). Cytological and isozyme variations have also been recorded in germplasm (Jain et al., Reference Jain, Zadoo, Roy, Kaushal and Malaviya2003b, Reference Jain, Roy, Kaushal, Malaviya and Zadoo2006). Because of such intra-species variability, the species shows adaptability to diverse agro-climatic conditions and abiotic stress tolerance. The grass is known for its shade and salinity tolerance (Kaushal et al., Reference Kaushal, Malaviya and Singh2000; Malaviya et al., Reference Malaviya, Kaushal and Kumar2006, Reference Malaviya, Roy, Anand, Choubey, Baig, Dwivedi, Kushwaha and Kaushal2019, Reference Malaviya, Baig, Kumar and Kaushal2020b) and can fit well into agroforestry systems especially under orchards (Seresinhe and Pathirana, Reference Seresinhe and Pathirana2000).
Genetic improvement of Guinea grass through hybridization has been limited with a few reports of breeding for seed production only (Noirot, Reference Noirot1985; Noirot et al., Reference Noirot, Messager, Dubos, Miqucl and Lavorez1986). The occasional presence of sexual plants in nature has been a source of wide variability present in natural forms. In earlier studies, variability in Guinea grass was also attributed to the occasional presence of sexual forms through diploid-tetraploid-dihaploid cycles (Savidan and Pernes, Reference Savidan and Pernes1982). A molecular diversity study on Guinea grass accessions did not indicate a significant association between genetic and geographical variation, thus indicating the role of the possible natural crossing among the species of agamic complex P. maximum, P. infestum and P. trichocladum (de Sousa et al., Reference de Sousa, Jank, de Campos, Sforça, Zucchi and de Souza2011). Based on molecular phylogenetic studies, Zuloaga et al. (Reference Zuloaga, Salariato and Scataglini2018) reported close relatedness among Guinea grass, P. trichocladum, Eriochloa punctata and Urochloa plantaginea. Sexual line in Guinea grass was first reported by Combes and Pernès (Reference Combes and Pernès1970) in diploid accessions from East Africa followed by sexual tetraploids from South African apomictic accessions (Smith, Reference Smith1971; Hanna et al., Reference Hanna, Powell, Mlillot and Burton1973). Manipulating apomixis was suggested as a tool in plant breeding by Savidan (Reference Savidan1986) and Savidan et al. (Reference Savidan, Jank, Costa and do Valle1989); and the concept of Hybridization-supplemented Apomixis-component Partitioning Approach (HAPA) provided a clear pathway for breeding apomictic grasses (Kaushal et al., Reference Kaushal, Agarwal, Malaviya, Siddique and Roy2008b). Apomictic seed development through various reproductive pathways, utilizing partitioning components of apomixis, led to the development of the largest ploidy series in Guinea grass (Kaushal et al., Reference Kaushal, Malaviya, Roy, Pathak, Agrawal, Khare and Siddiqui2008a, Reference Kaushal, Agarwal, Malaviya, Siddique and Royb, Reference Kaushal, Paul, Saxena, Dwivedi, Chakraborti, Radhakrishna, Roy and Malaviya2015, Reference Kaushal, Dwivedi, Radhakrishna, Srivastava, Kumar, Roy and Malaviya2019) which has been acknowledged in Limca Book of Records as the largest series in any particular crop (Kaushal et al., Reference Kaushal, Roy, Malaviya, Dwivedi, Chakraborti, Radhakrishna, Paul and Saxena2020). The recent studies have shattered earlier perceptions that polyploid species represent evolutionary dead-ends; instead ancient polyploidy events are often associated with major clades (Soltis et al., Reference Soltis, Visger and Soltis2014; Van de Peer et al., Reference Van de Peer, Mizrachi and Marchal2017). Polyploidy has been suggested to play a central role in shaping and restructuring plant genomes. These findings provide an insight into the variability in a polyploid and largely apomictic species. The other possible source of natural variation could be habitat divergence playing a driving role in speciation as suggested by Lavania (Reference Lavania2020) and spontaneous hybrid reported between Guinea grass and P. infestum (Embrapa, 2001).
Thus, the study successfully established the presence of wide diversity in the Guinea grass gene pool, which can be suitably utilized in the breeding programme of the grass. Evaluation of a large number of genotypes in artificially created abiotic stress conditions, such as shade, moisture and salinity, requires huge resources. Hence, evaluation of a limited number of germplasm, identified as a core sub-set of the genotypes in the study, will facilitate the breeders in saving resources. Intra-species hybridization among the diverse genotypes with desirable traits can lead to cultivars with improved biomass yield and quality. Interspecies compatible crosses, involving the agamic complex of P. maximum, P. infestum and P. trichocladum (de Sousa et al., Reference de Sousa, Jank, de Campos, Sforça, Zucchi and de Souza2011), can be another way to improve Guinea grass. Additionally, morphological characterization of the genotypes together with molecular studies, in future, can further enrich knowledge on genetic diversity and genetics of the complex traits by combining the phenotypic dissimilarity matrix and the genotypic dissimilarity matrix as followed in other species (Cruz et al., Reference Cruz, Ferreira and Pessoni2011; Alves et al., Reference Alves, Bhering, Rosado, Laviola, Formighieri and Cruz2013). Such studies have greater relevance in the grasses because of the limitation of generating a mapping population due to the prevalence of apomixis. Hence, considering a population genomics association approach along with a genotype–phenotype association analysis following Talbot et al. (Reference Talbot, Chen, Zimmerman, Joost, Eckert, Crow, Semizer-Cuming, Seshadri and Manel2017) can be an option for further genetic studies in the grass.
Acknowledgements
The authors thank the National Bio-resource Development Board, Department of Biotechnology, Government of India and Indian Council of Agricultural Research for financial support.
