Morphological variability and genetic diversity of Aman rice germplasm of Bangladesh cultivated in Mymensingh region

Fahmida Reza Emi1, Hadisa Khatun1, Fahmina Yasmine2, Ahmed Khairul Hasan1* & Mohammad Anwar Hossain3* 1Department of Agronomy, Bangladesh Agricultural University, Mymensingh-2202, Bangladesh 2Bangladesh Institute of Nuclear Agriculture, Mymensingh, Bangladesh 3Department of Genetics and Plant Breeding, Bangladesh Agricultural University, Mymensingh-2202, Bangladesh *Email: akhasan@bau.edu.bd; anwargpb@bau.edu.bd


Introduction
Rice (Oryza sativa L.) is popularly called as 'global grain'. It is the staple food of about 156 million people of Bangladesh (1) which constitutes 95% of food grain production in the country. It is the easiest source of food and commodity of significance for the people of Bangladesh (2). It belongs to the family Gramineae. The amount of rice grown in Asia is about 90% of the world's rice which alone covers the food demand for about 60% population of the world (3). Bangladesh ranks fourth among the 114 rice growing countries of the world with the production of 52.3 million metric tonnes whereas China is the first with the production of 211.09 million metric tonnes (4). As rice plays a very important role in traditional diets and in the livelihood of people, it earns a special position in many nations. Three types of rice grown in Bangladesh, viz., Aus, transplanted (T.) Aman, broadcast (B.) Aman and Boro depend on the wide environmental diversity which results from considerable variation in topographic and seasonal components. Land used for cultivating Aus, T. Aman and Boro season is 1.08, 4.86 and 5.68 million ha, respectively (4). Although the area under Aman rice is increased, the area for cultivation of local varieties is not increased due to its low yielding ability. Out of 5.68 million ha, only 0.94 million ha of land are used for cultivation of local varieties (5). As a result, the local varieties have been disappeared day by day.
Local germplasm of Aman rice plays an indispensable role in the local food security as well as sustainable development of agriculture (6) as germplasm provides adaptable genes for the changing environment and inclusion of these genes from landraces can ensure expected grain production.
Many of the local varieties adapting in different parts of the country have very good quality, fineness, aroma, taste and high protein content (7). These also include sources for simply inherited resistance or tolerance against different biotic and abiotic stresses and provide complex traits for further improvement of grain quality and yield. So, the conservation of germplasm of local Aman rice varieties is very important. On the other hand, food is one of the security problems faced by the world in the 21st century. Among all food items, the significance of rice is supreme. The global rice demand will be 880 million tonnes in 2025 which is about 70% more than present production (8). In order to meet up the demand, two ways can be effective: expanding the rice growing area and increasing productivity or both (9). As there is no possibility of increasing cultivated area, development of high yielding varieties is the only option to increase the yield of Aman rice.
The future strategies in the improvement of rice largely depend on the availability of genetic resources and their effective utilization. At present, varietal identification for rice are primarily based on morphological and physiological parameters. During the development of varieties, the plant breeders who uses several techniques for creating genetic variation and selecting from within that variation mainly uses the superior plants after final selection (10). Parental genotypes from random populations are selected based on the information about genetic diversity. Besides genetic diversity, crop yield not only depends on the different yield contributing traits but also on the factors involved about environment. In case of rice, higher yield also depends on the higher number of effective tillers/hill, number of filled grains/panicle and 1000-grain weight (11). These traits are also correlated among themselves. The magnitude of genetic variability like as, phenotypic and genotypic variances, phenotypic coefficient of variation (PCV) and genotypic coefficient of variation (GCV), broad sense heritability, genetic gain (GA) is highly responsible for the genetic improvement of any crop based on which the breeding methods are constructed for its additional improvement. The extent of variability is measured by GCV and PCV. Correlation analysis is used to determine the association among the different traits (12). Principal component analysis (PCA) marks out the plant characters that provide the most in creating variation within a group of entries. On the other hand, the relationships between yield and yield contributing traits can be analyzed through path-coefficient (13) which explains the real causes of relationship. The above tools are found to be useful for proper characterization in further breeding program. Besides genetic characterization, qualitative characterization of Aman rice germplasm is essential for utilizing the appropriate attribute based donors for further study and also crucial to the present era for protecting unique rice. Plant characterization has been effectively used in searching out appropriate attributes among individual genotypes which is possible through distinctness, uniformity and stability (DUS) test (14).
Elucidation of morphological and genetic characterization and assessment of diversity is very crucial for future varietal improvement. There is no comprehensive report regarding the morphological and genetic variability and diversity of cultivated Aman rice germplasm of Mymensingh region. Lack of access to phenotype information is still seen as a limiting factor for the use of plant genetic resources. Therefore, the present study was conducted to characterize Aman rice germplasm based on morphological traits, to assess the nature and magnitude of genetic diversity, to study the relationship, heritability and genetic advances of yield and yield contributing traits and also to identify the suitable traits for varietal improvement.

Plant materials, exploratory site and design
The experiment was set up at the exploratory Farm of BINA, Mymensingh following randomized complete block design (RCBD) with three replications. was used as plant materials which were collected from different places of Mymensingh region. The unit plot size was 2 m× 3 m. Twenty-five days old seedlings were used for transplanting @ one seedling hill -1 with the space of 15 cm × 20 cm. Recommended doses of fertilizers and manures were applied to the soil. Other intercultural operations were done whenever necessary.

Data recording
Five plants from each germplasm were selected randomly for data collection on different quantitative traits viz. plant height, days to 50% flowering, days to maturity, total tillers/hill, effective tillers/hill, panicle length, filled grains/panicle, unfilled grains/panicle, 1000-grain weight and yield/plant. Qualitative traits were collected in terms of blade pubescence, blade color, leaf sheath: anthocyanin color, basal leaf sheath color, leaf angle, flag leaf angle, ligule color, ligule shape, collar color, auricle color, culm anthocyanin color, culm angle, internode color, culm strength, panicle type, secondary branching, panicle exertion, axis, shattering, threshability, awn: distribution, awn color, apiculus color, stigma color, lemma and palea color, lemma and palea pubescence, sterile lemma color, seed coat color, endosperm type and decorticated grain: scent.

Statistical analysis
Statistical analysis was done for the data on different quantitative traits by using MINITAB ® 17 statistical software packages following a RCBD design. Estimation of the genotypic variance (GV) and phenotypic variance (PV), broad sense heritability (%), GA was done following the standard formulae (15); GCV and PCV values were calculated (16); GA (%) was also analyzed based on the formula (17).Transformed uncorrelated means of characters were used to calculate the Mahalanobis distance (D 2 ) values (18). Pearson correlation matrix, path analysis, PCA and clustering were also performed using statistical software R, version 3.3.2.

Results and Discussion
Evaluation of performance of the studied rice germplasm for yield contributing traits The differences among the germplasm for all the studied traits were highly significant (p<0.01) ( Table 1) which is an indication of genetic diversity of the germplasm. There are reports on highly significant variability for all the studied traits in their experiment (19,20). Significant (p<0.01) variations were seen for individual traits which were also observed by many other researchers for plant height (21)(22)(23); for days to 50% flowering (24,25); for days to maturity (26); for no. of total tillers/hill (27,28); for no. of effective tillers/hill (24,29); for grain yield/plant (30,31).
In this study, Ranishail was the tallest (134.67 cm) among the rice germplasm and Lalmatia (71.33 cm) was the shortest one. These variations in plant height are mostly due to different genetic makeup of the germplasm. In this investigation, it was found that Shongbinni was the late flowering (118 days) germplasm and Sentu-5 was the earliest (97.67 days) flowering one because of the influence of genetic inheritance and availability of nutrient content in soil. In case of maturity, Sentu-17 required the shortest time for maturity whereas Gaindha, Mery gold and Ranishail were the late maturing genotype. Days to maturity ranged from 135.33 to 151.33 days. The results were almost similar to the earlier findings (22). Germplasm viz. Sentu-17 might be useful in future breeding program to develop short duration variety.
With the advancement of growth stages, the tiller number of the varieties increased. But it was not PLANT SCIENCE TODAY 974 consistent because reduced number of tillers/hill was observed in some of the germplasm at harvest. The reason behind the reduction was the death of some tillers. Number of total tillers/hill ranged from 8.67 (Pairjaat) to 18.00 (Sentu-16) and the mean was 14.25.
On the other hand, effective tiller number depends on the number of tillers produced (32) which is the most important parameter for rice. The number of effective tillers/hill was found the highest (16.50) for Sentu-16 and the lowest (5.33) for Fullota in this study as the varieties are genetically different which was primarily influenced by heredity. Panicle length directly controls the yield of a particular variety (33). The longest panicle (28.83 cm) was recorded in Gaindha and the shortest (22.33 cm) in Rupashail. These variations in length of panicle might be due to their genetic makeup. The findings were in line with Hoque and Khanam (29,34).
Ranishail possessed the highest number of filled grains/panicle (250.11) which may be due to lower sensitiveness of this genotype to heat stress (e.g. excessive temperature, low sunshine hour) at grain filling stage. On the other hand, the lowest value for filled grains/panicle (70.03) was recorded for Gaindha. The results found for this trait was similar to an earlier work (24). In case of unfilled grains/panicle, Sentu-5 contained the maximum (66.67) unfilled grains/panicle and Purabinni had the minimum (9.11). Hoque and Khanam also observed similar type of result for this trait (29,34). The more the thousand grain weight, the more the grain yield. Thousand grain weight ranged from 9.38 to 31.420 g. This trait is used for characterizing rice varieties (35,36). Grain yield/plant is the ultimate goal of rice production. Grain yield/plant ranged from 4.91 to 10.33 g with a mean 8.42 g. Sentu-16 had the grain/plant in maximum quantity and the lowest amount was found in Gaindha. So, the findings of this study would be useful in finding out the suitable genotype for future research as characterization through different morphological traits is a key step for assessment of genetic potential of rice.

Estimation of genetic variability, heritability and genetic advance of the studied rice germplasm for yield contributing traits
The results of Table 2 express the result for phenotypic and genotypic variances, PCV and GCV, heritability, GA and GA% of the studied traits. The values of σ 2 G and σ 2 P were the highest for filled grains/panicle and the other traits had the lowest value in this case. Besides, the values of genotypic variance were inferior to phenotypic variance for all the studied quantitative traits, expressing the optimum environmental effect on yield/plant ( Table  2). Earlier works also reported the same result in their study (37)(38)(39). The highest differences for PCV and GCV were observed for total tillers/hill, panicle length and grain yield/plant; rest of the traits had lower differences between GCV and PCV. The results were consistent with the results of earlier works (37,(40)(41)(42). In order to predict the reliability of phenotypic value, heritability was measured. Table 2 presents high broad sense heritability (h 2 b) for all the studied traits except number of total tillers/hill and grain yield/plant. Many earlier researchers also reported high heritability for these traits (43)(44)(45). High GA in percentage of mean was observed in effective tillers/hill, panicle length, filled grains/panicle, unfilled grains/panicle and 1000-grain weight. Pandey and Anurag also recorded high GA% in their study for these traits (46). High heritability coupled with high GA (%) was observed for effective tillers/hill, panicle length, filled grains/panicle, unfilled grains/panicle and 1000-grain weight. Similar trend of high heritability with high GA for different traits in rice was also reported earlier (47). Traits having high heritability (h 2 b) accompanied by high GA might be more advantageous in predicting gain than heritability alone. So the findings of this study might be useful in order to find out the expected traits for crop improvement programmes. Table 3 indicates the relationship between the phenotypic and genotypic correlation coefficients among different yield and yield contributing traits. In most of the cases, genotypic correlation coefficients were greater than the phenotypic correlation coefficients suggesting that the association was mainly due to genetic reason. The findings were corroborated with the earlier report (40). Grain yield/plant was positively and significantly correlated with plant height, total tillers/hill, effective tillers/hill, filled grains/panicle and 1000-grain weight which indicate the simultaneous improvement of crops. Similar kind = Phenotypic variance, σ 2 g = Genotypic variance, PCV = Phenotypic coefficient of variation, GCV = Genotypic coefficient of variation, h 2 b= Heritability, GA= Genetic advance and GA (%) = Genetic advance in percentage of mean of association was also revealed in the works of other researchers (48,49). Plant height exhibited positive and highly significant correlation with total tillers/hill, effective tillers/hill, filled grains/panicle, 1000-grain weight and grain yield/plant. Similar trend of findings was also reported previously (48,50). In case of days to 50% flowering, the relationship was positively significant with only days to maturity (51,52). Total tillers/hill was positively and significantly correlated with effective tillers/hill, filled grains/panicle, 1000grain weight and grain yield/plant. A similar result was found by other researchers for effective tillers/hill (24,53) and for grain yield/plant (24,48) in previous experiments. Regarding effective tillers/hill, the correlation was highly significant and positive with filled grains/panicle, 1000-grain weight and grain yield/plant. On the other hand, panicle length showed negative but significant association with filled grains/panicle and grain yield/plant. Similar results were recorded for the association of panicle length with other yield contributing traits (48,54). For filled grains/panicle, positively significant correlation was found with 1000-grain weight and grain yield/plant which was also revealed for grain yield (50,55). Thousand grain weight expressed positive and highly significant correlation with grain yield/plant. The result was in line with the results of some earlier experiments (48,55). The positive and significant correlation between the traits suggested that the association was less affected by the environmental factors. So, these traits could be used as criteria for selection for higher yield.

Character association between yield and yield contributing traits
The correlation values are divided into direct and indirect effects through path coefficient analysis for measuring the relative importance of associated causal factors (Table 4 and 5). The values from the table implied that days to 50% flowering, effective tillers/hill, panicle length, filled grains/panicle and 1000-grain weight exerted positive direct effect on grain yield/plant at both phenotypic and genotypic level. Similar trend of result was reported earlier and found that number of effective tillers/hill and filled grains had positive direct effect on grain yield/plant (49,56). Direct positive effect of previously mentioned traits on grain yield/plant suggested that the selection of these traits would be effective for improving grain yield in rice. Number of effective tillers/hill exerted the highest direct positive effect on grain yield/plant at both phenotypic and genotypic level (57). In case of   plant height, days to maturity, total tillers/hill and unfilled grains panicle, the effect was direct and negative at both levels which suggested that the direct selection of these traits would not be helpful. The residual effect decides how the best causal factors explain the variability of resultant factor (grain yield/plant). The value for residual effect of phenotypic path coefficient was 0.103, which suggested that the nine characters contributed 89.7% of variability in grain yield/plant in phenotypic path analysis. The value for residual effect of genotypic path coefficient was 0.036 indicated that the nine characters contributed 96.4% of variability in grain yield/plant studied in genotypic path analysis. This is due to non-significant correlation of some traits with yield.

Nature and magnitude of genetic diversity for the yield and yield contributing traits
Different characters contributed differently to the total variation as indicated by their eigen vectors as well as their loading on the different principal axes were known from the results of the PCA ( Table 6). The first three components accounted for 79.57% of the total variation with eigen value of > 1.0 which was matched with the information provided (58, 59). The first component accounted for 48.46% of the total variation where grain yield/plant had the highest positive value followed by filled grains/panicle, effective tillers/hill, 1000-grain weight, plant height and total tillers/hill whereas the second component accounted for 19.94% of the total variation explained mostly by days to 50% flowering, days to maturity and unfilled grains/panicle. The findings were similar with the earlier studies (24,59,60). For PC3, days to maturity, total tillers/hill and panicle length contributed 11.18% of total variation with panicle length (0.63) given the highest contribution. The result was in consent with the exploratory results of some other researchers (61)(62)(63). So, the results expressed that extensive genetic variation was existed in the germplasms of this study. Resulting germplasm of this study will be more effective in finding out the parents for the improvement of yield in rice.
The germplasm were grouped into 5 clusters based on D 2 valus (Table 7 and Fig. 1) where cluster III and IV had 8 germplasm where each of them covering 25.81% of the total germplasm which were considered as the largest clusters whereas cluster V was the smallest cluster covering 12.90% of the total germplasm indicating the presence of genetic diversity. Similar clustering and high degree of genetic diversity were also reported by others in rice (64,65,66). The rice germplasm accumulated in the same cluster indicate that they are not sharply diversified. Cluster I and II covered about 16.13 and 19.35% of the total studied germplasm, respectively. The results also revealed that no duplication was present for the studied traits.
In most of the cases, the intra-cluster distance was smaller than the inter-cluster distance which indicates that the diversity among the germplasm of this group was broad (Table 8 and Fig. 2). The result was similar to earlier reports (30,67). Maximum intracluster diversity was observed in cluster II while minimum in cluster III. The results indicated that the germplasm belonged in cluster II had more heterogeneity among themselves and germplasm in cluster III were comparatively more closely related. As for the inter-cluster distance, maximum inter-cluster distance was shown from cluster II and cluster V suggesting that hybridization among the studied germplasm having high yield potential taken from these highly divergent clusters might produce heterotic combinations and wider variability in segregating generations. Minimum inter-cluster  distance was observed between cluster I and cluster V expressing the narrow diversity between these clusters. Although, the germplasms of these two clusters might not be used for crossing with each other in hybridization programme, it might be used in breeding programme for biparental crosses between the most diverse and the closest groups to break the awful linkage between yield and yield contributing traits. Table 9 presented the mean performance of different clusters for different quantitative traits.
The result revealed that cluster II was characterized by high abundance of plant height, total tillers/hill, effective tillers/hill and 1000-grain weight whereas low abundance for these characters was found in cluster V showing maximum contribution of these traits towards the divergence between cluster II and V. There are studies on mean performance of different clusters on rice germplasm using D 2 statistics (58,68). Cluster II included maximum days to 50% flowering, number of filled grains/panicle and grain yield/plant whereas cluster I had the minimum days to 50% flowering number of filled grains/panicle and grain yield/plant indicating divergence between cluster I and cluster II. Cluster II included germplasm with maximum days to maturity whereas cluster IV included germplasm with minimum days to maturity indicating  3 0 3 1

Cluster I
Cluster II Cluster III Cluster IV Cluster V Germplasm with the longest panicle were included in cluster I whereas cluster II had the shortest ones indicating divergence between cluster I and cluster II. Maximum number of unfilled grains/panicle was contained by cluster I while cluster III had minimum number of unfilled grains/panicle indicating divergence between cluster I and cluster III. The mean values of cluster groups revealed that cluster II was highly divergent among the all cluster groups. The traits included in this cluster viz., plant height, days to 50% flowering, days to maturity, total tillers/hill, number of effective tillers/hill, filled grains/panicle, 1000-grain weight and grain yield/plant contributed maximum towards the total divergence. The genetically diverse genotypes become more important for developing variety resistant to diseases and adverse climatic conditions as the modern variety with narrow genetic bases are vulnerable to these (2). So, traits having broad genetic bases found from this study can be used for developing a resistant variety as well as reducing yield losses.

Characterization based on qualitative characters using DUS test
Distinctness, uniformity and stability test was done to characterize the qualitative morphological traits of the studied rice germplasm (Table 10). Out of 31 qualitative traits, 5 showed homogeneity (blade pubescence, basal leaf sheath color, ligule shape, auricle color and decorticated grain: scent) and the rest 26 showed heterogeneity. All the germplasm possessed glabrous typed pubescence, green colored leaf sheath, 2-cleft shaped ligule, pale green colored auricle and non-scented decorticated grains.

Pigmentation
Regarding the leaf blade color, 9% germplasms had pale green, 66% germplasms had green and 25% had dark green leaf blade. There are reports on the highest frequency for green leaf blade (69). In case of leaf sheath and culm anthocyanin coloration, only Dudhbinni possessed it. Studies are also on greater variation for leaf sheath and culm anthocyanin coloration (70,71). Among the germplasms evaluated,

979
EMI ET AL
Traits having different pigmentation are very functional and helpful in sorting out of off-types and keeping the seed genetically pure.

Angle of leaf, flag leaf and culm and culm strength
Majority of the germplasm possessed horizontal type leaf angle (61%) whereas erect type of angle of leaves were found for 39% germplasm. On the other hand, flag leaf angle was found erect in 52% germplasm and semi-erect in 48%. Similar angle of flag leaf was also found by others (72). Erect type of leaf and flag leaf angle helps to accumulate more sunlight and contributes more in photosynthesis as well as producing more photosynthate which results in higher grain yield. Culm angle was found erect for 57% germplasm, intermediate for 30% germplasm, open for 10% germplasm and spreading for one germplasm. Wide variability for culm angle was also reported in rice (75). For 3% accession, culm strength was found strong, moderately strong for 32%, intermediate for 26%, weak for 36% and very weak for 3% accession. Strong and moderately strong culms are beneficial for rice production as it reduces lodging as well as yield losses.

Type of axis and panicle, panicle exertion, secondary branching and awn
It was found that 32% germplasm had straight axis whereas 68% had droopy type. Similar trait was also studied (74) who also found droopy axis in most of their studied genotypes. Among the tested germplasm, 87% accessions had intermediate type of panicle and only 13% had open panicle. None of them had compact type of panicle. About 36% germplasm with moderately well exerted panicle followed by 32% with just exerted panicles, 26% with partly exerted panicles and 6%with well exerted panicles. Variation was found for panicle exertion as it is affected by the agroclimatic condition and cropping seasons of this crop (73,76). Well exerted panicle is more advantageous than other types as it helps in successful pollination which is a must for higher yield in rice. Of the studied germplasm, 77% had light secondary branching. Only 19.36% germplasm possessed awn at grain (80.645%). Similar work was also reported (69). Earlier reports were also showed a wider variability for this trait (71). Awn at tip was found for 13% germplasm and 3.23% for each upper quarter and upper three quarters. Rice having awn at grain helps it to survive in the changing environment.
Lemma and palea pubescence, endosperm type, leaf senescence, shattering and threshability Among the studied germplasm, 71% showed glabrous pubescence of lemma and palea followed by 19% short hairs and 10% hairs on upper portion. The thickness of pubescence of lemma plays a very important role in identification of different varieties of rice (77). Majority of the germplasm possessed glutinous type of endosperm (74%) while 7% had non-glutinous or nonwaxy type (6%) and 19% had the intermediate type.
Out of the studied rice germplasm, 36% germplasm expressed very early leaf senescence, 26% late and slow, 16% of each intermediate and very late and 6% early leaf senescence. Shattering was very low in most of the selected germplasm accessions (64%) which is very helpful for reducing post-harvest loss of rice as it reduces the shed of grain. Rest of the accessions had low (26%) and moderate shattering (10%). Regarding threshability, 58% germplasm showed intermediate type of threshability followed by 13% moderately difficult, 13% loose, 13% easy and 3% difficult type of threshability. Rice germplasm having easy type of threshability is desirable as it reduces the post harvest loss of rice making the grain separation easy from its plant. High variability for threshability was also reported in rice (58,72).

Conclusion
Exploration and documentation of phenotype information is a priority research need in crop modeling, breeding, for association studies, gene discovery, choosing cultivars and planning crosses for maximizing genetic gain. To the best of our knowledge, the current study is the first phenotypic study involving yield and yield contributing traits including DUS traits and genetic analyses of the traits by utilizing the local rice germplasm. Significant variability and diversity were observed among the studied rice genotypes for both qualitative and quantitative traits. Among the studied genotypes, Sentu-16 was the highest yielding that can be used in future plant breeding program. Selection of the traits such as number of effective tillers/hill, number of filled grains/panicle and 1000-grain weight might be effective to improve grain yield in rice as these traits had high heritability coupled with high GA (%) and positive significant relationship with grain yield/plant. Maximum genetic diversity was found among the genotypes included in cluster II and cluster V and selection of these genotypes might give rise to high heterotic response in the segregating generation. The traits such as blade color, leaf angle, flag leaf angle, culm angle, internode color, culm strength, panicle exertion, shattering, threshability, apiculus color, stigma color, lemma and palea color, lemma and palea pubescence, sterile lemma color, seed coat color and leaf senescence found as suitable traits which might be used for further improvement of rice. Finally, the findings of the study open-up new horizon for developing high yielding Aman rice varieties by utilizing promising genotypes and trait associations.