Evaluation Genetic Variation and Diversity of Grain Yield and Quality Traits in Rice (Oryza sativa L.) Genotypes for Low Input NPK Fertilizers


 Determination of genetic variance in a large number of rice genotypes is an effective strategy for increasing yield. The goal of this research was to determine the genetic variability, phenotypic (PCV) and genotypic (GCV) coefficients of variation, broad-sense heritability, expected genetic advance and multivariate analysis for eight rice grain quality and yield traits, in twenty Egyptian and exotic genotypes under low NPK fertilizer input levels at RRTC, Sakha, Kafr El-Sheikh, Egypt and evaluated across two successive seasons. Results revealed highly significant mean squares for all traits. High estimates of both PCV and GCV were detected for grain elongation followed by gelatinization temperature and head rice. High estimates of heritability were noted for grain length, grain shape, hulling, milling, head rice, amylose content, grain elongation and grain yield. Results revealed that highly significant differences among different genotypes were observed for studied characteristics under different NPK levels. Cluster results revealed that genotypes from the same origin or taxonomy type were clustered together. Diversity analysis showed four clusters. Cluster I and III had maximum genotypes (70%) and Cluster IV showed the highest mean values for studied traits. The results revealed that PC1 and PC2 accounted for 65.6% of the diversity between genotypes investigated. These findings show that some genotypes have a lot of diversity, indicating an opportunity to breed for low-input genotypes without sacrificing grain production and quality. GZ10590-1-3-3-2 and IET1444, both of which have high grain yield, can be employed as hybrid parents and could help with further genetic research for reduced NPK input.


Introduction
Rice is one of the staple food crops for about half of the world's population. Therefore, rice production should be signi cantly increased to meet the needs of a growing world population. The global demand  Samonte et al. (2006) reported that nitrogen is one of the most essential macronutrients for rice production. Fertilizers with appropriate management practice help to increase the productivity of rice in farmers' elds (Gairhe et al. 2018; Timsina et al. 2012). The N, P and K are macro elements. Many studies have shown that the appropriate use of NPK fertilizers has enhanced the yield and substantially improved rice quality (Oikeh et al. 2008). The recommendation of chemical fertilizers should be based on soil analysis and crop response. A good fertilization strategy should be developed that combines the use of organic and chemical fertilizers, as well as improving crop productivity and environmental quality (Devkota et al. 2019). Nitrogen plays a vital role in determining the growth and yield potential of crops. The best mineral fertilizer rate is one that yields the highest economic return at the lowest expense (Ananthi et al. 2010).
Grain quality is de ned as a major factor that decides market values of agricultural products and foods in each phase from production through consumption. Rice grain quality has always been an important consideration in variety selection and development. Cooking and eating quality traits are a part of the factors in assessing the grain quality of rice. Grain quality will become even more essential in the future for many of whom rely heavily on rice as a staple diet, become better off and want higher quality rice (Lampe 1993). Rice has different cooking and eating properties depending on the variety and grain type.
The dry-aky cooking characteristic of rice is found in varieties with a high percentage of amylose, a medium gelatinization temperature and relatively low water absorption. The varieties with low amylose and low gelatinization temperature tend to be sticky and cohesive when cooked, absorb more water, and thus have long grain elongation after cooking.
Genetic parameters such as GCV and PCV measure the amount of genetic diversity in genetic resources as well as the degree to which genotype is modi ed by the environment. While the selection is made primarily on yield contributing characters, heritability and genetic advance are key selection factors. Heritability estimates combined with genetic improvements are usually more useful than heritability estimates alone in estimating the gain under selection (Paul et al. 2006). We apply cluster analysis when we need to categorize crop genotypes, in which the resulted clusters are important for selecting the best parents for breeders (Sanni et al. 2012). The main objectives of the present investigation are to evaluate the performance of grain quality and yield traits in twenty rice genotypes at low fertilizer input levels to i) estimate genotypic (GCV) and phenotypic (PCV) coe cients of variability, broad-sense heritability and genetic advance for grain quality and yield traits, ii) utilize multivariate analysis to better understand the relationships and patterns between genotypes and iii) examine the connections between grain yield and grain quality attributes.

Plant materials
In total, twenty Egyptian and exotic rice genotypes were chosen at random and employed in this investigation. The genotypes were provided by the Genetic Stock of Rice Breeding Program, Agricultural Research Center (ARC, Giza, Egypt). The name, origins and subspecies group of these rice genotypes have illustrated in Table 1.

Experimental layout
The present investigation was carried out at Research Farm of the Rice Research and Training Center (RRTC), Sakha, Kafr El-Sheikh, Egypt, during the 2019 and 2020 planting seasons. The selected genotypes were evaluated under four levels of NPK fertilizers i.e., full dose, two-third, one-third and zero of the recommended dose of NPK. The recommended doses of NPK are 165, 36 and 58 kg N, P 2 O 5 and K 2 O per hectare, respectively. At the depth of 0-30 cm from the soil surface, representational soil samples had been selected. The method of soil interpretation followed the procedure of Black et al. (1965). The results of soil analysis in the studied seasons are presented in Table 2.
The nursery was well prepared and fertilized with four kg/m calcium superphosphate (15.5 % P 2 O 5 ) before plowing and three kg urea (46.5% N) was applied after plowing as well as, one kg zinc sulfate (22% Zn) immediately applied before sowing and after puddling. Rice seeds at the rate of 60 kg/ha were soaked in freshwater for 24 hours and incubated for 48 hours to improve germination. The pregerminated seeds were broadcast on May 15 th in both seasons. The current investigation was applied in a split-plot design with three replications. NPK treatments were distributed over main plots and the genotypes were allocated in sup plots. The permanent eld was plowed and then well dry leveled.
Phosphorus fertilizer in the form of calcium superphosphate (15.5% P 2 O 5 ) was applied before land preparation according to the treatment schedule in this hypothesis. Potassium fertilizer in the form of potassium sulfate (48 % K 2 O) was incorporated in the dry soil before planting according to the treatments used in this study. Nitrogen fertilizer was added according to the treatments in the form of urea (46.5 % N). Two splits (2/3) were applied and incorporated in dry soil before planting as well as (1/3) has applied after thirty days from transplanting as topdressing. The permanent eld was immediately irrigated. Thirty days old seedling of each genotype was individually transplanted in 10-row per replicate with a spacing of 20 cm between rows and 20 cm between plants.

Data collection
All rice grain quality and grain yield traits were calculated according to the standard evaluation system for rice (IRRI 1996). At harvest, grain yield (t/ha) from randomly 10 square meters was measured. Moreover, the laboratory analysis was conducted for grain quality traits, the grain samples were milled and analyzed for physicochemical properties. Milled rice out turn was determined by husking 200 g rough rice and milled in Satake Rice Mill. Head rice recovery was determined by separating broken parts from milled rice. Milling % and head rice % were expressed as a percentage of rough rice and milled rice, respectively. Rough grain length and breadth were measured by slide calipers (IRRI 1996). In determining the rough grain shape, rough rice was rst classi ed into four classes based on length, very long (more than 7.5 mm in length), long (6.61 to 7.5 mm in length), medium (5.51 to 6.6 mm in length) and short (5.5 mm or less in length). The grain was again classi ed into four classes considering length to breadth ratio; long (ratio above 4), slender (ratio from 3.1 to 4), medium or (ratio from 2.1 to 3.0), bold (ratio 1.1 to 2.0) and round (ratio less than 1.1). Based on amylose content, milled rice was classi ed as waxy (1-2% amylose), very low (>2-9% amylose), low (>9-20% amylose), intermediate (>20-25% amylose) and high (25-33% amylose).
For measurement of grain elongation ratio, 10 measured (length and width) grains were taken into a 20 ml glass test tube and soaked for 20 minutes with 5 ml of tap water. The test tubes were then immersed in boiling water for around 30 minutes after soaking. When the grain was fully cooked, the water inside them was drained. The cooked grains were then placed on a glass sheet for a few minutes to drain excess moisture before being measured in length and width. The relative change in rice grain length after cooking is referred to as the grain elongation ratio. Six whole milled grains of rice from each plant were spaced evenly in small transparent plastic boxes containing 10 ml of 1.70% potassium hydroxide solution to determine gelatinization temperature (GT). In an incubator set to 30°C, the boxes are covered and left undisturbed for 23 hours. The GT, which was visually rated on a seven-point numerical scale, was represented by such alkali spreading and clearing of starchy endosperm (Table 3).

Statistical analysis
The data obtained for each trait was statistically evaluated over the two seasons according to Le Clerg et al. (1962), then it was subjected to analysis of variance, which was used to partition the gross phenotypic variability into the components due to genetic (hereditary) and non-genetic (environmental) factors. Genotypic variance is the part of the phenotypic variance that can be attributed to genotypic differences among the phenotypes. Similarly, phenotypic variance refers to the total variation among phenotypes where Vp, Vg and X are the phenotypic variances, genotypic variances and grand mean per season, respectively, for the traits under consideration. Broad sense heritability (h 2 B) expressed as the percentage of the ratio of the genotypic variance (Vg) to the phenotypic variance (Vph) was estimated on the genotypic mean basis as described by Allard (1999). Genetic advance (GA) expected and GA as a percent of the mean assuming selection of the superior 5% of the genotypes were estimated by the methods illustrated by Fehr (1987) as follows: where k is a constant (which varies depending upon the selection intensity and, if the latter is 5%, it stands at 2.06), S ph is the phenotypic standard deviation (√Vph), h 2 B is the heritability ratio and x refers to the season mean of the trait.
To create the dendrogram depending on squared Euclidean distance, the unweighted pair group method of arithmetic average linkage (UPGMA) was applied using SPSS version 15 (IBM Corporation 2010). The initial cluster distances in Ward's minimum variance method are therefore de ned to be the squared Euclidean distance between points: dij = d({ Xi}, {Xj} = ||Xi -Xj||2. The principal component analysis (PCA) was subsequently investigated using Genstat version 12.0 software.

Result
Genetic variability between rice genotypes The grand mean, genotypic and phenotypic coe cient of variability, broad-sense heritability and genetic advance as percentages of the mean are illustrated in Table 4. Low estimates of phenotypic and genotypic variances were recorded for all the studied traits except head rice % and grain elongation traits. The environmental variance was very low for all traits especially for grain shape and grain yield. The phenotypic coe cient of variability (PCV) was higher than the GCV for all the characters studied. PCV was ranged between 2.572 and hulling to 19.956 for grain elongation. PCV and GCV were lower than (10 %) for grain length, grain shape, hulling percentage, milling percentage, amylose content and grain yield (t/ha). Otherwise, high estimates of both PCV and GCV were detected for grain elongation followed by gelatinization temperature and head rice. Negligible values of the difference between PCV and GCV were recorded for grain length, grain shape, hulling, milling, head rice recovery, amylose, grain elongation and grain yield. In general, PCV values were higher than GCV for various characters studied. Among the desirable traits, low GCV and PCV were observed for grain length. High estimates of broad-sense heritability were noted for grain length, grain shape, hulling, milling, head rice, amylose content, grain elongation and grain yield. Their estimated values were ranged between 71.25% for grain length and 99.55% for both grain elongation and grain yield traits, while almost moderate estimates were exhibited for gelatinization temperature (67.78 %). However, high estimates of expected genetic advances were found for grain elongation (40.924 %), head rice recovery (27.336 %) and gelatinization temperature (25.603 %). Grain elongation, gelatinization temperature and head rice had shown high PCV and GCV along with high to moderate heritability (h 2 ) and genetic advance. The additive gene action governed the above-mentioned three traits. High heritability with high genetic advance as percent mean was observed for all the grain quality traits except for hulling percent and milling percent.
Genotypes Performance under different NPK treatments Data pointed out that there were highly signi cant differences among different genotypes in the studied characteristics under different NPK levels. Grain yield, grain quality and cooking quality traits were affected signi cantly by NPK treatments. The studied characteristics were increased gradually by increasing NPK levels from 0 up to the full dose of recommended doses of NPK fertilizers. The results revealed that the GZ10590-1-3-3-2 rice genotype produced the highest grain yield followed by IET1444 , while Nerica1 came in the last rank and gave the lowest value in this aspect (Fig. 1). Grain length was higher in IRAT170 which came in the rst rank and recorded the highest value of grain length while Giza178 recorded the shortest grains (Fig. 2). Grain shape was higher in IRAT170 and came in the rst rank followed by Giza182 (Fig. 3). Milyang 109 produced the lowest values of grain shape. Giza177 recorded the highest values of hulling percentage followed by Giza179. Korea1, Nerica1 and Milyang109 came in the last rank and recorded the lowest hulling percentage and recorded nearly the same value of hulling percentage (Fig. 4). The highest percentage of milling was observed by the rice genotypes Giza177, GZ10991-5-18-5-1, GZ10333-9-1-1-3 and GZ10101-5-1-1-1. While the lowest value of milling rice was observed with Nerica1 (Fig. 5). GZ10333-9-1-1-3 recorded the highest values of head rice while GZ10598-9-1-5-5 recorded the lowest value of head rice under this study (Fig. 6). Amylose content was high in Nerica 1, it came in the rst rank and recorded the highest value of amylose content, while, GZ10598-9-1-5-5 recorded the lowest value of amylose content (Fig. 7). The highest value of grain elongation was recorded by IET1444. While IRAT170 and GZ10590-1-3-3-2 recorded the lowest grain elongation (Fig. 8). Gelatinization temperature (GT) in IET1444 came in the rst rank and recorded the highest value while IRAT170 and GZ10590-1-3-3-2 recorded the lowest values (Fig. 9).

Diversity analysis
Principal component analyses Principal component analysis (PCA) is used to generalize the variation in different explaining factors. Our PCA was done for different grain quality traits of the twenty rice genotypes. Eigenvalues and a fraction of variation in each principal component are shown in Table 5. PC1 explained 37.6 % of total variations observed among the genotypes while it was 28.0% for PC2. The total diversity was explained by nine principal components ( Table 5). The results showed that only the rst three principal components jointly accounted for 80.9% of the total variation among the genotypes, while these three exhibited eigenvalues above unity. However, the large fraction of the divergence (88.2% variation) was explained by the rst four components. Hulling was correlated with Milling, in addition, gelatinization temperature was closely associated with grain elongation, as well as grain shape was found to be correlated with grain length. All Egyptian genotypes and promising lines were features with the high yielding ability and high milling and hulling relearn. IET 1444 was greater in gelatinization temperature and grain elongation compared with other rice genotypes. Both Nerica 1 and IRAT 170 were unique to grain length and grain shape traits (Fig.  10). The Principle components for grain yield and grain quality traits in twenty rice genotypes and their distribution of priority in resulting PCA were reported in Tables 6 and 7, respectively. The results demonstrated that grain length, milling and amylose content were classi ed in PC1, while, head rice, gelatinization temperature and grain elongation were located in PC2. On the other hand, grain yield and hulling were distributed in PC3.

Cluster analysis
Clustering analysis was performed to look for similarities between rice accessions and assess the possibility of hybridization. Mean performance of different clusters for the characters manifested that genotypes with maximum grain yield, high milling output, good grain shape and low amylose content were accumulated in cluster I, whereas genotypes with maximum hulling were grouped in cluster II.
Moreover, genotypes that took minimum head rice were clubbed into cluster III. Whereas low yielding genotypes with soft gelatinization temperature and long grain elongation were grouped into cluster IV (Table 8).
Among twenty rice genotypes, cluster analysis resulted in four clusters following Ward's method (Fig. 11). Genotypes from the same origin or taxonomic type were found to be clustered together in the cluster results. Four distinct clusters emerged from the diversity analysis. Clusters I (7 genotypes) and III (7 genotypes) had the most genotypes (70%) while Cluster IV (one genotype) had the highest mean values for the qualities being evaluated. In cluster I, out of 7 rice genotypes, 5 of them were Egyptian japonica type, with high yielding, shortest grains, bold grain shape, highest milling and head rice recovery and low amylose content, namely, Giza 177, Sakha 107, Gz 10101-5-1-1-1, Gz 10598 -9-1-1-5-5 and Gz10590 -1-1-3-9-1, while the other two genotypes Giza 178 and Korea 1 were (indica/japonica) type. With regarding cluster III, 7 genotypes were aggregated, out of them, 4 genotypes were Egyptian japonica type, namely, Sakha 108, Sakha 109, Gz 10333-9-1-1-3 and Gz 10590-1-3-3-2 with high hulling. However, the other three genotypes were exotic, namely Milyang 109 (Indica/japonica type), IRAT 170 (Indica type) and Nerica 1 (Indica type). On the other hand, 5 rice genotypes were grouped into cluster II, 0ut 0f them, 2 genotypes, Giza 179 and Suweon 375 were (Indica/Japonica) type and 2 genotypes, Fukunishiki and Gz 10991-5-18-5-1 were (japonica type) as well as only one genotype Giza 182 was (Indica type). Finally, only one rice genotype namely, IET 1444 tolerant to water stress and suitable to cultivate under drought conditions (Indica type) was located in cluster IV. The distribution of 20 rice genotypes in different clusters is illustrated in (Table 9) and the dissimilarity matrix according to Euclidean square based on studied traits in the tested rice genotypes were observed in Table 10.

Discussion
Low estimates of phenotypic and genotypic variances were recorded for all the studied traits except head rice and grain elongation traits. The environmental variance was very low for all traits especially for grain shape and grain yield, indicating less in uence of environment in the expiration of these traits. The PCV was higher than the GCV for all the characters studied. PCV and GCV were low for grain length, grain shape, hulling, milling, amylose content and grain yield indicating limited scope for further genetic improvement of these traits through selection. Otherwise, high estimates of both PCV and GCV were detected for grain elongation followed by gelatinization temperature and head rice indicating wide variability among the varieties studied for these traits and the possibility of genetic improvement of these traits through selection. Negligible values of the difference between PCV and GCV were recorded for grain length, grain shape, hulling, milling, head rice recovery, amylose, grain elongation and grain yield suggesting a lower environmental in uence on the expression of these traits. Furthermore, Bharath et al.
(2018) reported that quantitative characters had shown less difference between PCV and GCV. Also, they recorded that single plant yield and gelatinization temperature had shown high PCV and GCV. Moreover, Archana et al. (2018) exhibited that PCV values, in general, were higher than GCV for various characters studied. Grain elongation, gelatinization temperature and head rice had shown high PCV and GCV along with high to moderate heritability (h 2 ) and genetic advance, these traits are less in uenced by the environment. High heritability accompanied with high genetic advance as percent mean were recorded for grain yield per plant (Archana et al., 2018). This signi es that these characteristics are governed by additive gene action and selection for these traits would be effective. Similar ndings were recorded previously (Hammoud et al. 2006a;Sahu et al. 2017). It had been observed that the coe cient of variation ranges from 8.61% for hulling percentage to 45.01% for gelatinization temperature.
There were highly signi cant differences among different genotypes in the studied characteristics under different NPK levels. Grain yield, grain quality and cooking characteristics were affected signi cantly by NPK treatments. The studied characteristics were increased gradually by increasing NPK levels from 0 up to the full dose of recommended doses of NPK fertilizers. The advantageous impacts of NPK on rice productivity and quality were noted by several previous investigations. The favorable effect of NPK fertilizer application on rice grain yield might be due to the increase in NPK availability in soil and subsequently content in rice plants leading to produce more energy which enhances the photosynthetic rate that improved grain lling process (Biswas and Dravid 2001;Ibrahim 2001). The effect of NPK fertilizer application on hulling percentage was mainly due to the maximum storing of starch in the endosperm of grains which caused a reduction in the hull components such as palea, lemma, pericarp, aleurone layers and rachilla. The increase in milling percentage due to increasing NPK levels may be due to the increase in metabolite substances in grains (Asif et al. 1999;Metwally 2007;Naeem et al. 2010). Data revealed that the GZ10590-1-3-3-2 rice genotype produced the highest grain yield followed by IET1444 , while Nerica1 came in the last rank and gave the lowest value in this aspect. These results could be attributed to the superiority of GZ10590-1-3-3-2 in growth vigor as well as yield attributes. The differences among rice genotypes may be due to the genetically inherited variants. The differences among rice genotypes in yield and grain quality characteristics may be due to the genetically inherited variants. Ebaid and El-Rewainy (2005) and Metwally et al. (2020) reported that hulling, milling, head rice, gelatinization temperature, grain elongation, amylose content traits were varied among different Egyptian rice genotypes. Singh et al. (2011) indicated that rice grain quality characteristics variations among rice genotypes are dominated by the genetic background of those genotypes. Zhao et al. (2018) reported that the GS9 gene regulates grain shape and hull thickness by altering cell division. It means rice grain shape is mainly controlled by the genetic background. The variations among the studied genotypes in cooking qualities may be due to the differences in their shape and thickness. Mohapatra and Bal (2006) suggested that the thickness of the rice grain is an important factor in deciding the diffusion of water during cooking. The differences among rice genotypes may be due to the genetically inherited variants. PCA is used to generalize the variation in different explaining factors. The results showed that only the rst three principal components jointly accounted for 80.9% of the total variation among the genotypes, while these three exhibited eigenvalues above unity. While a large fraction of the divergence (88.2% variation) was explained by the rst four components. The results demonstrated that grain length, milling and amylose content were classi ed in PC1, while, head rice, gelatinization temperature and grain elongation were located in PC2. On the other hand, grain yield and hulling were distributed in PC3. . Mean performance of different clusters for the characters manifested that genotypes with maximum grain yield, high milling output, good grain shape and low amylose content were accumulated in cluster I. Whereas, genotypes with maximum hulling were in grouped cluster II. Moreover, genotypes that took minimum head rice were clubbed into cluster III. Whereas, low yielding genotypes with soft gelatinization temperature and long grain elongation were grouped into cluster IV.

Conclusions
High heritability of grain quality and yield traits suggests that the environment has only minor in uence, implying that simple breeding methods based on selection will be e cient in genetic improvement. Lowinput fertilizers in agricultural production will be considered one of the necessary agricultural production inputs in the next years allowing us to feed a growing population while keeping low costs. This study dealt with the possibility of using different rates of the necessary elements such as nitrogen (N), phosphorous (P) and potassium (K) to rice fertilize. Results exhibited highly signi cant mean squares for all characteristics suggesting the presence of genetic difference among rice genotypes for all traits. Low estimates of PCV and GCV were recorded for all traits except head rice and grain elongation. Grain yield, grain quality and cooking traits were in uenced signi cantly by NPK treatments. The studied characteristics were improved progressively by rising NPK levels from 0 up to the full dose of recommended NPK fertilizers. There were highly signi cant variations among genotypes in studied traits. GZ10590-1-3-3-2 rice genotype exhibited the highest grain yield followed by IET1444 . Also, it was concluded that origin and taxonomy were both responsible for clustering genotypes.

Declarations
Authors' contributions SGHRS contributed to supervision, encouragement, suggesting the problem, design, data analysis and writing up the manuscript. TFM contributed to support, design, performance eld experiments, data analysis and writing up the manuscript. SHAT contributed to designing, preparing, and scienti c advice, data analysis and writing up the manuscript. MMS contributed to design, data analysis and writing up the manuscript. KFMS design, data analysis and writing up the manuscript and following up the publication with the journal (correspondence). MABE design, performance eld experiments, data analysis and writing up the manuscript. All authors read and approved the nal version. .

Code availability
Not applicable.

Figure 5
Mean performance of milling rice % trait among twenty rice genotypes over two contrasting environments (2019 and 2020) as affected by NPK treatments Figure 6 Mean performance of head rice % trait among twenty rice genotypes over two contrasting environments (2019 and 2020) as affected by NPK treatments Figure 7 Mean performance of amylose content % trait among twenty rice genotypes over two contrasting environments (2019 and 2020) as affected by NPK treatments Figure 8 Mean performance of gelatinization temperature trait among twenty rice genotypes over two contrasting environments (2019 and 2020) as affected by NPK treatments Figure 9 Mean performance of grain elongation % trait among twenty rice genotypes over two contrasting environments (2019 and 2020) as affected by NPK treatments Figure 10 Principal component analysis (PCA) for twenty Egyptian and exotic rice and nine characters (grain yield, grain length, grain shape, hulling, milling, head rice, amylose content, gelatinization temperature and grain elongation) Page 25/25

Figure 11
Clustering pattern (pooled over 2019 and 2020 seasons, RRTC, Sakha, Egypt) of the twenty Egyptian and exotic rice genotypes using Ward`s method based on grain yield and quality traits