2.1. Study Site

The field trial was conducted at the Scientific Industrial Research and Development Centre in Harare, Zimbabwe, during the 2018/19 growing season. The area is located in the northern part of Harare and lies on longitude of 30°7′E and latitude of 17.8°55′S, at an altitude of 1488.35 m above sea level (MASL). The area falls under region IIa which normally receives annual rainfall with a range of 750–1000 mm [24]. Average summer temperatures during the day and night are 26.1°C and 15.7°C, respectively. Average winter temperatures during the day and night are 22.2°C and 6.3°C, respectively. The predominant soils in this area are Fersiallitic soils. Nyamapfene [25] described the soils as 40% clay, 80% sand separate, and 90% silt and being of good agricultural value.

The soil was sampled in the topsoil layer (0–30 cm) over the experimental blocks. It was sampled in both diagonals at the four corners of the block, at the middle, between the corners, and between the middle of diagonals and corners to give 10 subsamples. The samples were analyzed separately and since they had almost the same values, these were composited and averaged [26]. The soil analysis was carried out at Food and Biomedical Research Institute (FBTI) of the Scientific Industrial Research and Development Centre and had the properties shown in Table 1. Soil pH was determined by weighing a 15 g of soil sample in a 200 ml glass jar to which 75 ml 0.1 M CaCI₂ was added. The mixture was mechanically shaken for 30 minutes and pH was determined using a digital pH meter (model: Orion 701). The optimum phosphorus available to rice occurs at a pH lower than 6.5. Phosphorus is usually available when the soil Ph is between 6.0 and 6.5. In acid soils (pH < 6.0), phosphorus is associated with aluminium and iron compounds which are slowly available to most plants [27]. When pH is above 6.5, phosphorus is associated with magnesium and calcium [28].

Total nitrogen was measured by Kjeldahl method using concentrated KCI to digest the sample [29]. Soil phosphorus was extracted using anion exchange resin [30]. In this method, anion exchange resin and soil were suspended in water at the ratio of 4 : 5: 100 and shaken for 16 hours after which phosphorus sorbed by resin was eluted and estimated. The extraction of potassium, calcium, and magnesium was performed using NH₄Cl [31]. Each sample of the cations was deposited in a glass vessel to which 25 mL of the extraction was added and Mg, K, and Ca were estimated.

2.2. Experimental Design and Treatments

Forty doubled haploid elite rice lines were obtained from Rural Development Administration (RDA), Suwon, South Korea, under the Korea Africa Food and Agriculture Cooperative Initiative (KAFACI) and four other lines were obtained from Africa Rice Center, Senegal, which had acceptable yields, plant architecture, high grain quality, and resistance to pest and diseases and were early maturing (Table 2). Five local checks (Nerica 1, 3, and 7, Macheke land race, and Mhara 1) were used in this experiment (Table 2). These plant materials were planted during the rainy season of December 2018. The experiment was laid out in a simple alpha lattice design (7∗7) replicated three times.

2.3. Trial Establishment and Agronomic Management

Genotypes were first established in a nursery and transplanted, one seedling per hill at 30 days after sowing. The seedlings were transplanted in beds with a net plot of 1 m² maintaining a 20 cm∗20 cm plant spacing of individual plant and all recommended cultural practices for rice were observed. A thin film of water of 1-2 cm was maintained at the time of transplanting. Afterward, the water level was increased up to 5 cm. From flowering to maturity, water was allowed to stagnate in the field up to two weeks before harvest; later on, water was removed subsequently. A predrilling base application of 300 kg per ha of NPK (7 : 14 : 7) fertilizer was made followed by a total of 100 kg of urea (46% N) in two applications of 35 kg at 21 days after transplanting and 65 kg at panicle initiation.

2.4. Data Collection and Analysis

Data was collected on days to 50% heading (days from planting to the period when 50% of the rice plants were headed), panicle length (cm), culm length (cm), number of total tillers/panicle, number of effective tillers/plant, number of grains/panicle, number of filled grains/panicle, spikelet fertility/panicle (%), 1000-grain weight (g), grain yield (t/ha), grain length (cm), and grain shape. The data were recorded according to the International Rice Research Institute [32] procedures.

Analysis of variance was computed using general statistics (GenStat 14th addition) computer statistical package [33]. The correlation coefficient was calculated to determine the extent of relationship between grain yield and other important variables, using ARiS [34].

Grain shape is defined as the ratio of the brown rice length and brown rice width [35]. The scale shown in Table 3 was used to describe grain shape as suggested by IRRI [36].

3.1. Agronomic Performance of Rice

The results of the current study revealed significant amount of variability among the tested genotypes. Genetic variability in the initial breeding materials is important in producing the desired recombinants for crop improvement. Analysis of variance revealed significant and exploitable variability due to genotypes (p < 0.05) for days to 50% heading, panicle length, culm length, number of filled grains/panicle, spikelet fertility/panicle, 1000-grain weight, grain yield, grain length, and grain shape but not for number of tillers/plant, number of productive tillers/plant, and number of grains/panicle (Table 4). The variation for these traits could be due to genotype composition and differed to adapted current environmental conditions. Similar results were reported by Anis et al. [16] and Khalifa [38].

Mean agronomic performance of rice for the tested genotypes is shown in Table 5. Regarding days to 50% heading, which is a good indicator of maturity, the early heading genotype was 70462 (72 days), closely followed by 70497 (74 days) and 70033 (74 days). The elite rice lines 70540 and 70340 revealed the longest days to 50% heading (117 and 108 days, respectively). Li et al. [39] reported that average heading period of 97 days was ideal for high yielding cultivars. The mean heading period of this study was 82 days, suggesting the chance of selecting early maturity genotypes which can escape terminal moisture and drought stress. Genotypes that revealed early heading date would have prolonged grain filling duration to give high yields [6]. However, in this study, lines with prolonged grain filling duration did not always give higher yields. Sheehy et al. [40] suggested that yield is influenced by not only duration in grain filling but also the efficiency in partitioning of photoassimilates to their respective sinks. Lodging is one of the major factors limiting the yield potential of rice. The results for culm length varied from 55.3 cm to 125.7 cm. Elite line 60323 produced the shortest culm length (55.3 cm), while genotype 70462 produced the longest culm length (125.7cm). It was noted that a number of genotypes in this study have short sturdy culm length. A study by Mazarire et al. [6] has shown that such genotypes could be useful in crop improvement because they do not suffer much from both root and stalk lodging and will have higher response to nitrogen application. Panicle length is one of the important parameters assessed by rice breeders in yield related studies, because longer panicles with more filled grains led to higher yields. Furthermore, studies conducted previously [41, 42] have shown that panicle length alone does not determine high grain yield as traits such as percentage of filled grains per panicle, grain size, grain shape, and longer panicles ultimately contribute to higher grain yield. The mean values for panicle length ranged from 19.1 cm to 35.7 cm. Simultaneously, rice genotypes 70462, Sahel 177, Nerica 7 (check), and 70383 recorded the highest panicle lengths of 35.7, 32, 32, and 31 cm, respectively. On the other hand, genotypes 70528 and 70033 gave the lowest values of 19.1 and 19.8 cm for panicle length, respectively. Based on the IRRI irrigated rice panicle classification [36], the findings of the study revealed that there is enough variability for panicle length among lines for improving panicle architecture and grain yield due to high correlation of this trait to number of grains/panicle. Concerning the number of filled grains/panicle, rice genotype 70462 (204) produced the highest number, closely followed by 60409 (183) and Arica 3 (120.6). Elite line 60143 produced the highest (90.3%) spikelet fertility/panicle, followed by 70462 (90%) and Arica 3 (81.2). Generally, doubled haploid lines had high number of filled grains/panicle than local checks. These results were earlier reported by Hidayatullah et al. [12]. The mass of grains of individual plants directly determines the yield of a population [43]. Most of the lines were in the range of 20 g to 30 g for 1000-grain weight. Maximum 1000-grain weight was calculated for genotypes Sahel 177 (31 g) and Arica 3 (31 g). Adequate number of fertile grains/panicle and heavy grains are important characters, which should be exploited in the selection of high yielding genotypes [10, 44]. The genotypes significantly varied for grain yield (14.61-0 t/ha), with high yielding genotype 70462 producing 66% higher grain yield compared to high yielding local check (Nerica 7). On the basis of performance, elite line 70462 (14.61 t/ha) produced the highest grain yield, closely followed by 60143 (13.33 t/ha) and 70383 (10.80 t/ha) compared to local check Nerica 7 with 4.94 t/ha. On the other hand, genotypes Arica 3, Sahel 177, 60040, 70537, 60409, 70476, Sahel 134, and 70482 gave mean performances above the high yielding local check, with corresponding yields of 9.18, 8.38, 7.76, 7.58, 7.36, 7.11, 5.79, and 5.42 t/ha, respectively. This suggests the existence of enough variability among genotypes was highly significant. This gives a better scope for selection of promising genotypes in rice breeding program for further testing, because selection based on yield is reliable [23].

Grain size and shape are among the top rice quality properties that breeders consider when developing new varieties because preferences for these traits differ from one group to another [45]. Grain length ranged from 10.6 mm to 6.2 mm (Table 5). Arica 3 produced the highest grain length of 10.6 mm, followed by 60409 (10 cm) and 70462 (9.9 cm). According to Standard Evaluation System for rice on brown rice [36], genotypes were classified into three groups, namely, extra long (41 genotypes), long (7 genotypes), and medium (2 genotypes) grained types. Grain shape varied from 3.9 mm (Nerica 3) to 2.1 mm (70474). The classification for grain shape has led to two categories, namely, slender (31 genotypes) and long (18 genotypes) grained types out of four possible types [36]. Joachim [46] and Fitriah et al. [47] reported that long grained rice and slender grained rice are highly preferred by rice-consuming people and this is also in line with consumers in Zimbabwe; therefore the traits could be exploited to improve the grain properties of rice.

3.2. Correlation Analysis

Yield is a complex character being influenced by several quantitative traits. Some characters are highly correlated with grain yield. The association among these traits and their association with grain yield in rice is an important criterion in the selection of desirable genotypes [48]. However, if more of the characters are involved in correlation study, it becomes difficult to predict the traits which really contribute towards grain yield. Results on the correlation between 12 quantitative traits are presented in Table 6. Correlation revealed that grain yield was significantly positively correlated to filled grains/panicle (0.784), culm length (0.605), spikelet fertility/panicle (0.677), grain length (0.551), 1000-grain weight (0.518), panicle length (0.449), and number of grains/panicle (0.328) (Table 6). Similar research findings were reported by Kimani [5] who worked on correlation analysis of some important yield contributing traits in rice. This was further supported by Butta et al. [49]. Conversely grain yield was significantly negatively correlated to days to 50% heading (–0.37). Agahiet et al. [50] suggested that priority should be given to these yield contributing traits in selection for grain yield while 1000-grain weight and grain length are considered as copartners in the grain yield gain. The significantly positive association of these characters with grain yield suggests that these traits have an influence in improving grain yield; therefore direct selection towards these traits would be more effective in ensuring high grain yield. These results agree with the findings of Ogunbayo et al. [51] and Akbar et al. [11] who reported a significantly positive correlation of grain yield and number of filled grains per panicle. Kato et al. [52] also reported similar results. The negative correlation of grain yield with days to 50% heading indicates that grain yield can be improved by selecting early maturing lines.

3.3. Broad Sense Heritability

The broad sense heritability is the relative proportion of genotypic and phenotypic variances for the characters and is used as a predictive measure in selection criteria. Heritability of a particular trait assists plant breeders in determining the total variation ascribable to genotypic effects, which are useful portion of variation [19]. In this study, the magnitude of broad sense heritability ranged from 43% to 93% (Table 7). The characters that show high heritability estimates (above 60%) were culm length (92), panicle length (93), days to 50% heading (88), number of filled grains/panicle (67), 1000-grain weight (82), grain yield (78), grain length (78), and the grain shape (64). Similar results were reported by Liangaiah et al. [53] for days to 50% heading, Arpita et al. [54] for culm length, Vanisree et al. [55] for panicle length, Tuhina-Khatun et al. [56] for the number of filled grains/panicle, Kiani et al. [57] for grain length, Abdul Fiyazet et al. [58] for 1000-grain weight, and Ammar et al. [59] for grain yield. Meanwhile, moderate heritability estimates (50–60%) were recorded for the total number of tillers/plant (52) and spikelet fertility/panicle (57). Similar results were reported by Tuhina-Khatun et al. [56] for the number of total tillers/plant, Ammar et al. [59] for spikelet fertility/panicle, and Hussain et al. [60] for L/B (grain shape). High-to-moderate heritability estimates indicate that these traits are less likely to be influenced by the environmental conditions [61]. Therefore, selection based on the phenotypic expression of these characters can be effective by using simple selection methods. Low heritability estimates were recorded for the number of productive tillers/plant (43) and number of grains/panicle (43). This is in conformity with the finding of Rafii et al. [62] for the number of productive tillers/plant and that of Osekita et al. [63] for the number of grains/panicle, indicating ineffectiveness for direct selection.