Sesame (Sesamum indicum L.) is an annual plant of Pedaliaceae family considered to be the oldest of the oilseed crops cultivated by man, having been grown in the Near East and  Africa  for  over  5,000  years  for  cooking  and medicinal purposes (Sharma et al., 2014). Generally, 65% of world sesame production is used for edible oil extraction and 35% for confectionary purpose (Pham et al., 2011). The fatty acid composition is rather attractive, due to the high level of unsaturated fatty acids. Sesame seed is the single readily available source of protein high in sulfur containing amino acids (Bradley, 2002). It is the major cash crop for smallholder farmers and a valuable foreign exchange revenue commodity for different countries. The remaining cakes of sesame are used as a source of crude protein for animal feed.

In Ethiopia, sesame is used as cash crop, export commodity, raw materials for oil industries and as source of employment opportunity. Now a day, it becomes the primary export oil crop playing a role in the agricultural Gross Domestic Product (GDP) of the country. A sizable proportion of the population, therefore, generates income from oilseed farming, trade and processing (Abate et al., 2015).  However,, sesame  production and extension in Ethiopia is quite limited, particularly because of its low yield. One of the major problems facing sesame production in Ethiopia has been growing of inferior sesame varieties with low yield and poor quality. To overcome the problems of low productivity of sesame, there must be a sound procedure for selection of high yielding varieties adapted to the local environment (Tadele, 2005).

Seed yield in sesame like other field crops is a multifacet character and direct selection for this trait may often be misleading. The components that determine the yield are best indices for selection. Therefore, knowledge of relationship between important yield traits and seed yield may help the researchers to identify suitable donors for a potential and successful breeding program (Kumaresan and Nadrajan, 2002). Estimation of character associations could identify the relative importance of independent traits contributing to dependent ones and suggest upon the traits that may be useful as indicator for other traits. In other words, character associations between yield components can be used as the best guide for successful yield improvement by indirect selection. Achievement of such success depends upon sort and accuracy of estimated correlation coefficient, plant materials, environmental conditions and their interaction. Among several factors,  yield  related  traits  highly  influence  the  amount  of  grain  yield  that  can  be obtained (Salah et al., 2013). Some of the yield related traits in sesame include: days to flower, days to maturity, plant height, number of branches, capsules per plant, capsule length, number of seeds per capsule, thousand-seed weight and seed yield per plant. These traits affect yield positively and/or negatively and their effect on yield depends on the influence of environment on them.

The knowledge of nature and magnitude of genetic variability is of immense value for planning efficient breeding programme to improve the yield potential of crop species. Likewise, information on genetic association of plant traits with seed yield has great importance to breeder in selecting desirable genotypes (Parameshwarappa et al., 2009). Phenotypic selection  of parents based on their performance alone may not always be reliable procedure since phenotypic expression is highly influenced by environmental factors, which are non-heritable. It is therefore, essential to select genotypes on the basis of their genetic worth (Salah et al., 2013). Thus, correlation helps in selection of superior genotype from diverse genetic populations (Jogdhande et al., 2017). However, in correlation studies indirect associations become more complex and confusing but path analysis can avoid this complication by measuring the direct influence of one trait on other as well as permits the partitioning of a given correlation coefficients into its components of direct and indirect effects (Manisha et al., 2018; Jogdhande et al., 2017). The path coefficient analysis is an effective means of analyzing direct and indirect causes of association and permits critical examination of the specific traits that produce a given correlation. It provides information about magnitude and direction of direct and indirect effect of the yield components (Chaudhary et al., 2005; Bizeti et al., 2004).

However, lack of information (particularly agro-ecological based) on character association of yield and its contributing traits, believed to limit the genetic improvement of sesame in Ethiopia. Hence, the present investigation was focused to gather adequate information on genetic association of yield and yield related traits in sesame accessions collected from mid-altitude areas of Ethiopia.

Description of the study site

The experiment was conducted at Melkassa Agricultural Research Center (MARC), Ethiopia, during the 2014 cropping season. Melkassa is located along the Upper Awash valley between 8° 33’ N and 39° 17’ E. The altitude of the Center is 1550 m.a.s.l. and the minimum and maximum annual temperature ranges from 14.35 to 28.22°C, respectively. The mean annual rain fall in the area is 704.8 mm with verti-cambisol soil type of pH = 7.6.

Experimental materials and design

The materials for the study comprised 81 sesame accessions (including three released varieties as standard checks) representing the mid-altitude areas of Ethiopia (Table 1) that were obtained from Ethiopian Biodiversity Institute (EBI) and Werer Agricultural Research Center (WARC). The experiment was laid out in 9 × 9 simple lattice designs with two replications and each genotype was planted in a plot consisting of four rows of 2.5 m long at a distance of 40 cm between rows and 10 cm between plants. All cultural practices were applied as required throughout the season.

The correlation coefficient was analyzed based on the tabulated data, using META-R Version- 6. 01 (Alvarado et al., 2017). The path coefficient analysis was computed for 12 yield related traits using the phenotypic and genotypic correlation results following the procedure suggested by Dewey and Lu (1959). Seed yield per plant was used as dependent character in path coefficient analysis and the remaining traits were used as independent variables.

R_ij = P_ij + å _rik P_kj

Where, r_ij is mutual association between the independent character (i) and dependent traits (j) as measured by correlation coefficients, p_ij is components of direct effects of the independent traits (i) on the dependent traits (j), å _rik P_kj  = summation of components of indirect effect of a given independent character (i) on the dependent traits (j) via all other independent traits (k).

The residual effect (R) was computed using the formula suggested by Dewey and Lu (1959) as:

R = √1-R²   Where, R² = å r_ij p_ij

Analysis of variance and mean performance of accessions

The analysis of variance revealed significant difference among the genotypes for all traits, indicating the presence of sufficient variability among the tested accessions for the traits under consideration (Table 2). This result was in agreement with the previous results reported by, Parameshwarappa et al. (2009), Salah et al. (2013) and Ahmed and Ahmed (2013). Mean performance of 81 sesame genotypes for 13 traits (data not shown) showed that seed yield per plant was ranged from 2.6 g (Oromia-9) to 8.6 g (Oromia-22) with overall mean of 5.33 g/plant. Generally, 44% of the genotypes had greater mean seed yield than the overall mean of genotypes. The highest mean seed yield/plant was recorded for Oromia-22 (8.66 g) followed by Am-NW-13 (8.41 g), Am-SW-5 (8.23), Oromia-25 (7.62 g), Am-NW-14 (7.60 g), Am-SW-10 (7.53) and Oromia-1 (7.51 g). These genotypes were found to be superior even to the local checks released varieties. The highest (52.15%) and lowest (43.35%) mean oil content was recorded for Oromia-13 and Am-SW-7 genotypes, respectively, with overall genotypes mean of 47.1%. The genotypes Oromia-13 (52.15%) and Am-NSh-2 (51.20%) had the highest mean oil content and were found to be superior as compared to the standard checks. A high variation in number of capsules/plant, primary branches/plant and plant height was found among the studied genotypes. Similarly, high variation in number of primary branches, plant height and 1000 seed weight was reported by Gidey et al. (2013).

Correlation coefficient  

Phenotypic and genotypic correlation coefficients of the various traits are presented in Table 3. The phenotypic and genotypic correlations in general were higher than the environmental correlation for the studied traits.

Number of capsules/plant, biomass/plant, harvest index and 1000 seed weight exhibited highly significant (<0.001) and high positive association with seed yield/plant at both phenotypic and genotypic level, indicating that these traits are reliable yield components and seed yield can be improved through direct selection of these traits. Similar results were reported by Muhamman et al. (2010) and Haruna et al. (2012), who found highly significant correlation of sesame seed yield with number of capsules/plant, biomass/plant and 1000 seed weight. Plant height and number of seeds/capsule showed highly significant and moderate positive correlation with seed yield/plant at both phenotypic and genotypic level. Number of primary branch/plant showed significant (<0.05) and low positive genotypic association with plant height, biomass/plant, harvest   index and   seed yield/plant. Hence, indirect selection in favor of these traits can improve seed yield in sesame. Similar results were reported by Sumathi et al. (2007) and Sumathi and Muralidharan (2010) for plant height, number of branches, number of capsules, days to 50% flowering, days to maturity and 1000 seed weight. However, oil content showed significant (<0.05) negative genotypic correlation with most of the yield traits (Table 3). This implies that indirect selection for high oil content would reduce seed yield/plant. Hence, simultaneous improvement for seed yield and oil content is difficult in the studied germplasm. Similar result was found by Daniya et al. (2013). In the contrary, Onginjo and Ayiecho (2009) found insignificant positive correlation of oil content with seed yield and suggested that selection for oil content had no any effect on seed yield.

Generally, correlation analysis   revealed   the presence of highly significant positive association between seed yield and the yield related traits such as number of capsules, biomass/plant, harvest index, 1000 seed weight and number of seeds/capsule. This finding is in line with Muhamman et al. (2010), Tamina and Tapash (2011), Haruna et al. (2012) and Daniya et al. (2013) who reported significant positive correlations between yield traits and final seed yield in sesame.

Path coefficient analyses

According to Bhatt (1973), correlation analysis may not be sufficient to explain the extent of associations in a manner that will enable one to decide on either a direct or an indirect selection strategy. Because of  this,  path  coefficients analysis was carried out between the different traits at both phenotypic and genotypic levels to partition correlation coefficients into direct and indirect effects; permitting a critical examination of the specific forces acting to produce a given correlation and measuring the relative importance of the causal factors. The result of phenotypic path analysis (Table 4) revealed that biomass/plant (1.08) had maximum positive direct effect on seed yield/plant followed by harvest index (0.75) and capsules/plant (0.53). Similarly, at genotypic level (Table 5), biomass/plant (0.64), capsules/plant (0.60) and harvest index (0.34) imposed highest direct effect on seed yield/plant. These traits also had strong positive correlation with seed yield at both phenotypic and genotypic level. Therefore, these traits can be considered as the principal traits while selecting for seed yield. In other words, selection indices may be formed by considering all these traits for improvement of seed yield. This result was in agreement with previous studies on association of traits in sesame accessions of different countries (Mothilal, 2005; Ahadu, 2008; Goudappagoudra et al., 2011; Ibrahim and Khidir, 2012).

The traits viz., 1000 seed weight (0.90), capsules/ plant (0.85), harvest index (0.80) and number of seeds/capsule (0.73) showed maximum indirect effects on seed yield at phenotypic level. Thousand seed weight and seeds/capsule also showed high indirect effect at genotypic level (Table 5). Therefore, this finding strongly emphasized that these two traits (1000 seed weight and seeds/capsule) made the greatest indirect contribution to seed yield. However, both these traits had negative direct effect to seed yield/plant at phenotypic level, suggesting that direct selection in favor of these traits can affect seed yield. This result agreed with those of Azeez and Morakinyo (2011), Ibrahim and Khidir (2012).

The analysis of variance revealed significant difference among the sesame accessions for all traits considered, indicating the presence of sufficient variability among the tested genotypes. Mean performance of 81 sesame genotypes for 13 traits generally showed that 44% of the genotypes had greater mean seed yield than the overall mean of genotypes. Genotypes Oromia-22, Am-NW-13, Am-SW-5, Oromia-25 and Am-NW-14, exhibited the highest mean seed yield and were superior even to the check varieties. Whereas, the genotypes Oromia-13 and Am-NSh2 had the highest mean oil content and were found to be superior as compared to the checks. Hence, these genotypes should be given emphasis while intending to improve sesame yield (seed and oil) in mid-altitude areas of Ethiopia. The phenotypic and genotypic correlations in general were higher than the environmental correlation for the studied traits. Number of capsules/plant, biomass/plant, harvest index and 1000 seed weight exhibited highly significant and high positive association with seed yield at both phenotypic and genotypic level, suggesting that these traits are reliable yield components and seed yield can be improved through direct selection of these traits. On the other hand, Plant height and number of seeds/capsule showed highly significant and moderate positive correlation with seed yield/plant at both phenotypic and genotypic level; hence, indirect selection in favor of these traits can improve seed yield in sesame. Number of capsules/plant, biomass/ plant, days to maturity and harvest index imposed maximum positive direct effect on seed yield. Hence, they can be considered as the principal traits while selecting for seed yield. In other words, selection indices may be formed by considering all these traits for improvement of sesame seed yield.

The authors have not declared any conflict of interests.