Tef (Eragrostis tef [Zucc.] Trotter) is an allotetraploid crop with chromosome number (2n = 4x = 40) (Jones et al. 1978; Ponti 1978; Tavassoli 1986) and it is the most important staple cereal crop in terms of production and consumption, which grows extensively under various climatic and soil conditions (Neela and Solomon 2018). Tef is a crop for which Ethiopia is the center of origin and diversity (Vavilov 1951). The domestication of tef is one of the contributions of Ethiopian farmers to our world. As compared to other cereals grown in Ethiopia tef is the most preferred cereal by urban consumers as well as its producers. The reasons for its preferences include a source of the best quality human food and animal feed, tolerance to both high and low moisture stresses, high price for its grain and straw, low-post harvest pest and disease problems, and high longevity of the grain even under farmers’ traditional storage conditions (Ketema 1993, 1997; Assefa et al. 2001).

According to Assefa et al. (2001), tef is characterized by a versatile adaptation and grows in 11 of the 18 major agroecological zones of Ethiopia. They also noted that tef often exhibits a high level of phenotypic plasticity in phenology, morphology, and agronomic performance depending on the environment in which it is grown. In Ethiopia, national average grain yield of tef is about 18.5 t/ha (CSA 2020). However, improved varieties of tef produced grain yield of 1700–2200 kg/ha on farmers’ fields and 2200–2800 kg/ha on research fields and well managed large farms (Seyfu 1997; Anteneh et al. 2014).

Despite the preferences of the urban consumer community and the largest area coverage of tef, its national average yield is very low as compared to other cereals. The current low yield levels can be attributed to different production constraints such as susceptibility to lodging, moisture stress, water logging, weeds, seed shattering, and poor pre- and post-harvest agronomic management practices. Currently, the Poor genetic potential of cultivars under widespread production, and the problems of lodging and diseases are major causes for yield reduction of tef (Assefa et al. 2013). The objective of the current study on different aspects. To estimate the extent and pattern of genetic variability among selected tef genotypes with emphasis on yield and yield related components. To determine the magnitude and pattern of genetic variability, heritability, and genetic advance of tef genotypes. To determine the nature and extent of the interrelationship of yield and yield attributing traits; and relationship among tef genotypes.

Sirinka Agriculture Research Center served as the venue for the experiment. This center is situated in Eastern Amhara, within the North Wollo zone of Northeastern Ethiopia, prominently positioned along the main road connecting Addis Ababa to Mekelle. It lies approximately 509 km from Addis Ababa and 372 km from Bahir Dar, the regional capital. Geographically, it is pinpointed at coordinates 11°45'00"N, 39°36'36"E. The center’s elevation stands at 1850 meters above sea level, with annual temperature fluctuations ranging from a low of 13.6 °C to a high of 27.3 °C. The region receives an average annual rainfall of about 945 mm and features Eutric vertsoil, which is conducive for agriculture. Notably, the area is recognized for its tef production capabilities.

The study utilized sixty-four distinct tef genotypes sourced from Debrezeit Agricultural Research Centre, renowned for their superior cross-breeding attributes. The selection included a standard check (Mena) and a local check derived from Sirinka Agricultural Research Centre (refer to Table 1). The experimental layout adopted was a simple lattice design (8 × 8) with dual replications to ensure reliability. Random assignment of each treatment to designated plots within a block was meticulously carried out, with each plot measuring 1 m*1 m (1 m²). Planting during the first week of July 2022, aligning with the main cropping season and optimal soil moisture conditions. Adherence to recommended agricultural practices was observed by applying 121 kg ha^-1 NPS and 39 kg ha-1 urea as fertilizers, alongside a seeding rate of 10 kg per ha.

Data collection and analysis were meticulously carried out to evaluate fifteen phonological and morph-agronomic traits. The assessment included days to heading and maturity, biomass, grain yield per hectare, main panicle shoot weight, harvest index, and lodging index, all measured at the plot level. Additionally, plant height, culm and panicle length, number of fertile tillers per plant, and spikelets per panicle were evaluated on an individual plant basis.

To discern variations among genotype, an Analysis of Variance (ANOVA) was executed utilizing SAS version 9.4. The experiment’s significance was determined through Duncan’s Multiple Range Test (DMRT) at a 5% probability level.

The phenotypic and genotypic variances and coefficients of variation were estimated according to the method suggested by (Singh and Chaudhary 1985).

$\text{Genotypic variance (g)}=\frac{(MSPg-MSPe)}{r}$

Where, r = number replication, MSg = mean square due to genotypes, and MSe = mean square of error (environmental variance).

Phenotypic variance (p) = σ²e + σ²g

$\text{Phenotypic coefficient of variation (PCV)}=\frac{\sqrt{{\sigma }^{2}p}}{x}*100$

$\text{Genotypic coefficient of variation (GCV)}=\frac{\sqrt{{\sigma }^{2}g}}{x}*100$

Heritability (H²) in the broad sense for quantitative characters was computed using the formula suggested by (Allard1999).

$H^2=\frac{{\sigma }^{2}g}{{\sigma }^{2}p}*100$

As demonstrated by (Robinson et al. 1956), heritability can be categorized as low (0–30%), moderate (30–60%) and high (60% and above). The genetic advance (GA) with the selection intensity of the superior 5% (K = 2.06) of the plants was estimated in accordance with the methods illustrated by (Allard 1999):

$\mathrm{GA}=\mathrm{K}*\sqrt{{\sigma }^{2}p}*H^2\left(b\right)$

σ p = is phenotypic standard deviation on mean basis. The genetic advance as % of the mean (GAM) will be calculated to compare the extent of the predicted advance of different traits under selection using the formula:

$-\mathrm{GAM}=\frac{GA}{X}*100$

The genetic advance as % of the mean (GAM) was calculated. According to Johnson et al. 1955a, genetic advance as percent of mean was classified as low (<10%), moderate (10–20%), and high (>20%).

Based on the average data, wide ranges between the maximum and minimum mean values were observed for most of the traits evaluated (Table 2).The range of days to heading, days to physiological maturity and grain filling period from 37 to 47, 87 to 96.5, and 43.5 to 55 days and the mean performance values of days to heading, days to physiological maturity and grain filling period 42.3, 92.1 and 49.8 days, respectively. The results of the current investigation were in agreement with range values reported previously in other tef studies (Kebebew et al. 2001; Tsion et al. 2016). The range of plant height, panicle length, culm length and peduncle length from 103.2 to 125.5 cm, 40.9 to 53 cm, 57.38 to 74.7 cm and 12.46 to 26.49 (cm), respectively.

The mean performance values of plant height, panicle length, culm length and peduncle length 114.2, 46.3, 67.9 and 19.9 (cm), respectively (Table 2).The mean performance values for number of spikelets per panicle and main panicle weight 535 and 2.4 and the range of number of spikelets per panicle and main panicle weight 387.5 to 735.4, 1.6–3.6(g) and respectively (Table 2). The range of above-ground biomass, grain yield and harvest index are from 15500 to 29500 (kg/ha), 2095 to 4205 (kg/ha), and 10.6 to 24.15(%), respectively.

The mean performance values for above-ground biomass, grain yield, and harvest index were 20875 (kg/ha), 3271 (kg/ha), and 15.9 (%), respectively (Table 2). Twenty-five genotypes had mean values greater than the standard check (Mena) for grain yield (kg/ha). Plaza-Wuthrich et al. (2013); Tsion (2016) and Abebe (2019) reported similar ranges for most of the traits in their studies.

Phenotypic and genotypic coefficients of variation are critical metrics in the study of population genetics. They provide insight into the variability present within a population’s traits. A high genotypic coefficient of variation (GCV) signifies a substantial amount of genetic diversity, which is essential for the adaptability and evolution of species. The GCV values can vary significantly, as observed in the range from 1.1% for days to physiological maturity up to 17.9% for lodging index. Similarly, phenotypic coefficients of variation (PCV) also offer valuable information, with observed ranges from 3.4% for days to physiological maturity to 30.1% for lodging index.

The categorization of these coefficients have in to low (0–10), moderate (10–20), and high (>20%) by Sivasubramaniah and Menon (1973) aids in understanding the extent of variation in different traits. For instance, moderate PCV values were noted for traits such as the number of spikelets per panicle, grain yield, and harvest index, aligning with findings by Worku et al. (2019). However, contrasting reports by Abebe (2019) and Worku et al. (2019) highlight lower PCV values for the number of total tillers per plant and above-ground biomass yield.

On the flip side, certain traits exhibited low PCV values, including days to heading, days to physiological maturity, grain filling period, plant height, panicle length, and culm length. These variations underscore the complex interplay between genotypic and phenotypic factors that influence trait expression in plants.

Previous studies such as those by Solomon et al. (2009), Habtamu et al. (2011a), and others have shown findings similar to current results regarding phenotypic coefficients of variation for most tef traits. However, this contrasts with the findings of Getahun et al. (2021) and Girma et al. (2022), which indicated moderate values for traits like total tillers per plant and above-ground biomass. Additionally, genotypic coefficients of variation (GCV) were moderately high for traits such as spikelets per panicle (10.3%), grain yield (12.3%), harvest index (14.1%), and lodging index (17.9%) (Table 4).

The phenotypic coefficient of variation often reflects both genotype and environmental effects, with higher PCV than GCV suggesting a greater environmental influence on trait expression, as noted by Habte et al. (2015). High GCV values indicate a potential for trait improvement through selection, emphasizing the importance of genotypic variation, which is more accurately assessed by GCV, particularly for traits like spikelets per panicle, grain yield, harvest index, and lodging index (Solomon et al. 2013).

Estimates of broad sense heritability and expected genetic advance are pivotal in plant breeding. Heritability estimates, as delineated by Robinson et al. (1949), are classified into three categories: low (0–30%), moderate (30–60%), and high (60% and above). Analyses of variance reveal broad sense heritability values ranging from 2.9% for the number of total tillers per plant to 74% for panicle length. Notably, panicle length (74%) and culm length (68.9%) exhibit high heritability, corroborating with Habte et al. (2015) who reported similar findings for tef genotypes; also reported high heritability estimates for panicle length in tef genotypes (Tsion 2016).

High heritability signifies minimal environmental influence relative to genetic factors in trait determination, implying that progenies are likely to resemble the parent in performance. Conversely, traits such as plant height, number of spikelets per panicle, above-ground biomass, grain yield harvest index, and days to heading exhibit moderate heritability (30–60%). Traits with low heritability values (20%) include grain filling period, number of total tillers per plant, number of effective tillers per plant, days to maturity, and main shoot panicle weight. In this study, harvest index (22%) and lodging (21.8%) demonstrated high genetic advance as a percentage of the mean. Meanwhile, grain yield (17.8%), number of spikelets per panicle (17.07%), and above-ground biomass (13.66%) presented moderate genetic advance estimates as a percentage of the mean (Table 4).

Low genetic advance as a percentage of mean were observed for several traits in plant breeding, indicating the extent of variability that can be attributed to genetic differences. For days to heading a relatively low percentage of 5.9% was noted, and days to physiological maturity were even lower at 0.7%.while. The grain filling period showed a 3.01% genetic advance, with plant height and culm length at 4.82% and 7.37%, respectively. Panicle length presented a higher percentage of 9.35%, whereas the number of total tillers per plant was at 2.92%. Main shoot panicle weight and the number of fertile tillers per plant were also low, at 1.7% and 4.84% respectively.

High heritability coupled with significant genetic advance as a percentage of the mean suggests that additive genes predominantly influence the expression of traits, making selection an effective strategy for improvement. Johnson et al. (1955b) posited that the estimate of genetic advance as a percentage of the mean becomes an even more potent selection tool when considered alongside heritability estimates.

Furthermore, Denton and Nwangburuka (2011) emphasized that to enhance traits of interest, it is crucial to consider estimates of genotypic and phenotypic coefficients of variation in conjunction with heritability and genetic advance as a percentage of the mean. Conversely, high heritability estimates paired with low genetic advance as a percentage of mean imply that non-additive gene action significantly influences trait expression, as noted by Fatema et al. (2011). Lastly, Akbar et al. (2003) highlighted that high genotypic coefficients variance (GCV), heritability, and genetic advance as a percentage of mean are invaluable for selecting high-performing genotypes for trait improvement through selection.

Many associations observed from this experiment among yield and yield related traits are discussed as follows. In this experiment estimate of genotypic (Rg) and phenotypic (Rp) correlation coefficients between each pair of studied traits are presented in (Table 5). The genotypic correlation coefficient is found to be relatively higher in magnitude than their corresponding phenotypic correlation coefficient, except in a few cases, which indicated presence of inherent association among considered traits.

Grain yield showed positive and highly significant both genotypic and phenotypic correlation with number of spikelets per panicle, plant height and harvest index. Correlation coefficient analysis also revealed that grain yield had a positive and significant genotypic correlation with panicle length and plant height. Grain yield also showed a positive and highly significant phenotypic correlation with the number of spikelets per panicle, aboveground biomass and harvest index whereas positive and significant phenotypic correlation with plant height. This positive association of grain yield with panicle length, culm length, number of spikelets per panicle, above-ground biomass, and harvest index would assist breeders in identifying high-performing genotypes through selection for these traits. While traits with significant negative associations indicate that improving one may reduce another, the lodging index an essential trait in tef improvement exhibits a negative yet significant correlation with grain yield.

Many authors have reported the positive correlations of grain yield with above ground biomass and harvest index (Solomon 2010; Abel et al. 2013; Dagnachew and Girma 2014; Habte et al. 2015; Chekole et al. 2016). In line with this, Habte and Likyelesh (2013) found that above-ground biomass, panicle length, and harvest index had a positive and highly significant correlation with grain yield. Conversely, Wondewosen et al. (2012) observed a negative correlation between grain yield and grain filling period under stressful conditions. This aligns with previous findings by Solomon (2010) and Habte et al. (2015), who also noted a negative and significant genotypic and phenotypic correlation between grain yield and lodging index. Such negative correlations may arise from the influence of different genes or pleiotropic genes that exert dominance over the traits, steering them in opposing directions (Kearsey and Pooni 1996).

Correlation between days to heading and other traits was extensively studied. Days to heading exhibited highly significant phenotypic and genotypic correlations (p < 0.01) with several key traits such as days to physiological maturity, plant height, panicle length, culm length, above-ground biomass yield, and grain yield. Conversely, it showed non-significant correlations (P > 0.05) with the number of total tillers per plant, number of effective tillers per plant, and number of spikelets per panicle (Table 5).

Negative phenotypic correlations were observed for traits like days to heading to grain filling period, harvest index, peduncle length, and lodging index. The positive correlations between days to heading and traits like plant height were consistent with findings by Adenew (2002). However, contrasting results were reported by Ayalneh et al. (2012a), where days to heading had a negative correlation with above ground biomass, which differs from the positive correlation observed in this study. Days to maturity also showed a positive and highly significant phenotypic correlation (p < 0.01) with most traits studied. However, traits such as the number of fertile tillers per plant, number of spikelets per panicle, harvest index, and lodging index exhibited negative and significant correlations (p < 0.01). Traits like grain filling period, culm length, plant height, panicle length, and above-ground biomass had highly significant and positive genotypic correlations with days to physiological maturity. These results align with Habte and Likyelesh (2013), who also reported positive and significant correlations between days to heading and days to physiological maturity with plant height, panicle length, above ground biomass, and harvest index.

Correlation between plant height and other traits has been extensively studied, revealing significant associations with various agronomic characteristics. Genotypic and phenotypic analyses indicate that plant height is positively correlated with culm length, panicle length, days to heading, days to physiological maturity, and above-ground biomass. Notably, both plant height and panicle length demonstrate significant genotypic correlations with yield related traits such as grain yield and number of spikelets per panicle. These findings are supported by Chekole et al. (2016), who observed similar trends in correlations at both genotypic and phenotypic levels.

The relationship between above-ground biomass and other traits also presents significant positive phenotypic correlations, particularly with panicle length and culm length. This aligns with the observations of Ayalneh et al. (2012b) and Chekole et al. (2016), who noted correlations with lodging index and harvest index. Solomon et al. (2010) further highlighted a strong correlation between above-ground biomass, grain yield, plant height, and panicle length. In contrast, lodging showed highly significant and negative genotypic and phenotypic correlation with days to heading, days to physiological maturity and panicle length, whereas significant and negative genotypic correlation with culm length.

Number of total tillers per plant showed that highly significant and positive genotypic correlation with number of spikelets per panicle and the number of fertile tillers per plant. Number of spikelets per panicle was highly significant with panicle length, number of total tillers per plant, grain yield and negative genotypic and phenotypic correlation with peduncle length. Whereas number of fertile tillers per plant was highly significant and negative genotypic and correlated with main shoot panicle weight. Harvest index was high significant and negative genotypic and phenotypic correlation with days to heading and panicle length, whereas high significant and negative genotypic correlation with days to physiological maturity and plant height (Table 5).

Lodging is the most important trait that can play very great role in the productivity of tef crop. It brings direct and indirect effect resulting in both quantity and quality loss. The presence of significant (p < 0.01) phenotypic correlation above-ground biomass was non-significant (0.05). Generally, lodging index showed a negative phenotypic correlation with all traits of tef under consideration except above-ground biomass was significant.

In line with the present finding Chekole et al. (2016), reported that lodging index showed high significance and negative correlation with days maturity and plant height at genotypic and phenotypic levels and high significance and negative correlation with plant height Habte et al. (2015) also reported that lodging index had negative correlation with days to heading, days to physiological maturity, plant height, and culm length.