MATERIALS AND METHODS

Breeding materials and testing sites: Thirty six advanced barley landraces and crosses were planted in Bale highlands at Agarfa, Robe and Sinana locations in the southeast of Ethiopia during 2006 Bona (August-December) cropping season. Seven of these barley lines were developed from local crossing, whereas, the other materials were landraces developed in barley landrace enhancement programs (Table 1). These evaluation sites are characterized by bimodal rainfall with two cropping seasons; the Bona season extends from August to December; and the Ganna season extends from March to July. The average rainfall of Bona ranges from 270-560 mm and that of Ganna ranges from 250-560 mm. Agarfa is located at 460 km in the south east of Addis Ababa which is 38° 40^’ to 46° 3’ E longitude and 4° to 8^o 11’ N latitude with the altitude of 2350 meters above sea level (m.a.s.l).

Table 1:	Description of breeding materials used in the experiment
NA: Not available, m.a.s.l: Meter above sea level,*: Line obtained from crossing in barley breeding program at sinana agricultural research center

The annual temperature ranges from a minimum of 5.5 to 8.7°C to a maximum of 17.5 to 32°C. In addition, the mean annual rainfall varies from 800 to 1000 mm and the soil type is Cambisol. On the other hand, Sinana is located at 07° 07’N latitude and 40° 10’E latitude (Geremew et al., 1998) with 2400 m.a.s.l and 463 km from Addis Ababa. Average annual minimum and maximum temperature of Sinana is 9 and 21°C, respectively and the soil is with light clay type. Robe site is situated nearest to Robe (main town) which have somewhat similar climatic, topographic and edaphic conditions with Sinana.

Experimental design and agronomic practices: Barley seeds were planted in 3 m² plot area of six rows with 20 cm spacing between rows and of 2.5 m length in 6x6 simple lattice design with two replications at each location during 2006 Bona season. Recommended rate of 125 kg ha^-1 barley seeds and 50 kg ha^-1 DAP (Di-Ammonium Phosphate) fertilizer was used. Planting was conducted in mid August 2006 during Bona rainy season. Data was collected from the central four rows. The normal crop management and cultural practices were carried out uniformly at all locations.

Data collection and statistical analysis: Data was recorded on eighteen characters on plot and plant basis, however, test of homogeneity of error variance showed that the error mean squares were homogeneous only for the following eight characters such as number of spikelets per spike, spike length, spike weight, number of kernels per spike, biomass yield per plant, days to maturity, thousand kernel-weight and grain yield per plot for which combined analysis of variance was done (Table 2). Data were recorded on plot basis for days to maturity, grain yield per plot and thousand kernel weight, while, the other characters were taken on plant basis by considering an average of ten randomly sampled plants from each plot. Analysis of variance was done after subjecting mean values of the recorded data to MSTATC (Freed et al., 1988). Since the relative efficiency of lattice design was very similar with randomized complete block (RCB) design so that for the flexibility of lattice design (Cochran and Cox, 1957) data were analyzed as per RCB design.

The phenotypic (δ²p) and genotypic (δ²g) variances were estimated according to the method suggested by Burton and de vane (1953) whereas, broad sense heritability (h²) was estimated following Allard (1960) as:

Table 2:	Combined analysis of variance, heritability and genetic advance (GA) for number of spikelets per spike (NSPK), spike length (SPL), spike weight (SPW), number of kernels per spike (NK), biomass yield per plant (BM/pl),days to maturity(DM), thousand kernel weight (TKW) and grain yield per plot (GY/P) studied in barley landraces and crosses across locations (Sinana, Robe and Agarfa) in Bale during 2006 Bona cropping season
*,**Significance test at 0.05 and 0.01 level, respectively, ns: Non significant, CV: Coefficient of variation, SE: Standard error

and Genetic advance (GA) was calculated in accordance with the methods illustrated by Johnson et al. (1955) as:

where, K is the standardized selection differential at 5% selection intensity (K = 2.063).

RESULTS

Combined analysis of variance across locations showed that there was significant (p<0.01) variation among barley landraces and crosses tested for all characters considered. On the other hand, genotype x environment interaction revealed significant (p<0.01) difference only for grain yield per plot, thousand kernel weight and number of kernels per spike (Table 2). Overall mean grain yield for genotypes ranged from 1.1 to 2.6 t ha^-1 (data not shown). Aruso/EH/956 F₂ 8H-6-4 SNR FBC 99G 0003-21 advanced line which was obtained from crossing gave the highest mean grain yield (2.6 t ha^-1), whereas, Aruso/HB42 SNR FBC 99B 0003-9 and Acc.1718-15 barley lines gave the lowest mean grain yield per (1.1 t ha^-1). When the different rowed types were compared, the six and two rowed types showed different response for some characters such that six rowed barleys tended to produce higher mean grain yield per plot, grain yield per plant, number of kernels per plant and spike weight. As opposed to this, lowest mean values were recorded for thousand kernel weight and spike length in six rowed barleys. Whereas, two rowed barleys tended to give highest values for thousand kernel weight and spike length but low values were obtained for number of kernels per plant and spike weight (data not shown).

Phenotypic Coefficient of Variability (PCV) and Genotypic Coefficient of Variation (GCV) across locations was relatively high for grain yield per plot (23.27, 19.50%), number of kernels per spike (20.70, 18.30%), biomass yield per plant (22.10, 14.20%) and spike weight (15.20, 11.80%) (Table 3), respectively. On the other hand, days to maturity showed very minimum GCV (0.90%) and PCV (1.30%) values across locations.

Table 3:	Estimates of genotypic, phenotypic, and environmental coefficient of variability for number of spikelets per spike (NSPK), spike length (SPL), spike weight (SPW), number of kernels per spike (NK), biomass yield per plant (BM/pl),days to maturity (DM), thousand kernel weight (TKW) and grain yield per plot (GY/P) studied in barley landraces and crosses at Sinana , Robe, Agarfa and across locations in 2006 Bona cropping season
GCV: Genotypic coefficient of variability (%), PCV: Phenotypic coefficient of variability (%) and ECV: environmental coefficient of variability

Broad sense heritability and expected genetic advance (at 5% selection intensity) was estimated across locations for all agronomic traits considered (Table 2). Therefore, for pooled analysis, heritability ranged from 0.401 to 0.868. It was high for spike length (0.868) which was followed by thousand kernel weight (0.856), number of kernels per spike (0.785), grain yield per plot (0.714), spike weight (0.600), days to maturity (0.461), biomass yield per plant (0.413) and number of spikelets per spike (0.401).

Genetic Advance (GA) was relatively high for spike length (15.59) and number of kernels per spike (11.27) and thousand kernel weight (5.71) and this was coupled with high heritability (Table 2).

DISCUSSION

Ethiopian barley landraces are contributing as major sources of genes in barley crop improvement which has many advantages than exotics. In crop genetic improvement, breeding gain requires heritable variation in important agronomic characters of the crop. Therefore, the available genetic variation, heritability and expected genetic gain in important agronomic characters are useful to design better effective breeding strategies in barley landraces.

The presence of significant (p<0.01) variation among barley landraces and crosses for all characters considered implies that there is high variability among genotypes tested. Genotype x environment interaction revealed significant (p<0.01) difference only for grain yield per plot, thousand kernel weight and number of kernels per spike (Table 2) indicating barley genotype and crosses showed inconsistent performances across locations for these characters. Similarly, Ceccarelli, (1989), Ceccarelli and Grando (1991), Jackson et al. (1993) and Van Oosterom et al. (1993) reported genotype-environment interaction in barley for grain yield. The presence of difference in yield performance of the six and two rowed types may be considered during selection. The maximum yield was obtained from the advanced line, Aruso/EH/956 F₂ 8H-6-4 SNR FBC 99G 0003-21 (2.6 t ha^-1) which was developed through crossing with Aruso (local) in landrace enhancement program. This indicates the requirement of crossing program with landraces as one option to improve barley yield.

Phenotypic Coefficient of Variability (PCV) and Genotypic Coefficient of Variation (GCV) across locations was relatively high for grain yield per plot, number of kernels per spike, biomass yield per plant and spike weight (Table 3). In addition to this, these characters were accompanied by high heritability (Table 2) expressing the presence of more additive gene effects for possible improvement. Days to maturity showed the least variable trait. Generally, the differences between phenotypic and genotypic coefficient of variability for all the corresponding characters was small indicating that these characters were less influenced by the environment. Most of the research reports on major crop species indicate that genetic variability of important agronomic traits is predominantly additive genetic variance, while the non-additive genetic variance is generally smaller than the additive genetic variance (Moll and Stuber, 1974). Genotypic coefficient of variability estimate gives good implication for genetic potential in crop improvement through selection (Johnson et al., 1955). For all characters measured, PCV value was greater than GCV. The result of genotypic and phenotypic coefficients of variation obtained suggests that there is a good scope for yield improvement through phenotypic selection for grain yield per plot, number of kernels per spike, spike weight and biomass yield per plant. In other reports, Singh et al. (1986) found small PCV for days to maturity. Progress from selection depends on genetic variability existing in the population and selection is more effective when the genetic variation in relation to environmental variation is high (Allard, 1960; Poehlman and Sleeper, 1995). The genetic effects are not independent of non-genetic effects (Moll and Stuber, 1974) in crop improvement. In this study, comparably the contribution of non-genetic factors was small for important quantitative traits such as grain yield per plot, thousand kernel weight and number of kernels per spike.

Heritability (broad sense) was high for all important quantitative characters measured Dabholkar (1992) classified heritability estimates as low (5-10%), medium (10-30%) and high (>30%). Accordingly, all of the agronomic characters considered for analysis showed high heritability constituting high breeding value which has more additive genetic effects which is important for crop improvement. In similar study, Rytger et al. (1967) reported smaller heritability for grain yield than this result. Singh and Ceccarelli (1995) found smaller heritability than this report for grain yield in F₆ barley crosses conducted at Atlas, Tel Hadya, Berda and Bouider in Syria. In another report, Singh et al. (1986) also found heritability estimates for thousand kernel weight and spike length which is relatively lower as compared to this report. While, Vimal and Vishwakarma (1998) reported higher heritability for spike length, grain yield and number of spikelets per spike. Low heritability was reported by Ceccarelli (1994), Ceccarelli (1996) on barley in low yielding environments. Atlin and Frey (1990) also concluded that heritability in low yielding environment is lower than high yielding environment in Oat (Avena sativa L.) and Abdelmula et al.(1999) found similar value in Faba bean (Vicia faba L.). Because, the magnitude of heritability is affected by the type of genetic material, character to be measured and environmental conditions to which the individuals are subjected (Dabholkar, 1992; Falconer and Mackay, 1996). In this study, the heritability estimate is relatively high may be due to conducive rainfall and other environmental conditions including the adaptability of barley landraces and crosses used. In characters with highest heritability, phenotypic is a good index of genotypic merit; so that genetic gain can be made easily through selection (Johnson and Frey, 1967).

Genetic advance (GA) was relatively high for spike length (15.59) and number of kernels per spike (11.27) and thousand kernel weight (5.71) and this was coupled with high heritability (Table 2). According to Johnson et al.(1955), characters with high heritability accompanied with high genetic advance also result in better genetic gain through selection as high heritability will not always be associated with high genetic advance (Amin et al., 1992). Beside this, high heritability coupled with high genetic coefficient of variation estimates also gives a reliable estimate of the amount of genetic advance through phenotypic selection (Burton, 1952) which indicates the effectiveness with which selection of genotypes can be based on phenotypic performance (Johnson et al., 1955).

CONCLUSION

This result demonstrates that there exist variability among barley landraces and crosses tested for important quantitative traits indicating high potential for effective crop improvement and/or for further manipulation of the genetic resources through breeding as Ethiopian landraces are good sources of genes for many desirable traits. The advanced line, Aruso/EH/956 F₂ 8H-6-4 SNR FBC 99G 0003-21 which was obtained from barley crossing gave highest grain yield (2.6 t ha^-1) which is encouraging and showed the possibility of improvement of landraces through crossing. High heritability values obtained for most of quantitative traits in this study indicate the presence of high yielding condition during testing. In general, the study revealed that direct selection scheme would be more promising and encouraging than indirect selection for improving yield. This could be the nature of landraces because of their better adaptive traits to variable environmental conditions which have important implications for sustainable crop production.

ACKNOWLEDGMENTS

The authors thank the Oromia Agricultural Research Institute: Sinana Agricultural Research Center for their financial assistance to carryout the experiment. Amare Biftu and Tilahun Bayisa are also acknowledged for their technical support in field data collection.