Associations of traits with yield in Dekoko ( Pisum sativum var . abyssinicum ) accessions in the highlands of southern Tigray , Ethiopia

Dekoko (Pisum sativum var. abyssinicum) is a unique crop developed and cultivated in Ethiopia. The objectives of the study were; understanding the genetic variability present in Dekoko population, determining the correlation between grain yield, protein content and other traits and understanding traits that can be used for indirect selection for high grain yield and high protein content. Local collections of Dekoko were planted in 3 replications of the RCBD design at Mekhan farmers’ Training Centre in Endamekhoni during 2010.Traits such as days to flowering and maturity and leaf width had low phenotypic coefficient of variation and genotypic coefficient of variation and low genetic advance (<20%). The genotypic correlation between and the direct effect of days to flowering, plant height and biomass on seed yield was positive suggesting that the traits can be used for indirect selection of high yielding accessions. Seed yield and protein content had perfect negative genotypic correlation (-1.00). The direct effect of protein content on seed yield was also negative implying that simultaneous selection towards increased seed yield and increased protein content at the same time may be difficult.


INTRODUCTION
Field pea (Pisum sativum L.) is the fourth most important legume crop in Ethiopia after faba bean, haricot bean and chick pea in terms of both area and total amount of production.Field pea covers over 203,990.64 ha with a total production of 257,031.41tons which accounts for 13% of the total grain legume production (CSA, 2011).
Both field pea and Dekoko are consumed as a protein supplement in the cereal-based diets of Ethiopia Sentayehu (2009).However , the average yield is very low, that is, 1.26 t/ha (CSA, 2011) for field pea and there is no record for Dekoko.The low yield in field pea is because of limited number of high yielding and disease resistant varieties, growing of pulses on marginal soils, poor management practices, absence or low fertilizer rate of application, and high insect and disease problems.
Even though the origin of field pea is controversial, Ethiopia is undoubtedly the centre of diversity for this crop since wild and primitive forms are known to exist in the high elevations of the country.Ethiopia is one of the major Vavilovian centers of diversity for several grain legume crops including lupine, field pea and wild ancestors of cow pea (Ali et al., 2003).
Cultivated Pisum is dominated by P. sativum species but P. sativum species abyssinicum (or simply P. abyssinicum) is a unique sub-species independently developed and cultivated in Ethiopia.The existing germplasm in the country shows tolerance to disease (IBC, 2007); (Sentayehu, 2009); (Jing et al., 2010).Pisum sativum is widespread across the Middle East and has affinity with the wild P. elatius while P. abyssinicum is restricted to highland regions of Ethiopia (South Tigray and North Wello) and Southern Yemen and shows a greater affinity to P. fulvum Yemane and Skjelvåg (2002); Jing et al. (2010).However, P. fulvum is found around the eastern edge (Syria, Lebanon, Israel, Palestine and Jordan) and not common in Ethiopia (Maxted and Ambrose, 2001).
P.s.abyssinicum is locally known as Dekoko (minute seeded) in Tigrigna and Yagere Ater (pea of my country) or Tinishu Ater (the smallest pea) in Amharic.According to Yemane and Skjelvåg (2002) Dekoko is capable of producing seed yield of up to 1.95 t/ha under phosphorus fertilization and is known for its high market price and for its food preference.Farmers and consumers in the study area call it as the "Dero-Wot of the poor" (chicken stew of the poor) probably to express its high nutritional value.Most often, the dry seeds of Dekoko are decorticated and split ('split peas') before boiling.Similarly, the seeds are boiled without decortications and consumed as soup and is preferred by most users (personal observation).In Ethiopia the annual consumption per person of field pea including Dekoko seeds is estimated at 6 to 7 kg (Messiaen et al., 2006;Sentayehu, 2009).
The genetic diversity of a species is the outcome of cumulative mutation, recombination and selection on individuals by the environment and selection by man for traits desirable for cultivation or consumption (Ali et al., 2007).The largest collection of P. sativum germplasm in Africa is located at the Institute of Biodiversity Conservation, Addis Ababa, Ethiopia, with over 1600 accessions (Messiaen et al., 2006).
A large genetic diversity has been found in P. sativum collections from both Africa (e.g.Ethiopia) and Asia.High to medium field pea genetic diversity in Ethiopia was observed in collections from Shoa, Gojam, Gondar, Wello, and Tigray while low to trace genetic diversity was observed in collections from Arsi, Gamo-gofa, Wellega, Illubabur and Kafa (Ali et al., 2003).
An understanding of morphological characters facilitates the identification and selection of desirable traits, designing new populations, transferring the desirable genes into widely grown food legumes through biotechnological means, resistance to biotic and abiotic stresses that are known to individual accessions increase the importance of the germplasm (Santall et al., 2001;Tar'an et al., 2005;Jorge, 2006).
Land races are the genetic wealth that a crop acquires over many years of its existence and have considerable breeding values as they contain valuable adaptive genes to different circumstances (Messiaen et al., 2006;Ali et al., 2003).In Ethiopia, more than 15 cultivars of field pea, with better yield potential, seed size, seed color and disease resistance than the farmers' varieties, have been released for different agro-ecological conditions (MoARD, 2008).Some of these varieties were obtained from local collections while others were obtained through hybridization of landraces with introduced germplasm.
Even though Dekoko (P.sativum var.abyssinicum) is important both for the local farmers and consumers, the existing germplasm was not studied for its diversity; neither was there any improvement work on this crop so far.The study was, therefore, conducted with the objectives of determining the correlation between the various quantitative traits and identifying those traits having high correlation with grain yield so that they can be used in indirect selection.

Description of the study area
The research was conducted in Southern zone of Tigray regional state, Wereda Endamekhoni at tabia Mekhan farmers' training center which is one of the mandate areas of Alamata Agricultural Research Center (AlARC).
Mekhan is only five kilo-meters South of Maichew and located about 660 km North of Addis Ababa and about 120 km south of Mekelle.Tabia Mekhan has a total population of 1249 households of which 741 are male headed and 508 are female headed.Major crops grown in the area include wheat, barley, maize, lentil, field pea and faba bean.The soil of the experimental site is black clay loam.The Wereda has a temperature range of 9 to 18°C with a mean annual rain fall of 600 to 700 mm (Endamekhoni BoANR, 2010).

Accessions evaluated
Twenty four local collections of Dekoko collected from two regions; South-Tigray and North-Wello were tested at Mekhan farmers training center (FTC) at an elevation of 2100 m above sea level.The accessions were collected in 2008 by Alamata Agricultural Research Center from weredas: Alamata, Ofla, Endamekhoni, Alaje, and Hintalo-Wejerat in South Tigray, and Kobo, Guba-lafto, Srinka and Habru in North-Wello (Table 1).

Experimental design and trial management
The trial was conducted using Randomized Complete Block Design with three replications and the plot size for each accession was 1.5 m 2 with inter-and intra-row spacing of 25 and 5 cm, respectively.Accessions were sown in six rows each 1m long.Phosphorus and nitrogen fertilizers with normal recommendation rates to other pulse crops 46 kg P2O5 and 18 kg N ha -1 , that is, 100 Kg DAP (Di-Ammonium Phosphate ha -1 ) and seed rate of 150 kg ha -1 were applied. exy = MSCP exy = environmental covariance between trait x and y;  2 g xy = (MSCP gxy -MSCP exy) /r = Genotypic covariance between trait x and y;  2 pxy = phenotypic covariance between traits x and y =  2 axy+  2 exy/ r; 2  gxy = genotypic covariance between character x and y.

Analysis of covariance (ANCOVA), correlations and path coefficient analysis
Analysis of covariance (ANCOVA) was conducted for the quantitative data (Table 2).

Phenotypic and genotypic correlation coefficients
Phenotypic correlation, the observable correlation between two variables, which includes both genotypic and environmental components between two variables, was estimated using the formula suggested by Miller et al. (1958):   gxy are phenotypic covariance and genotypic covariance between characters x and y, respectively.
2  px and 2  gx are phenotypic and genotypic variances for character x and 2  py and 2  gy are phenotypic and genotypic variances for character y.
The coefficient of correlation at phenotypic level was tested for its significance with table for simple correlation coefficient using n-2 df as suggested by Gomez and Gomez (1984)  Where, rgxy = genotypic correlation coefficient, SErgxy= standard error of genotypic correlation coefficient:


Where; h 2 1 and h 2 2 are broad sense heritability for character 1 and 2. The calculated 't' value was compare with the tabulated 't' value at the 5 and 1% level of significance using n-2 DF (where n is the number of accessions).

Path coefficient analysis
Partitioning of the cause and effect relationship of different traits will help to see what is contributing to the observed correlation.In some conditions, correlation alone does not give the exact picture of direct and indirect effect of characters up on each other, thus path coefficient analysis is preferable, since it can identify the direct and indirect causes of associations and can measure the relative importance of each (Singh and Chaudhary, 1977;Sharma, 1998).Association of yield with its components was determined by the application of correlation and path analysis.The use of path analysis requires a cause and effect situation among the variables.Path coefficient analysis is usually calculated using the formula suggested by Dewey and LU (1959) to assess direct and indirect effects of different traits on grain yield (dependent trait j) as: Where rij is mutual association between the independent trait (i) and the dependent trait (j) as measured by the correlation coefficient rij, pij is component of direct effect of the independent trait (i) on the dependent variable (j); and rik pkj is the components of indirect effect of a given independent trait (i) on the dependent traits (j) via all other independent traits (k).
The residual effect (U) which is the unexplained variation of the trait that is not accounted for by path coefficient is calculated using the formula of Dewey and LU (1959) as:

Genotypic correlation coefficients
Genotypic correlations between traits indicate the direction and magnitude of correlated responses to selection, the relative efficiency of indirect selection, and permit calculation of optimal multiple trait selection indices (Falconer and Mackay, 1996).Plant breeders traditionally have estimated genotypic and phenotypic correlations between traits using the method of moments on the basis of a multivariate extension of ordinary least squares referred to as multivariate analysis of variance (MANOVA) (Anderson, 1958;Mode and Robinson, 1959).
Phenotypic values of different traits in the same plant are often correlated, such as height and yield.Environmental factors and genetic effects are two reasons for correlations.
A set of closely linked genes present on one chromosome tend to be inherited together (not easily separable by recombination).If two genes are in linkage disequilibrium, a genetic covariance may arise between traits X and Y.In this study, 24 Dekoko accessions were analyzed for genotypic coefficients of 13 characters.Grain yield exhibited significant and positive associations with days to emergence (1.00***), days to flowering (1.00***) and maturity (1.00***), and biomass (0.985***) (Table 3).These positive and strong associations with grain yield revealed the importance of these characters in determining grain yield and indicated that selection for one or all of these traits would result in superior yield (Pandey and Gritton, 1975).It showed significant and negative associations with leaf width (-1.00***), pods per plant (-1.00***) and crude protein content (-1.00***), indicating the difficulty of improving both yield and protein content simultaneously.Days to emergence exhibited significant and positive association with grain yield, days to flowering and maturity, and plant height while it expressed significant and negative associations with pest and pod per plant (Table 3).Similarly crude protein content showed significant and positive associations with number of pods per plant (1.00***), and 1000 seed weight (1.00***) while it exhibited significant and negative association with grain yield (-1.00***) and biomass (-1.00***).
Even though, there is no report on breeding of Dekoko to cross check, this result is in agreement with the reports of various researchers on field pea that indicated the existence of strong associations between agronomic characters and grain yield (Muhammad et al., 2009;Ali et al., 2007).

Phenotypic correlations coefficients
Phenotypic correlation is a function of genetic and environmental correlation.The expression can be simplified by substituting the square root of variances as suggested by Walsh (1981).In this study, 24 Dekoko accessions were analyzed for phenotypic correlation coefficients of 12 quantitative characters (Table 4).Accordingly, grain yield exhibited highly significant and positive associations with days to emergence (0.944***), flowering (0.967***) and maturity (0.977***), biomass (0.894**) and plant height (0.771**).However, it expressed highly significant and negative associations with pods per plant, leaf width, 1000 seed weight and crude protein content.Generally, the phenotypic correlation coefficients were lower than their corresponding genotypic values indicating that the influence of environmental factor up on the accessions is lower than the inherent genetic effects.However, grain yield showed no significant relationships with harvest index and pest score.Results of similar trend in field pea have been reported by many researchers (Muhammad et al., 2009;Singh et al., 2007;Sharma et al., 2007).

Path coefficient analysis
Information on correlation and path coefficient analysis is of much use to plant breeders for selection and breeding genotypes with increased yield potential.Correlation analysis for seed yield provides opportunity for selection and leads to a directional model based on yield and its components in field experiments.Path coefficients have been used for complex characters in several crop species to provide information on interrelations of complex characters and to develop selection criteria (Kang et al., 1993;Gravois et al., 1991;Diz et al., 1994).In this investigation, 11 characters of Dekoko were analyzed for their direct and indirect effects up on grain yield (Table 5).

Direct effects of various characters on grain yield
Five out of 11 characters had positive direct effect on grain yield.They were number of pods per plant, biomass, plant height, days to flowering and pest score (11.358, 7.182, 7.148, 5.182, and 1.292) respectively (Table 5).Characters with negative direct effects were days to emergence, crude protein content, days to maturity, harvest index, leaf width, and thousand seed weight (-8.801, -8.057, -1.292, -0.762, -0.394, -0.232), respectively.Number of pods per plant had the highest (11.358) direct effect while days to emergence had the highest negative (-8.801) direct effect up on grain yield.The direction of the correlation coefficient between grain yield with the trait and the direct effect of the trait on seed yield coincide and are positive for biomass yield, plant height, and days to flowering.Selecting taller plants that flower late and produce high biomass can lead to increase in seed yield.However the indirect negative effects of these traits through other traits must be taken into account.Although the genetic correlation between seed yield and pods per plant is negative, number of pods per plant has big positive direct effect on seed yield.The negative correlation is due to its big negative indirect effects through other traits which will be discussed later.
Although days to emergence and days to maturity had positive genotypic correlation with seed yield their direct effect is negative.One should be cautious in using these traits for indirect selection for seed yield.The positive genotypic correlation of these traits with seed yield is through their positive indirect effect through other traits.
Crude protein had negative genotypic correlation with seed yield and also a big negative direct effect on seed yield.This indicates that it is impossible to increase crude protein content and seed yield of Dekoko simultaneously.
In a similar study on field pea, grain yield was found to have a highly significant positive correlation with number of pods per plant and above ground biological yield (Togay et al., 2008;Ali et al., 2007).Number of pods per plant showed highest degree of diversity and can be used directly for the improvement of the crop.The same results were reported by the work of Turi ( 2004) quoted in Ali et al. (2007) and Mehrani (2002).Malik et al. (1987) and Ghafoor et al. (1998) also reported positive correlation of grain yield with the above ground biological yield, which proved the complete association of the two traits.Many researchers like Donald (1962), Lal (1967), Sing et al. (1980), and Khan and Malik (1989) have already suggested that selection on the basis of best performance is vital in improving field pea.Leleji et al. (1972) examined significant negative correlations between grain yield and protein percent in dry beans.Kelly and Bliss (1975) found negative correlation between grain yield and percent protein in beans.Similarly, Camacho (1978) explained that protein concentration in legumes had negative correlation with grain yield.In this study, it was seen that there exists an association between the different quantitative traits of Dekoko.

Days to emergence
Days to emergence had negative indirect effect up on grain yield at genotypic levels through number of pods per plant, pest and days to maturity (Table 5).

Days to flowering
Days to flowering had positive direct effect up on grain yield at both genotypic and phenotypic levels exhibited via biomass, plant height and crude protein content.

Days to maturity
Days to maturity had positive direct effect up on grain yield at both genotypic and phenotypic levels which was expressed via days to flowering, biomass, plant height and crude protein content.

Leaf width
Leaf width had a positive indirect effect up on grain yield at both genotypic and phenotypic levels that was exhibited via days to emergence and maturity, number of pods per plant and pest.

Number of pods per plant
Number of pods per plant had a positive direct effect up on grain yield at both genotypic and phenotypic levels that is exhibited through days to flowering, biomass, plant height, and crude protein content.

Biomass
Biomass had positive direct effect up on grain yield at both genotypic and phenotypic levels that is expressed via days to flowering, plant height, and crude protein content.

Thousand seed weight
Thousand seed weight had negative direct effect up on grain yield at both genotypic and phenotypic levels.It was revealed via days to flowering, biomass, pest and crude protein content.

Plant height
Plant height had positive direct effect up on grain yield at both genotypic and phenotypic levels expressed via days to flowering, biomass, and crude protein content.The indirect effect of plant height up on grain yield was exhibited through days to emergence and maturity, number of pods per plant, and pest.

Harvest index
Harvest index had negative direct effect at genotypic level and positive direct effect at phenotypic level.The indirect effect of harvest index was revealed via days to emergence and maturity, biomass, 1000 seed weight, pest and crude protein content.

Pest
Pest score had negative direct effect up on grain yield at both genotypic and phenotypic levels expressed via days to flowering, number of pods per plant, and plant height.
The positive and indirect effect of pest up on grain yield was expressed via days to emergence, days to maturity, and crude protein content.

Crude protein content
Crude protein content had negative direct effect up on grain yield at both genotypic and phenotypic levels exhibited via days to emergence, biomass, and plant height.

Table 1 .
Accessions of Dekoko Included in the Study and Their Sources with their Altitude

Table 2 .
Form of analysis of covariance between quantitative characters.
exy Df= degree of freedom, r= number of replications, a= number of genotypes; MSCP rxy = Mean sum of cross product of replication for variable x and y; MSCP gxy = Mean sum of cross product of accessions for variable x and y; MSCP exy = Mean sum of cross product of error for variable x and y;

Table 5 .
Direct (bold and underlined diagonal)and indirect (out of diagonal) correlation coefficients of 11 Characters of Dekoko Accessions grown at Mekhan (Endamekhoni).DE=days to emergence, DF=days to flowering, DM=Days to maturity, LW=leaf width, PPP=pods per plant, BIO=above ground biomass, TSW= thousand seed weight, PH= plant height, HI=harvest index, PEST= Insect pest, PROT= percent of Crude protein.Number of pods per plant, biomass, plant height and days to flowering have significant and positive direct effects up on yield while days to emergence, protein content and days to maturity had significant but negative effect up