Genetic Variation and Inheritance of Early Growth Characteristics in Three Wild Pistachio Populations

Introduction

The genus Pistacia is a member of the Anacardiaceae family and consists of at least 11 species (Kafkas and Perl-Treves 2001; Kafkas et al. 2001; Golan-Goldhirsh et al. 2004). Three species of the genus Pistacia, P. vera L., P. atlantica Desf. and P. khinjuk Stocks, grow naturally in Iran (Ahmadi Afzadi et al. 2007). Pistacia atlantica Desf. (wild pistachio), locally known as “Baneh”, is economically the most important species for rurals in the forest areas of the Zagros mountains in Iran (Pourreza et al. 2008). It is grown with almond, oak and other forest trees in alpine areas, foothills and at different altitudes between 600-3000 m. The resin of wild pistachio, called Saqez, is used for a variety of industrial and traditional purposes, including food and medicine (Pourreza et al. 2008). P. atlantica is also used as a rootstock for the edible pistachio tree, P. vera L., (Kafkas and Perl-Tereves 2001; Kafkas et al. 2001; Ranjbar Fordoei et al. 2002; Barazani et al. 2004; Ahmadi Afzadi et al. 2007). Adaptation of Pistacia trees to harsh desert conditions and their longevity make them ideal candidates for reforestation in arid zones (Golan-Goldhirsh et al. 2004). In spite of its various uses, no efforts for the genetic improvement of this species have been made (Golan-Goldhirsh et al. 2004; Javanshah et al. 2006).

Knowlegde about the genetic diversity within P. atlantica  is useful in establishing management  strategies for conservation of this species and improving its production potential. Najowa et al. (2009) reported wide genetic diversity among P. atlantica  genotypes of Suweida province in Syria. Kafkas and Perl-Treves (2001) investigated the inter- and intra-specific variability among the wild Pistacia genotypes from Turkey on the basis of 30 phenotypic characters and found that 24 characters were polymorphic among and within species. Furthermore, Kafkas et al.  (2002) characterized the phenotypic and morphologic diversity within the important wild Pistacia species of Turkey, including P. atlantica, at the inter and intra-specific level, using a larger number of qualitative and quantitative traits and found high diversity at both levels. In order to study the morphological variability within and among P. atlantica  Desf. populations of Algeria, Belhadj et al. (2008) undertook a comparative analysis of morphological characters of leaves and fruits, as well as micromorphlogical aspects of leaves, in eight wild populations grown under different climatic conditions. Analysis of variance for the morphological traits under study showed significant diversity within and among the populations. Pazouki et al. (2010) studied the genetic variability within P. atlantica and P. vera from Iran by microsatellite markers. However, they indicated that variability within wild pistacia species was considerably lower than the cultivated species. In addition, Salehi Shanjani et al. (2009) analyzed amplified fragment length polymorphisms in a total of 216 pistachio accessions which included seven populations from three wild species (P. vera, P. khinjuk, P. atlantica  subsp. kurdica). The lowest amount of polymorphism was observed within P. atlantica  subsp. kurdica which indicated strong genetic erosion of this species in Iran due to over-exploitation and consequently severe decline in habitat.

 Selection of the best provenance of desired species for a given site or region is necessary for achieving maximum productivity in plantation forestry, agroforestry and future breeding work (Dhanai et al. 2003). Moreover, detecting the best genotype in such populations is breeding art, which is performed through various methods (Chaturvedi and Pandey 2004). Progeny tests have been established on multiple sites to evaluate the genetic merits of the selected trees (Gulcu and Celik 2009).

Heritability estimates have been found to be useful in revealing the relative value of selection based on phenotypic expression of different characters (Safavi et al. 2011). High estimates of heritability indicate that selection for the characters of study will be effective because of being less influenced by the environment. Based on our knowledge, no studies are available in the literature about the inheritance of morphological characters in wild pistachio. Therefore, the aims of this study were to estimate genetic variation, heritability and genetic advance for some growth characters of P. atlantica Desf., using half-sib progenies obtained from mother plants of three different locations.

Materials and Methods

Progeny test data

In September 2008, seeds were collected from 60 randomly selected trees of wild Pistacio (P. atlantica  var. kurdica) from three natural populations in northwest of Iran. The fruit and resin of wild pistachio are important economical sources for rural people in these areas. Trees under study were mostly located in natural forest stands and were at least 100 meters apart to decrease the possibility of being the components of the same parent (Ginwal et al. 2005). Furthermore, the distances between selected populations were more than 200 km. The sampling areas ranged from 44° 35′ to 46° 42′ E and 36° 04′ to 38° 06′ N  (Table 1 and Figure 1). Some characteristics of the mother trees such as height, stem diameter, maximum seed diameter, minimum seed diameter and seed weight were reported to estimate the phenotypic correlation coefficient of characters of these trees with traits of their half-sib progenies.

Figure 1.  Location map of three Pistacia atlantica populations in Iran that were

 sampled for this study

Table 1. Description of selected wild pistachio populations under study

Thirty seeds from each mother plants were sown in plastic pots in the Javanshir forest nursery in December 2008, which is located at 36˚39́ N, 45˚08́ E and 1387 meters above sea level. The resulting half-sib families were evaluated in a randomized complete block design with three replications. Ten seeds from each mother tree were randomly allocated in the experimental plots. Collar diameter (cm), seedling height (cm), number of leaves and length of sprout (cm) were recorded in the first and second growing seasons.

Data analysis

To test the statistical differences among the populations and progenies nested within the populations, analysis of variance was carried out on the data with the GLM procedure of SAS software, based on the recorded characteristics. A linear random model was used for the combination of the two-year data based on split plot in time. The statistical model and its components were as follow:

Where = the phenotypic value of the progeny of the i-th mother tree nested within the n-th population in the j-th block; = the grand mean; = the effect of  j-th block; = the effect of n-th population; = the effect of i-th progeny nested within the  n-th population; = the interaction effect of the j-th block and n-th population; = the interaction effect of the j-th block and i-th progeny nested within the n-th population; = the effect of k-th year; = the interaction effect of k-th year and n-th population; = the interaction effect of k-th year and i-th progeny nested within the n-th population; = the interaction effect of the j-th block and k-th year; = the interaction effect of the j-th block, k-th year and n-th population; = the interaction effect of the j-th block, k-th year and i-th progeny nested within the n-th population; = the random error.

In addition, an analysis of variance was performed on all progenies, disregarding the populations, to estimate the components of

variance and narrow sense heritability of the characters under study. The model was as follows:

Since the number of surviving seedlings was different in the plots and several experimental units were lost before data recording, the data were not balanced and thus were analyzed by GLM procedure of SAS.

Progenies of each mother tree were assumed as half-sib families and additive genetic variance ( ) was estimated as =  where is the among-family genetic variance (Falconer and Mackay 1996). Narrow sense heritability ( ) among families, on the entry mean basis, was estimated as follows (Falconer and Mackay 1996):

Where,  is variance due to the random error and is variance of interaction of genotype with year and   and  are the number of years and half-sib progenies, respectively.

      Percent of genetic gain among families was also calculated using the following formula (Kown and Torrie 1964): 

Where,  is the standardized selection  differential, is the square root of phenotypic variance and  is the grand mean for each trait.

  Phenotypic correlation coefficients between all pairs of the studied traits were determined by the following equation (Kown and Torrie 1964):

Where  is phenotypic correlation coefficient,  is the mean products,  and  are mean squares for traits  and , respectively.

Results and Discussion

Results of the analysis of variance over two years for the studied characters are shown in Table 2. Significant differences were obtained among half-sib progenies within the populations for all of the characters under study. There were no significant differences among the populations for the measured characters. As was expected, significant differences were observed between years for all growth characters. However, the only significant genotype by year interaction was that of population*year for collar diameter indicating that differences among populations were not similar in two years for this trait. Differences between the trees based on the studied characters recorded on their progenies would imply the possibility of selection for investigated traits in breeding programs (Sebbenn et al. 2003; Lesser and Parker 2004). Furthermore, a large within-population variation provides the buffering capacity to cope with the change in environmental conditions. Based on the results, genetic variation for plant height was more than other traits. Squared phenotypic coefficient of variation was also high for plant height (51.1%) and number of leaves (20.21%). These results reflect superiority of selection for plant height in the early growth stages (Table 3).

Table 2. Analysis of variance of the three populations of

 Pistacia atlantica over two yearsfor the characters under study

                   ** Significant at 0.01 probability level, respectively

Table 3. Grand mean and squared phenotypic coefficient of variation  (PCV²) for seedling

characteristics of Pistacia atlantica half-sib families

Estimates of additive genetic variance, narrow sense heritability and expected genetic gain from selection are presented in Table 4. Collar diameter, number of leaves and sprout length had moderate heritability. Their expected genetic gains were not also high. On the other hand, narrow sense heritability of plant height was relatively high (0.71). It means that additive gene effects are important in determining this character (Ajmal et al. 2009).  However, there are some possibilities of pollination within small groups of related trees giving room for a certain degree of full-sib mating, ending to over-estimation of the narrow sense heritability. In this case, as Baliuckas et al. (2005) argued, the estimates must be regarded as upper limits heritability. The expected genetic gain of plant height (with 10% selection intensity) was also substantially high which indicated again the possibility of improving wild pistachio genotypes for early plant growth. Very little data have been published regarding genetic structure of wild pistachio tree (Pedersen et al. 2007).

Table 4. Estimates of additive genetic variance, narrow sense heritability and genetic gain among

half-sib families of Pistacia atlantica

Based on the half-sib progenies evaluated,  genotypes number 103, 109 and 125 from Salmas population, number 204, 206, 209, 215 and 221 from Shahindezh population and number 301, 303, 306, 309, 312, 313, 315 and 322 from Baneh population were regarded as good general combiners for seed collection. Among these, entries 204, 309 and 312 were the most promising genotypes to be used in the genetic improvement programs of wild pistachio (data not shown).

Phenotypic correlation coefficients of seedling traits from half-sib progenies with the characteristics of the mother trees are presented in Table 5. Collar diameter showed positive significant correlation with all of the characters of mother tree, indicating that trees with larger stem diameter, taller height, larger and heavier seeds had progenies with larger collar diameter. Also, taller seedlings belonged to taller mother trees with larger stem diameter and maximum seed diameter. Sprout length of the half-sib progenies were also significantly correlated with the tree height, stem diameter and minimum seed diameter of the mother trees. On the other hand, leaf number showed no significant phenotypic correlation coefficient with the characteristics of mother tree. The largest positive phenotypic correlation was found between collar diameter of the half-sib progenies and seed weight of the mother trees (r = 0.42). Knowledge of relationship between vegetative characters of wild pistachio tree would be essential to select effectively superior mother plants for developing seed orchard of the species by which fast growing seedlings could be produced. Phenotypic correlation coefficients are mainly used for indirect selection in plant breeding programs (Chaturvedi and Pandey 2004; Blada and Popescu 2007). Therefore, significant positive phenotypic correlation of seedling plant height with mother tree height, as well as high heritability and large genetic gain, suggest that seedling height may be regarded as the potential predictor of the tree height in wild pistachio. However, the correlation coefficients were not high and, therefore, more data are needed to make decisive conclusions about these relationships. 

Table 5. Phenotypic correlation coefficient between seedling traits and characteristics of mother

 trees in Pistacia atlantica

         *’** Significant at 0.05 and 0.01 probability levels, respectively.

Introduction

The genus Pistacia is a member of the Anacardiaceae family and consists of at least 11 species (Kafkas and Perl-Treves 2001; Kafkas et al. 2001; Golan-Goldhirsh et al. 2004). Three species of the genus Pistacia, P. vera L., P. atlantica Desf. and P. khinjuk Stocks, grow naturally in Iran (Ahmadi Afzadi et al. 2007). Pistacia atlantica Desf. (wild pistachio), locally known as “Baneh”, is economically the most important species for rurals in the forest areas of the Zagros mountains in Iran (Pourreza et al. 2008). It is grown with almond, oak and other forest trees in alpine areas, foothills and at different altitudes between 600-3000 m. The resin of wild pistachio, called Saqez, is used for a variety of industrial and traditional purposes, including food and medicine (Pourreza et al. 2008). P. atlantica is also used as a rootstock for the edible pistachio tree, P. vera L., (Kafkas and Perl-Tereves 2001; Kafkas et al. 2001; Ranjbar Fordoei et al. 2002; Barazani et al. 2004; Ahmadi Afzadi et al. 2007). Adaptation of Pistacia trees to harsh desert conditions and their longevity make them ideal candidates for reforestation in arid zones (Golan-Goldhirsh et al. 2004). In spite of its various uses, no efforts for the genetic improvement of this species have been made (Golan-Goldhirsh et al. 2004; Javanshah et al. 2006).

Knowlegde about the genetic diversity within P. atlantica  is useful in establishing management  strategies for conservation of this species and improving its production potential. Najowa et al. (2009) reported wide genetic diversity among P. atlantica  genotypes of Suweida province in Syria. Kafkas and Perl-Treves (2001) investigated the inter- and intra-specific variability among the wild Pistacia genotypes from Turkey on the basis of 30 phenotypic characters and found that 24 characters were polymorphic among and within species. Furthermore, Kafkas et al.  (2002) characterized the phenotypic and morphologic diversity within the important wild Pistacia species of Turkey, including P. atlantica, at the inter and intra-specific level, using a larger number of qualitative and quantitative traits and found high diversity at both levels. In order to study the morphological variability within and among P. atlantica  Desf. populations of Algeria, Belhadj et al. (2008) undertook a comparative analysis of morphological characters of leaves and fruits, as well as micromorphlogical aspects of leaves, in eight wild populations grown under different climatic conditions. Analysis of variance for the morphological traits under study showed significant diversity within and among the populations. Pazouki et al. (2010) studied the genetic variability within P. atlantica and P. vera from Iran by microsatellite markers. However, they indicated that variability within wild pistacia species was considerably lower than the cultivated species. In addition, Salehi Shanjani et al. (2009) analyzed amplified fragment length polymorphisms in a total of 216 pistachio accessions which included seven populations from three wild species (P. vera, P. khinjuk, P. atlantica  subsp. kurdica). The lowest amount of polymorphism was observed within P. atlantica  subsp. kurdica which indicated strong genetic erosion of this species in Iran due to over-exploitation and consequently severe decline in habitat.

 Selection of the best provenance of desired species for a given site or region is necessary for achieving maximum productivity in plantation forestry, agroforestry and future breeding work (Dhanai et al. 2003). Moreover, detecting the best genotype in such populations is breeding art, which is performed through various methods (Chaturvedi and Pandey 2004). Progeny tests have been established on multiple sites to evaluate the genetic merits of the selected trees (Gulcu and Celik 2009).

Heritability estimates have been found to be useful in revealing the relative value of selection based on phenotypic expression of different characters (Safavi et al. 2011). High estimates of heritability indicate that selection for the characters of study will be effective because of being less influenced by the environment. Based on our knowledge, no studies are available in the literature about the inheritance of morphological characters in wild pistachio. Therefore, the aims of this study were to estimate genetic variation, heritability and genetic advance for some growth characters of P. atlantica Desf., using half-sib progenies obtained from mother plants of three different locations.

Materials and Methods

Progeny test data

In September 2008, seeds were collected from 60 randomly selected trees of wild Pistacio (P. atlantica  var. kurdica) from three natural populations in northwest of Iran. The fruit and resin of wild pistachio are important economical sources for rural people in these areas. Trees under study were mostly located in natural forest stands and were at least 100 meters apart to decrease the possibility of being the components of the same parent (Ginwal et al. 2005). Furthermore, the distances between selected populations were more than 200 km. The sampling areas ranged from 44° 35′ to 46° 42′ E and 36° 04′ to 38° 06′ N  (Table 1 and Figure 1). Some characteristics of the mother trees such as height, stem diameter, maximum seed diameter, minimum seed diameter and seed weight were reported to estimate the phenotypic correlation coefficient of characters of these trees with traits of their half-sib progenies.

Figure 1.  Location map of three Pistacia atlantica populations in Iran that were

 sampled for this study

Table 1. Description of selected wild pistachio populations under study

Thirty seeds from each mother plants were sown in plastic pots in the Javanshir forest nursery in December 2008, which is located at 36˚39́ N, 45˚08́ E and 1387 meters above sea level. The resulting half-sib families were evaluated in a randomized complete block design with three replications. Ten seeds from each mother tree were randomly allocated in the experimental plots. Collar diameter (cm), seedling height (cm), number of leaves and length of sprout (cm) were recorded in the first and second growing seasons.

Data analysis

To test the statistical differences among the populations and progenies nested within the populations, analysis of variance was carried out on the data with the GLM procedure of SAS software, based on the recorded characteristics. A linear random model was used for the combination of the two-year data based on split plot in time. The statistical model and its components were as follow:

Where = the phenotypic value of the progeny of the i-th mother tree nested within the n-th population in the j-th block; = the grand mean; = the effect of  j-th block; = the effect of n-th population; = the effect of i-th progeny nested within the  n-th population; = the interaction effect of the j-th block and n-th population; = the interaction effect of the j-th block and i-th progeny nested within the n-th population; = the effect of k-th year; = the interaction effect of k-th year and n-th population; = the interaction effect of k-th year and i-th progeny nested within the n-th population; = the interaction effect of the j-th block and k-th year; = the interaction effect of the j-th block, k-th year and n-th population; = the interaction effect of the j-th block, k-th year and i-th progeny nested within the n-th population; = the random error.

In addition, an analysis of variance was performed on all progenies, disregarding the populations, to estimate the components of

variance and narrow sense heritability of the characters under study. The model was as follows:

Since the number of surviving seedlings was different in the plots and several experimental units were lost before data recording, the data were not balanced and thus were analyzed by GLM procedure of SAS.

Progenies of each mother tree were assumed as half-sib families and additive genetic variance ( ) was estimated as =  where is the among-family genetic variance (Falconer and Mackay 1996). Narrow sense heritability ( ) among families, on the entry mean basis, was estimated as follows (Falconer and Mackay 1996):

Where,  is variance due to the random error and is variance of interaction of genotype with year and   and  are the number of years and half-sib progenies, respectively.

      Percent of genetic gain among families was also calculated using the following formula (Kown and Torrie 1964): 

Where,  is the standardized selection  differential, is the square root of phenotypic variance and  is the grand mean for each trait.

  Phenotypic correlation coefficients between all pairs of the studied traits were determined by the following equation (Kown and Torrie 1964):

Where  is phenotypic correlation coefficient,  is the mean products,  and  are mean squares for traits  and , respectively.

Results and Discussion

Results of the analysis of variance over two years for the studied characters are shown in Table 2. Significant differences were obtained among half-sib progenies within the populations for all of the characters under study. There were no significant differences among the populations for the measured characters. As was expected, significant differences were observed between years for all growth characters. However, the only significant genotype by year interaction was that of population*year for collar diameter indicating that differences among populations were not similar in two years for this trait. Differences between the trees based on the studied characters recorded on their progenies would imply the possibility of selection for investigated traits in breeding programs (Sebbenn et al. 2003; Lesser and Parker 2004). Furthermore, a large within-population variation provides the buffering capacity to cope with the change in environmental conditions. Based on the results, genetic variation for plant height was more than other traits. Squared phenotypic coefficient of variation was also high for plant height (51.1%) and number of leaves (20.21%). These results reflect superiority of selection for plant height in the early growth stages (Table 3).

Table 2. Analysis of variance of the three populations of

 Pistacia atlantica over two yearsfor the characters under study

                   ** Significant at 0.01 probability level, respectively

Table 3. Grand mean and squared phenotypic coefficient of variation  (PCV²) for seedling

characteristics of Pistacia atlantica half-sib families

Estimates of additive genetic variance, narrow sense heritability and expected genetic gain from selection are presented in Table 4. Collar diameter, number of leaves and sprout length had moderate heritability. Their expected genetic gains were not also high. On the other hand, narrow sense heritability of plant height was relatively high (0.71). It means that additive gene effects are important in determining this character (Ajmal et al. 2009).  However, there are some possibilities of pollination within small groups of related trees giving room for a certain degree of full-sib mating, ending to over-estimation of the narrow sense heritability. In this case, as Baliuckas et al. (2005) argued, the estimates must be regarded as upper limits heritability. The expected genetic gain of plant height (with 10% selection intensity) was also substantially high which indicated again the possibility of improving wild pistachio genotypes for early plant growth. Very little data have been published regarding genetic structure of wild pistachio tree (Pedersen et al. 2007).

Table 4. Estimates of additive genetic variance, narrow sense heritability and genetic gain among

half-sib families of Pistacia atlantica

Based on the half-sib progenies evaluated,  genotypes number 103, 109 and 125 from Salmas population, number 204, 206, 209, 215 and 221 from Shahindezh population and number 301, 303, 306, 309, 312, 313, 315 and 322 from Baneh population were regarded as good general combiners for seed collection. Among these, entries 204, 309 and 312 were the most promising genotypes to be used in the genetic improvement programs of wild pistachio (data not shown).

Phenotypic correlation coefficients of seedling traits from half-sib progenies with the characteristics of the mother trees are presented in Table 5. Collar diameter showed positive significant correlation with all of the characters of mother tree, indicating that trees with larger stem diameter, taller height, larger and heavier seeds had progenies with larger collar diameter. Also, taller seedlings belonged to taller mother trees with larger stem diameter and maximum seed diameter. Sprout length of the half-sib progenies were also significantly correlated with the tree height, stem diameter and minimum seed diameter of the mother trees. On the other hand, leaf number showed no significant phenotypic correlation coefficient with the characteristics of mother tree. The largest positive phenotypic correlation was found between collar diameter of the half-sib progenies and seed weight of the mother trees (r = 0.42). Knowledge of relationship between vegetative characters of wild pistachio tree would be essential to select effectively superior mother plants for developing seed orchard of the species by which fast growing seedlings could be produced. Phenotypic correlation coefficients are mainly used for indirect selection in plant breeding programs (Chaturvedi and Pandey 2004; Blada and Popescu 2007). Therefore, significant positive phenotypic correlation of seedling plant height with mother tree height, as well as high heritability and large genetic gain, suggest that seedling height may be regarded as the potential predictor of the tree height in wild pistachio. However, the correlation coefficients were not high and, therefore, more data are needed to make decisive conclusions about these relationships. 

Table 5. Phenotypic correlation coefficient between seedling traits and characteristics of mother

 trees in Pistacia atlantica

         *’** Significant at 0.05 and 0.01 probability levels, respectively.