Introduction

The genus Pistacia in the Anacardiaceae family contains 13 or more species, among which Pistacia vera L. produces commercially valuable edible nuts. The other species grow in the wild and their seedlings are used mainly as rootstocks for pistachios (Kafkas et al. 2002a). There are two main centers of diversity for Pistacia: one comprises the Mediterranean region of Europe, Northern Africa and the Middle Eastern countries. The second is the Eastern part of Zagros Mountains and Caucasus region ranging from Crimea to the Caspian Sea (Zohary 1952).

In the first monograph study of Pistacia species, Engler (1881) listed eight species and a few subspecies, however he did not suggest any sectional subdivisions for such species and some species were not fully described by him (Zohary 1952). So far the most comprehensive taxonomic study of the Pistacia genus was reported by Zohary (1952), who divided the genus into four sections and 11 species according to leaf characters and nut morphology. However, he found no justification in retaining mutica (Fisch. et C. A. Mey.) Rech. f. and cabulica (Stocks) Rech. f. as species or subspecies. Kafkas and Perl-Treves (2001) characterized Pistacia species in Turkey by morphological and molecular data. They revised the miss-identification of a sample as P. eurycarpa Yalt. that was previously described by Yaltirik (1967) as P. khinjuk. Recently two monophyletic groups have been proposed in this genus by using cpDNA, Terebinthus and Lentiscus, representing deciduous and evergreen species respectively (Parfitt and Badenes 1997). Three important wild Pistacia species including P. vera, P. khinjuk and P. atlantica grow in Iran. Forests of wild P. vera spread to an area of about 75,000 ha, in central Asia, near the boarders of Turkmenistan, Afghanistan and Northeast of Iran. In Iran P. vera grows predominantly in the Sarakhs region covering roughly 17,500 ha (Behboodi 2003).

The atlantica species has been postulated to have three subspecies in Iran, mutica, kurdica (Zoh.) Rech. f. and cabulica (Rechinger 1963). Zohary (1952) described P. atlantica as comprising of two subspecies (kurdica, latifolia DC.) and he did not find any justification in retaining P. mutica and P. cabulica as either a species or subspecies and thought instead they should be considered as P. atlantica Desf. subsp. latifolia DC. Zohary (1972) later suggested P. atlantica to be as an Irano-Turanian species with three subspecies; Asiatic (subsp. cabulica), Mediterranean (subsp. atlantica) and Asiatic-Mediterranean (subsp. mutica).

P. atlantica Desf. subsp. mutica (Fisch. et C. A. Mey.) Rech. f. is a highly resistant rootstock to root-knot nematodes compared with P. vera, P. palestina Boiss. and P. khinjuk (Farivar-Mehin 1995).

P. atlantica subsp. kurdica (Zoh.) Rech. f. is mainly centered in Iran and Afghanistan, overlapping with P. vera in some areas. The subspecies cabulica is more tolerant in comparison to other P. atlantica subspecies, to exothermic and warm weather conditions. It has been suggested that this subspecies or its hybrids could be used as rootstocks to improve domestic pistachio production (Behboodi 2003). P. khinjuk trees are widely distributed at elevations ranging from 700 to 2000 m on hills and mid-height mountains (Behboodi 2003). Although able to withstand some of the most inconvenient weather conditions it is nevertheless sensitive to the fungus Phytophthora spp. (Banihashemi 1995) but has moderate resistance to root-knot nematodes (Farivar-Mehin 1995).

Leaf characteristics and nut morphology are the main diagnostic traits used in distinguishing various species of Pistacia. Flower characteristics have been less widely used for characterization of Pistacia species except at the genus and higher levels. Wood anatomy has also been used as a tool in identification (Grundwag and Werker 1976).

Yaltirik (1967) classified Pistacia species in Turkey and introduced P. eurycarpa as a new species. Kafkas et al. (2002b) studied the morphological diversity of wild Pistacia species in Turkey and found that nut weight and width was in significant correlation with the terminal leaflet lengths in P. terebinthus L. and P. atlantica. It was also reported that the width and length of the terminal leaflets were in negative correlation with the number of leaflets present. Upon examining P. terebinthus it was showed that leaf length was in significant correlation with the number of leaflets present.

Modern objectives in plant breeding may be achieved by the evaluation of traits amongst genetic resources and combining those of interest in one cultivar. Although new methods of molecular markers for genotype description have been proved useful, these methods are however expensive. Morphological characters must be recorded for selection of parents and are also the first choice used for describing and classifying the germplasm. Statistical methods including principle components or cluster analysis can be used as useful tools for screening the accessions. Additionally, morphological characteristics sometimes correlate or are associated with characteristics that are difficult to evaluate such as disease susceptibility and therefore may be useful as markers in breeding programs.

The study undertaken aimed to establish if any morphological relationships existed between 11 types of wild and cultivated pistachios in Iran that are mainly used as rootstocks. The results achieved could be positively applied in the characterization of pistachio species and in breeding programs.

Materials and methods

A total of 11 Pistacia types, each comprising three samples, growing in situ from Kerman and Fars provinces and ex situ from the Iranian Pistachio Research Institute (IPRI) were labeled to enable recording of their morphological specifications. The areas for in situ collection were selected according to Rechinger (1963). The genotypes were described based on the Pistacia descriptor developed by the International Plant Genetic Resources Institute (IPGRI 1998) with minor modifications. Each type (species or subspecies) comprised of three replicated samples. One type was unidentifiable and labeled as unknown. Thirty one characteristics (17 quantitative and 14 qualitative) were identified for evaluating the chosen samples (Tables 1, 2). Ten rachises were harvested of each tree to measure the rachis length and the number of fruits per rachis. The flower buds were dried and then soaked in water for 12 h to allow the bud scales to separate to enable counting of the number of scales in the flower buds. Ten fully developed leaves were removed from each tree to evaluate the characteristics of the leaves and leaflets. The shape of the terminal leaflet, terminal leaflet apex, terminal leaflet base, and the nut shape were scored according to the descriptor. One hundred nuts per tree were randomly selected to measure their weight and dimensions. Analysis of variance, means comparison, simple correlations, factor and cluster analysis were carried out using SPSS and SAS software to reveal the relationships between the genotypes.

Table 1 The list of qualitative traits of Pistachios and their classification according to IPGRI descriptor
Full size table
Table 2 Pistachio genotypes used for morphological classification and their measured quantitative characteristics
Full size table

Results

Analysis of variance

Significant differences (P ≤ 0.05) were detected among the species for all the noted characteristics by analysis of the variance. Leaf length, terminal leaflet length and nut characteristics such as nut length and width were significantly different between P. atlantica and the other species. However, there were no differences between three subspecies of P. atlantica from Iran for the afore mentioned characteristics apart from nut length. Also P. khinjuk was significantly different in comparison to other species in nut characteristics and number of fruits per rachis (Table 3). Mean values of the studied morphological characteristics showed large variations between the genotypes for all of the measured traits. Mean values and the range of variability for the different characteristics of each genotype are presented in Table 4. P. vera was shown to have the heaviest nuts and the largest nut dimensions, whereas P. khinjuk were found to be the lightest nut examined (Table 2). Also the number of leaflets was least amongst P. vera and P. khinjuk and most in P. atlantica. The largest dimensions of leaves and leaflets in P. atlantica were found in those growing in Fars province and least for those in found in Kerman province.

Table 3 Means comparison of quantitative traits in different Pistacia genotypes
Full size table
Table 4 Pistachio characteristics, range of variability, mean and coefficient of variations for qualitative and quantitative traits
Full size table

Correlations

The correlation between each pair of traits was calculated (Table 5). It was found that several leaf characteristics were in significant correlation with nut characteristics. Nut characteristics such as length (r = +0.80), width (r = +0.67), thickness (r = +0.66) and weight (r = +0.77) positively correlated with the width of the leaves. Also nut thickness (r = +0.51), nut width (r = +0.50) and nut length (r = +0.81) were in significant correlation with the width of the terminal leaflet. Leaflet characteristics, terminal leaflet width (r = −0.78) and terminal leaflet length (r = −0.54), were in negative correlation with the number of leaflets.

Table 5 Bivariate correlations among quantitative and qualitative traits in pistachio genotypes
Full size table

In P. vera split nut percentage correlated with width of the leaf (r = +0.65), terminal leaflet length (r = +0.82), terminal leaflet width (r = +0.72), nut length (r = +0.93), nut width (r = +0.60), nut thickness (r = +0.75) and nut weight (r = +0.93). Also split nut percentage was in negative correlation with the number of leaflets (r = −0.55).

Factor analysis

Factor analysis was used to determine the number of main factors for reducing the number of effective characteristics to discriminate genotypes (Table 6). Based on factor analysis the characteristics of leaves and nuts accounted for 40% of the variance as the first main factor with the other six factors, explaining 94% of the total variance. For each factor, a factor loading of more than 0.65 was considered as being significant. For the first factor, characteristics including leaf width, length of terminal leaflet, nut length, nut width, nut thickness, one hundred nuts dry weight and split nut percentage had a loading of more than 0.65 and defined 40% of the overall variance. The width of the terminal leaflet, petiole length of terminal leaflet, terminal leaflet base shape, terminal leaflet size, number of leaflets, arrangement of scales in flower bud, flower bud width and leaf rachis wing were significant for the second factor with 20.42% of overall variance. The third factor with 13.52% of the overall variance contributed to characteristics such as rachis length, leaf length, number of scales in the flower bud and flower bud length. The remaining factors were leaf texture (4th factor), petiole shape and nut shape (5th factor), growth habit (6th factor) and trunk color (7th factor).

Table 6 Eigen values and cumulative variance for seven major factors obtained from factor analysis and the characteristics within each factor for Pistacia genotypes
Full size table

The pistachio genotypes were grouped according to these seven factors. Cluster analysis divided accessions into three sub-clusters each consisting of genotypes belonging to the species P. vera, P. khinjuk and P. atlantica. Based on the results, P. khinjuk was found to be an in-between species, but more resembled P. atlantica than P. vera. P. atlantica, with P. atlantica subsp. mutica was located in the same group while P. atlantica subsp. kurdica was separated from them. Hybrid accession located between P. atlantica subsp. kurdica and P. atlantica subsp. mutica (Fig. 1).

Fig. 1
figure 1

Dendrogram representing morphological relationships among Pistacia genotype using ward method. AKK: P. atlantica subsp. kurdica, Kerman. UNK: Unknown, Kerman. AKF: P. atlantica subsp. kurdica, Fars. BBI: Garden mastic, Institute ACF: P. atlantica subsp. cabulica, Fars. KHI: P. khinjuk, Institute. AAI: P. atlantica subsp. atlantica, Institute. QZI: P. vera cv. Qazvini, Institute. BDI: P. vera cv. Badami Riz, Institute. SRI: P. vera var. Sarakhs, Institute. AMI: P. atlantica subsp. mutica, Institute

Full size image
Table 7 Classification subspecies of P. atlantica Desf. according to different researchers
Full size table

Discussion

Comparison of means showed that there were significant differences between P. atlantica and other species for many leaf and nut traits. In similar studies, Kafkas et al. (2002b) reported that average leaf length, terminal leaflet length and width, leaf petiole length and all nut characteristics were significantly different between each of the three species P. terebinthus, P. atlantica and P. eurycarpa.

Correlations between quantitative traits of pistachio accessions showed that several leaf characteristics were in significant correlation with nut characteristics. Kafkas et al. (2002b) also reported that nut weight in P. atlantica was in significant correlation with terminal leaf length and nut thickness, which is in accordance with the finding of the study carried out. Results showed that split nut percentage correlated with the dimension and weight of the nut. It was deduced that the splitting suture may develop better with increasing the nut dimensions and it seems that the kernel mechanical force is higher in this case. Leaf rachis wing was absent in the P. vera, P. khinjuk and the unknown genotype whereas it was present in the other genotypes. In a similar study, Zohary (1952) reported the absence of the leaf rachis wing in P. vera, P. terebinthus and P. khinjuk, but found it to be present in P. atlantica and P. lentiscus. The shape of the petiole cross-section was found to be round or flat with the exception of the unknown sample in that it was semi-round. In previous studies Zohary (1952); Yaltrik (1967) and Kafkas et al. (2002b) reported that the shape of the leaf petiole cross- section was round and angled in P. terebinthus and P. eurycarpa respectively.

Factor analysis showed that the characteristics of the leaves and nuts provide the main factor confirming 40% of the total variance, which must be taken into consideration when distinguishing pistachio rootstocks. According to Talhouk et al. (2000), nut characteristics in Amygdalus communis had the highest loading values for the first component in component analysis.

Cluster analysis could be used easily divide accessions with regards to species P. vera, P. khinjuk and P. atlantica. The P. atlantica along with P. atlantica subsp. mutica were located in a similar group and P. atlantica subsp. kurdica, P. atlantica Desf. subsp. cabulica (Stocks.) Rech. f. and the unknown genotype separated from them, therefore P. atlantica subsp. kurdica could be considered as being a distant species from P. atlantica. Yaltirik (1967) described and introduced P. eurycarpa as a new species. Zohary (1952) treated this accession as a subspecies of P. atlantica (subsp. kurdica), but it differs from the latter in at least two characteristics, having light green leaves on both sides and depressed fruit. Furthermore, its leaves are usually thicker and not numerous or neither does it have narrow leaflets as in P. atlantica. He also reported that P. eurycarpa is intermediate in terms of leaf characteristics between section Eu-Terebinthus and section Butmela. According to Yaltirik (1967), P. eurycarpa is a widespread species and often dominant in Iran, North of Iraq, Afghanistan and Southeastern Turkey. In this study it was found that P. atlantica and P. atlantica subsp. mutica formed the closest pairs genotypes. Zohary (1952) came to no conclusion as to whether P. mutica was a separate species or a subspecies of P. atlantica. The ovate leaflets and their reduced number were the only characteristics in which mutica differs from P. atlantica. Al Yafi (1978) described P. mutica based on leaf characteristics using herbarium samples and retained it within P. atlantica; also hybrid accessions located between P. atlantica subsp. kurdica and P. atlantica subsp. mutica (Table 7).

According to previous studies (Parfitt and Badenes 1997), P. vera and P. khinjuk were the two closest species while in the study undertaken P. khinjuk was closer to the P. atlantica species it can therefore be concluded that the present finding confirm the Kafkas et al. (2002b) report. According to Zohary (1972), P. atlantica subsp. kurdica with larger fruits, ovate and few paired leaflets closely resembled P. vera. He also reported that P. atlantica and P. atlantica subsp. latifolia (P. atlantica subsp. mutica and cabulica) are merely derivatives of P. atlantica subsp. kurdica, showing a trend towards increasing the number of leaflets per leaf and decreasing in nut dimensions.

Vegetative morphological characteristics of P. khinjuk, especially the leaves are very similar to P. vera. However, it has the smallest nuts between the species examined in the undertaken study and more resembles P. atlantica. According to Zohary (1952), fewer leaflets relate to a more ancestral species, therefore it can be concluded that P. vera is the most ancestral species followed by P. khinjuk and P. atlantica.

Using molecular characterization of the studied genotypes would allow for more obvious and clear distinguishing showing the genetic distances and the relationships of the accessions.