*Data sets examined*

All data on the olive and oleaster can be analyzed as i) botanical samples to look for a key to differentiate them ;ii) similarity records to try to differentiate the two forms from a statistical point of view, and iii) genetic relationships using Bayesian methods.

The botanical differentiation between the oleaster and the olive, although with numerous attempts, does not reveal key traits neither morphologic nor molecular [31, 47, 67].

Similarity records have been shown more effective in differentiating between the two forms. Clear cut separation has occurred though some trees remained not clustered, which may indicate that there are hybrid forms between trees from the two groups or that all trees did not share all the traits recorded either morphological or molecular. Low frequency markers should be eliminated in such analyses as they may weigh too much and may distort the results. These methods lead to references of phylogenetic relationships between different levels of taxa, even though, they are not accurate for distant taxa (species, genera), but they have shed light on variety relationships for the olive [47, 64, 68, 69].

[57] Breton has examined a wide sampling of about 1900 trees including wild olives (950 from 55 sites), old trees (50) with undetermined status, and cultivars (about 900, either abandoned, feral forms or from collections) sampled in most places around the Mediterranean basin (Figure 3). Using 16 nu-DNA loci (Single Sequence Repeat or SSR) and 3 cp-DNA loci (single base repeats), the comprehensive data set was examined using structure software [63] by different a priori packages made of oleaster trees, cultivars, oleaster and feral trees. All clusters revealed by this study were checked by other aggregation methods (FCA, Dendrograms) to verify their consistency. However, these methods cannot ensure the biological existence of such groups. [70] have examined about 250 cultivars and two oleaster populations with AFLP markers (there are mostly dominant markers) from the central Mediterranean with similar methods. [71] has examined 171 wild trees with 8 SSR from the north-western Mediterranean [30] has studied about 32 cultivars and 70 oleaster trees from Tunisia, with morphological and molecular methods.

12 Olive Germplasm – The Olive Cultivation, Table Olive and Olive Oil Industry in Italy

the end of the last ice age may explain its present distribution.

point of view, and iii) genetic relationships using Bayesian methods.

have shed light on variety relationships for the olive [47, 64, 68, 69].

new information.

*Data sets examined* 

dominant markers but are estimated with dominant markers, i.e., Correspondence (FCA) and Hierarchic analyses (leading to dendrograms). Bayesian methods for nuclear DNA (nu-DNA) appeared in 2000 to analyze the data sets and they enable to constitute clusters (based on inferences of allele frequencies) and to check in each individual under examination the proportion of the genome that is coming from the different groups made by the software [63]. These clusters under some hypotheses may correspond to ancestor origins. However, Bayesian method can mix data from nu-, and cp- or mt-DNA. All the methods have contributed in eliciting the olive's origin and they have opened the way to use more adequate methods [64]. Obviously, old data sets could be treated with new methods to get

By the year 2000, after several completed projects (European projects and country projects), the molecular diversity in the wild form appeared deeply structured, that means the geographic distribution of the molecular markers in the wild tree was not homogenous [1, 24, 65, 66]. The genetic structure (estimated by the Fst) was stronger with mt- and cp-DNA markers than with the nu-DNA. Moreover, the mt- and cp-DNA distribution in the eastern and the western halves of the Mediterranean Sea appeared strongly structured. Even if sampling problems for all the studies had biased their data, the trend from the whole data supports that clines for allele frequencies do exist in the wild olive diversity. The clines could be due to different causes and as for other tree species the spread of the wild olive at

All data on the olive and oleaster can be analyzed as i) botanical samples to look for a key to differentiate them ;ii) similarity records to try to differentiate the two forms from a statistical

The botanical differentiation between the oleaster and the olive, although with numerous

Similarity records have been shown more effective in differentiating between the two forms. Clear cut separation has occurred though some trees remained not clustered, which may indicate that there are hybrid forms between trees from the two groups or that all trees did not share all the traits recorded either morphological or molecular. Low frequency markers should be eliminated in such analyses as they may weigh too much and may distort the results. These methods lead to references of phylogenetic relationships between different levels of taxa, even though, they are not accurate for distant taxa (species, genera), but they

[57] Breton has examined a wide sampling of about 1900 trees including wild olives (950 from 55 sites), old trees (50) with undetermined status, and cultivars (about 900, either abandoned, feral forms or from collections) sampled in most places around the Mediterranean basin (Figure 3). Using 16 nu-DNA loci (Single Sequence Repeat or SSR) and 3 cp-DNA loci (single base repeats), the comprehensive data set was examined using

attempts, does not reveal key traits neither morphologic nor molecular [31, 47, 67].

**Figure 3.** Anti Atlas Morocco elevation 1525masl. © Catherine Breton
